Idea Transcript
Modelling Knowledge, Attitudes, Self-Management, and Quality of Life in Type 2 Diabetes Mellitus
Yee Cheng Kueh Student ID: 3766442
College of Sport and Exercise Science Victoria University
Submitted in fulfilment of the requirements of the degree of Doctor of Philosophy by Research
March, 2014
ii
Doctor of Philosophy Declaration “I, Yee Cheng Kueh, declare that the PhD thesis entitled ‘Modelling Knowledge, Attitudes, Self-Management, and Quality of Life in Type 2 Diabetes Mellitus’ is no more than 100,000 words in length including quotes and exclusive of tables, figures, appendices, bibliography, references, and footnotes. This thesis contains no material that has been submitted previously, in whole or in part, for the award of any other academic degree or diploma. Except where otherwise indicated, this thesis is my own work”.
Signature
Date 12/03/2014
iii
ABSTRACT The aim of this thesis was to investigate the relationship between diabetes knowledge, attitudes, self-management, and quality of life (QoL) of people with Type 2 Diabetes Mellitus (T2DM). The main hypothesis was there would be significant relationships between diabetes knowledge, attitudes, selfmanagement, and QoL in the path model. I conducted two studies on two different populations, one in Australia and the other in Malaysia. In addition, these two samples represented different cultures. In research on the Australiabased population, I examined the relationship between diabetes knowledge, attitudes, self-management, and QoL in people with T2DM living in Melbourne, Australia. In research in the Malaysia-based population, I examined the relationship between diabetes knowledge, attitudes, self-management, and QoL in people with T2DM living in Kelantan, Malaysia. Next for the purpose of crosscultural comparison, I compared the differences between the Australia-based sample and the Malaysia-based sample in terms of their demographic background, health-related information, diabetes knowledge, attitudes, selfmanagement, and QoL. The recruitment process on the Australia-based sample took approximately seven months at the Alfred Hospital and Western Hospital. During this recruitment process, questionnaire packs containing an invitation to participate from hospitals, information to participants, demographic information, the Diabetes Knowledge (DKN) scale, Diabetes integration (ATT19), Summary of Diabetes Self Care Activities (SDSCA), and Diabetes Quality of Life (DQoL) were distributed. This resulted in data for 291 participants that were complete and usable for analysis. There were 192 male and 99 female participants in the
iv
Australia-based sample with mean age of 55.84 years (SD = 11.10) and mean duration of diabetes since diagnosis of 11.91 years (SD = 9.01). The path model for the Australia-based sample showed the relationship between the main variables (diabetes knowledge, attitudes to T2DM, self-management of T2DM, and QoL of T2DM) and extraneous variables (duration of diabetes since diagnosis and age). For the main variables, it was found that diabetes knowledge was a significant predictor for attitudes to T2DM and self-management of T2DM in blood glucose testing. Attitudes to T2DM was also a significant predictor for self-management of T2DM in terms of diet. In addition, self-management of T2DM in terms of blood glucose testing was a significant predictor of impact of QoL of T2DM, and self-management of T2DM in terms of diet was a significant predictor of satisfaction and impact of QoL of T2DM. Self-management of T2DM in exercise was a significant predictor of satisfaction in QoL of T2DM. However, self-management of T2DM in terms of foot care was not significantly related to either satisfaction or impact of QoL of T2DM. Among the extraneous variables, duration of diabetes since diagnosis was significantly associated with diabetes knowledge, attitudes to T2DM, self-management of T2DM in exercise and blood glucose testing. Age of participants was significantly associated with diabetes knowledge, attitudes to T2DM, self-management of T2DM in diet and blood glucose testing, and satisfaction of QoL of T2DM. The direction of the coefficient in the Australia-based sample path model indicated that those who were older scored lower in diabetes knowledge, but held more positive attitudes to T2DM, undertook more regular self-management of T2DM of diet and blood glucose testing, and had a higher level of QoL of T2DM in terms of satisfaction
v
than those who were younger. A similar approach was used for the Malaysiabased population. The research questionnaires used in the Australia-based sample were translated into the Malay language. The recruitment process for the Malaysia-based participants took approximately six months. Packs that included information to participants, consent form, demographic information, DKN, ATT19, SDSCA, and DQoL, were distributed to people with T2DM during their visit to the diabetes clinic in Hospital Universiti Sains Malaysia (HUSM), Malaysia. During the recruitment process, 276 usable questionnaire packs were received for analysis. In Malaysia, 129 male and 147 female participants with mean age of 57.10 years (SD = 8.47) and mean duration of diabetes since diagnosis 10.44 years (SD = 7.53) participated in the research. The findings for the Malaysia-based sample highlighted the relationships among the main study variables and the extraneous variables in a path model. Diabetes knowledge was found to be a significant predictor of attitudes to T2DM and self-management of T2DM in foot care, whereas attitudes to T2DM was a significant predictor of impact of QoL of T2DM. Among the self-management of T2DM components, diet and exercise were found to be significant predictors for satisfaction of QoL of T2DM. Self-management of T2DM in diet and foot care were significant predictors of impact of QoL of T2DM. For the extraneous variables, duration of diabetes since diagnosis was found to be significantly associated with diabetes knowledge and self-management of T2DM in foot care. Age of participants was significantly associated with diabetes knowledge, attitudes to T2DM, selfmanagement of T2DM in diet, and satisfaction of QoL of T2DM. The direction of the coefficient in the Malaysia-based sample path model indicated that those
vi
who were older scored lower in diabetes knowledge, but possessed more positive attitudes to T2DM, undertook more regular self-management of T2DM in terms of diet, and had a higher level of QoL of T2DM in terms of satisfaction than those who were younger. A cross-cultural comparison was conducted with the two different samples. The Australia-based sample and the Malaysia-based sample were compared based on their demographic background, type of treatment, and the main study variables. The differences between the samples were examined using chi-square and independent samples t-tests. The variables of gender and type of treatment (using insulin or not using insulin for treatment) based on groups were also analysed using one-way analysis of variance. These results highlighted some similarities and differences between the Australia-based sample and the Malaysia-based sample. In general, the Australia-based participants scored significantly higher in diabetes knowledge and reported more regular self-management of T2DM in exercise, blood glucose testing, and foot care. The Australia-based sample also scored higher on attitudes to T2DM compared to the Malaysia-based participants. On the other hand, Malaysia-based participants reported a lower level of impact of QoL of T2DM. There was no significant difference between self-management of T2DM in diet and satisfaction of QoL of T2DM between the two samples. Overall, the findings of the present thesis highlighted the relationships between diabetes knowledge, attitudes to T2DM, self-management of T2DM in terms of diet, exercise, blood glucose testing, foot care, and satisfaction and impact of QoL of T2DM. Additionally, the Australia-based sample and the Malaysia-based sample showed different path models with different path relationships in terms of the main study variables and
vii
the extraneous variables. The samples from these two countries showed differences in most of the study variables. Important areas that warrant further research were also suggested in the present thesis, including examination of other variables that may significantly contribute to the explanation of the path models that developed for both country samples. The implications of the present thesis emphasise the importance of continually increasing knowledge about diabetes, and enhancing positive attitudes to T2DM, regularity of self-management of T2DM, and levels of QoL of T2DM among people with T2DM. These aims are relevant to researchers and health care providers who work with people with T2DM, including diabetes specialists, psychologists, spouses, family members, and friends of people with T2DM. Finally, the findings of the present thesis are also applicable to health planners in Australia and Malaysia, who need to implement appropriate health programs and to consult people with T2DM and their families.
viii
ACKNOWLEDGEMENTS I would like to take this opportunity to thank and express my appreciation to many wonderful individuals who contributed to the successful completion of this dissertation. I would like to express my foremost gratitude to my principal supervisor Professor Tony Morris. Professor Tony, thank you for your insight, research expertise, enthusiasm, constant support and feedback throughout my PhD, it has been greatly appreciated. I am greatly thankful for the thoughtfulness you expressed, your guidance in improving my writing, and your perseverance in helping me to produce high quality academic research writing. I honestly feel very privileged and honoured to have this great opportunity to work with you and have such an exceptional and dedicated supervisor throughout my PhD journey. I would also like to thank you for giving me the opportunity of collaborating with you in several conference presentations and also your encouragement to motivate me to present several research papers in various conferences. Thank you. I would also like to acknowledge the support given to me from my associate supervisors, Doctor Erika Borkoles, Doctor Himanshu Shee, and Professor Aziz Al-Safi Ismail. Doctor Erika, thank you for the confidence and encouragement you placed in me to complete my PhD thesis writing. Thank you for your kind words, emotional support to keep me motivated and inspired, especially during challenging times of completing the thesis writing. Your support, guidance and valuable feedback on my thesis writing were important and highly appreciated. Doctor Himanshu, I would like to thank you for your expertise in statistical analysis especially on the structural equation modeling analysis. Acknowledgement is also gratefully given to Professor Aziz Al-Safi
ix
Ismail, Hospital Universiti Sains Malaysia, Malaysia, for your supervision in the Malaysian participants’ recruitment process. It had been a real honour to have this opportunity working with you especially in my data collection in Malaysia. Thank you for your kind support, knowledge, expertise, and guidance in helping me to go through the challenging recruitment processes in Hospital Universiti Sains Malaysia. Thank you for your willingness to take me through this challenge, making my Malaysia-based sample research a success. Thank you for your willingness to share your knowledge with me, and your support and encouragement throughout my intellectual journey. Without all your continuous assistance and constructive comments, this thesis would not have been completed successfully. My deepest gratitude goes to the Alfred Hospital and Western Hospital, Melbourne for allowing and supporting me to make the Australia-based sampling process a success. My sincere appreciation goes to Professor Duncan Topliss and Ms. Patricia Nugent, from the Department of Endocrinology and Diabetes, Alfred Hospital, for your kind help and support for making the recruitment of participants in the Alfred Hospital a success. Professor Duncan, thank you for seeing potential in my research proposal, and believing in my research contribution to the community. Thank you for your generous support in recruiting participants via your department at the Alfred Hospital. I would also want to thank Ms. Patricia Nugent for your kind assistance and support during my data collection at the Alfred Hospital especially in preparing the questionnaire packs for the participants. I would also like to express my deepest appreciation to Ms Cheryl Steele, Department of Diabetes Education, Western Hospital, Melbourne,
x
for your supervision especially in recruiting participants in Western Hospital. Ms Cheryl, thank you for your knowledge, support, and thoughtful suggestions especially in the process of recruitment of participants in Western Hospital. Your kindness and generosity to help made my recruitment process in Western Hospital a pleasurable experience. My appreciation also goes to all the participants in this research. This thesis would not have been possible without all the participants who volunteered their time and knowledge to take part in the research reported in this thesis. I also would like to extend my gratitude to all the staff members in the hospitals who directly or indirectly gave assistance in this research. You all have made my PhD research journey a meaningful experience. To the most important people in my life – my parents, my beloved husband Garry Kuan, my baby daughter Gabriella Kuan. There are no words that can describe how blessed and thankful I am to have such a loving and supportive family throughout my PhD journey in Melbourne. To my beloved parents, thank you for your unceasing prayer for me to complete my study in Melbourne. Garry, I want to particularly thank you for your continuous support, motivation, and inspiration throughout my journey as a PhD student in Melbourne. Your unfailing love has accompanied me throughout this journey, making it a joyous one. I am thankful for my lovely daughter, Gabriella who was born during the last stage of thesis writing. Gabriella, you have been the most wonderful reward that I received in my PhD completion. Last, but not least, my sincere thankfulness also goes to all my friends who have given their constant support and countless prayers in my PhD. You made my stay in Melbourne a pleasurable and unforgettable experience.
xi
TABLE OF CONTENTS STUDENT DECLARATION...…………………..…….…………………...
ii
ABSTRACT....…............................………………….…………...………… iii ACKNOWLEDGEMENTS………………………….………………...........
viii
TABLE OF CONTENTS…….……………………….…………………….. xi LIST OF TABLES …………….....…………………….…………………...
xx
LIST OF FIGURES ………….……………………….………………….…
xxiv
CHAPTER 1: INTRODUCTION ………….……………...………………..
1
CHAPTER 2: LITERATURE REVIEW….……………………….……......
8
Type 2 Diabetes Mellitus……….……….……………………………….. 9 Clinical Course….………….....……………………………………...
10
Diagnosis…….…….………………………………………………....
11
Treatment………...…………………………………………………... 12 Prevalence of Type 2 Diabetes Mellitus...................…..……………….... 13 Global Prevalence and Trends……….....…….........……………….... 13 Australia…………….....…………………………………………......
15
Malaysia………….............….……………………………………...... 18 Diabetes Knowledge………………..……………………………………. 21 Diabetes Attitudes……………..…………………………………………. 30 Diabetes Self-Management…………......………………………………... 38 The Relationship between Diabetes Knowledge, Attitudes, and SelfManagement…...........................................................................................
50
Knowledge and Attitudes……………………………………..…...…
52
xii
Knowledge and Self-Management………...……………..……..……
54
Attitudes and Self-Management………………..…...……………..…
56
Knowledge, Attitudes, Self-Management……………………............
62
Diabetes Quality of Life…………………………..........…………….......
65
Relationship of Diabetes Knowledge, Attitudes, Self-Management, and Quality of Life.................................................................................……...
70
Cross-cultural Comparison between Australia and Malaysia……………. 77 Summary of Literature………......……………………………………….. 81 The Present Thesis…………..…………………………………………… 83 CHAPTER 3: CONCEPTUAL FRAMEWORK…..……………………......
86
Conceptual framework……..………………………………………….…
86
Rationale for Conceptual Framework of Self-Management in Type 2 Diabetes Mellitus ..............................................................………….…… 93 Summary of Hypotheses and Research Questions………......…….……..
101
CHAPTER 4: METHOD……......………………………………….....…….
103
Study Design……..………………………………………………….…… 103 Study Population…..……………………………………………….…….. 104 Sampling Method…..………………………………………….……..
104
Australia…………...……………………………………………... 104 Malaysia…………...……………………………………………... 104 Participants………………..………………………………………….......
105
Measures…………………………..........………………………….…...... 106 Demographic and Health Measures……………..……...........………. 106
xiii
Diabetes Knowledge (DKN) Scale…………...………….…………... 106 Diabetes Integration Scale-19 (ATT19)......…………….…………....
108
Summary of Diabetes Self-care Activities (SDSCA).....………..…… 109 Diabetes Quality of Life (DQoL) Scale……......…………..………… 111 Revision and Translation of Questionnaires…...……………………….... 114 Australia…......………………………………………………..…….... 114 Malaysia…………..…………………………………………...……... 115 Translation of questionnaires…..………………………………… 116 Reliability of Malay version of questionnaires……………...…… 117 Procedure ……………......……………………………………….………
118
Australia…………......…………………………………………..…… 119 Malaysia…………......………………………………………..……… 120 Data Management……………..…………………………………………. 120 Diabetes Knowledge (DKN)………………..…………………..……
121
Diabetes Integration (ATT19)…………..……………………………
121
Summary of Diabetes Self-care Activities (SDSCA)………………... 121 Diabetes Quality of Life (DQoL)……..………………………..….…
122
Data Screening……......……………………………………….…….…… 122 Analytical Procedures……..……………………………………...……… 124 Exploratory Factor Analyses (EFA)………..………………...………
124
Confirmatory Factor Analyses (CFA)……………..…………………
126
Model Assessment……………..………………………………..…… 128 Absolute fit indices…………..………………………………...… 128
xiv
Normed chi-square (χ2/ df)…………………………………..…… 128 Root mean square error of approximation (RMSEA)…………
129
Standardized root mean square residual (SRMR)………..……
129
Goodness-of-fit index (GFI)………………………………………
129
Incremental fit indices………………...…………………….……
130
Comparative fit index (CFI)……………………………………… 130 The Tucker Lewis index (TLI)…………………...………..……...
130
Summary of fit indices………………..…………………..……...
131
Descriptive Statistics………………..………………………..………
132
Assessment of Path Model…………………..……………….………
132
Cross-cultural Comparison……………..………………….…………
133
Measurement invariance test…………………..………....………
134
Differences between two cultural groups………......….…………
135
CHAPTER 5: RESULTS AND DISCUSSION FOR AUSTRALIABASED SAMPLE........................................................................................... 137 Sample Overview………………………………………………...………
138
Measurement Model Assessment………………………………….……..
140
Exploratory Factor Analysis (EFA)……………………..…………… 140 Diabetes self-management (SDSCA)……………..……...………
141
Diabetes quality of life (DQoL)-satisfaction…………………......
144
Diabetes quality of life (DQoL)-impact………………..….……..
146
Confirmatory Factor Analysis (CFA)……………………...………… 148 Diabetes quality of life (DQoL)–satisfaction……………….....…
148
xv
Diabetes quality of life (DQoL)–impact……………......……...… 154 Descriptive Analysis…………………......………………………….…… 160 Diabetes Knowledge and Attitudes to Type 2 Diabetes Mellitus...….. 160 Self-Management of Type 2 Diabetes Mellitus.....………….......…… 161 Quality of Life of Type 2 Diabetes Mellitus..……………..…………
162
Correlation Analysis……………..…………………………….…………
163
Hypotheses for Path Analysis……………..………………….......……… 166 Extension of Hypotheses and Research Questions………...………… 166 Results of Assessment of the Path Models……………..……...………… 171 Discussion…………………......……………………………….………… 184 Relationship between Diabetes Knowledge, Attitudes, and SelfManagement for the Australia-Based Sample…………………..……
184
Factors Associated with Quality of Life of Type 2 Diabetes Mellitus for the Australia-based Sample…......................................................... 192 Significance of Extraneous Variables for the Australia-Based Sample ….............................................................................................
198
Methodological Issues…………......………………………………....
202
Future Research…………………………………………........………
209
Summary………………………………......…………………...………… 215 CHAPTER 6: RESULTS AND DISCUSSION FOR MALAYSIA-BASED SAMPLE.........................................................................................................
216
Sample Overview……………..………………………………………….
216
Measurement Model Assessment……………………………….......……
219
xvi
Exploratory Factor Analysis (EFA)……………………………......… 219 Diabetes self-management (SDSCA)………………………….....
219
Diabetes quality of life (DQoL)-satisfaction……………………..
222
Diabetes quality of life (DQoL)-impact…………………...…......
224
Confirmatory Factor Analysis (CFA)………………………………... 226 Diabetes quality of life (DQoL)-satisfaction……………..……....
226
Diabetes quality of life (DQoL)-impact………..…....…………...
231
Descriptive Analysis………………………………......…….…………… 237 Diabetes Knowledge and Attitudes to Type 2 Diabetes Mellitus......... 237 Self-Management of Type 2 Diabetes Mellitus.......…….........……… 238 Quality of Life of Type 2 Diabetes Mellitus......………………......…
239
Correlation Analysis…………………......…………………………….…
240
Hypotheses for Path Analysis……………......………………….......…… 243 Extension of Hypotheses and Research Questions……………...…… 243 Results of Assessment of the Path Models…………….............………… 247 Discussion…………………...…………………………………………… 259 Relationship between Diabetes Knowledge, Attitudes, and SelfManagement for the Malaysia-Based Sample…………....…..........…
259
Factors Associated with Quality of Life of Type 2 Diabetes Mellitus for the Malaysia-Based Sample............................................................ 265 Significance of Extraneous Variables for the Malaysia-Based Sample..................................................................................................
269
Methodological Issues……………..………………………................
272
xvii
Future Research……………………......……………………......……
277
Summary…………………………..…………………………….......…… 281 CHAPTER 7: RESULTS OF CROSS-CULTURAL COMPARISON …….
282
Multiple-Group Confirmatory Factor Analysis on Diabetes Quality of Life (DQoL) Scale………....................................................……......……
282
Characteristics of Respondents……………………......………….……… 286 Results Regarding Gender………………………………………......……
291
Results Regarding Treatment with Insulin……………......……...………
295
Discussion…………………..……………………………….....………… 300 Methodological Issues…………………………......…………………
309
Future Research……………………………………..………......……
318
Summary…………………………………………......…………...……… 323 CHAPTER 8: DISCUSSION AND CONCLUSIONS……………………...
325
Overview of Results and Major Findings………………………………... 326 Australia-Based Sample………………......………………….......…..
326
Malaysia-Based Sample…………………......………………….........
336
Comparison between Australia-Based and Malaysia-Based Sample...
342
Implications for Practice………………......…………………………....... 349 Concluding Remarks………………………………..……………………
355
REFERENCES.........……….………………………..…………………...
357
Appendix A - Demographic Information Form (English Version)............
422
Appendix B - DKN Questionnaire (English Version)................................ 423 Appendix C - ATT19 Questionnaire (English Version)............................. 425
xviii
Appendix D - SDSCA Questionnaire (English Version)...........................
427
Appendix E - DQoL Questionnaire (English Version)..............................
428
Appendix F - Australia-Based Sample – Support Letter (The Alfred Hospital)............................................................................... 430 Appendix G - Australia-Based Sample – Participants Information Sheet (The Alfred Hospital).........................................................
431
Appendix H - Australia-Based Sample – Participants Information Sheet and Consent Form (Western Hospital)...............................
433
Appendix I - Demographic Information Form (Malay Version)................ 437 Appendix J - DKN Questionnaire (Malay Version)................................... 438 Appendix K - ATT19 Questionnaire (Malay Version)..............................
440
Appendix L - SDSCA Questionnaire (Malay Version).............................. 442 Appendix M - DQoL Questionnaire (Malay Version)...............................
443
Appendix N - Malaysia-Based Sample – Information to Participants and Consent Form (in Malay Language)................................... 446 Appendix O - Calculation of Construct Reliability and Variance Extracted Estimate.............................................................. 450 Appendix P - Australia-Based sample - Histogram Graphs of Variables (Diabetes Knowledge Scores, Exercise, and Diet).............
453
Appendix Q - Malaysia-Based Sample - Histogram Graphs of Variables (Diabetes Knowledge Scores, Exercise, Diet, and Blood Glucose Testing)................................................................. 455
xix
Appendix R - Australia-Based Sample – Scatter Plots of Variables (Duration of Diabetes since Dagnosis, Age, and Diabetes Knowledge Scores)............................................................. 458
xx
LIST OF TABLES Table 4.1
Criteria for Model Fit Assessment………………………....….
Table 5.1
Mean (M) and Standard Deviation (SD) of Age and Duration
132
of Diabetes since Diagnosis (Australia-Based Sample).....…… 138 Table 5.2
Results of Exploratory Factor Analyses for SDSCA Scale (Australia-Based Sample)..............……………………......…..
Table 5.3
143
Results of Exploratory Factor Analyses for DQoLSatisfaction Scale (Australia-Based Sample)………….……… 145
Table 5.4
Results of Exploratory Factor Analyses for DQoL-Impact Scale (Australia-Based Sample)....……………………………. 147
Table 5.5
Structure Coefficients for DQoL-Satisfaction Factor 1S and Factor 2S (Australia-Based Sample)…...…………………....... 154
Table 5.6
Structure Coefficients for DQoL-Impact Factor 1I and Factor 2I (Australia-Based Sample)……………………....……..........
Table 5.7
Percentage Scores for Diabetes Knowledge and Attitudes to T2DM (Australia-Based Sample).......................................…...
Table 5.8
159
161
Self-Management of T2DM Practice Score Measured out of 7days a Week (Australia-Based Sample)..................................... 161
Table 5.9
QoL of T2DM Score in Percentage Measured by Satisfaction Scale and Impact Scale of DQoL (Australia-Based Sample)..... 162
Table 5.10 Descriptive Statistics and Correlations of Variables (AustraliaBased Sample)…….................................................................... 165
xxi
Table 5.11 Hypotheses (Initial Structural Model/ Model 1, AustraliaBased Sample).............................................................................
172
Table 5.12 Hypothesised Structural Model 1 Fit Indices (Australia-Based Sample)........................................................................................ 175 Table 5.13 Hypothesised Structural Model 2 Fit Indices (Australia-Based Sample)........................................................................................ 177 Table 5.14 Respecified Structural Model 3 Fit Indices (Australia-Based Sample).................................................................................…... 179 Table 5.15 Level of Support for the Hypotheses (Australia-Based Sample)….......................................................................……..... 181 Table 5.16 Final Path Model Hypotheses (Australia-Based Sample)........... 183 Table 6.1
Mean (M) and Standard Deviation (SD) of Age and Duration of Diabetes since Diagnosis (Malaysia-Based Sample)……......
Table 6.2
Results of Exploratory Factor Analyses for SDSCA Scale (Malaysia-Based Sample)........…………………………....…
Table 6.3
223
Results of Exploratory Factor Analyses for DQoL-Impact Scale (Malaysia-based Sample)….....………....…………...….
Table 6.5
221
Results of Exploratory Factor Analyses for DQoLSatisfaction Scale (Malaysia-based Sample)……….….......….
Table 6.4
217
225
Structure Coefficients for DQoL-Satisfaction Factor 1S and Factor 2S (Malaysia-Based Sample)………………………….. 231
Table 6.6
Structure Coefficients for DQoL Impact Factor 1I and Factor 2I (Malaysia-Based Sample)…………....………………….....
236
xxii
Table 6.7
Percentage Scores for Diabetes Knowledge and Attitudes to T2DM (Malaysia-Based Sample)……….……………….........
Table 6.8
238
Self-Management of T2DM Practice Score Measured by 7days a Week (Malaysia-Based Sample)………....…….......….. 239
Table 6.9
QoL of T2DM Score in Percentage Measured by Satisfaction Scale and Impact Scale of DQoL (Malaysia-Based Sample)..... 240
Table 6.10
Descriptive Statistics and Correlations of Variables (Malaysia-Based Sample)…......................................................
Table 6.11
Hypotheses (Initial path model/ model 1, Malaysia-Based Sample)..................................................................................…
Table 6.12
252
Respecified Structural Model 3 Fit Indices (Malaysia-Based Sample)…..................................................................................
Table 6.15
250
Hypothesised Structural Model 2 Fit Indices (Malaysia-Based Sample)......................................................................................
Table 6.14
248
Hypothesised Structural Model 1 Fit Indices (Malaysia-Based Sample)…..................................................................................
Table 6.13
242
255
Level of Support for the Hypotheses (Malaysia-Based Sample)….....................................................................……….
257
Table 6.16
Final Path Model (Malaysia-Based Sample)……......……...…
258
Table 7.1
Assessment of Measurement Model Fit for the AustraliaBased and Malaysia-Based Samples DQoL-Satisfaction in the Second-Order Level Model……................................................ 284
xxiii
Table 7.2
Assessment of Measurement Model Fit for the AustraliaBased and Malaysia-Based Sample DQoL-Satisfaction in the Second-Order Level Model with Item QS11 Deleted………… 285
Table 7.3
Assessment of Measurement Model Fit for the AustraliaBased and Malaysia-Based Samples DQoL-Impact in the Second Order Level Model……….......................................….
Table 7.4
Independent t-test of Differences in Characteristics of Respondents with T2DM (n=567)……………………….........
Table 7.5
286
288
Chi-square Test of Differences in Characteristics of Respondents with T2DM (n=567)……….......………………..
290
Table 7.6
One-Way ANOVA Results for Differences Based on Gender..
293
Table 7.7
Tukey’s Post-Hoc Multiple Comparison for Gender…………. 294
Table 7.8
One-Way ANOVA Results for Differences Based on Insulin Treatment...................................................................................
Table 7.9
298
Tukey’s Post-Hoc Multiple Comparison for Insulin and NonInsulin Users…………………………………………………..
299
xxiv
LIST OF FIGURES Figure 2.1
Model of the possible causal links between diabetes knowledge, attitudes, self-management, and QoL…....……….
70
Figure 3.1
Proposed conceptual framework................................................
92
Figure 3.2
Hypotheses and research questions illustrated in a path diagram....................................................................................... 95
Figure 5.1
Number and percentage of types of treatment (Australia-based sample).......................................................................................
Figure 5.2
Participants’ education background (Australia-based sample).......................................................................................
Figure 5.3
150
Improved DQoL-satisfaction Factor 2S measurement model (without QS10, QS12, Australia-based sample)........................
Figure 5.6
149
DQoL-satisfaction Factor 2S measurement model (Australiabased sample).............................................................................
Figure 5.5
140
DQoL-satisfaction Factor 1S measurement model (Australiabased sample).............................................................................
Figure 5.4
139
151
DQoL-satisfaction second order level measurement model (Australia-based sample)............................................................ 152
Figure 5.7
Improved DQoL-satisfaction second order measurement model (Australia-based sample)................................................. 153
Figure 5.8
DQoL-impact Factor 1I measurement model (Australia-based sample).......................................................................................
155
xxv
Figure 5.9
Improved DQoL-impact Factor 1I measurement model (without QS17 and QS18, Australia-based sample)..................
156
Figure 5.10 DQoL-impact Factor 2I measurement model (Australia-based sample).......................................................................................
156
Figure 5.11 DQoL-impact second order level measurement model (Australia-based sample)............................................................ 157 Figure 5.12 Improved DQoL-impact second order level measurement model (Australia-based sample)................................................. 158 Figure 5.13 Initial hypothesised structural model (Australia-based sample)………...........................................................................
174
Figure 5.14 Hypothesised structural Model 1 with standardised regression weights (Australia-based sample)…......................................…
176
Figure 5.15 Hypothesised structural Model 2 with standardised regression weights (Australia-based sample)..............................................
178
Figure 5.16 Respecified structural Model 3 with standardised regression weight (Australia-based sample)................................................ 180 Figure 6.1
Number and percentage of types of treatment (Malaysia-based sample).......................................................................................
218
Figure 6.2
Participants’ education background (Malaysia-based sample)... 218
Figure 6.3
DQoL-satisfaction Factor 1S measurement model (Malaysiabased sample).............................................................................
Figure 6.4
226
DQoL-satisfaction Factor 2S measurement model (Malaysiabased sample).............................................................................
227
xxvi
Figure 6.5
Improved DQoL-satisfaction Factor 2S measurement model (without QS7 and QS12, Malaysia-based sample)....................
Figure 6.6
228
DQoL-satisfaction second order level measurement model (Malaysia-based sample)............................................................ 229
Figure 6.7
Improved DQoL-satisfaction second order measurement model (Malaysia-based sample)................................................. 230
Figure 6.8
DQoL-impact Factor 1I measurement model (Malaysia-based sample).......................................................................................
Figure 6.9
232
Improved DQoL-impact Factor 1I measurement model (without QI18 and QI19, Malaysia-based sample)....................
233
Figure 6.10 DQoL-impact Factor 2 measurement model (Malaysia-based sample).......................................................................................
233
Figure 6.11 DQoL-impact second order level measurement model (Malaysia-based sample)............................................................ 234 Figure 6.12 Improved DQoL-impact second order level measurement model (without QI7, Malaysia-based sample)...........................
235
Figure 6.13 Initial hypothesised structural model (Malaysia-based sample).......................................................................................
249
Figure 6.14 Hypothesised structural Model 1 with standardised regression weights (Malaysia-based sample)..............................................
251
Figure 6.15 Hypothesised structural Model 2 with standardised regression weights (Malaysia-based sample)..............................................
253
xxvii
Figure 6.16 Respecified structural Model 3 with standardised regression weights (Malaysia-based sample)..............................................
256
Figure P1.1 Histogram of diabetes knowledge in percentage (Australiabased sample)……………………………………………….....
453
Figure P1.2 Histogram of mean self-management of T2DM in terms of diet (Australia-based sample).............................................................
454
Figure P1.3 Histogram of mean self-management of T2DM in terms of exercise (Australia-based sample)…………………………...…
454
Figure Q1.1 Histogram of diabetes knowledge in percentage (Malaysiabased sample).............................................................................
455
Figure Q1.2 Histogram of mean self-management of T2DM in terms of diet (Malaysia-based sample)………………...……………….. 456 Figure Q1.3 Histogram of mean self-management of T2DM in terms of exercise (Malaysia-based sample).............................................. 456 Figure Q1.4 Histogram of mean self-management of T2DM in terms of blood glucose testing (Malaysia-based sample)………………. 457 Figure R1.1 Scatter plot of duration of diabetes since diagnosis and age (Australia-based sample)............................................................ 458 Figure R1.2 Scatter plot of duration of diabetes since diagnosis and diabetes knowledge (Australia-based sample)...........................
459
Figure R1.3 Scatter plot of age and diabetes knowledge (Australia-based sample).......................................................................................
459
1
CHAPTER 1 INTRODUCTION The World Health Organization (WHO, 2006) classifies diabetes into two major categories, type 1 and type 2. Around 90% of people with diabetes in developed and developing countries have type 2 diabetes mellitus (T2DM), which is largely the result of excess body weight and physical inactivity (WHO, 2011). Diabetes is classified among the most serious non-communicable diseases in the world, and projections suggest that the epidemic of T2DM will become worse in the near future. It was estimated that in 2010, the world prevalence of diabetes among adults was 6.4%, affecting 285 million and this will increase to 7.7%, affecting 439 million adults by 2030 (Shaw, Sicree, & Zimmet, 2010). Shaw et al. (2010) pointed out that in between 2010 and 2030, it is estimated that there will be a 69% increase in numbers of adults with diabetes in developing countries and a 20% increase in developed countries. Countries that are more developed and having western cultures have shown a steady increase in T2DM. For example, in Australia, approximately 1 million Australians have been diagnosed with diabetes and 90% of those had T2DM (Diabetes Australia [DA], 2012a). Developing Asian countries are showing rapid increases in diabetes. For example, in Malaysia, it has been reported that nearly 1.2 million people have diabetes and more than 98% of these have T2DM (Zanariah et al., 2009). T2DM is primarily related to loss of insulin sensitivity or dysfunctions resulting from the combination of resistance to insulin action and inadequate insulin secretion (Khardori, 2011). It is different from type 1 diabetes mellitus (T1DM) because people with T2DM are not absolutely dependent on insulin for
2
life, whereas those with T1DM must inject insulin daily. Although many people with T2DM are ultimately treated with insulin, they are considered to require insulin for treatment, but not to depend on insulin (Khardori, 2011). It can be a frightening and overwhelming experience for a person being diagnosed with T2DM because it affects their long-term health and everyday life. T2DM is becoming virtually an epidemic in many countries where the prevalence of a sedentary lifestyle and obesity is increasing. More complex description of T2DM is presented in Chapter 2. T2DM has a large lifestyle component, mainly diet and physical activity. T2DM develops when a lifestyle called “diabetogenic” (i.e., excessive caloric intake, inadequate caloric expenditure, obesity) is adopted by individuals especially those with a susceptible genotype (Khardori, 2011). However, the risk for diabetes associated with excess weight varies with different ethnic groups. For instance, compared to a European, an individual of Asian origin is at higher risk of diabetes at lower levels of overweight (WHO Expert Consultation, 2004). Therefore, considering and understanding cultural and ethnic variation in response to T2DM is very important. However, regardless of which culture and ethnicity a person belongs to, an important aspect of the treatment for T2DM is lifestyle related. Increasing mechanisation is reducing the extent to which people are physically active at work and at home. Low cost, energy-dense foods are readily accessible through “fast-food” chains and processed foods in supermarkets. These factors increase the prevalence of overweight and T2DM (Drewnowski & Specter, 2004). This is why T2DM is called a lifestyle disease.
3
Research has shown that T2DM can be prevented by changes in the lifestyles of both men and women who are at high risk for the illness (Tuomilehto et al., 2001). Thus, behavioural and lifestyle interventions have become a major focus in research on the prevention and management of T2DM (Ramadas, Quek, Chan, & Oldenburg, 2011). Lifestyle interventions promote healthy behaviours that have multiple health benefits in diabetes prevention. Implementation of lifestyle interventions with people who already have T2DM continues to be a challenge for many medical professionals. Changing lifestyle behaviour depends on people changing many aspects of their own behaviour (e.g., exercise and diet). A large proportion of diabetes health improvement depends on the willingness of people to change their lifestyle behaviour. Changing lifestyle behaviour among people with T2DM can depend on how much knowledge they have about the illness and their self-belief that they can make the proposed lifestyle changes, which is influenced by their attitudes toward the illness. For example, in a study in Egypt, Kamel, Badawy, El-Zeiny, and Merdan (1999) found that diabetes practices or preventive measures among the general population were related to level of knowledge and attitudes. Lange and Piette (2006) reported that the higher was the level of education that people with diabetes had, the greater was their belief in the importance of treatment for controlling their diabetes and, thus, the lower were levels of fatalistic beliefs. In a Malaysian study, Ambigapathy, Ambigapathy, and Ling (2003) found that knowledge of and attitudes toward diabetes among participants were significantly correlated. When a person has positive attitudes and is willing to practise according to the knowledge they have about the illness, they are more likely to
4
engage in diabetes self-management regimens compared to others. In a survey of elderly people with diabetes in Hong Kong, J. Tang et al. (1999) found that attitudes to diabetes influenced self-management. Diabetes is associated with many complications and has major impact on QoL and life expectancy (Catanzariti, Faulks, & Waters, 2007). Diabetes can contribute to poor QoL, if it is not well controlled. Individuals with diabetes are at greater risk than the general population of being diagnosed with other complications, such as kidney disease, peripheral vascular disease, lower extremity ulcers and amputations, retinopathy, and neuropathy (Ciechanowski, Katon, Russo, & Walker, 2001). About 95% of diabetes management is conducted by the individuals who have diabetes (Anderson, 1985). Thus, it is important for people to change their lifestyle behaviour in order to manage their health condition. However, changing behaviour to manage T2DM can then affect QoL. T2DM is also considered one of the most psychologically and behaviourally demanding of the chronic medical illnesses (Cox & GonderFrederick, 1992). The psychosocial burden faced by those having diabetes is likely to negatively affect their QoL and is likely to impact on their selfmanagement behaviour (Glasgow, Ruggiero, Eakin, Dryfoos, & Chobanian, 1997). Kaplan (1990) argued that behavioural outcomes are the most important consequences in studies of health care and medicine. These outcomes can be longevity, health-related QoL, and symptomatic complaints. Traditional measures using biomedical variables often have limited reliability and validity (Kaplan, 1990).
5
The prevalence of, and QoL concerns with, diabetes have been widely studied and knowledge, attitudes, or self-management practices have been shown to relate to each other in particular populations (e.g., Ambigapathy et al., 2003; Fey-Yensan, English, & Genova, 2003; Palaian et al., 2006; J. Tang et al., 1999). However, little research has demonstrated the network of associations between knowledge, attitudes, self-management, and QoL in people with T2DM. Further, no research has been identified that has examined the relative influence of these important factors. Thus, it is important to know the level of knowledge, attitudes, and self-management among people with T2DM in a study population. Furthermore, this should lead to examination of the inter-correlation of knowledge, attitudes, and self-management, and whether or how these variables influence the QoL of people with T2DM. Developments in the diagnosis and management of diabetes have considerably improved health outcomes. These outcomes are most likely influenced by the awareness about the disease of people with diabetes and their efforts to achieve the goals associated with effective self-management particularly related to lifestyle. Although lifestyle varies in different countries, both developed and developing countries are experiencing an increase in the number of people with T2DM. In some western developed countries, such as Australia, where many people engage in sedentary lifestyles and there have been population-based changes in diet, these lifestyle factors contribute to an increase over time in the number of people with T2DM. This is associated with a large health, social, and economic burden for individuals with the disease, their families, and the community.
6
Some developing countries especially in Asia, Africa, and South America, have experienced rapid development associated with changes of lifestyle, physical activity, and eating habits. Researchers have pointed out that these lifestyle changes create a rapid increase in the number of people with diabetes (Horton, 2009). Although, the prevalence of diabetes is generally higher in developed countries than in developing countries, it is predicted that the greatest impact of increased prevalence of diabetes by 2025 will be experienced in developing countries, including many in Asia (Hossain, Kawar, & Nahas, 2007). Unfortunately, in many Asian countries, awareness about diabetes and its complications is not as advanced as it is in Western regions (Jabbar, Hameed, Chawla, & Akhter, 2007). Malaysia is an Asian developing country that has experienced a surge of diabetes incidence estimated to lead to a total of 2.48 million people with diabetes by the year 2030 (Zanariah et al., 2009). To address this, it is important to increase understanding of the major factors that influence lifestyle related to the prevention and management of T2DM. In this thesis, I focus on examination of the relationships between knowledge, attitudes, self-management, and QoL among people with T2DM. The main aim is to examine not only the levels of variables like knowledge, attitudes, self-management, and QoL in people with T2DM, but also the ways in which they relate to each other. Although researchers have examined relationships between two of these variables, such as knowledge and self-management or selfmanagement and QoL, no research has been found that systematically examines the relationships between all of these variables at the same point in time in the same people. Further, I aim to explore cultural differences among these variables
7
and their relationships between Australia, a developed western society, and Malaysia, a rapidly developing Asian society. Whereas studies have been conducted in a range of western countries and also in a number of Asian cultures, no studies have been identified that have directly compared the levels of and relationships between these key variables for the self-management of T2DM in different cultures. Knowledge gained in this research will help experts to provide more opportunity for the community to understand their level of diabetes knowledge, attitudes, self-management, and QoL. The results of the research in this thesis will inform hospital staff and community health professionals of the current level of diabetes knowledge, attitudes, self-management, and QoL among people with T2DM, which will help them to decide which of these areas they should focus on. The results will also help health professionals to identify ways to enhance the QoL of people with T2DM by implementing tailored programs that improve their diabetes knowledge, attitudes, or self-management.
8
CHAPTER 2 LITERATURE REVIEW In this chapter, I describe the pathology of T2DM, as well as issues of diagnosis and treatment. Then I explore the key concepts of knowledge of diabetes, attitudes, self-management, and psychological well-being of people with T2DM. Current views of the relationships between knowledge of diabetes, attitudes, and self-management in people with T2DM are then considered. Next, the impact of the relationships between knowledge of diabetes, attitudes, and selfmanagement on T2DM QoL is reviewed. Then, the relationships between knowledge, attitudes, self-management, and QoL are demonstrated in a model that is visually illustrated. Finally, I review the importance of cultural comparison between people with T2DM from two different countries, one with a long history of urban living and another with rapid ongoing transition to an urban lifestyle. All literature searches were conducted systematically using appropriate database search engines (e.g., PubMed, Cochrane library, MEDLINE, ScienceDirect, and Psychology and Behavioral Sciences Collection) with keywords that focused on the central topics of the present research. These were diabetes knowledge, attitudes, self-management, quality of life, diabetes, T2DM, prevalence, and structural equation modelling. I identified, appraised, selected, and synthesised all high quality research evidence relevant to the present research areas. The literature in this chapter was updated throughout the thesis writing. This chapter concludes with a summary of the literature review and a statement of the purpose of this thesis, leading to the development of a conceptual framework for this research, which is outlined in Chapter 3.
9
Type 2 Diabetes Mellitus Diabetes is a condition in which there is disregulation of glucose metabolism (Peyrot, McMurry, & Kruger, 1999). There are two major types of diabetes that are named insulin-dependent or Type 1 diabetes (T1DM) and noninsulin-dependent or Type 2 diabetes (T2DM). T1DM was also previously known as “juvenile or childhood onset” diabetes (WHO, 2011). People with T1DM cannot secrete sufficient amounts of insulin due to dysfunction of the pancreas. Thus, T1DM requires daily administration of insulin (WHO, 2011). T1DM is not caused by lifestyle, but it may be inherited with strong family links, which cannot be prevented (DA, 2008). However, maintaining a healthy lifestyle, regular blood glucose testing, and medicating with insulin are very important to manage the health condition of people with T1DM (DA, 2008). T2DM is often called “adult-onset” because it usually affects older adults (DA, 2011). However, more and more young people, even children are getting T2DM (DA, 2011). T2DM accounts for 90% of people with diabetes around the world and it is associated with excess body weight and physical inactivity (WHO, 2011). People with T2DM do not require insulin treatment to remain alive, although up to 20% of people with T2DM are treated with insulin to control blood glucose levels. T2DM is characterized by a variable combination of insulin resistance with relative insulin deficiency (Gallwitz & Haring, 2006). Normally, insulin secretion acts on cells to transport glucose from the blood into the cell. In insulin resistance, however, the sensitivity of the cells to insulin is decreased. As a result, higher levels of insulin are needed in order for insulin to have its usual effect on sugar in the blood (Valberg & Firshman, 2009). Over time, the
10
necessity of producing this extra insulin puts a strain on the insulin producing cells in the pancreas. Eventually, the insulin producing cells are exhausted, and insulin production falls resulting in insulin deficiency. A person with T2DM therefore may eventually experience relative insulin deficiency and require insulin if other medications fail to control glucose levels adequately (Alberti, 2010). Although researchers study and debate the reasons, it is still uncertain what causes T2DM. In a report on the epidemic of T2DM in the twenty-first century, Cusi (2009) stated that most researchers agree that increasingly sedentary lifestyles coupled with excessive calorie intake, in particular of calorie rich foods high in carbohydrates and saturated fats, have led to an explosion in the prevalence of T2DM. Although there are some treatments associated with reduced severity and duration of symptoms in T2DM, currently, no pharmacological therapies clearly arrest the progression of T2DM in the longterm. Clinical Course Symptoms of diabetes include thirst, polyuria, weight loss, recurrent infections, and in severe cases, precoma (Alberti, 2010). It is a chronic, metabolic disease with significant morbidity and mortality due to its major and severe complications (Abdullah, Margolis, & Townsend, 2001), including staphylococcal skin infections, retinopathy noted during a visit to the optician, polyneuropathy causing tingling and numbness in the feet, impotence, and myocardial infarction or peripheral gangrene due to arterial disease (Kumar & Clark, 2002).
11
Risk factors of T2DM appear to vary between different ethnic populations. Regardless of ethnicity, a family history of diabetes increases the risk of future T2DM (Lyssenko et al., 2005). However, the relative effect of family history decreases with the increasing prevalence of T2DM in the population. The predictive value of family history is also relatively low in young individuals, whose parents may not have yet developed the disease. A low level of physical activity, abdominal obesity, and the presence of the metabolic syndrome also increase the risk of T2DM (Laaksonen et al., 2002; Lorenzo, Okoloise, Williams, Stern, & Haffner, 2003). Metabolic syndrome is not a disease itself, but is a number of disorders that occur together which increase the risk of developing T2DM. Being overweight or obese and physically inactive can increase the risk of metabolic syndrome (Better Health, 2011). The development of T2DM can be due to lifestyle factors, such as smoking, sedentary lifestyle, and high dietary fat intake (Gomersall, Madill, & Summers, 2011). Thus, T2DM can be brought under control if people with T2DM moderate such factors themselves. Diagnosis T2DM presents with the symptoms of excessive thirst and urination, and tiredness. The symptoms of T2DM usually develop very gradually. In general, people with T2DM in its early stages have no symptoms at all and are only diagnosed after routine medical screening, which finds a high blood glucose level or glucose in the urine. Due to late diagnosis, people with T2DM may have had the disease for many years unknowingly and can have significant complications already present at the time of diagnosis (Ambler, Ambler, Barron, Cameron, & May, 2008).
12
Diagnosis of diabetes requires the identification of a glycaemic cut-off that discriminates between normal individuals and those with diabetes. The cutoff is based on the blood glucose level (American Diabetes Association [ADA], 2012). Considering the Diagnosis and Classification of Diabetes Mellitus, in 2012, the ADA recommended the following for a diagnosis of diabetes. Someone is diagnosed with Diabetes Mellitus when symptoms of diabetes have occurred and fasting plasma glucose is over 126 mg/dl (7.0 mmol/l), or 2-hour plasma glucose value is over 200 mg/dl (11.1 mmol.l). As mentioned previously, the symptoms of diabetes include polyuria, polydipsia, and unexplained weight loss. Polyuria refers to abnormally large amounts of urine produced in the body and polydipsia refers to excessive fluid intake. Thus, one may experience exceeding thirst due to the loss of fluid. Treatment Diabetes is a chronic disease for which there is currently no cure. However, diabetes can be managed with various treatments that are available in most developed countries (Ambler et al., 2008). In the health system of many countries, there are multidisciplinary diabetes teams that assist people to make changes to healthier lifestyle and other aspects of diabetes management. Diabetes management includes weight control, food planning and healthy eating, exercise, monitoring blood glucose, and follow-up screening for other complications (Ambler et al., 2008). Some people with T2DM can successfully control their blood glucose level with dietary measures, exercise, and weight loss, so they may not require pharmaceutical treatment, at least not for a number of years (Ambler et al., 2008). Oral medication and injections of insulin are used to control blood
13
glucose level when the diet and exercise diabetes management methods are unsuccessful for people with T2DM (Ambler et al., 2008). In general, most people with T2DM will eventually require pharmaceutical measures to manage their diabetes. The three main types of medications used for T2DM treatments are: metformin and thiazolidinediones, which increase insulin sensitivity, sulphonylureas and meglitinides, which increase pancreatic secretion of insulin, and acarbose, which alters the absorption of carbohydrates from food and reduces rises in blood glucose levels (Ambler et al., 2008). When the combination of lifestyle measures and oral medication inadequately control T2DM, insulin treatment is commenced with or without oral medication (Ambler et al., 2008). Prevalence of Type 2 Diabetes Mellitus Diabetes is a major and growing health problem affecting all ages in most countries around the world. The prevalence of diabetes mellitus (predominantly in T2DM) is increasing globally (Kuller, 1997), so diabetes has emerged as one of the World's biggest health problems and its prevalence is increasing at an alarming rate (Asha, Pradeepa, & Mohan, 2004). The greatest increases in the future are expected to occur in developing countries such as Asia and Africa following the trend of urbanization and lifestyle changes (WHO, 1999). The following sections provide an overview of the global prevalence of diabetes, and an overview of diabetes in Australia and Malaysia, specifically for T2DM. Global Prevalence and Trends The prevalence of diabetes is dramatically rising worldwide. It was estimated that 194 million people worldwide, or 5.1% of the adult population is
14
suffering from diabetes and this is expected to increase to 333 million, or 6.3%, by 2025 (International Diabetes Federation [IDF], 2011). It was estimated that at least 30 million people in the Western Pacific region have diabetes. This number is expected to double by 2025 (WHO, 2000a). The WHO (2005) predicts that 60% of the worldwide population with diabetes will be in Asia in less than a decade. The rapidly developing Asian nations, such as Singapore, Malaysia, and Thailand are projected to experience the biggest surge associated with an increase in the incidence of diabetes complications. Diabetes is also identified as the fourth or fifth leading cause of death by disease in most high-income countries (IDF, 2011). It was estimated that 25 million years of life is lost each year due to diabetes-related mortality (IDF, 2006). This increasing trend of diabetes worldwide has coincided with the changes of environment and lifestyle due to more advanced industrialization and globalization. These lifestyle changes result in more sedentary jobs, aging populations, and increased availability of food and beverages containing high content of sugar, salt and fat, and less consumption of fibre (Kolb & Mandrup-Poulsen, 2010). Diabetes mellitus is a chronic medical condition. Given the pandemic potential of diabetes and the fact that it is usually diagnosed late, there is a wide concern about the global burden and the enormous cost of treating the debilitating complications of T2DM. Furthermore, it is estimated that one in three diabetics is undiagnosed (Young & Mustard, 2001). This statistic is based on a study conducted by Young and Mustard in Canada, where health facilities are plentiful and health professionals are well trained. Thus, there may be even larger proportion of people who are undiagnosed for diabetes in developing countries,
15
which may even exceed the number of diagnosed cases, such as in South Africa with 85% of those with diabetes are thought to be undiagnosed (World Diabetes Foundation, 2012). Australia Diabetes is one of the leading threats to the health of Australians. Between 2011 to 2012, it was estimated that 4.0% of the Australian population (875 400 people) reported to have diabetes, of which around 12.4% had T1DM, 85.3% had T2DM and the other 2.3% were unknown (Australian Bureau of Statistics [ABS], 2012). T2DM is the predominant form of diabetes in Australia and worldwide. There are about two million Australians who are categorised as having risk of developing T2DM (Diabetes ACT, 2012), and approximately 280 Australians develop diabetes every day (DA, 2012a). King, Aubert, and Herman (1998) predicted an increase in diabetes cases between 1995 and 2025, of which 42% will occur in the developed countries, which includes Australia. Compared with other diseases, diabetes is the sixth leading cause of death in Australia, especially T2DM which, continues to be the fastest growing chronic disease in Australia (DA, 2012a). In the National Diabetes Register (NDR) it was reported that 25% of eligible National Diabetes Services Scheme (NDSS) registrants with derived T2DM live in the state of Victoria, which is the second highest percentage after New South Wales with 40% (NDR, 2006). Over the last 10 years, the prevalence of diabetes in Victoria has increased dramatically. The figures released recently by Diabetes Australia Victoria (DAV) showed that 252, 000 Victorians were known to have diabetes in 2011 and this number skyrocketed from 95, 000 in
16
2001 (Betts, 2011). In term of percentages, about 2% of Victorians were diagnosed with diabetes in 2001, but this figure had more than doubled to 4.5% in 2011 (DAV, 2011). In 2001, there was only one area considered a diabetes hotspot, but in 2011 the number of diabetes hotspot areas had risen to 64, a dramatically increase of 63 hotspots in 10 years (DAV, 2011). An area is considered a diabetes hotspot when more than 4% of its population has diabetes either T1DM or T2DM. The main driver of this epidemic of diabetes in Victoria is due to the increasing prevalence of risk factors in the population, such as obesity (Betts, 2011). T2DM is a common chronic disease among people 40 years and over, and is characterised by a relative insufficiency of insulin and resistance to its action (National Health Priority Areas, 1998). Recent evidence indicates that people are developing T2DM at younger ages (Catanzariti et al., 2007). In Australia, based on a survey of participants’ self-reported information, between 2007 and 2008, approximately 96% of people with diabetes were aged 35 years or more and 43% were aged 65 years or more (Australian Institute of Health and Welfare [AIHW], 2011a). More than 70% of Australian adults aged 15 years or over did little or no exercise as reported by AIHW (2011b). The rates of overweight or obesity have risen from 39.2% in 1990 to 55.4% in 2008 (AIHW, 2011b). This indicates that Australians are susceptible to being overweight and tend to develop obesity at a younger age than they used to, which is the key risk factor for the development of T2DM. This is because an increase in body weight will lead to increased insulin resistance and effects in insulin secretion.
17
Australia is a multicultural nation. According to the ABS report on the 2006 census, 69.6% of residents in Victoria stated that they were born in Australia and others were from England (3.3%), Italy (1.7%), New Zealand (1.3%), Vietnam (1.2%), and China (1.1%) (ABS, 2007). The top five languages spoken at home, other than English, in the 2006 census were Italian (2.7%), Greek (2.4%), Vietnamese (1.5%), Cantonese (1.4%), and Mandarin (1.3%) (ABS, 2007). Holdenson, Catanzariti, Phillips, and Waters (2003) reported that a larger percentage of overseas-born people had diabetes than Australian-born. In 2004-2005, among those born overseas people, those from Southern and Central Asia had the highest rates of diabetes with 8.7%, followed by people from North Africa and the Middle East, with 6.6%, and those who came from South East Asia with 5.7% (ABS, 2006). The rate of diabetes for Australian-born people was 3.3% (ABS, 2006). Results from the Melbourne Collaborative Cohort Study indicated that Greek and Italian-born immigrants aged 40-69 years had higher diabetes prevalence rates with 9.7% and 9.4% respectively than their Australian counterparts with 2.9% (Thow & Waters, 2005). Based on the National Health Survey (NHS; 2001), Holdenson et al. (2003) reported that the prevalence rates of diabetes among immigrants from both the UK and Ireland, and other Northern and Western European countries were not statistically significantly different from the prevalence in the Australian-born population. It is, however, still difficult to assess the prevalence of T2DM in various culturally and linguistically diverse (CALD) communities due to data limitations. In particular, data are often based
18
only on regional groups, rather than on specific country or language-based classification (Thow & Waters, 2005). Diabetes can cause personal, social, and financial burdens for a country. Due to the increasing prevalence of diabetes in the country, Australia spent almost AU$990 million in 2004 to 2005 on diabetes expenses, including hospital admission, medical services, pharmaceuticals prescription and research (AIHW, 2011c). Moreover, people with diabetes had other disabilities and were more likely to experience psychological distress than those without diabetes. For example, in 2003, 56% of the people with diabetes had disability, such as severe core activity limitations, which reduced their capacity to function well in terms of self-care, mobility, and communication (AIHW, 2011c) and 41.6% of adult Australians with diabetes reported having medium, high, or very high levels of psychological distress (AIHW, 2011d). Therefore, there are still greater needs for research to be conducted on diabetes, especially T2DM, in Australia with an objective that the prevalence and burden of disease can be minimized and psychological well-being of people with T2DM can be understood and improved. Malaysia Malaysia has undergone an enormous socioeconomic and demographic transformation over the last two decades as a result of massive industrialization, globalization, mechanization, and an improved educational system. In this transition period, the country has experienced rapid improvement in the standard of living and the QoL, a reduction in death rate, and concomitant ageing of the population. Therefore, Malaysia, a multiethnic nation of three major Asian races (Chinese, Indians, and Malays) with a population of about 25 million, is highly
19
prone to an epidemic of diabetes, considering the global trend, the socioeconomic transition in the country, and the well-known risk factors of the disease (Yun, Hassan, Aziz, Awaisu, & Ghazali, 2007). The greatest increases in T2DM are likely to occur in developing countries, such as Malaysia, as it becomes industrialized and subjected to the influence of globalization. Data from the Western Pacific region declaration in 2000 (WHO, 2000a) stated that the prevalence of diabetes in Malaysia had reached 8.9% and this figure had risen to 14.9% in 2006 (3.4 million people; Ali & Jusoff, 2009). Furthermore, many of them were not aware of their condition (Malaysian Diabetes Association, 2007). In 1993, the Second National Health and Morbidity Survey reported that the prevalence of diabetes among adults was 8.2% in urban areas and 6.7 % in rural areas of Malaysia (Public Health Institute, 1997). The overall prevalence of diabetes among adults above 30 years increased by 80% over the decade between 1996 to 2006 from 8.3% to 14.9% (Zanariah et al., 2009). During this period, the prevalence of newly diagnosed diabetes increased from 2.5% (National Health and Morbidity Survey [NHMS] II) to 5% (NHMS III), a rise of 12% per year (Zanariah et al., 2009). In Kelantan, a predominantly Malay region in the north-eastern state of the Malaysia Peninsula, the prevalence of diabetes among people above 18 years old was 11.7% in 2006 (Ismail & Nui, 2007). Now, this illness has extended to younger people, so more children and adolescents in the Asian-Pacific region are being diagnosed with T2DM (IDF, 2003). The Second Malaysian NHMS estimated that 14.6% of people with diabetes develop diabetic retinopathy, with 10% developing kidney diseases and
20
50% suffer neuropathy (Public Health Institute, 1997). According to a report in the Diabetes Care Data Collection Project Study in Malaysia that involved 6,836 participants, the mean age of onset of T2DM was 56.4±12.7. Of people with T2DM, 70% were between 30 and 60 years of age, and 12% were over 65 years (Mustaffa, 1999). The diabetes prevalence seemed to increase with age and the highest prevalence of undiagnosed diabetes was among the age group of 60-64 years old (Zaini, 2000). Treatment comprises non-pharmacological methods, such as diet and exercise, and pharmacological methods with oral hypoglycaemic agents and insulin injection. Malaysia is a country with a multiethnic population with three main racial and cultural populations, namely Chinese, Indian, and Malays, all well represented in this country. China and India are braced to exceed the world’s prevalence rate of T2DM (Ramachandran, Ma, & Snehalatha, 2010; Zaini, 2000). This is likely to be reflected in the large Chinese and Indian sections of the population of Malaysia showing increasingly high prevalence rates of T2DM. This is because migrant status itself, together with socio-economic and lifestyle changes, are strong predictors to diabetes (Zaini, 2000). Malaysia will certainly be affected by an epidemic of diabetes, because there has been a major change in the lifestyles and longevity of the population. Furthermore, diabetes is usually diagnosed late and results in an unacceptably high cost of care due to complications, such as cardiac and vascular diseases, renal, and neurological dysfunctions (Zaini, 2000). It was estimated that health services diagnoses one case of T2DM every minute (Mahathevan, 2003). Mahathevan also claimed that Malaysians with
21
diabetes are not knowledgeable about this disease and its complications. The low level of awareness of diabetes among the public, decision makers, and health professionals, a low priority in national health policy and plans, and limited availability and access to appropriate preventive and curative care, have all been identified as major issues for the management of diabetes in Malaysia (Ooyub, Ismail, & Daud, 2004). Malaysia has started to adopt various strategies to reduce the incidence of diabetes through joint action with other Western Pacific region countries (WHO, 2000b). However, there is a lack of published research or health promotion programmes being carried out in rural communities compared to the urban community in Malaysia. Mohan et al. (2005) pointed out that large diabetes education programmes are urgently needed both in urban and rural areas of India. In reporting on their study of people with diabetes in Malaysia, Ambigapthy et al. (2003) suggested that rural areas of Malaysia should not be left out in health educational and promotional activities concerning diabetes. In this matter, more attention should be given to the rural area, as rural health care delivery is often inferior to that of urban areas (Schorr, Crabtree, Wagner, & Wetterau, 1989). Diabetes Knowledge The recommendations given to people with T2DM have increasingly emphasized self-management. Diabetes education programs, which started in the late 1970s, were designed to ensure people had sufficient knowledge and understanding of T2DM and its treatment to manage key aspects of their own diabetes (Beeney, Dunn, & Welch, 1994). Many studies have been conducted using different questionnaires to evaluate diabetes knowledge among people with
22
T2DM (e.g., Dougherty, Schiffrin, White, Soderstrom, & Sufrategui, 1999; M. Rose, Fliege, Hildebrandt, Schirop, & Klapp, 2002; Rothman et al., 2005; Samuel-Hodge et al., 2009; Torres, Virginia, & Schall, 2005). There is also increasing complexity in the recommendations and information given to people regarding their chronic disease as medical knowledge develops. Therefore, there is a great need to evaluate and understand the level of diabetes knowledge among people with T2DM. Lack of knowledge among the public may also contribute to increasing the number of people who develop diabetes. A survey that was conducted by the Department of Pharmacy of the National University of Singapore to evaluate the general public’s knowledge of diabetes in Singapore, showed that the public was generally well informed about diabetes (Wee, Ho, & Li, 2002). However, in terms of specific knowledge about diabetes, the majority of people did not know about the types of diabetes and were not aware of the risk factors for diabetes. Awareness of the diabetes risk factors is essential for prevention of T2DM. In the same country, Wong and Toh (2009) reported that respondents with greater understanding of diabetes mellitus were more likely to have favourable health practices. Therefore, it is proposed that knowledge and understanding of diabetes may encourage people to take precautions to prevent them from developing the disease and participate in health screening available for the public. In measuring diabetes knowledge, some researchers have reported unfavourable outcomes of knowledge about diabetes among people diagnosed with either T1DM or T2DM. For example, in their study in Pakistan, Naeema, Abdul, Zafar, and Rubina (2002) showed that overall knowledge regarding
23
diabetes was not satisfactory, with only 13% of people with diabetes attaining high scores on an objective test of diabetes knowledge. On the other hand, researchers from Hong Kong found that half of the people with diabetes who they studied had adequate knowledge scores (J. Tang et al., 1999). Lowea and Bowen (1997) found that knowledge of diabetes was high in a sample from Newcastle, NSW. In a Massachusetts sample, Carbone, Rosal, Torres, Goins, and Bermudez (2007) reported that Latino people with diabetes demonstrated some knowledge about diabetes, but misunderstood causation. Examining other areas of knowledge about diabetes among Malaysian people with diabetes, Ambigapathy et al. (2003) stated that 91% of the respondents correctly thought an increase in thirst was one of the symptoms of diabetes. Eighty-nine percent of this sample also answered correctly to frequent urination and 58% of them mentioned increased appetite as a symptom of diabetes. A question regarding weight loss, however, showed poor response with only 27% of the sample aware that untreated diabetes is associated with loss of weight (Ambigapathy et al., 2003). Knowledge was found to be deficient regarding risk factors for diabetes among the general population. A study done in India by Mohan et al. (2005) showed that only 31.2% of the respondents recognised family history as a risk factor and 21.2% of participants answered that eating food with a high level of sugar and fat content was a risk factor. Other risk factors that were assessed in the Mohan et al. study included obesity and physical inactivity, but only 11.9% of the respondents were aware of these risk factors. Teufel-Shone, Drummond, and Rawiel (2005) conducted a study in Arizona which revealed that there was a significant improvement in the level of knowledge about risk factors for diabetes
24
among people with diabetes and their family members after they had gone through a family-based diabetes program. This finding demonstrated that giving education about risk factors associated with diabetes to family members of individuals with diabetes improved their level of knowledge and may potentially encourage self-management practices to address the modifiable risk factors. In a study by Kamel et al. (1999), the researchers found that 83.7% of the participants scored very poorly for level of knowledge regarding complications. This was supported by the study done by Mohan et al. (2005), in India, in which only 19% of those studied had knowledge regarding complications of diabetes. Among these respondents, 21.8% recognised foot problems as a complication, while 15.9% and 16.3% recognised kidney disease and eye disease as complications respectively. However, few people realized there were other major complications of diabetes, such as heart disease and stroke (Mohan et al., 2005). Thus, diabetes knowledge about self-management of T2DM is an important area for further research and for intervention, which could reduce the burden of managing the disease complications that may develop if it is not well managed and controlled. Assessment regarding treatment of diabetes also showed a varied level of knowledge in different studies among the general population. Kamel et al. (1999) found that 96.3% had poor awareness of the disease control aspect of diabetes. Knowledge about prevention of diabetes was also assessed in most of the studies together with other aspects. A total of 66% of the study population in Egypt were revealed to have good knowledge about prevention of diabetes regarding dietary
25
control in the study by Kamel et al., while only 2.7% of the sample had awareness of the importance of exercise (Kamel et al., 1999). Tham, Ong, Tan, and How (2004) used a pre-tested diabetes questionnaire with 43 questions to explore risk factors, treatment, management, and monitoring of the illness in their study of a convenience sample of people with diabetes and without diabetes who attended a hospital in Singapore. Tham et al. reported that people with diabetes had an overall mean score of 68.1% on diabetes knowledge, which was slightly higher than people without diabetes, who had an overall score mean of 65.9%, however this difference was not statistically significant. In other words, people with diabetes did not show significantly higher overall knowledge scores than those without diabetes. On the other hand, the results in the Singaporean sample showed that people with diabetes who were younger (≤ 54.99 years) scored higher than those who were older (≥ 55 years). Rafique, Azam, and White (2006) found a similar result in their study in Karachi, Pakistan, where 92.5% of their participants had been diagnosed with T2DM. Their results showed that younger participants scored higher in diabetes knowledge than those who were older. This indicates that younger people with T2DM had greater knowledge about diabetes than older people with T2DM. In addition, duration of diabetes has also been found to be associated with level of diabetes knowledge. In a study conducted among people with diabetes in north Malaysia, Yun et al. (2007) found that educational level and numbers of years with diabetes were the most important predictors of knowledge about diabetes. It has also often been reported that fear of chronic complications increased with diabetes duration (Gåfvels, Lithner, & Börjeson, 1993). This may
26
negatively or positively influence the way people act toward their illness, whether they are motivated to know more about the disease, whether they maintain a positive attitude, or whether they show greater adherence to diabetes self-care regimens. Gåfvels et al. (1993) reported that people with shorter diabetes duration were more concerned about the management of their diabetes than people who had been diagnosed with diabetes for longer. A study conducted among people with diabetes in a hospital in Malaysia, using a questionnaire developed by the researchers, showed that 81% of the sample had good knowledge (Ambigapathy et al., 2003). This questionnaire evaluated the level of diabetes knowledge among people with diabetes in the areas of its symptoms, complications, prevention, dietary habits, and the importance of exercise. Eighty percent of the respondents in the study answered correctly to healthy diet with low fat, cholesterol, and carbohydrate as a preventive measure for diabetes. Among these respondents, 90% also realized the importance of exercise for diabetes prevention. In relation to other measures, 93% of these Malaysians with diabetes agreed that weight control is important in diabetes prevention (Ambigapathy et al., 2003). Ambigapathy et al. (2003) also reported that the majority (87%) of their respondents scored 50% and above in the diabetes knowledge questionnaire. Level of knowledge had a negative association with age among a sample from Singapore, where level of knowledge was found to be higher in those who were younger (≤54.99 years old). Tham et al. (2004) suggested that this might be due to poor education of the older adults in their study during their younger years, which caused them to have a limited understanding of the information given by
27
health care authorities or friends regarding diabetes or any other illness. This was supported by the study carried out by J. Tang et al. (1999), which showed similar poor results regarding age in relation to the level of knowledge in a sample from Hong Kong. The research literature has shown inconsistent results in regard to the relationship between gender and diabetes knowledge. For example, no significant differences were found between males and females in the study conducted by Tham et al. (2004) in Singapore. However, a study carried out in Egypt by Kamel et al. (1999) showed a poorer level of knowledge with regard to to illness control among females compared to the male population. A similar study, done by Rafique et al. (2006) on people with diabetes in Karachi, Pakistan illustrated that female respondents had poorer levels of knowledge compared to males. This may be explained by the illiteracy rate where women (18.3%) showed a significantly higher literacy rate than men (10.5%). Yet, in the Indian study done by Mohan et al. (2005), the researchers revealed that 77.6% of males knew about diabetes compared to 73.6% of females, suggesting there was little difference in knowledge between genders. In addition, a study done by Garrett et al. (2005) in Minnesota showed that there was no significant difference in the level of knowledge between male and female respondents after they offered them an intervention, although, the pre-intervention level of knowledge proved to be higher in male respondents. This suggested that females were able to increase their level of knowledge to equal that of males when given the opportunity (Garrett et al., 2005).
28
Level of education has a direct relationship with the level of knowledge regarding diabetes. In a study conducted in Singapore, respondents with high educational background were found to have a higher level of knowledge compared with persons with a low educational level (Wee et al., 2002). Kamel et al. (1999) also demonstrated from their study in Egypt that university graduates had a greater level of knowledge than respondents with little formal education. Knowledge about the prevention of diabetes also increased with level of education (Mohan et al., 2005). In another study done in the United States, Sharp and Lipsky (1999) reported on the ability of continuous medical education programmes to improve level of awareness regarding diabetes. In this study, logistic regression analysis confirmed that educational level was the most influential predictor variable correlated with increase in knowledge of the disease. This was in agreement with research done by Yun et al. (2007), showing that educational level was an important predictor of knowledge and awareness, irrespective of disease status. Personal history or family history has been found to display a significant relationship with level of knowledge among the general population. Respondents from Singapore with a personal history of diabetes showed greater knowledge regarding diabetes than those who had no such history (Tham et al., 2004). A study investigating the significance of family history in Canada by McManus, Stitt, and Bargh (2006) also concluded that people with a family history of diabetes were more knowledgeable on risk factors for diabetes than those without a history. Pre-intervention studies showed a significant relationship between personal history or family history of diabetes and level of knowledge among the
29
general population (McMurray, Johnson, Davis, & McDougall, 2002). However, in a study on respondents who attended a diabetes education programme in Minnesota, Garrett et al. (2005) stated that personal and family history did not play an important part in awareness regarding diabetes. A study by Siminerio and Koerbel (2000) in Pittsburgh, USA, showed considerable increase in level of knowledge on this aspect after a diabetes education programme was held among school personnel. Knowledge regarding prevention of diabetes can be reflected through preventive practices after the educational programmes. In Glasgow, Scotland, Hamid, Knill-Jones, Sunita, and Rodgers (2006) showed a 20% increase in preventive practices regarding diabetes among the general population, which was a significant improvement on level of knowledge related to this aspect of diabetes care. Late diagnosis of the disease or absence of knowledge about the symtoms of diabetes have always been the biggest challenge for people with T2DM. For example, Tessaro, Smith, and Rye (2005) who conducted a qualitative study in West Virginia, United Sates, reported that most of their participants with diabetes never distinguished the symptoms before diagnosis. This means that the actual beginning of the disease may have been even further in the past for most people with T2DM. People with diabetes knew little about diabetes before diagnosis, unless there was a family member with diabetes. Lack of diabetes knowledge hinders the adoption of preventive health behaviours among the general population (Tessaro et al., 2005). There is still lack of knowledge about types of diabetes among people with diabetes in some developing countries. In Malaysia, a study examining this
30
concluded that the majority of both the diabetic and healthy groups did not know that there are different types of diabetes, possibly due to lack of understanding of the information they received during education and counseling (Yun et al., 2007). However, programmes such as diabetes education and self-monitoring of blood glucose are still important sources of improving knowledge about diabetes in people with T2DM (Centers for Disease Control and Prevention, 1997). Diabetes is a silent disease, so many people are not aware of it until they develop one of its life-threatening complications or they have medical tests for other reasons that identify high blood sugar levels (Wee et al., 2002). Knowledge of diabetes can assist in early detection of diabetes and thus early treatment and prevention can be adopted by people at risk. This should reduce the incidence of complications caused by diabetes. Assessing the level of diabetes knowledge is important especially when such ignorance becomes a threat to the health of the population. Diabetes Attitudes The attitudes people hold are important for the effectiveness of efforts to improve the results of treatment in chronic disorders, like diabetes mellitus. There has been great progress in identifying factors that predict psychological adjustment among people with diabetes (Cox & Gonder-Frederick, 1992). Many people with certain diseases may experience chronic frustration, being discouraged by the disease that often does not seem to respond to their efforts to manage it. Some may feel hopeless about their efforts and despondent about the possibility of preventing long-term complications (Polonsky, 2000). People with diabetes, especially T2DM, must make changes in habitual behaviours, in order
31
to successfully regulate the metabolic processes, such as monitoring their blood glucose levels by regular testing. It is important for them to make changes to their diet and to their physical activity levels in order to help them adjust their blood sugar levels. This can be difficult and involves an emotional struggle to find a way to include diabetes in their life and to confront the knowledge of mortality that diabetes may represent. Therefore, coping with diabetes can cause a psychological burden when people are diagnosed and many consider themselves responsible for their illness (Karlsen & Bru, 2002). Attitudes of people with T2DM can play an important role in their emotional response and coping with their illness, as well as affecting their efforts to manage their diabetes in everyday life. Although many people with diabetes are able to cope well with their illness and live normal, healthy lives, there are still substantial numbers who suffer emotionally, without receiving basic psychosocial support from their health community (Skovlund & Peyrot, 2005). Researchers have pointed out that psychosocial issues play an integral role in all aspects of diabetes care (R. J. Anderson, Freedland, Clouse, & Lustman, 2001). Monitoring of psychological health may also improve the outcomes for people with diabetes as part of ongoing diabetes care (Pouwer, Snoek, Ploeg, Ader, & Heine, 2001). Ajzen (1989) defined attitudes as the disposition to respond favourably or unfavourably to an object, person, institution, or event. Even though the definition of attitude may vary in different disciplines, most social psychologists agree that the characteristic attribute of attitude is its evaluative nature, such as pleasant versus unpleasant (Ajzen, 1989). These characteristics can be verbal
32
measureable responses, which are expressions of beliefs, feelings, and behavioural intentions. Some of the major theories of health behaviour, such as the health belief model, underline that attitudes and beliefs are major factors influencing health behaviour (Ajzen & Fishbein, 1980). A positive attitude toward diabetes illness can be an important factor that affects whether individuals with T2DM are in agreement with the general selfcare recommendations for people with T2DM. A study by Ambigapathy et al. (2003) in Malaysia showed that 95% of the respondents had an overall positive attitude towards prevention of diabetes. This demonstrated that more than three quarters of the respondents agreed with the positive statements on willingness to exercise (96%), ease in changing food habits (90%), and maintaining suitable body weight (88%). All the respondents (100%) agreed that people with diabetes should take responsibility for self-care in diabetes, while 86% of the participants agreed that most people with diabetes can enjoy life, while at the same time maintaining blood-sugar control within the guidelines recommended by medical advisors. However, 56% of respondents agreed that no diet restriction is needed if they felt well (Ambigapathy et al., 2003). Positive attitudes toward diabetes illness will also motivate people with T2DM to stay active in their everyday self-care activities. In research done in Malaysia, Yaacob, Ismail, and Bebakar (2007) reported that the majority of the participants in the study had a positive attitude toward their own responsibility to control and monitor their T2DM even during Ramadan fasting. This also showed that most of the people with diabetes who were participants in this study practiced Ramadan fasting regardless of their illness. Yaacob et al. also
33
concluded that with a positive attitude among people with T2DM, regarding disease control and monitoring, physicians were able to encourage more people with diabetes to do home blood glucose monitoring. People with positive attitudes toward their T2DM are also more likely to have the desire to seek additional knowledge about their illness. A study conducted by Tessaro et al. (2005) among the population in West Virginia showed that participants with diabetes wanted to obtain more information about the illness. They had positive attitudes toward seeking knowledge regarding diabetes from the physicians (Tessaro et al., 2005). However, the participants reported that there was a lack of diabetes education by physicians to help them deal with the illness. Physicians were not able to spend sufficient time with their clients to explain why and how they should manage their diabetes. Singaporeans were found to be not active in seeking information about diabetes. Only a small proportion made the effort to obtain information from health professionals, talks, and seminars (Wee et al., 2002). In the study, they found that most of their respondents’ information was obtained through friends, relatives, or books and magazines. Communication through laypeople and the impact of information presented in the mass media cannot be underestimated. Wee et al. (2002) also suggested that health professionals must actively deliver the information to the general public when they come into contact with them either as a practitioner or a community pharmacist. There is no clear evidence on how education can improve the attitudes that people have towards seeking knowledge, especially regarding diabetes. However, a study done on smoking revealed that basic advice given by doctors
34
had motivated smokers to seek more detailed advice from other counsellors at a later date (Lancaster, Stead, Silagy, & Sowden, 2000). Thus, simple advice and encouragement from doctors may also improve the attitudes of people with T2DM to seek more knowledge about their illness. In their study of 46 people in India with diabetes, Palaian et al. (2006), found that counselling by pharmacists was effective in improving the knowledge of people with diabetes, but not in improving their attitudes and practices. Research suggests that negative perceptions regarding medical treatment of diabetes still exist among the general population. This was shown in the study carried out by Rafique et al. (2006) in Karachi, where the respondents believed in traditional remedies as treatment for diabetes, with 25% of the participants thinking regular use of bitter-gourd could cure diabetes. Sixty percent of these participants also agreed that no restriction was required for bread made up from gram flour (Rafique et al., 2006). Cultural beliefs also played an important role in people seeking treatment for their diabetes. For example in a focus group study conducted by Tessaro et al. (2005) in West Virginia, diabetes diagnosis was often associated with blame and guilt and the perception of diabetes was self-induced. Tessoro et al. proposed that these experiences were related to cultural beliefs. Tessoro et al. also found that most of the general urban white population in West Virginia, considered that diabetes was a burdensome disease and a fearful condition with severe complications. The fear was stated as a reason to avoid seeing physicians. As for gender differences, the results varied in different studies. According to Mahathevan (2003), among people in Glasgow, Scotland there was
35
no significant difference between males and females regarding their attitudes towards diabetes. However, in the report of their study in Pakistan, Rafique et al. (2006) stated that male respondents had higher levels of positive attitudes in comparison with female respondents. The impact of interventions was compared among different genders by Nielsen, Olivarius, Gannik, Hindsberger, and Hollnagel (2006). Female participants showed higher adaptive attitudes towards lifestyle modification as part of diabetic management compared to male respondents. Level of education has been shown to influence the attitudes of the general population to diabetes. This showed that the level of attitudes increased with higher levels of education (Rafique et al., 2006). This finding was also supported by a study in Ontario. Pomerleau, Pederson, Østbye, Speechley, and Speechley (1997) stated that less educated individuals reported poorer attitudes about health behaviour. Nevertheless, in a recent study conducted in Brazil, Rodrigues, Santos, Teixeira, Gonela, and Zanetti (2012) found no statistically significant relationship between attitude scores and education level. In the Rodrigues et al. study, there were 123 participants with T2DM, whose mean years of education was 4.54 (SD = 3.66). They reported that 68.29% of the participants did not finish primary education. Hence, it is possible that the relationship was not significant due to the large proportion of participants who had low education background. Rodrigues et al. concluded that education level is still crucial in establishing a positive attitude when people are diagnosed with diabetes.
36
Tessaro et al. (2005) stated that psychological measures of attitudes towards diabetes varied between individuals with diabetes on the basis of socioeconomic status. Many of those with lower household income in an Appalachian population study thought that physicians did not understand their financial difficulty or they themselves were unwilling to deal with cost issues. From this, the researchers concluded that people with less household income would keep their problem to themselves instead of telling the doctors, and tended to ignore the prescriptions of physicians unless the physicians were willing to prescribe the medicine for free (Tessaro et al., 2005). Interventions have been shown to have an impact on one’s attitude and belief. Various studies in different fields have demonstrated the importance of doing interventions towards improving level of attitudes of people with diabetes, their family members, and practitioners. For instance, a study was done in Taiwan to assess the impact of continuous education programmes among pharmacists. It showed that their overall level of attitudes regarding diabetes improved significantly after they had completed the intervention (Chen, Lee, Huang, Chang, & Chen, 2004). Rickheim, Weaver, Flader, and Kendall (2002) pointed out that a group intervention regarding diabetes education worked as effectively as an individual intervention in their study in Minnesota, as both groups showed overall improvements in attitudes towards diabetes. Snoek and Skinner (2002) emphasized the need for effective, wellevaluated psychosocial interventions to assist people in dealing with the daily demands of diabetes. A study done by Hamid et al. (2006) among a multi-ethnic population in Glasgow, Scotland revealed that participants reported a significant
37
increase in attitudes that recognised the seriousness of diabetes and the value of tight control of diabetes after the educational intervention program. Donohoe et al. (2000) found significant improvements in the attitudes of people with diabetes towards foot care after an education intervention. These findings suggest that education and knowledge are the first step in helping people changes their attitudes and subsequently their behaviour to successfully manage their diabetes. Studies that have evaluated diabetes education programs have demonstrated that the educational models or treatment programs need to include behaviour change, rather than just improving knowledge and attitudes of participants (Dunn, 1990). Research has also shown that people with diabetes who have positive attitudes toward managing their diabetes will be more likely to change their behaviour in controlling their blood glucose level compared to people with negative attitudes (R. M. Anderson, Donnelly, & Dedrick, 1990; DeWeerdt, Vissar, Kok, & van der Veen, 1990; Dunn, 1990; Lockington, Powles, Meadows, & Wise, 1988; Masaki, Okada, & Ota, 1990). Fey-Yensan et al. (2003) also concluded that negative attitudes toward T2DM were significantly associated with barriers to self-care and diet management. Diabetes has been considered one of the most psychologically and behaviourally demanding chronic illnesses (Cox & Gonder-Frederick, 1992). Diagnosis and management of diabetes always involve aspects of lifestyle change, which requires long-term adjustment that can be very disruptive for people with T2DM and their families. Moreover, the regimen for diabetes management is complex and demanding. This includes complex nutritional practices, weight management, frequent monitoring of blood glucose, foot care,
38
in many cases insulin injections or oral medication (E. Fisher, Delamater, Bertelson, & Kirkley, 1982). Thus, measuring the degree of psychological adjustment to diabetes management is a key component for addressing psychosocial and emotional adjustment among people with diabetes, either T1DM or T2DM. Psychological adjustment measures that involve attitude statements should help researchers and practitioners to assess the perception of diabetes and its treatment. Diabetes Self-Management Self-management is defined as the knowledge and skills necessary for an individual to take care of oneself, manage crises, and change lifestyle to manage illness successfully (Clement, 1995). Self-management is an important part of daily life for people with diabetes. It has been reported that approximately 95% of diabetes care is self-treatment or self-management (R.M. Anderson, 1995). To control diabetes, individuals must monitor their daily lifestyle behaviour and often they must change long-held habits. Therefore, although self-management is vital in people with T2DM, it is not always as effective as health professionals would like. It is important that individuals adhere to self-management, to prevent further complications associated with diabetes in order to maintain or achieve a positive QoL. However, health professionals have to recognize that long-term behaviours are very hard to adjust or change. Thus, understanding factors that are associated with individual diabetes self-management behaviours is important for health professionals. These health professionals can then take appropriate action to help people with T2DM to improve self-management of the condition.
39
Controlling diabetes requires excellent self-management skills, where individuals are required to perform various treatments or activities by themselves. The major goal set for self-management among people with diabetes and their health care providers is tight control of blood sugar levels through adherence to a protocol of blood glucose self-monitoring, diet, exercise, and taking medication as prescribed (Paterson, Thorne, & Dewis, 1998). Hence, self-management activities demand a great effort, which many people find difficult to incorporate into their daily life (Glasgow & Eakin, 1998). Flexibility in self-management is important. Performing diabetes self-management every day may be considered to be a burden, frustrating and even overwhelming for some people with diabetes due to the large effort required to engage in various activities, and the need for flexibility (Polonsky, 2000; Watson, Briganti, Skinner, & Manning, 2003). This may become a frustrating task for many people with the condition in the longer term (Polonsky, 2000). In addition, people with diabetes have difficulty in sustaining their self-care activities when they are experiencing stress (M. F. Peyrot, et al, 1999). Diabetes self-management is an essential task in diabetes care, as good self-management will lead to improvement of glycemic control, metabolic control, blood pressure, and weight control, and will reduce complications. Some studies reported that there is a lack of information concerning what foods are nutritionally appropriate. For example, managing carbohydrate intake is a key factor in maintaining a healthy diet, and exercise of an appropriate intensity, duration, and frequency could improve diabetes prognosis in a person (Horowitz, Williams, & Bickell, 2003; Von Goeler, Rosal, Ockene, Scavron, & De Torrijos,
40
2003). Although the physician can suggest strategies to improve diabetes control, individuals with T2DM still have the responsibility to carry out the treatment regimen on a daily basis, usually for many years. Because most diabetes care tasks are conducted by the people with diabetes, health professionals have limited control over the management of each individual’s illness (Funnell & Anderson, 2000). The responsibility for managed their diabetes condition lies mostly with the individuals with T2DM. However, different individuals may have different preferences in seeking diabetes knowledge and in their use of health care information. Longo et al. (2010) in their qualitative study reported that people with diabetes make decisions about diabetes self-management depending on their current needs, seeking and incorporating diverse information sources not traditionally viewed as providing health information. Longo et al. found that most of the participants relied on sources of information from health professionals. Thus, health professionals still play important roles in influencing self-management decision-making among people with diabetes. The participants with diabetes in this study looked for information that was consistence with their own knowledge and experience of self-management, and concurrently sought clinicians to reinforce the information they had obtained and the appropriateness of undertaking certain behaviours that the individuals with diabetes proposed. Management of diabetes usually requires demanding behaviour change when people are diagnosed with T2DM, commonly including diet, exercise, medication, and blood sugar testing as the key components of self care (Glasgow & Eakin, 1998; Gomersall et al., 2011). The effectiveness of diabetes self-
41
management lies in the consistent application of preventive and clinical interventions. This includes efforts to detect the disease, promote effective selfcare, reduce the incidence of complications, and improve the coping skills of people with diabetes and related conditions (Ooyub et al., 2004). Therefore, lifestyle plays an important role in diabetes management. In order to manage this illness, people with diabetes must make a combination of medication and lifestyle changes. Diabetes management not only depends on medication treatment or drug therapy, but also on physical exercise, diet, and other lifestyle changes (ADA, 1997). Diet is considered the cornerstone of treatment for people with T2DM and is commonly used as first step therapy (UK Prospective Diabetes Study [UKPDS] Group, 1995). The risk of developing diabetes can be reduced by about 91%, by engaging in all these practices together. Even in persons with a positive family history of diabetes, the risk is reduced by 88% (Pinkney, 2002). Therefore, people with T2DM should continue to act on the advice from health professionals in regards to their diabetes self-management and retain a positive attitude that their diabetes is controllable. In a clinic-based study on diabetes self-management in Malaysia, about 99% of the respondents were practising at least two out of four diabetes regimens, which were regular exercise, healthy diet, monitoring blood glucose and monitoring body weight. However, only about half of the respondents (56%) were practising all these four regimens (Ambigapathy et al., 2003). It is very common for people with T2DM to have difficulty practising all the self-care recommended by their health practitioner. People with diabetes must comply with
42
demanding requirements of their treatment regimens and this can create fear of failure (Maes, Leventhal, & de Ridder, 1996). These regimens can involve daily behavioural tasks, such as changes of habits on diet and exercise with which people with diabetes must comply for the rest of their life (Karlsen & Bru, 2002). Practices such as diet, exercise, smoking cessation, and yearly checkups will help to prevent the incidence of diabetes and delay the progression of its complications in people who are already known to have diabetes (Oladele & Barnett, 2006). Handley et al. (2006) reported that the majority of the 228 study participants (75%) from San Francisco, California, with CHD risk factors, including those diagnosed with diabetes, were carrying out an action plan that focused on diet or exercise for self-management of their illness condition. Moreover, in a qualitative study, the researchers demonstrated that most people with diabetes selected diet and or exercise for their behaviour change (Seligman et al., 2007). The study that examined people with diabetes attending a Malaysian clinic, showed 94% of these individuals were following a healthy diet (Ambigapathy et al., 2003). On the other hand, R. M. Anderson, Fitzgerald, and Oh (1993) suggested that diabetes self-management, such as diet planning, requires a life-long commitment. This can only be sustained with genuine internalization of the purposes and the value of good diabetes self-care. Epidemiological data has consistently connected increased physical activity to reduced risk of all cause mortality and morbidity in individuals with T2DM. For instance, study has shown that people with T2DM who reported walking for at least two hours per week had a 39% lower all cause mortality rate than those who reported no walking (F. B. Hu, 2001). In diabetes health care, it is
43
recommended that people should exercise aerobically or do physical activities at least once every two days. This is because insulin sensitivity, which increases on every single bout of aerobic exercise, lasts 24-72 hours (Sigal, Kenny, Wasserman, Castaneda-Sceppa, & White, 2006). According to Ambigapathy et al. (2003), 64% of the total 100 respondents in their study did exercise at least three times per week for a minimum of 20 minutes each time. In addition, weight management also plays an important role in preventing the development of diabetes. Weight loss can be achieved effectively through interventions in nutrition and exercise (Foliaki & Pearce, 2003). Ninety percent of the respondents attending the Malaysian clinic in the study by Ambigapathy et al., reported that they practised regular weight monitoring, at least once every three months. Australian Diabetes Management in General Practice guidelines recommended self-monitoring of blood glucose for all people with T2DM (Diabetes Australia and RACGP, 2010). These guidelines favour diabetes selfmanagement as an appropriate and effective intervention. The aim of this monitoring is to lower blood glucose and to optimise glycaemic control to minimize risk for long-term diabetes complications and, subsequently, enhance QoL (Ciechanowski et al., 2001). V. Rose, Harris, Ho, and Jayasinghe (2009) reported that factors, such as patient self-efficacy and general practitioner (GP) communication in self-monitoring of blood glucose, were more likely to optimise diabetes self-management behaviours. People with diabetes, who had high selfefficacy (i.e., belief that they were able to manage their diabetes) and perceived their GPs to have good communication skills, showed high levels of blood
44
glucose testing (V. Rose et al., 2009). However, in the V. Rose et al. study, there was no effect of GP communication on general diet and exercise. A systematic review by Clark revealed that there are a number of barriers to diabetes self-management. Barriers identified included characteristics of people with diabetes, socio-environmental context, the disease itself, and interaction with diabetes care and education providers (Clark, 2008). In a study conducted by Cameron, Leventhal, and Leventhal (1993), participants were asked to create their own models as representations of their illness, which would give personal meaning to symptoms and experience of illness. Researchers have shown that illness representations are associated with self-management behaviours and treatment outcomes across a range of chronic diseases (Glasgow, Hampson, Strycker, & Ruggiero, 1997; Griva, Myers, & Newman, 2000; S. E. Hampson, Glasgow, & Strycker, 2000). Self management behaviour of people with diabetes is also likely to be affected by their social milieu (e.g., spouse, family, friends). For instance, support from spouses is often the most influential form of support for individuals with diabetes (Thompson & Pitts, 1992). Another study suggested that partners’ representations partially mediated the influence of representations of people with diabetes about their exercise and dietary behaviours (Searle, Norman, Thompson, & Vedhara, 2007). A study conducted by Weijman, Ros, Rutten, Schaufeli, Schabracq, and Winnubst. (2005a) to identify relationships between work experience, personal factors, and self-management in a diabetes working population, found that family and friends support as important for employees with T2DM. They would eat more frequently at regular times when they
45
experienced this support. These findings were in line with Toljamo and Hentinen (2001a) study, where they reported that support from family and friends was associated with adherence to self-care. Other factors that are important in diabetes control are motivation, selfesteem, and approach to disease management (MacLean & Lo, 1998). In a study conducted by MacLean and Lo (1998) positive self-esteem influenced the success of adherence to self-care behaviours (i.e., diet, blood sugar testing, and exercise) among 95 respondents, comprising 42 males and 53 females from New South Wales, Australia. They further explained that success in complying to wellness behaviours was a function of positive attitudes and an acceptance of the challenges of the illness, ability to utilise family support and have positive selfesteem. Failure to comply with these causes stress, chronic, and transient mental distress. However, in a study by Johnston-Brooks, Lewis, and Garg (2002) that examined people with T1DM in comparing self-esteem and self-efficacy, selfefficacy was a stronger predictor of all aspects of self-care than self-esteem. National Standards for Diabetes Self-Management Education (DSME; 2002), published by the ADA, indicated that diabetes self-management education is “the cornerstone of care for all individuals with diabetes who want to achieve successful health-related outcomes” (Mensing et al., 2002, p. S140). Research indicates that the educational background of T2DM people plays an important role in diabetes self-management. Those who have lower educational attainment usually have greater difficulty in understanding physicians’ explanations about diabetes and its management (Larme & Pugh, 2001). However, in Mahathevan’s study (2003) concerning diabetes mellitus among a rural population in Malaysia,
46
level of education was found to have no significant relationship with the level of practice regarding diabetes. In Mahathevan’s study, more than half the participants had either no schooling or just Primary school level education. However, Mahathevan did not report how regular of the self-practices were among his participants or the participants’ levels of self-practice. Therefore, whether all participants were undertaking useful self-practices or whether they just had poor self-practices is unknown. However, a systematic review on diabetes self-management conducted by Steed, Cooke, and Newman (2003) revealed that self-management interventions were more frequently associated with improvement than educational interventions. As for gender differences related to self-management practices, a study carried out in Pakistan showed overall exercise among the participants were found to be poor and only a few had good exercise practices (>30 minutes per day). However, they reported that females were less involved than men in practices regarding physical activity (Naeema et al., 2002). In Mexico, research showed that men were more likely than women to be in good control, showing closer compliance with dietary recommendations (Mercado & Vargas, 1989). In their study they found that all the male participants had their food prepared specially for them by another family member, but only 13% of the females had this type of support from family members, which may explain why men were more likely to be in good control. Age may also influence individuals’ awareness about their health condition and the urge to have regular medical check-ups. For example, a study executed among people with diabetes aged 35 years and above showed older
47
persons in the age group 55-65 years made more annual visits to their doctor compared to the middle and young age groups (Oladele & Barnett, 2006). Another survey conducted among Montana residents showed that the majority of the respondents who were screened for diabetes during the previous year comprised individuals aged 65 and older (Harwell, Smilie, McDowall, Helgerson, & Gohdes, 2000). Age may also associate with diabetes self-management among people with T2DM. Savoca, Miller, and Quandt (2004) found that people in good control were diagnosed at an older age and had superior coping skills, especially in relation to managing their diet. This can be due to the past experience learned by the older age people with T2DM, such as lessons learned from family experiences. The persistence of individuals with T2DM may also reflect their experience in facing challenges in the past (Savoca et al., 2004). Weijman et al. (2005a) reported that people with diabetes who displayed an avoidance coping style were less likely to perform self-management activities frequently. In their study, they also concluded that greater age and lower level of education were associated with reduced frequency of self-management. Weijman et al. also found that the burden of self-management was related with workload and seriousness of disease. In their study, workers with T2DM who had higher workloads were more likely to perceive injecting insulin as a burden than workers who had lower workloads. According to the literature (Heins, Arfken, Nord, Houston, & McGill, 1994; Toljamo & Hentinen, 2001b) it has been argued that having control in one’s work is crucial for the performance of selfmanagement activities.
48
Economic status or household income affects preventive care practices that are required to manage diabetes. According to the report from the Western Pacific Declaration on Diabetes in 2000, poverty has a negative impact on the level of practice of diabetes self-management among the general population (IDF, 2000). However, in a study conducted in a rural area of Malaysia with high levels of poverty and little education, Mahathevan (2003) showed that there was no significant difference in diabetes-related practice, according to household income. Mahathevan did not report what is the level of diabetes-related practice among the study population. Therefore, it can be possible that all his respondents generally did not have regular diabetes-related practice due the their background. People with diabetes need to receive structured self-management education in order to make choices about their diabetes self-management, such as medication taking, diet, physical activity, and blood glucose monitoring (Skinner et al., 2006). Although audits of diabetes education in the UK showed that most people received some form of “education” services when they were diagnosed with diabetes, there was a large variation between the content of services. A UK health report explained that most educational programs were unstructured, few programs were formally evaluated, and not many individuals who delivered the education were formally trained for this purpose (Audit Commission, 2000; Department of Health, 2003). However, some researchers have found that educational intervention programs increased participants’ knowledge of diabetes and self-care activities in the short term (S. A. Brown, 1990; Norris, Engelgau, & Venkat, 2001). Hence, education programs for people with T2DM should be properly designed in order to enhance their knowledge about diabetes self-
49
management, which should help those people to adhere more effectively to their self-management regimen. Overall, it would appear that many factors may influence the self-care or self-management behaviour of people with diabetes. Diabetes self-management is not simply a question of taking medication as prescribed, but it also involves substantial lifestyle modification for people who are diagnosed with T2DM. Selfmonitoring of blood glucose among people with T2DM plays an important role in diabetes care and it is also widely recommended. This self-monitoring helps people with diabetes to adjust their insulin dosage, diet, and exercise regimens, and also enables those people to detect and prevent hypoglycaemia (Kaeter, Ferrara, Darbinian, Ackerson, & Selby, 2000). The main components of the diabetes self-management regimen also include diet, physical activity, and self-monitoring of blood glucose (Cox & Gonder-Frederick, 1992; Grauw et al., 1999; Ruggiero et al., 1997; Toljamo & Hentinen, 2001b). Self-management in diabetes is the crucial part of diabetes treatment among people with T2DM and the importance of it should not be overlooked. This is because diabetes is a lifelong and progressive illness, in which complications increase and become severe if the illness is not in control. Managing diabetes through lifestyle factors can affect this progression, but selfmanagement can be challenging for people with T2DM.
50
The Relationship between Diabetes Knowledge, Attitudes, and SelfManagement Researchers have examined the relationships between knowledge, attitudes, and self-management related to T2DM using a range of questionnaire designs. The Diabetes Knowledge (DKN) scale developed by Dunn et al. (1984) and the ATT39 questionnaire developed by Dunn, Smartt, Beeney, and Turtle (1986) were adopted by Menard et al. (2007) to assess respectively, knowledge about, and attitudes to, diabetes. Menard et al. stated that no suitable measure of diabetes self-management was available; therefore, they developed a new selfmanagement scale based on various expert recommendations. Similarly, several researchers have developed their own questionnaires for assessing knowledge, attitudes, and self-management related to diabetes (e.g., Ambigapathy et al., 2003; Carbone et al., 2007; Naeema et al., 2002; Rafique et al., 2006). The term “practice” of diabetes was used in studies by Ambigapathy et al., Naeema et al., and Rafique et al., which assessed what is more frequently called selfmanagement in diabetes. The reliability and validity of the questionnaires used in each of these studies has not been adequately examined. Measures for which there appears to be consensus among researchers, concerning reliability and validity, are the DKN scales for assessing knowledge (Beeney et al., 1994), the Diabetes Integration Scale (ATT19) for assessing attitudes (Welch, Beeney, Dunn, & Smith, 1996), and Diabetes Self-Care Activities (SDSCA) subscales for assessing self-management (Toobert, Hampson, & Glasgow, 2000). There are some other validated questionnaires assessing diabetes knowledge, such as the Audit of Diabetes Knowledge (ADKnowl; Speight &
51
Bradley, 2001) and the Diabetes Knowledge Test (DKT; Fitzgerald et al., 1998), questionnaires assessing diabetes self-management, such as the Self-Care Inventory-Revised (SCI-R; Weinger, Butler, Welch, & Greca, 2005), and questionnaires assessing diabetes QoL, such as the Audit of Diabetes Dependent QoL (ADDQoL; Bradley, Todd, Gorton, Symonds, Martin, & Plowright, 1999). Although the ADKnowl is built on the DKN scale and claimed to be suitable for people with T1DM and T2DM, it consists of large number of items (104 items), which takes a longer time to complete by participants compared to the DKN scale. This could lead to boredom or fatigue which could affect the reliability and validity of responses. The DKT was developed in the USA. It is another measure that is designed to assess diabetes knowledge. The DKT also has some weaknesses that make it less suitable for the present research participants than the DKN. One of the weaknesses is that incorrect answers are similarly worded, but inaccurate versions of the correct responses. Although the DKT is designed to ‘catch out’ respondents who are guessing, it may confuse respondents who know the answer. The SCI-R is a newly developed questionnaire that assesses diabetes self-care. It includes only one item for exercise and one of the important areas of self-management, foot care, is not included. Weinger et al. acknowledged that the SCI-R may need further refinement to improve its content validity and clarify its factor structure. Another questionnaire that assesses diabetes QoL is the ADDQOL measure. In the ADDQoL, respondents are provided options to indicate the item is not applicable for them (e.g., aspects of life, such as sex life). Although this “inapplicable” option is considered by Bradley et al. to be essential to allow respondents to restrict their responses to items relevant to them, it could
52
create items without ratings and cause missing observations in the statistical analyses. Bradley et al. (1999) suggested rescoring the inapplicable responses as zero to retain all data, but this may create outliers with unwanted distortion of results. When there are many missing observations and outliers in the data, it would pose a serious challenge to Structural Equation Modelling analysis (Kline, 2005). The ADDQoL measure also lacks generalisability due to its individualized nature (Burroughs, Desikan, Waterman, Gilin, & McGill, 2004). Researchers have assessed the relationships between knowledge, attitudes, and self-management or practices related to diabetes in several studies (Ambigapathy et al., 2003; Mahathevan, 2003; Palaian et al., 2006). The relationships between these variables are discussed in the following sub-sections. The questionnaires used by researchers to measure diabetes knowledge, attitudes, and self-management may be different from one study to another, because different researchers used their preferred questionnaires to measure the same variables in various contexts with diverse populations. Regardless of the different questionnaires used by researchers in measuring relationships between knowledge, attitudes and self-management, the findings in the literature give a general idea about, and evidence of, the relationships between these three variables among people with diabetes, specifically T2DM. Knowledge and Attitudes Studies have revealed that the high prevalence and the increasing trend of diabetes are due in large part to lack of knowledge and poor attitudes towards diabetes (e.g., Ambigapathy et al., 2003; Kamel et al., 1999; Naeema et al., 2002). This may be because health-seeking behaviours are mainly determined by
53
individuals’ knowledge, attitude, beliefs and health care practice. Research has been conducted to improve education programs that aim to enhance diabetes knowledge among people with T2DM. It is common in diabetes education to presume that improvement in knowledge, attitudes, and skill will lead to more informative treatment advice and will be followed by more effective metabolic control (Dunn, Beeney, Hoskins, & Turtle, 1990). In a study conducted by Dunn et al. (1990), the impact of such a diabetes education programme was assessed by examining changes in knowledge and attitude. The diabetes knowledge (DKN scores) and attitude (ATT39 scores) were improved after a 2-day diabetes education programme over a 15-month period. The mean scores on the DKN scales were increased by 25% of one standard deviation. This improvement was consistent with the literature (Dunn, 1988). In their study conducted in a clinic in Malaysia, Ambigapathy et al. (2003), concluded that knowledge of and attitudes toward diabetes among people in their study were significantly correlated. One hundred people with DM who attended the clinic were interviewed using a questionnaire constructed by Ambigapathy et al.. The diabetes knowledge measure included questions about diabetes symptoms, complications, prevention, diet, and exercise. The attitudes of the participants were evaluated based on their willingness to undertake diabetes control activities, such as exercise, diet, weight control, and self-care. Ambigapathy et al. reported that there was a positive correlation between level of knowledge and attitude concerning diabetes, which showed that people with higher levels of knowledge had more positive attitudes toward living with their diabetes.
54
The way of transferring diabetes knowledge from health officers might have an influence on individuals’ attitudes toward diabetes and its treatment. When a person lacks diabetes knowledge, they may be less likely to develop a strong attitude toward diabetes and self-care. In a qualitative study, Dietrich (1996) conducted interviews among a group of women with T2DM. Dietrich found that the participants had low levels of diabetes knowledge. This category of women expected their physician to provide guidelines and advice to them on their diabetes management. They were more interested to know where and how knowledge had been transmitted to them, rather than what they learned. This may be due to their low educational level and the fact that they grew up in a time when physicians’ instructions were accepted as authority (Dietrich, 1996). Knowledge and Self-Management Self-management implies an intrapersonal understanding of the knowledge about diabetes control. Hernandez (1996) drew attention to the three-stage theory of “integration” for people with T1DM, in which it is proposed that people gain incremental knowledge and then interest in managing diabetes. Gradually, people who accept the integration process will become experts in their diabetes, although they may not adhere to the regimen prescribed by health care providers. This theory provides an understanding of how knowledge can directly or indirectly influence the adherence of people with diabetes in their daily self-management. Knowledge of diabetes can become a cornerstone in decision making on diet, exercise, blood glucose monitoring, use of medication, weight control, and foot care (Murata et al., 2003). In a study to determine the management behaviour of people with diabetes, Kamel et al. (1999) observed a linear
55
relationship between overall knowledge level and diabetes management. People with diabetes lacked knowledge and consequently had low levels of self-care practices. This is expected, as some health information of some kind may be necessary before personal health action is carried out. In their study, Kamel et al. reported that they observed a significant relationship between people’s knowledge about diabetic retinopathy and their attendance at eye screenings. People who scored higher in knowledge about diabetic retinopathy tended to have greater satisfaction about self-care related to their eyes. However, no significant relationship was observed between knowledge about foot problems and foot selfcare, knowledge about dietary regimens and compliance with the required diet regimen, and knowledge of medications and compliance with prescribed medicines. Norris et al. (2001) conducted a systematic review of the effectiveness of self-management training in T2DM. They reviewed 72 of the 84 published research articles on randomised controlled trials, in which all or most participants had T2DM. From the review, they found out that when follow-up was short (less than 6 months), the effect on outcomes was more obvious or demonstrated greater effectiveness. The effectiveness of interventions lasting less than 6 months was not just observed in terms of an improvement in diabetes knowledge, but also by improvements in the skills of self-monitoring blood glucose, and improvement of dietary habits, weight and physical activity levels, and glycaemic control.
56
Attitudes and Self-Management Researchers have proposed that individuals who have positive attitudes toward managing their diabetes will be more likely to adjust their self-care behaviour in order to control their blood glucose levels than those who have negative attitudes (Dunn, 1990; Lockington et al., 1988; Masaki et al., 1990). Likewise, Greene, Beaudin, and Bryan (1991) reported that improving attitudes can positively affect diabetes management. Norris et al. (2001) even suggested that interventions are needed to address attitudes and motivation, rather than just knowledge about diabetes in order to achieve behavioural change. Attitudes can be an important factor in motivating people to adhere to their diabetes selfmanagement. The relationship between attitudes and self-management has been found not to be significant in some studies. For example, a total of 46 people with T2DM enrolled in a study conducted by Palaian et al. (2006). Of these participants, 19 were assigned for diabetes counselling as the test group. At the end of the study, Palaian et al. found that knowledge scores were significantly higher in the counselling group than the control group, those who did not have diabetes counselling. However, they found no significant difference between the counselling group and control group in terms of their attitude and practice scores. It was also proposed that attitude and practice among people with diabetes were not correlated (Palaian et al., 2006). Ambigapathy et al. (2003), in their study in a clinic, Malaysia, showed no significant correlation between participants’ attitudes and self-management, which indicated that an increase in knowledge, can
57
increase attitudes, but this does not necessarily lead to a corresponding improvement in self-management. Some people with a family history of diabetes may face demotivation in practising healthy lifestyle to prevent T2DM. In a Canadian study, respondents with a family history of diabetes who realized there was a higher risk of having diabetes in their life were not practising protective behaviours, such as smoking cessation and exercise (McManus et al., 2006). McManus et al. (2006) argued that a negative attitude toward diabetes prevention might have caused people not to take active precautions. This claim needs to be further study. Family behaviour toward diabetes could also have an effect on people’s diabetes selfcare. Studies have shown that family attitudes can either support or challenge a person’s psychosocial adaptation to illness. Subsequently, this could affect their confidence, intent, and willingness to implement disease management strategies (L. Fisher et al., 2000; Trief, Himes, Orendorff, & Weinstock, 2001). J. Tang et al. (1999) reported that participants’ attitudes were associated with practice (diabetes self-management) from the results of their questionnaire survey. There were four attitude questions. A point was given to each question, which indicated a positive attitude on diet and medication control, and for body weight perception. Diabetes practice questions included areas of diet control, use of herbal or food remedies, and self-monitoring. J. Tang et al. categorised attitude into low and high score. An adequate score for attitude was defined as scoring >2/4. They found that attitude and practice scores were correlated. Those who scored higher in self-management were more likely to have adequate attitudes compared to those who scored lower on practices. Similarly, Fey-Yensan and
58
English (2003) concluded that negative attitudes toward T2DM were significantly associated with barriers to self-care and diet management. The terms beliefs and attitudes are widely used in social psychology research in the study of human behaviour. For instance, belief and behaviour were found to correlate significantly with attitude (Bruvold, 1973). According to Bruvold, attitude is defined as general affective response to an object. Beliefs are defined as statements about existing states of nature that the individual accepts as true or factual. Beliefs are regarded as positive if they suggest that the attitude object can cause desirable consequences. Ajzen and Fishbein (1977) argued that an individual’s attitude toward an object influences the overall pattern of their responses to the object. They also proposed that attitude is a sound predictor of action. Commitment to diabetes self-care requires positive attitude and behaviour changes (R. M. Anderson et al., 1990; De-Weerdt et al., 1990; Dunn, 1990; Lockington et al., 1988; Masaki et al., 1990). Thus, attitude is an important indicator in the study of diabetes self-management. In diabetes management, attitude has been identified as an important psychosocial variable to consider (Greene et al., 1991). Studies of psychological adjustments and behaviour of people with diabetes, have always involved the study of their belief and/or their attitude. In some studies, both terms have been used to describe the association with diabetes self-management or behaviour (Coates & Boore, 1996; Kurtz, 1990; Polonsky, Fisher, Guzman, Villa-Caballero, & Edelman, 2005; Wdowik, Kendall, Harris, & Auld, 2001). Several studies have shown that patients’ beliefs about treatment effectiveness for diabetes were associated with the degree of self-management
59
(Glasgow, Hampson, et al., 1997; Hampson, Glasgow, & Toobert, 1990; Skinner & Hampson, 2001). Skinner and Hampson (2001) suggested that clinicians, educators, and interventionists should consider adolescents’ beliefs about their diabetes and its treatment as components influencing self-care, emotional wellbeing, and glycaemic control. In their personal models of diabetes study, Hampson et al. (1990) found that personal-model dimensions (i.e., personal beliefs and emotions about the cause, symptoms, course, treatment, and consequences of their disease) increased the prediction of diet level and exercise. There are some health behaviour and health education theories, such as the Health Belief Model (Becker & Janz, 1985), the proponents of which have pointed out that attitudes and beliefs are major components of health behaviour. Strong attitudes are proposed to predict manifest behaviour (Ajzen, 2001). Ajzen also described how perceived quantity of attitude-relevant knowledge was stronger in middle adulthood than during early or late adulthood. Readiness to change in people’s attitudes declines from early to middle adulthood and then improves again in late adulthood. In their survey on attitude theory among research articles published between 1996 and 1999, Ajzen claimed that the ability of attitudes to predict behavioural intentions has become a major focus of theory and research in the past decade and, in general, attitudes are recognized as predicting social behaviour (Ajzen, 2001). Other theories such as the Theory of Planned Behaviour (TPB) and the Theory of Reasoned Action (TRA) were used as a framework in studies concerned with the prediction of behaviour from attitudinal variables (Ajzen, 1991, 2001). Therefore, understanding the self-care behaviour of people with diabetes and providing appropriate diabetes education
60
require some awareness of the attitudes toward the disease and diabetes care (R. M. Anderson, Nowacek, & Richards, 1988). In a study of people with T2DM, based on the theory of planned behaviour framework, Gatt and Sammut (2008) reported that the behavioural intentions of people with T2DM were strongest for medication compliance, which is the most performed self-care behaviour compared to physical activity and dietary adherence, which were the least performed behaviours. This also suggests that people with T2DM are more likely to perform the self-care activities that require the least effort and lifestyle change. Gatt and Sammut concluded that attitudes and perceived behavioural control are important predictors of intent to carry out self-care behaviour in people with T2DM, which is paralleled with the theory of planned behaviour. According to TPB, the formation of behavioural intentions, or how much effort an individual is willing to put into the action, is influenced by the attitudes towards a behaviour, perceived subjective norm, and perceived behavioural control (Ajzen, 1985). Therefore, understanding the self-care behaviour of people with diabetes in responding to their needs requires some knowledge of their attitudes toward the disease and diabetes care. Research indicates that the attitudes to T2DM of people with diabetes depend on the perceived difficulty of the behaviour changes they are required to make (R. M. Anderson et al., 1993). In a survey about fundamental behaviours carried out by R. M. Anderson et al. (1993), the results showed that people with diabetes differ in their performance of behaviours that are difficult in comparison to behaviours that are easier. For instance, self-care, such as foot care, involves a
61
series of distinct behaviours, which probably present a moderate level of difficulty compared to following a diabetes diet, exercising, and blood glucose testing (which are considered difficult), and carrying diabetic identification and sweets (which are considered easy). Anderson et al. reported that the attitude of people with diabetes who were adherent in regular foot inspection differed from those who had low levels of adherence. The adherers had more positive attitudes to having and coping with T2DM. This showed that attitudes are more relevant to adherence when the level of difficulty of self-care is considered. Attitudes toward diabetes also depend on the disease severity. Lange and Piette (2006) concluded that people with moderately serious diabetes strongly believed that treatment is very important in preventing complications. Thus, they endorsed medical advice. Diabetes is a major risk factor for stroke, myocardial infarction, and many other life-threatening complications. Because of this, it is crucial to create awareness concerning diabetes. Living with diabetes may cause a lot of stress for individuals. Experiencing diabetes-related stress may also affect people’s ability for self-care, which can impact metabolic control (Berlin, Rabideau, & Hains, 2012; Peyrot & McMurry, 1992). As people with diabetes are continuously facing challenges to cope with their problems, their ability to copeaffects their response to diabetes treatment regimen (Davidson, Boland, & Grey, 1997). Karlsen and Bru (2002) studied coping styles among people with T2DM and T1DM. They found that coping styles were independent predictors of dietary behaviour, but did not explain physical activity participation. This study illustrated that an active form of coping is expected to have a positive association with diet self-management.
62
Some people with T2DM prefer a traditional, hierarchical, and authoritarian approach to diabetes care, while others prefer an active approach of autonomy and self-direction (R. M. Anderson et al., 1990). Promoting adherence to diet or other self-care activities should be related to the needs and expectations of individuals with T2DM. A study conducted using the revised DAS (Diabetes Attitude Scale; R. M. Anderson et al., 1990) to explore the relationship between attitudes and self-reported adherence to recommended diabetes self-care (R M Anderson et al., 1993). This study confirmed that, in general, people with T2DM with high levels of adherence to various diabetes self-care recommendations had more positive attitudes toward diabetes. Although studies have employed various types of measure to address the level of attitudes and self-management among people with either all types of diabetes or specifically T2DM, research has shown that individuals’ attitudes can be an important indicator of their diabetes self-management practices. Therefore, it is crucial that researchers examine the level of attitudes and diabetes selfmanagement among people with T2DM using the specific measures that have been designed to address each variable. It will be noteworthy if the level of and the association between attitudes and self-management among people with T2DM can be understood in populations where these variables have not yet been addressed. Knowledge, Attitudes, and Self-Management The way diabetes knowledge is communicated may have an influence on attitudes toward diabetes and its treatment. When people know little about the disease, they are less likely to develop strong attitudes toward diabetes and self-
63
care (Dietrich, 1996). An individual’s knowledge, attitudes, belief and health practice can largely determine their health seeking behaviours. Some hospitals provide counselling for people with T2DM in order to improve their health condition. In an evaluation of the impact of counselling in people with diabetes in India, Palaian et al. (2006) found that there was an improvement in their knowledge, but not in their attitudes and practices. They also reported that there was no correlation between attitude and practice after a counselling program. Thus, it could not be confirmed whether improvement in knowledge about diabetes among people with diabetes resulted in higher levels of diabetes selfmanagement practices. However, very often people with diabetes will adjust their lifestyle within a framework that is also influenced by their cultural, sociodemographic, knowledge and resource base (Wong & Toh, 2009). Attitude can be considered the strength of belief a person has. A person can have an attitude (positive or negative) toward a phenomenon based on the overall evaluation of the person’s beliefs (Ajzen, 1991). Different cultural groups may have different beliefs and attitudes about their illness condition. For example, in a study by Lange and Piette (2006) in United States on Hispanic/Latino participants with diabetes, those who showed high levels of fatalism reported the belief that diabetes is a non-serious illness. Lange and Piette proposed that this was due to their belief in fate, which led them to take a passive role in their diabetes management. Research has demonstrated that strong belief regarding the effectiveness of treatment for diabetes was associated with positive self-management behaviours. In their study of Malaysian people with diabetes, Ambigapathy et al. (2003) concluded that attitudes and self-management were
64
not correlated, but they found a significant positive correlation between knowledge and attitudes. The success of self-management requires a combination of factors, such as personal knowledge, attitudes, skill, support from others and resources in the community (E. Fisher et al., 1982). People in rural areas are less likely to have a combination of these factors which are necessary for good diabetes management (Irvine, 1989). Thus, investigating the factors affecting self-management, such as knowledge and attitudes, should enhance understanding of their differences and their impact. The relationship between knowledge and attitudes, the relationship between knowledge and self-management, and the relationship between attitudes and self-management have been observed in separate studies. Although there are a few studies that examined levels of knowledge, attitudes, and self-management (e.g., Ambigapathy et al., 2003; Badruddin, Basit, Hydrie, & Hakeem, 2002; Mahathevan, 2003; Naeema et al., 2002), only the relationships between pairs of variables were tested, either knowledge and attitudes, knowledge and selfmanagement, or attitudes and self-management. It is crucial to understand how these three variables associate with one another in one model. Correlation or regression analyses employed by previous research could not address these relationships. However, advance statistical analysis, such as path analysis is able to address the relationships between these three variables together by having more than two hypotheses tested in one model. It would be interesting to examine the relationship between knowledge, attitudes, and self-management all together, which is lacking in the current literature.
65
Diabetes Quality of Life Diabetes is a major progressive and life-threatening disease with many complications. It is likely that people with diabetes will experience lower levels of QoL as the disease progresses. This will affect their motivation in trying to maintain their health. Thus, it is important to know how key variables, such as knowledge, attitudes, and self-management, affect QoL among people with diabetes, especially with T2DM. A number of researchers from different countries have employed measures to assess QoL in people with diabetes, including the Medical Outcomes Study Short Form 36 (MOS SF-36; Ware & Sherbourne, 1992; Wu, Tomar, Tolios, Fryback, & Klein, 1998), the Medical Outcomes Short Form 20 (MOS SF-20; Hanninen, Takala, & KeinanenKiuKaanniemi, 1998), the Quality of Well-Being Index (QWB; Wu et al., 1998), and the Diabetes Quality of Life questionnaire (DQoL; Menard et al., 2007). The SF-36 and the SF-20 comprise a number of questions related to specific tasks (G. C. Brown et al., 2000). Redelmeier and Detsky (1995) suggested that these tests potentially miss certain confounding variables, which contribute to measuring QoL, such as the wealth of the individual, spousal or family support, and specific psychosocial relationships. According to Jacobson, Groot, and Samson (1994), the QWB is designed for comparison across studies and disease groups, whereas the DQoL is sensitive to lifestyle issues and contains special questions in measuring satisfaction and impact. According to a worldwide research group organized by the WHO, QoL can be defined as individuals’ perceptions of their position in life in the context of culture and value systems in which they live, and in relation to their goals,
66
expectations, standards, and concerns (The The World Health Organization Quality of Life [WHOQoL] Group, 1993, 1995). This definition reflects that QoL is subjective and directly related to internal psychological and physiological mechanism that results in satisfaction with life (Hornquist, 1982). Furthermore, in a study of QoL in different cultural settings, Skevington, Sartorius, Amir and The WHOQoL Group (2004), explained that the measurement tool should be simple, easily used, and demonstrably applicable in different cultural settings. They suggested that such measures would be useful in multinational clinical trials, epidemiological studies, and human health and satisfaction research. Race and ethnic background might be expected to affect the expression of QoL. This is because individuals’ culture and value system intrinsically influence their perceptions of their life position (Skevington et al., 2004). At present, general measures of QoL are not sensitive to capture important gains in health outcomes due to either new therapeutic interventions or programs in diabetes (Testa, Somonson, & Turner, 1998). Therefore, it is important to have a reliable and valid questionnaire to assess QoL in diabetes, which is suitable for a variety of diverse populations. Data had suggested that people with diabetes experience a decrease in their QoL compared to healthy individuals (Wändell, Brorsson, & Åberg, 1997). Most studies also report that QoL among people with diabetes is worse than people in the general population (Rubin & Peyrot, 1999). A number of studies have shown the association between diabetes and QoL (e.g., G. C. Brown et al., 2000; Goldney & Phillips, 2004; Hart et al., 2003; Menard et al., 2007). Goldney and Phillips reported that severe depression affected QoL in people with diabetes.
67
They concluded that the direct effect of depression on QoL is greater than the direct effect of diabetes on QoL. G. C. Brown et al. (2000) found that only T2DM had a large negative effect on QoL, whereas having T1DM did not affect QoL. Pouwer et al. (2001) proposed that there is indirect support for the assumption that monitoring psychological well-being in diabetes care could improve clinical outcomes. For instance, if prevalence of depression is increased among people with diabetes, it could have an adverse effect on individuals’ QoL, affect treatment adherence, and increase health care costs and risk for diabetes complication (R. J. Anderson et al., 2001). Pouwer et al. investigated whether diabetes care of outpatients can be improved by adding a monitoring procedure for psychological well-being to standard care. In their study, they found that psychological treatment for people with diabetes may help to improve mood of individuals with diabetes. Thus, monitoring psychological health in people with T2DM may improve reported clinical outcomes. Pouwer et al. proposed that general well-being and positive mental health should be targeted as part of the care of people with T2DM. In a United Kingdom study, the UK Prospective Study Group (1999) reported that T2DM had a negative impact on QoL, if the people with diabetes presented with long-term complications or with hyperglycaemic complaints (Hart et al., 2003). Authors of several studies found that older women with T1DM and without a partner had lower QoL than those with a partner (e.g., Aalto, Uutela, & Aro, 1997; Bott, Muhlhauser, Overmann, & Berger, 1998; Hart et al., 2003). In terms of gender, Glasgow, Ruggiero, et al. (1997) reported that men had a higher QoL than women in a survey that employed the well-being questionnaire SF-20.
68
This may be due to the observation that men and women differ in the way they cope with problems. Moreover, the prevalence of reported depression is generally higher in women (Kornstein et al., 2000; Piccinellie & Wilkinson, 2000) and women reported twice as much depressive symptoms as men in a survey conducted among people with T1DM (Enzlin, Mathieu, & Demyttenaere, 2002). These variations in reported depression between genders may contribute to the differences in QoL reported among different genders with T2DM. Some personal lifestyle factors are associated with improved health in terms of QoL among people with diabetes. The combination of reduced fat and sugar in the diet and increased exercise have not only been shown to improve glycosylated haemoglobin measures, which indicate positive control of blood glucose levels among people with diabetes, but these lifestyle changes also improve general QoL significantly (Kaplan, Hartwell, Wilson, & Wallace, 1987). Examining a group of primary care patients with T2DM, Weinberger et al. (1995) revealed that diabetes education was associated with physical functioning. People with diabetes who were satisfied with their education scored higher in their physical functioning. Weinberger et al. suggested that diabetes education motivated people with diabetes to do more exercise. R. M. Anderson et al. (1993) also suggested that people with diabetes who felt satisfied in their effort at selfcare especially in exercise and diet, were less likely to feel that the disease had a negative impact on their lives. According to a study conducted by Wredling et al. (1995), however, neither blood glucose self-monitoring nor participation in educational courses was associated with higher health-related QoL.
69
Lifestyle changes in people with T2DM may influence their QoL. Some external factors, such as a personal relationship with their physician may also affect QoL in people with diabetes. Hanninen et al. (1998) conducted a study to identify the factors in diabetes care that were associated with health-related QoL. Their results showed that people who had been treated by the same GP for at least two years were more likely to have better mental health and less pain. Therefore, they felt healthier compared to those who consulted with different doctors during the same period. Hänninen, Takala, and Keinänen-Kiukaanniemi (2001) concluded that continuity of relationship between physician and people with T2DM seemed to be an important factor that promoted good health related to QoL. There are some other factors that influence the QoL of people with T2DM. For example, researchers have reported that people treated with insulin experienced greater impact on diabetes-specific QoL compared to people not treated with insulin (Bradley et al., 1999; Davis, Clifford, & Davis, 2001). In the Diabetes Attitudes, Wishes, and Needs (DAWN) program, which is a programme designed to improve diabetes care outcomes, Skovlund and Peyrot (2005) reported that people with a poor psychological reaction to diabetes had poorer QoL and poorer health. A community survey carried out in Australia for T1DM and T2DM people showed that about 60% of the respondents who were diagnosed with diabetes were emotionally distressed, and 23% of them had wanted to have more emotional support (Beeney, Bakry, & Dunn, 1996). Diabetes is considered one of the most psychologically and behaviourally demanding of the chronic illnesses (Cox & Gonder-Frederick, 1992). Diabetes
70
does not just affect socioeconomic aspects in countries, but it also causes loss in personal income and productivity due to early deaths, early retirements because of stroke, heart attacks, amputation, chronic renal failure and blindness (Mustaffa, 2002). The psychological impact of diabetes on the individual and the affected family cannot be graded, but must also be taken into consideration (Mustaffa, 2002). Furthermore, cultural aspects of different populations can influence illness behaviour because individuals learn “approved” ways of being ill in their society as a normative experience (Kleinman, Eisenberg, & Good, 2006). For example, people from different cultures can show variations in terms of how a disorder is defined and coped with. Relationship of Diabetes Knowledge, Attitudes, Self-Management, and Quality of Life Research examining the links between knowledge, attitudes, selfmanagement, and QoL is scarce. The possible links between these four variables are presented in Figure 2.1 Researchers have mostly conducted cross-sectional, correlation studies between pairs of variables in Figure 2.1, in which they have not tested the path relationship between these variables in one model. Attitudes
Self-management
QOL
Knowledge Figure 2.1. Model of the possible causal links between diabetes knowledge, attitudes, self-management, and QoL.
71
A Canadian prospective study on QoL in people with T2DM revealed that diabetes self-management is one of the important keys to improve the quality of diabetes control and, thus, enhance the QoL of people with diabetes (Menard et al., 2007). Menard et al. also stated that an improvement in the attitudes of people with diabetes positively influenced QoL. In a study conducted in a rural clinic in Malaysia, Ambigapathy et al. (2003) found that attitude score increased with an increase in knowledge scores. Therefore, it is important for research to examine the links between knowledge, attitudes, and self-management, and the impact of observed relationships on QoL in T2DM. Faulkner and Chang (2007) reported that basic demographic factors like age, sex, socio-cultural orientation, health status, family systems, and individuals’ self-care involvement may ultimately affect their health outcomes, which may also influence the perception of personal QoL. According to Glasgow, Ruggiero, et al. (1997), self-reported exercise was associated with QoL after demographic and medical variables were controlled in a regression model. In their study, selfmanagement behaviours, which include diet, physical activity and exercise, and glucose self-testing, were assessed using a composite score. Items for the scale were drawn from the Summary of Diabetes Self-Care by Toobert and Glasgow (1994). The QoL was assessed using SF-20, which has three dimensions (physical functioning, social functioning, and mental health), each with a score range of 0-100, with higher scores reflect a higher QoL. Glasgow, Ruggiero, et al. (1997) revealed that only physical activity and exercise were significant predictors of all three QoL dimensions, physical functioning, social functioning, and mental health.
72
In a study on QoL of 36 people with T2DM under intensive multitherapy, Menard et al. (2007) concluded that improvement in diabetes selfmanagement certainly improved the quality of diabetes control among participants. In a study conducted in Thailand to determine the effects of a diabetes self-management program, Wattana, Srisuphan, Pothiban, and Upchurch (2007) found that QoL among people with T2DM was increased by selfmanagement. The self-management program organised by Wattana et al. included giving a small group diabetes education class, discussion classes, and home visit sessions to the participants with T2DM. Results showed that the self-management program had not only improved participants’ QoL, but also decreased the risk of having coronary heart disease and promoted more effective glycemic control. In addition, Menard et al. illustrated that a change in QoL, assessed by DQoL, was associated with more positive attitudes, assessed by ATT34 [Menard et al. probably meant ATT39] (Dunn et al., 1986). This confirmed that attitudes play an important role in QoL in people with T2DM. Therefore, people with T2DM who had more positive attitudes and more diligent adherence to diabetes selfmanagement taught by their health care providers, were more likely to report higher levels of QoL. Badruddin et al. (2002) provided an argument for the links between knowledge, self-management, and QoL based on diet. Badruddin et al. proposed that knowledge of diet is essential for people with diabetes, so that they understand how food can affect diabetic control, which should encourage them to follow good self-management practices related to diet. Knowledge of dietary principles may improve self-management and, thus, enhance QoL among people
73
with diabetes. Steed et al. (2003) revealed that psychosocial outcomes of diabetes self-management supported positive outcomes from dieting. By using the SF-20 quality of life survey form in their study, Weijman et al. (2005b) reported that participants with T2DM who followed dietary guidelines had fewer diabetic symtoms, were less fatigued, less depressed, and had higher QoL. Various researchers had reported the effects of self-management on QoL among people with diabetes. For instance, in a meta-analysis, Cochran and Conn (2008) reported that people with diabetes experience increased QoL after receiving interventions that were designed to improve diabetes self-management behaviours. The diabetes self-management intervention examined by Cochran and Conn was from 20 research articles that included aspect of diet, physical activity, physical education and training, health education, education program in self-management training. Smith and McFall (2005) found that dieting was unrelated to improved QoL, but they reported that exercise was associated with improved QoL. Glasgow, Ruggiero et al. (1997) indicated that lack of physical activity was related to lower QoL in adults with diabetes. Researchers have also demonstrated that a modest increase in physical activity, such as walking, can bring beneficial effects not just to healthy people, but also to those with chronic diseases (Pate et al., 1995; Tapp et al., 2006). Diabetes is associated with lower QoL which affects physical functioning, sleep, sexual functioning, health perception, daily activity, and causes pain and depressive symptoms (Egede, Zheng, & Simpson, 2002; Rubin & Peyrot, 1999; Smith, 2004; Valdmanis & Smith, 2001; Wandell, Brorsson, & Berg, 1992). In a report on National Standards of Diabetes Self-Management Education (DSME),
74
the ADA acknowledged that diabetes education is effective for improving clinical outcomes and QoL at least in the short term (Funnell et al., 2009). However, in a systematic review for the Cochrane Collaboration, Deakin, McShane, Cade, and Williams (2005) reported group-based self-management strategies for people with T2DM improved self-efficacy, self-management, treatment satisfaction, and QoL in long-term follow-up (more than 12 months). In a review of published articles on diabetes self-management education, Clark (2008) reported that knowledge and skills are necessary, but not sufficient to ensure good diabetes control in the long term. According to this report, there was no concrete evidence that resolves the question whether diabetes education is effective in promoting self-management to prevent diabetic morbidity and mortality or in improving QoL of people with diabetes (Clark, 2008). Misra and Lager (2008) developed a path model, which showed that self-care behaviour of people with T2DM was associated with higher QoL. In this study, Misra and Lager also reported that knowledge was related to self-care behaviour, but was not associated with QoL. Knowledge of diabetes and its complications increased adherence behaviours (Coates & Boore, 1996; Speight & Bradley, 2001; van den Arend, Stolk, Rutten, & Schrijvers, 2000) and compliance with physician recommendations and medication procedures (Gazmararian, Williams, Peel, & Baker, 2003). Evidence indicates that self-management programmes have positive effects on attitudes, self-management behaviour, glycaemic control, and overall QoL in people with T2DM (Bastiaens et al., 2009). Research has shown that group-based training for self-management strategies in people with T2DM is
75
effective by improving blood glucose, diabetes knowledge, systolic blood pressure levels, body weight, and in reducing the need for diabetes medication (Deakin et al., 2005). Improved diabetes self-management is critical to achieving metabolic control and reducing diabetes complications, subsequently leading to a better QoL (Rubin & Peyrot, 1999; Toljamo & Hentinen, 2001a, 2001b). The ADA (2003) also stated that adherence to self-management improves QoL. It involves complicated lifestyle changes, as many people with T2DM struggle with adherence behaviours (Ruggiero et al., 1997; Toljamo & Hentinen, 2001a, 2001b). For instance, Bonds et al. (2004) indicated that perceived difficulty with adherence to self-management behaviour is connected to poor blood glucose control and QoL. Successful diabetes control depends upon the extent of people’s skill with T2DM self-care, which can influence QoL (Gilden, Casia, Hendryx, & Singh, 1990; Luker & Caress, 1989). Usually, when people are diagnosed with diabetes, they may experience emotions, such as fear and anger, which affect their behaviour (Masaki et al., 1990). If people’s feelings and beliefs about their diabetes can be addressed, more effective self-care leading to higher levels of QoL can be delivered to people with T2DM. People who feel they are successful in their diabetes care on a regular basic may be less likely to feel that the disease has a negative impact on their lives (R. M. Anderson et al., 1993). According to DAV (2008), more than 60% of the Australian population are overweight or obese and being overweight is one of the major factors contributing to the increase of T2DM. Thus, efforts to lose weight are common among people with diabetes. In a health and retirement survey in the USA,
76
Nothwehr and Stump (2000) reported that about 58% of participants with diabetes were trying to lose weight, 80% were on a special diet, and 64% reported exercising. However, whether weight loss through dieting or exercise affects QoL of people with diabetes still remains unclear. In the general population, although there is a positive association between physical activity and QoL (McAuley & Katula, 1995; Rejeski & Mihalko, 2001), dieting is associated with lower QoL (Brownell, 1991; French & Jeffery, 1994). In weight loss intervention studies, researchers often find an increase in QoL is associated with weight loss (Kaukua, Pekkarinen, Sane, & Mustajoki, 2002; Rejeski et al., 2002). The demand of lifestyle changes to achieve weight control to manage diabetes may contribute to lower QoL (Smith & McFall, 2005). Smith and McFall examined the impacts of diet and exercise on QoL among people with diabetes. Their results showed that exercise was associated with improved QoL among people with diabetes, while diet was unrelated. Smith and McFall added that participants with diabetes who reported exercising to lose weight had QoL that was as high as or even higher than those without diabetes who did not exercise. In conclusion, the prevalence of, and QoL associated with, diabetes has been widely studied and knowledge, attitudes, and self-management have been shown to relate to each other and to QoL in particular populations. It is important to understand the various factors that may impact on the QoL of people with T2DM. The relationship between diabetes knowledge, attitudes, self-management and their impact on QoL still remains a huge area to be explored. Furthermore, there are limited published studies examining all the relationships together, with most research to date examining the relationships between pairs of variables only.
77
In an intensive review of the literature, I found no studies examining the path relationships between knowledge, attitudes, self-management, and QoL in one model. Although research has been conducted all over the world on all these key variables, there has been no systematic study of variations in the relationships between these key variables in different cultural settings. Cross-Cultural Comparison between Australia and Malaysia The prevalence of T2DM is estimated to rise over the next few decades globally and it causes a major public health burden of diabetes across the world (Wild, Roglic, Green, & King, 2004). Diabetes illness is a health, social, and economic burden for individuals with diabetes, and for their families and the community. It is also associated with various disease complications, so it impacts on the QoL and life expectancy for people diagnosed with diabetes. Diabetes mellitus has become one of the top ten leading causes of death in Malaysia since 2002 (WHO, 2006). The latest statistics released by DAV (2011) stated that diabetes is the sixth leading cause of death in Australia. Australia and Malaysia have approximately equal size of total population with 22 million people living in Australia (ABS, 2011) and 28 million people in Malaysia (Department of Statistics, 2011). Malaysia was reported to have approximately 1.2 million people with diabetes (Zanariah et al., 2009), which is higher than Australia with 898,800 people reported to have diabetes (AIHW, 2011a). However, DAV has reported that an estimated total of 1.7 million Australians have diabetes but 50% are undiagnosed (DAV, 2011). Diabetes has become the fastest growing chronic disease and one of the leading causes of death in Australia and Malaysia.
78
There is an interaction between environmental and ethnic-related factors, which influences changes to the risk burden of diabetes (Abate & Chandalia, 2003). Various studies have shown that the prevalence of diabetes is influenced by environmental factors, such as urbanization. For instance, the prevalence of diabetes was found to be about 4% in a small rural area of Japan (Toyota, Kudo, Goto, Taya, & Komatzu, 1976). However, another survey showed that the prevalence of diabetes among Japanese people living in Seattle, Washington reached over 21% (Fujimoto et al., 1987). Similar patterns have been observed for people of ethnic Chinese backgrounds, living in Hong Kong, Singapore, and Taiwan, who had higher prevalence of diabetes (Chou, Chen, & Hsiao, 1992; Cockram et al., 1993; Thai et al., 1987) compared to ethnic Chinese people who lived in Beijing (Chi, 1983). Although these studies were all done in the 1980s and 1990s, they show an interesting trend that prevalence of diabetes has dramatically increased worldwide, especially in rapidly developing countries. The same trend found in those studies might not be true today because of the rapid increase in prevalence of diabetes in Chinese cities, as China has modernised more quickly than almost any country in recent years. One reason that this difference may have occurred is variations in lifestyle factors that relate to diabetes. For example, diets that are high in animal fats and carbohydrates are more commonly seen in ‘westernized’ societies, such as the USA. This kind of dietary regimen has been associated with diabetes partly through the development of obesity (F. B. Hu, Van Dam, & Liu, 2001; Meyer, Kushi, Jacobs, & Folsom, 2001). Besides, lower levels of physical activity have been observed to be
79
associated with the “urbanization” and “westernization” process, increasing the prevalence and the risk of diabetes (Abate & Chandalia, 2003). One factor that has been widely cited as distinguishing between cultures in western countries and many Asian countries is how strongly individualist or collectivist they are (Hofstede, 2001). Most western countries have individualist cultures in which people think first about themselves. Many Asian countries have collectivist cultures in which personal interests are subservient to the needs of the community. This creates cultural differences that can indirectly influence people’s self-management, as well as how they interact with health professionals (Tripp-Reimer, Choi, Kelley, & Enslein, 2001). Individualistic cultures, such as Australia, which are associated with independent ideas of self, prioritize personal goals and uphold individual attitudes as key determinants of behaviour including health-related behaviour (Matsumoto, Yoo, & Fontaine, 2008). Therefore, privacy and assertiveness are highly valued in this culture. Conversely, collectivistic cultures, such as Malaysia, are associated with interdependent selves and group-focused goals. They value the group’s or community’s norms as primary determinants of behaviour, and they value group harmony. Therefore, I argue that these distinctive characteristics that are influenced by one’s culture may affect how people act on their illnesses and value their life as an individual with T2DM. Culture can also strongly influence the illness experience because it is an important part of social systems that influence the meaning and behaviour of the population. Thus, cultural factors influence perception, explanation, and valuation of the discomforting experience (Kleinman, Eisenberg, & Good, 2006).
80
For diabetes health, an important question is whether existing diabetes knowledge and practice, developed largely in the West, is directly applicable to nonWestern, and particularly developing countries, and how knowledge and practice impact on the QoL of people with T2DM in different cultural settings. Crosscultural research on diabetes psychological adjustment, self-management, and QoL at present is very scarce. Cross-cultural research on diabetes health is valuable as it highlights differences related to culture. It is needed to enhance understanding of how people from different cultures perceive their diabetes illness, and act upon the illness. Research has shown that cultural beliefs can have a major impact on people who are sick and the practitioners who treat them in terms of explanations of illness, goals for clinical management, and evaluation of therapeutic efficacy (Kleinman et al., 2006). Kleinman et al. (2006) stated that culture beliefs can affect how health problems are communicated, the manner in which people present their symptoms, when and to whom people go for care, and how those people evaluate that care. When it comes to defining a disorder and how to cope with it, cross-cultural and historical variations can be important. These variations can be equally great or different across ethnicity and culture. Doctors in different countries may have dissimilar explanations of disease and activities during the treatment of their patients are culture specific. In addition, diabetes treatment, such as insulin therapy, can have a significant impact on the life of people with T2DM. Insulin therapy usually occurs against a background of many years of habitual diabetes self-management that may have to be changed to improve the health condition of people with
81
T2DM. This can be a reminder of the progressive nature of the condition that requires more complex and invasive treatment. Thus, the level of impact of insulin therapy on people with T2DM can vary in different cultural contexts. Some studies have reported that insulin therapy was not associated with QoL (e.g., UKPDS Group, 1999), but other studies have shown apparently discordant results, with the effects of insulin on well-being or QoL ranging from beneficial (e.g., Chow, Tsang, Sorensen, & Cockram, 1995; Pibernik-Okanovic, Szabo, & Metelko, 1998) to deleterious (e.g., Goddijn et al., 1999; Hanninen et al., 1998). One reason why the results of studies undertaken in different cultures are not consistent may be that the research methods varied between studies. Therefore, it is important to undertake cross-cultural studies in which the same methods are employed in different cultures, allowing direct comparison. Summary of Literature In summary, researchers have identified separate bivariate relationships between diabetes knowledge, attitudes, and self-management. Each of these variables has a significant impact on health and QoL among people with T2DM. Diabetes knowledge can play an important role in influencing people’s attitudes to having diabetes and/or improving self-management among people with diabetes. In general, there is substantial information about diabetes available in most of the hospitals and in diabetes organisations. Unfortunately, not everybody with T2DM obtains that information to help them improve their knowledge about diabetes and self-management. Moreover, researchers have also found that knowledge about diabetes is related to attitudes. A positive attitude is helpful in enhancing motivation of people with T2DM to conduct their diabetes self-
82
management more regularly. Diabetes knowledge, attitudes, and selfmanagement have been shown to correlate with each other. Research on these three variables is restricted by looking at bivariate relationships only, which are knowledge and attitudes, knowledge and self-management, or attitudes and selfmanagement. No study has examined these three components together in one model which would allow information about how these variables influence each other; in particular examining their relative influence. Diabetes QoL remains an important issue for people with T2DM. Numerous interventions and programs conducted by researchers, health care providers, or health organisations aim to improve people’s QoL while living with diabetes. T2DM is a chronic disease where people are required to self-manage and to be largely responsible, for their own health condition. This requires positive attitudes among people with T2DM in handling their health condition, which may influence their self-management practices and also their QoL. In addition, knowledge about diabetes can play an important role in people’s attitudes and self-management. No studies were identified that examined the relationships between diabetes knowledge, attitudes, self-management, and QoL among people with T2DM in one model. Therefore, this remains an important area for future quantitative research. Diabetes is a worldwide epidemic and it is well-established in developed countries and increasing very quickly in the newly developing nations. Thus, it is of value to compare the key issues related to diabetes knowledge, attitudes, selfmanagement and QoL in both types of culture. Little research has explored the differences between two cultures in diabetes knowledge, attitudes, self-
83
management, and QoL among people with T2DM. Diabetes has remained the most important health problem faced by societies in many parts of the world. Moreover, variations in culture between countries may have different influence on the behaviours of people with T2DM. Thus, comparing a culture that reflects highly developed, western individualist culture and another that represents rapidly-developing Asian collectivist culture will provide further insights about the role of culture in the management of the global health problems associated with the rise of T2DM. The Present Thesis Based on the research examined in this literature review, it is clear that knowledge about diabetes, attitudes, self-management, and QoL among people with T2DM are key variables. It is also demonstrated by research that bivariate relationships exist between these key variables: knowledge to attitudes, knowledge to self-management, attitudes to self-management, self-management to QoL. However, all the research has examined bivariate relationships and more can be explored and learned by examining all the relationships together in one model, which can be achieved by using a Structural Equation Modelling (SEM) approach. Some relationships do not seem to have been studied at all, based on the literature identified in several thorough searches. These include the relationships of attitudes to QoL and knowledge to QoL. In addition, researchers have not compared different causal paths from the same starting variable to the same outcome variable, which is another valuable aspect of SEM. For example, it is of interest to determine whether the direct causal relationship between
84
knowledge and QoL is stronger than the indirect causal relationship between knowledge and QoL through self-management. Diabetes is a global illness which affects people in developed and developing countries. Australia is a well-developed country with high prevalence of diabetes. On the other hand, in many developing countries, such as Malaysia, the prevalence of diabetes has rapidly increased due to urbanization, which is associated with increasingly sedentary lifestyles among the people, as well as change of diet to more energy-dense foods. Thus, it is important to understand the relationship between these four variables in one model separately for both countries. It is also important to know whether the way these key variables operate is different between developing and developed cultures. This thesis will add new information to the current literature to increase understanding of the cultural differences between two distinct cultural groups in terms of their diabetes knowledge, attitudes, self-management, and QoL. In the present thesis, in Chapter 3, I discuss the development of the model that relates the four key variables and related demographic variables. In Chapter 4, the methods used to collect data in Australia and Malaysia are described. The next three chapters present results and discussion. In Chapter 5, I present the results of Phase 1, the cross-sectional study using Australia-based participants to examine the relationship between diabetes knowledge, attitudes to T2DM, selfmanagement of T2DM, and QoL using an SEM approach. Then, in Chapter 6, I present Phase 2, in which I examined the relationship of the same variables for the Malaysia-based participants, using the same SEM approach. For Phase 1 and Phase 2, extraneous variables were also considered during the analyses. Chapter 7
85
involves comparison of the key diabetes variables between two different cultural groups, namely Australia-based and Malaysia-based samples, in terms of their levels of diabetes knowledge, attitudes to T2DM, self-management of T2DM, and QoL. The present thesis was designed to provide insight into these current gaps in research knowledge. Thus the primary aims of the present thesis are: (1) to examine the relationships between all these four variables, namely diabetes knowledge, attitudes to T2DM, self-management of T2DM, and QoL among people with T2DM using an SEM approach; (2) to identify the strength of the path relationships between diabetes knowledge, attitudes to T2DM, self-management of T2DM, and QoL among people with T2DM using an SEM approach, in order to examine whether causal relationships are universal or whether there are differences between cultures; In order to achieve the primary aims, the specific aims were derived and listed as below: (3) to conduct a separate analysis for Australia-based and Malaysia-based samples examining the relationships between all these four variables using an SEM approach; (4) to compare the differences between Australia-based and Malaysiabased samples in terms of their levels of diabetes knowledge, attitudes to T2DM, self-management of T2DM, and QoL.
86
CHAPTER 3 CONCEPTUAL FRAMEWORK Based on the literature review presented in Chapter 2, in this chapter, I present the conceptual framework for the research conducted in this thesis and then derive specific hypotheses and research questions. In the first section, I discuss the proposed conceptual framework for this study based on previous research that has considered self-management of T2DM, including the influence of knowledge about T2DM, and attitudes to T2DM on self-management, and studies that have examined the impact of self-management on QoL. The research questions and hypotheses of the thesis that are generated from the conceptual framework and that will be tested using statistical analyses are stated in the second section. The research questions and hypotheses are expressed in the form of relationships supported by empirical findings from the literature review. Conceptual Framework The conceptual framework presented in this section of the study (see Figure 3.1) is based on the synthesis of the literature as presented in Chapter 2. Based on the literature, the major dimensions addressed in the conceptual framework are the relationships between diabetes knowledge, attitudes, selfmanagement, and QoL. According to the literature review in Chapter 2, research has shown a positive relationship between diabetes knowledge and attitudes (Ambigapathy et al., 2003; Lancaster et al., 2000; Tessaro et al., 2005). For instance, diabetes knowledge and attitudes toward diabetes among people in the study conducted by Ambigapathy et al. were significantly correlated; people with higher levels of
87
knowledge had more positive attitudes toward living with their diabetes. A higher level of diabetes knowledge has also been reported to relate to more positive selfmanagement related to their condition among people with diabetes (Kamel et al., 1999). Increases in knowledge about diabetes were found to be related to greater interest in managing diabetes (Hernandez, 1996). Conversely, lack of diabetes knowledge has been found to hinder preventive health behaviours in the general population (Tessaro et al., 2005). Researchers have also reported that people with diabetes who had positive attitudes were more likely to do well in their selfmanagement to control their diabetes (R. M. Anderson et al., 1993; Lange & Piette, 2006; J. Tang et al., 1999). Furthermore, in accord with Ajzen and Fishbein’s (1977) theory on the linkage between knowledge, attitudes, and behaviour, Grady, Entin, Entin, Brunye (2011) predicted that changes in knowledge affect attitudes and that attitudes are direct predictors of long-term behaviour. Their hypothesis of this prediction was tested among 155 people with diabetes, using regression analysis, and results had support the hypothesis. Research on major aspects of diabetes suggests that the variables, such as knowledge, attitudes, and self-management could contribute to predicting diabetes QoL. For example, Menard et al. (2007) illustrated that a change in QoL, which was assessed by the DQoL, was associated with more positive attitudes which were assessed by ATT34. Some researchers have reported that self-care behaviour of people with T2DM was associated with QoL (Misra & Lager, 2008), adherence to self-management has been shown to improve QoL (ADA, 2003), and it has been reported that success in diabetes control through self-care can influence the QoL of people with T2DM (Gilden et al., 1990; Luker
88
& Caress, 1989). QoL in people with T2DM can be considered to be associated, in many ways, with the presence of satisfaction with diabetes treatment, which is the positive aspect of living with diabetes, and the impact of diabetes, which is the negative aspect of living with diabetes. The range of satisfaction and impact in diabetes QoL includes: physical health, health-care relationships, domestic relationships, social experience and its environment, and self-efficacy . There could be a possible link between diabetes self-management, individual satisfaction and impact. How diabetes knowledge and attitudes can link to satisfaction and impact remains open to question. Research examining the links between knowledge, attitudes, selfmanagement, and QoL together in one model is scarce. Most research in diabetes has examined the relationship or correlations between two of these variables in separate simple regression or correlation models. However, more information will be added to the research if all these variables could be examined together in one model. The links from knowledge and attitudes to self-management are unclear. There might be a direct link between knowledge and self-management or between attitudes and self-management. Alternatively, it is possible that knowledge affects self-management primarily through attitudes, that is, greater knowledge leads to more positive attitudes and those attitudes lead people to pay more attention to managing their diabetes. Another possible pathway is that positive attitudes to coping with their diabetes could lead people to seek more knowledge, and increased knowledge might enhance people’s self-management of their diabetes. No research was identified in the literature that examined these alternative pathways. It is also important to determine how self-management can
89
have a direct effect on QoL especially on the satisfaction and impact of illness subscales. In turn, how knowledge and attitudes affect self-management and, thus, affect the QoL subscales or whether there is a direct effect of knowledge and attitudes on QoL subscales should be examined. To summarise these multivariable relationships, diabetes knowledge could have a direct effect on attitudes among people with T2DM, and both variables could have direct effects on selfmanagement and, in turn, self-management, knowledge, and attitudes could have direct effects on QoL. These variables are the main variables in the conceptual framework presented here. There are other extraneous variables that might affect the main variables in the model. A number or studies discussed in Chapter 2 identified variables, such as gender, level of education, age, time since diagnosis, and type of treatment that could have an impact of the main variables in the model. However, age and time since diagnosis of diabetes (duration of diabetes since diagnosis) are the extraneous variables on which I focus in the conceptual framework. The model will be used to examine how these two variables relate to diabetes knowledge, attitudes, and self-management. Age and duration of diabetes tend to covariance because, generally, when age increases, duration of diabetes will increase as well. The reasons for selecting these extraneous variables are now presented in the next paragraph. The age of people with T2DM is an extraneous variable that affects key aspects of diabetes care, such as knowledge, attitudes, self-management, and QoL, according to research. Studies have shown a correlation between diabetes knowledge and age, such that people with diabetes or specifically T2DM who
90
were younger tended to score higher in diabetes knowledge than those who were older (Rafique et al., 2006; Tham et al., 2004). On the other hand, people with T2DM who were younger, and diagnosed at a younger age, tended to have poor control over their diabetes condition (Savoca et al., 2004). With reference to duration of diabetes, researchers have reported that duration of diabetes was associated with level of diabetes knowledge, with people whose diabetes was of longer duration having greater knowledge about diabetes than people with shorter duration of diabetes (Yun et al., 2007). It has also been found that people with shorter diabetes duration were more concerned about their management of diabetes than those with longer duration (Gåfvels et al., 1993). Although there is a lack of empirical research looking at the relationship between diabetes knowledge and QoL among people with T2DM, it is possible that knowledge can have a direct effect on QoL or an indirect effect on QoL through other main variables. Figure 3.1 illustrates the possible links between the main and extraneous variables. This figure demonstrates the possible effects of knowledge and attitudes on self-management, self-management on QoL, knowledge on QoL, and attitudes on QoL. It also illustrates the possible effects of extraneous variables, age and duration of diabetes, on knowledge, attitudes, and self-management. Figure 3.1 presents the conceptual framework that reflects the possible relationships between the variables. There is a link between knowledge and attitudes with an arrow pointing toward attitudes suggesting that knowledge affects attitudes. The arrows pointing toward self-management from knowledge and attitudes suggest that knowledge and attitudes influence self-management. In
91
addition, there are arrows that point from knowledge, attitudes, and selfmanagement toward QoL suggesting that knowledge, attitudes, and selfmanagement all influence QoL. The effects of the extraneous variables, duration of diabetes since diagnosis and age, are represented by arrows pointing toward knowledge and self-management suggesting that influence of the extraneous variables affects knowledge and self-management. By creating a conceptual framework as demonstrated in Figure 3.1, it is possible to examine the relative influence of all these variables at the same time and in the same model, something which has not been done before in diabetes research that has been identified in the literature. The linkages shown in this conceptual framework have been translated into hypotheses and research questions that can be tested using statistical analyses. The research questions in the present research refer to the broad issues being examined, such as age and duration of diabetes since diagnosis. The hypotheses in the present research refer to specific predictions based on previous research that are each tested statistically and accepted or rejected. The hypotheses and research questions are explained in the next section.
92
Figure 3.1. Proposed conceptual framework.
93
Rationale for Conceptual Framework of Self-Management in Type 2 Diabetes Mellitus Based on the research reviewed in Chapter 2 and the propositions from that research, I developed a conceptual framework of the self-management of T2DM that links self-management to knowledge about diabetes, attitudes to diabetes, and diabetes related QoL, as well as considering the extraneous influence of age and duration of diabetes since diagnosis. That conceptual framework is depicted in Figure 3.1. Figure 3.1 refers to a conceptual framework, where the relationships of the variables are illustrated. Figure 3.2 refers to a path diagram that reflects the results of the analyses testing the relationships shown in Figure 3.1. In Figure 3.1, the paths are arranged in the way that I think best depicts the relationships, particularly in a causal flow from left to right. In Figure 3.2 it was necessary to reorganize the format to include all aspects of the causal paths identified statistically. Research indicates that there is a relationship between knowledge about diabetes and attitudes to T2DM. People with T2DM who have greater knowledge of diabetes typically report more positive attitudes to T2DM. In the conceptual framework, this is represented by an arrow linking knowledge and attitudes. The arrowhead points from knowledge to attitudes indicating that it is predicted that knowledge about diabetes influences people’s attitudes to T2DM, in such a way that people are likely to report more positive attitudes when they have greater knowledge about diabetes. This is reflected in formal hypothesis (H1): Hypothesis 1 (H1): Greater knowledge about diabetes leads to more positive attitudes to T2DM.
94
Hypothesis H1 is depicted in Figure 3.2 where the arrow that links knowledge to attitudes is labelled H1. Central to the conceptual framework is the relationship between knowledge about diabetes. Researchers have consistently reported that greater knowledge about diabetes is associated with more regular self-management of diabetes. In the conceptual framework, this is represented by the arrow linking
95
Figure 3.2. Hypotheses and research questions illustrated in a path diagram.
96
knowledge to self-management. This arrow has its arrowhead pointing to selfmanagement, which indicates that it is predicted that knowledge about diabetes influences self-management practices, so that people are more likely to undertake self-management practices when they have more knowledge about those practices. This is reflected in a formal hypothesis that will be tested, namely: Hypothesis 2 (H2): Greater knowledge about diabetes leads to more regular selfmanagement of T2DM practices. Hypothesis 2 is illustrated in Figure 3.2 where the arrow linking knowledge about diabetes is labelled H2. Another central part of the conceptual framework is the relationship between attitudes to T2DM and its self-management. Research indicates that there is a relationship between attitudes to T2DM and self-management. People with more positive attitudes to T2DM tend to practise more regular selfmanagement. In the conceptual framework, an arrow linking attitudes and selfmanagement represents this relationship. The arrowhead points from attitudes to self-management indicating that it is predicted that attitudes to T2DM influence the extent to which people with T2DM adopt self-management practices, so that people are more likely to undertake self-management practices when they have more positive attitudes to T2DM. This is reflected in a formal hypothesis namely: Hypotheses 3 (H3): More positive attitudes to T2DM lead to more regular selfmanagement of T2DM practices. Hypothesis H3 is depicted in Figure 3.2 where the arrow that links attitudes to self-management is labelled H3.
97
Crucial to the conceptual framework is the relationship between diabetes self-management and QoL. Research indicates that there is a relationship between self-management and QoL. People with T2DM who practise more regular selfmanagement report higher levels of QoL. In the conceptual framework, an arrow linking self-management and QoL represents this relationship. The arrowhead points from self-management to QoL indicating that it is predicted that selfmanagement influences the QoL of people with T2DM, such that people are likely to report greater QoL when they undertake more regular self-management practices. This is reflected in a formal hypothesis, namely: Hypotheses 4 (H4): More regular self-management of T2DM practices lead to higher levels of QoL of T2DM. Hypothesis H4 is depicted in Figure 3.2 where the arrow that links selfmanagement to QoL is labelled H4. The relationship between knowledge about diabetes and QoL is implied in this conceptual framework. Despite a lack of past research on this relationship, it is possible that knowledge influences QoL. Therefore, I propose that people with T2DM who have greater knowledge of diabetes are more likely to have higher levels of QoL. In the conceptual framework, this is represented by an arrow linking knowledge and QoL. The arrowhead points from knowledge to QoL, indicating that it is predicted that knowledge about diabetes influences the QoL of people with T2DM, so that people are more likely to report higher QoL when they have more knowledge about diabetes. This is reflected in a formal hypothesis, namely:
98
Hypotheses 5 (H5): Greater knowledge about diabetes leads to higher levels of QoL of T2DM. Hypothesis H5 is illustrated in Figure 3.2 where the arrow that links knowledge to QoL is labelled H5. Research indicates that there is a relationship between attitudes to T2DM and QoL. People with T2DM who have more positive attitudes to T2DM report greater QoL. In the conceptual framework, this is represented by the arrow linking attitudes to QoL. This arrow has an arrowhead pointing to QoL, which indicates that it is predicted that attitudes to T2DM influence QoL of T2DM, so that people are likely to have higher levels of QoL of T2DM when they have more positive attitudes to T2DM. This is reflected in a formal hypothesis, namely: Hypotheses 6 (H6): More positive attitudes to T2DM lead to higher levels of QoL of T2DM. Hypothesis H6 is illustrated in Figure 3.2 where the arrow that links attitudes to QoL is labelled H6. Besides the hypotheses discussed, I propose that additional relationships are developed to link the extraneous variables to the main variables in this current research. These relationships are derived into research questions that will be tested. Research has reported that longer duration of diabetes since diagnosis is associated with greater knowledge about diabetes. In the conceptual framework, this is represented by the arrow linking duration of diabetes since diagnosis to knowledge about diabetes. This arrow has an arrowhead pointing to knowledge, which indicates that it is predicted that duration of diabetes since diagnosis
99
influences knowledge about diabetes, so that people are more likely to have greater knowledge about diabetes when they have had a longer duration of diabetes since diagnosis. This is reflected in a research question, namely: Research question 1 (R1): Is longer duration of diabetes since diagnosis associated with greater knowledge about diabetes? Research question 1 is depicted in Figure 3.2 where the arrow linking duration of diabetes since diagnosis to knowledge is labelled R1. Although there is a lack of research reported on the relationship between duration of diabetes since diagnosis and self-management, there is a possible impact of duration of having diabetes on self-management practises of people with T2DM. People with longer duration of diabetes since diagnosis may be more likely to have regular self-management practises. In the conceptual framework, this is represented by an arrow linking duration of diabetes since diagnosis and self-management. The arrowhead points from duration of diabetes since diagnosis to self-management, which indicates that it is predicted that longer duration of diabetes since diagnosis influences people’s frequency of selfmanagement practices, such that people are likely to report more regular selfmanagement practices when they have longer duration of diabetes diagnosis. This is reflected in a research question, namely: Research question 2 (R2): Is longer duration of diabetes since diagnosis associated with more regular self-management of T2DM practices? Research question 2 is depicted in Figure 3.2 where the arrow linking duration of diabetes since diagnosis to self-management is labelled R2.
100
Research has also indicated that there is a relationship between age and knowledge about diabetes. People with T2DM who are younger have reported that they have higher knowledge about diabetes. In the conceptual framework, this is represented by an arrow linking age and knowledge. The arrowhead points from age to knowledge, indicating that it is predicted that age influences people’s level of knowledge about diabetes in such a way that younger people are more likely to report higher levels of knowledge. This is reflected in research question (R3). Research question 3 (R3): Do younger people with T2DM display greater knowledge about diabetes? Research question 3 is illustrated in Figure 3.2 where the arrow linking age to knowledge is labelled R3. Research also indicates that there is a relationship between age and selfmanagement of T2DM. Younger people with T2DM tend to have poorer selfmanagement of their diabetes. A research question that looks at the relationship between age and self-management of diabetes is proposed. People with T2DM who are older are more likely to have more regular self-management practices. In the conceptual framework, this is represented by an arrow linking age and selfmanagement. The arrowhead points from age to self-management, indicating it is predicted that age influences people’s self-management practices, so that people are more likely to undertake self-management practices when they are older. This is reflected in a formal hypothesis, namely: Research question 4 (R4): Do older people with T2DM display greater regularity of self-management of T2DM practices?
101
Research question 4 is depicted in Figure 3.2 where the arrow linking age to selfmanagement is labelled R4. Summary of Hypotheses and Research Questions In this summary section, I present the formal hypotheses and research questions based on the conceptual framework of variables that I developed from research on these variables. These hypotheses and research questions that examine the links between self-management of T2DM and the variables of knowledge, attitudes, and QoL, as well as the role of age and duration of diabetes since diagnosis are presented in the form of a path diagram illustrated in Figure 3.2. Hypothesis 1 (H1): Greater knowledge about diabetes leads to more positive attitudes to T2DM. Hypothesis 2 (H2): Greater knowledge about diabetes leads to more regular selfmanagement of T2DM practices. Hypotheses 3 (H3): More positive attitudes to T2DM lead to more regular selfmanagement of T2DM practices. Hypotheses 4 (H4): More regular self-management of T2DM practices lead to higher levels of QoL of T2DM. Hypotheses 5 (H5): Greater knowledge about diabetes leads to higher levels of QoL of T2DM. Hypotheses 6 (H6): More positive attitudes to T2DM leads to higher levels of QoL of T2DM. Research question 1 (R1): Is longer duration of diabetes since diagnosis associated with greater knowledge about diabetes?
102
Research question 2 (R2): Is longer duration of diabetes since diagnosis associated with more regular self-management of T2DM practices? Research question 3 (R3): Do younger people with T2DM display greater knowledge about diabetes? Research question 4 (R4): Do older people with T2DM display greater regularity of self-management of T2DM practices? In the research that is described in the chapters that follow, the variables are measured in samples from Australia and Malaysia and the model is tested separately in each sample, using SEM approach. This is the first research that I can find that examines all these important variables together in one model, allowing hypotheses to be tested and questions to be addressed that have not been examined in the same people at the same time or, in some cases, have not been examined at all.
103
CHAPTER 4 METHOD This chapter outlines the methods used in this study. In the current study, I measured the level of diabetes knowledge, attitudes to T2DM, self-management of T2DM, and QoL of T2DM. I aimed to examine the relationships between knowledge about diabetes, attitudes to T2DM, self-management of T2DM, and QoL of T2DM in people with T2DM. A subsidiary aim of this study was to confirm whether the questionnaires relating to diabetes were reliable and valid for use with the samples employed in this study. Critical to this research is the treatment of the data that addresses the aims. Thus, the latter part of this chapter focuses on the analyses employed. In this Method chapter, I also discuss the statistical analyses used to compare the differences between Australia-based and Malaysia-based samples in terms of their demographics, types of treatment, levels of diabetes knowledge, attitudes, self-management, and QoL. The term ‘cross-cultural comparison’ used in this study refers to the comparison between Australia-based and Malaysia-based samples in terms of those study variables. Study Design A cross-sectional design study was conducted. The cross sectional design is useful for generating quantitative data that can be used to establish wider generalisation. It also allowed quantitative data to be collected on a single occasion using each selected sample. This approach allowed data collection from different groups and from different sources. Four questionnaires (i.e., DKN, ATT19, SDSCA, DQoL) were administered to obtain participants’ responses to the variables under investigation, which were diabetes knowledge, attitudes to
104
T2DM, self-management of T2DM, and QoL of T2DM. Then the data collected on these variables were analysed. Study Population The source populations for this study were people diagnosed with T2DM who lived in Melbourne, Australia, and Kelantan, Malaysia. Sampling Method Two study populations were identified, which were Australia-based and Malaysia-based. The sampling method used with each study population is described here. Australia. The Australia-based participants of this study were people with T2DM who visited the Alfred Hospital and the Western Hospital in Melbourne, Australia for their checks related to diabetes. Due to the participant recruitment procedure, which was approved by the Hospitals’ ethics committees, a convenience sample was chosen from people with T2DM who registered with the hospital between May and November 2009. The data collection took approximately seven months included: application for ethics approval in each hospital, which took 1 to 2 months in each case; screening eligible participants from hospital databases which took a month in each case; and distribution of the questionnaire packs, which took 1 to 2 months. Some participants took 2 to 3 months to complete and return the questionnaire packs. Malaysia. The Malaysia-based participants were recruited through the Diabetes Clinic of the Hospital Universiti Sains Malaysia (HUSM) in Kelantan, Malaysia. The Diabetes Clinic was open once a week for people with T2DM. There were approximately 20 people with T2DM would visited the clinic each
105
week. Thus, witin four months, there would be approximately 320 eligible potential participants. The process of data collection in the Malaysia-based sample took approximately six months included: application for research ethics which took 2 months and data collection in the Diabetes Clinic which took 4 months. In this cross-sectional study, the application for ethics approval was done between January to February 2010, and a convenience sample was chosen to recruit people with T2DM who visited the diabetes clinic during the period from March to June 2010. Malay language is the only language spoken by most of the people with T2DM who visited the diabetes clinic. Therefore, the questionnaires were translated into Malay for administration. Participants Participants for this study were males and females, aged over 18 years, only those who were diagnosed with T2DM by medical practitioners and were registered with the specific hospitals utilised. For the Australia-based sample, the partcipants diagnosed with T2DM for at least a year were identified through the hospital’s patient database. For the Malaysia-based sample, participants diagnosed with T2DM for at least a year was referred to researcher by the staff nurses during their visits to the Diabetes Clinic for clinical check-up. They had to possess sufficient knowledge in English for the Australia-based participants and Malay for the Malaysia-based participants, to be able to read, understand, and answer the items in the four questionnaires. Total usable set of response data from the participants were: N = 567, with n = 291 in Australia and n = 276 in Malaysia. The mean age for participants in the Australia-based and Malaysiabased samples were 55.84 (SD = 11.10) and 57.10 (SD = 8.47) respectively.
106
Measures Demographic/Health Measure Several demographic and health treatment questions were administered. These questions assessed personal attributes of the participants (e.g., age, gender) and their health condition (e.g., how long they had been diagnosed with T2DM, how it was being treated) at the time when they completed the measures (see Appendix A). Diabetes Knowledge (DKN) Scale The Diabetes Knowledge (DKN; Dunn et al., 1984) scale was developed in the early 1980s. It meets the need for a short theory-based knowledge test and the psychometric criteria of reliability and validity. Furthermore, it is easily administered, being short (Beeney et al., 1994). This scale is a reliable measure of diabetes knowledge for researchers investigating the relationships between knowledge, psychological and social factors, health status, and metabolic control (Dunn et al., 1990). The DKN scale has been used with a group of participants ranging in age from teenagers to the elderly, and with individuals from a variety of ethnic backgrounds. It has been used with individuals diagnosed with insulin dependent diabetes mellitus (IDDM) or non-insulin dependent diabetes mellitus (NIDDM). It is designed to be self-administered by respondents (Beeney et al., 1994). The DKN questionnaire is an objective test that consists of three parallel forms (i.e., DKNA, DKNB, DKNC), with 15 multiple-choice items in each. Dunn et al. (1984) developed the DKN in response to the need for a short, theoretically-based knowledge test that met psychometric criteria of reliability
107
and validity and could be easily administered. Dunn et al. developed the 15 questions on each form to cover five broad issues, which are basic physiology of diabetes, including insulin action, hypoglycaemia, food groups and food substitutions, sick day management, and general diabetes care. Dunn et al. retained 45 multiple-choice questions (15 on each form) from a larger pool and gave each item a score of 1 for a correct response and 0 for an incorrect response. All items require a single correct answer, except items 13 to 15 for which several answers are correct and all must be checked to obtain a score of 1. The total score for each form is the sum of correct answers in a score range of 0 to 15, with higher scores indicating higher levels of diabetes knowledge. The DKN scales have alpha coefficients above 0.82 (Dunn et al., 1984). For this study, DKNA was used for measuring diabetes knowledge and the term DKN is used to refer to the DKNA for the rest of the thesis (see Appendix B). The DKN scale is still widely used by researchers in assessing diabetes knowledge. A recent study was conducted by Samuel-Hodge et al. (2009) to test a culturally appropriate, church-based intervention in central North Carolina, to improve diabetes self-management. Participants’ diabetes knowledge scores assessed by DKN were shown to improve after an 8-month intensive intervention phase. Researchers also extensively used the DKN scale in a randomized controlled trial study to measure the outcome of an intervention in a study population where ethnicity varied from African American (Keyserling et al., 2002) to UK Caucasian (Channon et al., 2007). The DKN scale was also used in a Canadian population study to assess diabetes knowledge (Dougherty et al., 1999). The DKNA scale was adopted by several researchers to measure the level
108
of diabetes knowledge among people with diabetes (Channon, Smith, & Gregory, 2003; Ciechanowski et al., 2001; Torres, Franco, Stradioto, Hortale, & Schall, 2009). Ciechanowski et al. (2001) conducted a study in the US that determined that the DKN scale is a valid and reliable instrument in assessing patient knowledge about diabetes and its treatment. They expressed the DKN scores as a percentage of the correct answer. In this thesis, total score of correct answers is reported as a percentage to reflect the level of diabetes knowledge among the respondents. Diabetes Integration Scale-19 (ATT19) The Diabetes Integration Scale-19 (ATT19; Welch et al., 1996) is an abridged version of the ATT39, measuring psychological adjustment and attitudes toward diabetes using a 19-item self-report questionnaire with a corresponding 5-point Likert scale. The ATT39 is the first questionnaire to measure diabetes attitude developed specifically to provide clinicians and researchers with a measurement tool to assess psychological adjustment to diabetes (Welch, Dunn, & Beeney, 1994). The early studies of psychological adjustment among people with diabetes had focused on identifying a diabetes personality and compliant personality, which had proved unproductive (Dunn, 1986; Wilkinson, 1987). The ATT39 items came from three sources, which were: results of two factor analyses on demographic, personality and treatment data; a review of the diabetes psychological literature; and patient interviews and clinical experience on patient adjustment in diabetes (Dunn et al., 1986). ATT19 measure consists of 19 self-report items (see Appendix C). Participants rate their agreement or disagreement with each item on a 5-point
109
Likert scale ranging from 1 (I disagree completely) to 5 (I agree completely). Sixtheen items are reverse scored (i.e., items 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 13, 14, 16, 17, and 19), so a high score on these items reflects a positive attitude toward having diabetes. The total score is calculated by summing the scores from each of the 19 items. This gives a potential score range of 19 to 95. The ATT19 has internal reliability of Cronbach’s alpha 0.84 (Welch et al., 1996). The ATT19 is a scale shorter form of the ATT39 scale that has been widely used by researchers in the past 10 years to measure the psychological adjustment and attitude of people with diabetes (e.g., Cooper, Booth, & Gill, 2008; Dempster, McCarthy, & Davies, 2011; Enzlin et al., 2002; New, 2010; Rickheim et al., 2002; Sridhar et al., 2007; Torres et al., 2009). The ATT19 scale has also been translated into Portuguese to evaluate the attitude of Brazilians with diabetes (Torres et al., 2005). Torres et al. (2005) stated that this scale is a benchmark for comparative studies in QoL and validation of instruments for assessing attitudes of people with diabetes. The internal consistency of the Portuguese ATT19 scale version reached alpha = 0.79, which was considered capable of assessing the psycho-emotional aspects of people with diabetes in Brazil (Torres et al., 2005). This Portuguese ATT19 scale version was then used in a study on evaluation of diabetes education programs in South-eastern Brazil (Torres et al., 2009). Summary of Diabetes Self-Care Activities (SDSCA) The Summary of Diabetes Self-Care Activities (SDSCA; Toobert et al., 2000) is a brief self-report measure of the frequency of completing different diabetes regimen activities over the preceding seven days (Toobert et al., 2000). The original version of the SDSCA questionnaire was first used in the 1980s by
110
Schafer, Glasgow, McCaul, and Dreher (1983) and Glasgow, McCaul, and Schafer (1987). Researchers have continued to use the SDSCA questionnaire over the past decades to measure the levels of self-management across different components of diabetes practices (e.g., Glasgow et al., 1992; S. Kim, Love, Quistberg, & Shea, 2004; Polonsky et al., 2005). The SDSCA questionnaire has undergone various modifications. The most recent revision was conducted by Toobert et al. (2000). Toobert et al. reviewed the reliability, validity, and normative data from seven different studies, which involved 1,988 people with diabetes. They also provided a revised version of the SDSCA questionnaire. In the review, they assessed seven articles with different samples, where all the participants were adults, and the majority had been diagnosed with T2DM, with mean ages ranged from 45 to 67 years. The review indicated that the SDSCA questionnaire is a multidimensional measure of diabetes self-management that has adequate internal and test-retest reliability, evidence of validity and sensitivity to change, and it was proved to be remarkably stable over the years (Toobert et al., 2000). The revised SDSCA (Toobert et al., 2000) consists of 11 core items (see Appendix D). The core items assess self-management behaviour in diet, exercise, blood glucose testing, foot checks, and smoking status. It also consists of 14 additional items that may be useful for researchers or clinicians. In the review, Toobert et al. (2000) set strict criteria for selecting items for the revised version. The criteria are: consistency in mean values across studies, sufficient variability and lack of ceiling or floor effects, temporal stability, internal consistency, predictive validity, sensitivity to change, ease of scoring, and ease of
111
interpretation. In this revised questionnaire, some of the items were excluded. For example, medication was not included in the core items because it has strong ceiling effects and has low test-retest reliability. Toobert et al. also simplified the scoring. The best items were retained and the response format was changed by using the metric of “days per week” instead of percentages. Based on the previous version of the SDSCA, Toobert and Glasgow (1994) reported that the average inter-item correlations were high and generally exceeded 0.5. The SDSCA questionnaire has been widely used, especially in T2DM research for assessing diabetes self-management behaviours (e.g., Lin et al., 2004; Osborn et al., 2010; Shigaki, Kruse, Mehr, & Sheldon, 2010; S. Smith, Paul, Kelly, & Whitford, 2011; T. S Tang, Brown, Funnell, & Anderson, 2008). It is essential to have a self-administered questionnaire that is a reliable and valid measure of diabetes self-management (Goodall & Halford, 1991), because the majority of diabetes care activities are handled by people with T2DM or even their families (Etzwiler, 1994; Funnell et al., 1991). The SDSCA selfadministered questionnaire was also considered to be the most practical and costeffective approach to self-care assessment (Toobert et al., 2000). Diabetes Quality of Life (DQoL) Scale The Diabetes Quality of Life questionnaire (DQoL; Diabetes Control and Complications Trial [DCCT] Research Group, 1988) was developed in the early 1980s and was intended to evaluate the relative burden of an intensive diabetes treatment regimen in a DCCT (Jacobson & DCCT, 1994). The items in this measure cover a series of issues directly relevant to diabetes and its treatment (DCCT Research Group, 1988). The DCCT Research Group contributed their
112
expertise in the development of this measure and the items were derived from the literature on psychosocial aspects of diabetes, as well as from input from patients and clinicians. Then, the DQoL measure was repeatedly reviewed by diabetologists, diabetes educators, nurses, and mental health professionals familiar with diabetes (Jacobson & DCCT, 1994). The DQoL measures the relative burden of diabetes treatment, with the goal of maintaining blood glucose levels as close as possible to those of people without diabetes. The QoL section consists of 46 items divided into three subscales: patient satisfaction, diabetes impact, and diabetes-related worries, including anticipated effects of diabetes and social worries. Answers are given on a 5-point Likert scale rated from 1 (very satisfied, no impact, no worry) to 5 (very dissatisfied, very impacted, very worried). The mean scores for each subscale using this method are interpreted against a 5-point scale, where 1 is equivalent to the highest QoL and 5 is considered the poorest QoL. The DQoL questionnaire used in this study contains 35 self-report items with 15 items measuring satisfaction, 20 measuring impact, and one items measuring self-rated general health (see Appendix E). The 11 items that form the worry subscale were found to be not applicable for most of the participants with T2DM in this research. It is more suitable for people with T1DM (Jacobson & DCCT, 1994). For example, item 1, 2, and 5 in the worry subscale “How often do you worry about whether you will get married?”, “How often do you worry about whether you will have children?” and “How often do you worry about whether you will be able to complete your education?” are more applicable to people with T1DM and of younger age than the present sample. Most participants with T2DM in the present
113
research would leave these items blank because they are not applicable to them. Therefore, the 11 items from the worry subscale were excluded from this study. The DCCT Research Group (1988) evaluated the DQoL measure, reporting high test-retest correlations in the 0.78 to 0.92 range in both adults and adolescents with diabetes. The DQoL scale has been widely used by researchers in the past 10 years in assessing the QoL of people with T2DM (e.g., M. T. Kim et al., 2009; Latham & Calvillo, 2009; Trento et al., 2004). Although the DQoL measure was developed in the United States, it has been widely used by researchers in different ethnic populations and different countries. For instance, the DQoL measure was used in a community-based based study on Korean Americans with T2DM (M. T. Kim et al., 2009). It was translated and re-validated into different language versions and used in different countries, such as Hong Kong (Chinese version; Shiu, Thompson, & Wong, 2008), Taiwan (Chinese version; Huang et al., 2008), Thailand (Thai version; Wattana et al., 2007), Turkey (Turkish version; Yildirim et al., 2007), and Italy (Mannucci, Mezzani, Conti, & Rotella, 1994). Watkins and Connell (2004) reported that this measure tends to be more sensitive to diabetes-related lifestyle issues. Jacobson et al. (1994) proposed that the measure is acceptable, easy to use, and that people with diabetes have little difficulty understanding the items.
114
Revision and Translation of Questionnaires The questionnaires used in this thesis were revised to suit the local study populations. As English is the main spoken language among the participants in Australia, the English versions of questionnaires were used for the Australiabased sample. Due to the main language spoken by Malaysians, particular the Kelantan participants, being Malay, a Malay version of the questionnaires was used for the Malaysia-based study. The revision and translation of the questionnaires explained here. Australia In the first stage of the study conducted in Australia, the four questionnaires (DKN, ATT19, SDSCA, and DQoL) were revised to be appropriate for the study populations. Face and content validation were conducted with five professionals (2 medical doctors, 1 nurse, 2 psychologists), who had expertise in questionnaire design, and experience of working with people with T2DM. The experts were asked to examine whether (a) the items on each questionnaire appeared to relate to the name and purpose of that questionnaire (face validity), and (b) all important aspects of the construct were adequately covered for each questionnaire (content validity). Experts were also required to comment on the cultural sensitivity of the original questionnaires, so that they considered whether the measures were appropriate for the Culturally and Linguistically Diverse (CALD) population in Melbourne. The experts endorsed the face and content validity of the measures. The questionnaires were confirmed to be suitable for use to evaluate the diabetes knowledge, attitude, self-
115
management, and QoL of people with T2DM of the Australia-based sample in their standard form. Malaysia For Malaysia, the appropriateness of items on all four questionnaires was considered by experts in the local culture. Based on the experts’ opinion, the satisfaction item-10 and impact item-10 were excluded from DQoL. These items are ‘How satisfied are you with your sex life?’ and ‘How often does your diabetes interfere with your sex life?’. Sensitive questions such as sex life were deemed to be inappropriate for the culture in Malay population especially in Kelantan state. These items were considered to be inappropriate to be included in a self-administered questionnaire for the local population in Kelantan, who are mostly traditional people from a Muslim background. Thus, they would feel uncomfortable answering these questions because they would be regarded as personal. The experts’ opinion in regard to this issue is important as they had numerous experiences with the local culture and they had to look after their patients’ feelings. Their inclusion might have led to many Malaysian participants not completing the study. Furthermore, Yildirim, Akinci, Gozu, Sargin, Orbay, and Sargin (2007) conducted an adaptation and validation study on DQoL also excluded the sex related questions among the Turkish participants. The reason given by Yildirim et al. was that the respondents were offended and did not feel comfortable socially to answer such question. Therefore, in the present research, these two items were not included in the DQoL questionnaire for data collection. In terms of language, most of the people in Kelantan speak fluently in “Bahasa Melayu” (Malay language). A majority of the people with T2DM who visit the
116
Diabetes Clinic in HUSM do not have fluent knowledge of English. Therefore, the measures used for this study were translated into Malay, which is understood by the local people. Translation of questionnaires. All questionnaires (DKN, ATT19, SDSCA, and DQoL), were forward-translated from the original English versions into the Malay versions by me and were then reviewed by a psychologist, two medical doctors, and a native Malay translator who were bilingual (Malay and English), to ensure they would be readily comprehended by the local population in Kelantan, Malaysia. Back-translation of the questionnaires was then carried out in order to uncover any discrepancies of meaning between the original English language versions and the Malay versions of the questionnaires (Brislin, 1986). To do the back-translation, the translated versions were distributed to a linguistic expert at the Department of Language (Unit Bahasa), Universiti Sains Malaysia and two diabetes researchers who have experience of working with people with T2DM in HUSM. Ambiguous items were identified and modified retaining the meaning consistent with the English version. These versions of measures were then back-translated by another group of bilingual experts without looking at the original English version. The final versions of the questionnaires were again given to a psychologist, a diabetes researcher, and a medical doctor who were of the opinion that the questionnaires had good face and content validity in measuring diabetes knowledge, attitudes, self-management, and QoL among people with T2DM.
117
Although debriefing of the measures with a small sample of people with type 2 diabetes process was not conducted officially (e.g., taking 10 people with T2DM to debrief the measures), I personally distributed the questionnaires to eligible participants, and they were asked to identify any unclear items and if they do not understand any item in the questionnaires, they are asked to inform the researcher. It appeared that participants understood all items and the items were meaningful to them. This is not surprising because health officers who work with people with T2DM on an everyday basis screened the questionnaires. Reliability of Malay version of questionnaires. The numbers for scoring each item of the subscales for ATT19 and DQoL were tested for reliability (Cronbach’s alpha). This was to ensure the stability of the Malay versions of the questionnaires. It was found that the Malay versions were compatible with the original English versions. The results are reported in the following section. The DKN questionnaires with multiple-choice items were not tested for internal consistency because responses to the items are multiple choices, each with a single correct answer except, items 13 to 15. Cronbach’s alpha could not be determined because it is proportional to the numbers of items in an index as well as the magnitude of their covariance. Cronbach alpha is not suitable to use to reflect the internal consistency of the DKN questionnaire. DKN is a multiple choice questionnaire and the correct answer was given a value (e.g., 1), whereas the incorrect answer was given a zero. Therefore, the Cronbach alpha for DKN reflects the magnitude of how easy or difficult it is to get the correct answer, rather than the internal consistency of the questionnaire. Furthermore,
118
participants in this research were asked to answer the DKN once only. The translated questionnaires are presented in Appendices I to M. All the analyses were done using SPSS 17.0 (2010 software). Reliability of the questionnaires was tested to determine the extent to which the items in each questionnaire were related to each other. The internal consistency reliability of the ATT19 and DQoL scales was evaluated by computing the corrected itemtotal correlation and Cronbach’s alpha. Nunnally (1978) has indicated that Cronbach’s apha of 0.70 or greater can be considered an acceptable reliability coefficient. The internal reliability test illustrated that the Malay version of ATT19 had a total internal reliability as measured by Cronbach’s alpha of 0.78, which is very close to the result reported by Welch et al. (1996; Cronbach’s alpha range from 0.82 to 0.84). The internal reliability tests for the Malay version of the DQoL revealed Cronbach’s alpha coefficients of 0.81 and 0.80 for the satisfaction scale and impact scale respectively. These results were close to the results reported by Jacobson et al. (1994). They reported that Cronbach’s alpha for satisfaction was 0.88 and impact was 0.77. Thus, the internal consistency reliability of the Malay versions of these questionnaires was considered acceptable. Procedure In the current study, I examined two study populations, Australia-based and Malaysia-based adults 18 years and over, who had been diagnosed with T2DM for at least a year. Research ethics approval was granted by Victoria
119
University Human Research Ethics Committee prior to the commencement of the study. Australia Following ethics approval, an application to conduct data collection was sent out to various hospitals in Melbourne. Two hospitals, the Alfred Hospital and Western Hospital, agreed to take part in this study. Following ethics approval from the Alfred Hospital, eligible participants were identified via the patient list held by the Department of Endocrinology and Diabetes. Through the Department, a questionnaire pack that included measures of DKN, ATT19, SDSCA, DQoL, a demographic and health form, a plain language information statement (see Appendices F and G) , and a reply paid envelope was sent out to eligible participants. Five-hundred questionnaire packs were mailed out. Return to me of the questionnaires from possible participants implied consent to participate in the study. A total of 500 questionnaires packs were sent out of which 205 (41%) completed questionnaire packs were returned. Following ethics approval from the Western Hospital, potential participants with T2DM were invited to participate in the study during their visits to the hospital. People were invited to participate in this study during their routine clinical appointments with their diabetes educator. During the study session, consent was gained from the participants using standard consent procedures before asking them to complete the questionnaires individually. A copy of the plain language information statement and consent form (see Appendix H) were provided for participants. The completed questionnaire packs were returned to me in the hospital. From the 125 questionnaire packs that were handed out, 100
120
(80%) completed questionnaire packs were returned by the end of the data collection phase in this hospital. Malaysia Research ethics approval from the Universiti Sains Malaysia Human Research Ethics Committee was granted before data collection was conducted. Participants were recruited via the Diabetes Health Clinic in HUSM. People with T2DM were invited to participate in this study during their routine clinical appointment with their physician. Participants were selected who had sufficient knowledge in Malay and would be able to read and understand the questionnaires provided to them. Participants who agreed to fill in the questionnaires were given a copy of the information statement and consent form (see Appendix N). Consent was gained from the participants using the standard consent procedures approved at HUSM before they completed the questionnaires. From the 300 questionnaires packs that were handed out, I received 285 (95%) completed questionnaire packs by the end of the data collection period in this hospital. Data Management In order to identify the level of diabetes knowledge, attitude, selfmanagement, and QoL of people with T2DM, the total score of each measure was used. The procedures for calculating scores are explained in the following section. The results are reported in the descriptive sections in Chapters 5 and 6 for Australia-based and Malaysia-based samples respectively. The total score of each measure was used for path analyses to identify the relationships between the studied variables.
121
Diabetes knowledge (DKN) The DKN questionnaire contains 15 multiple-choice items: a score of 1 is assigned for a correct response and 0 for an incorrect response. The total score is calculated by summing the scores from the 15 items. This gives a potential score range of 0 to 15. For analysis purposes, in this study, the total scores were converted to percentages. For example, a raw score of 12 out of 15 corresponds to 80%. Higher scores on this measure indicate a higher level of diabetes knowledge (Beeney et al., 1994). Diabetes Integration (ATT19) The ATT19 questionnaire contains 19 self-report attitudinal items that are scored on a 5-point Likert scale that ranges from 1 to 5. For analysis purposes, and as recommended by the authors of the ATT19 (Welch et al., 1996), the score of each item was summed. Higher scores on this measure indicate a more positive attitude toward diabetes and better adjustment compared to lower scores. Summary of Diabetes Self-care Activities (SDSCA) The SDSCA questionnaire consists of 11 self-report items. The scoring is in the metric of “days per week”. Average scores are calculated for each of the four areas assessed by the SDSCA, which are diet, exercise, blood-glucose testing, and foot care. For items 1 to 10, the number of days per week on a scale of 0 – 7 is used except for item 4, which is about of the number of days during the last seven days on which a person ate high fat foods, such as fried food or full-fat dairy products. These items are reverse scored (Toobert et al., 2000). Item 11, which is about smoking status of participants, was reported as well.
122
Diabetes Quality of Life (DQoL) For reporting purposes in this study, all the 15 response scores on the items on the satisfaction subscale were reversed, so that positive satisfaction is equivalent to a higher score. For the impact subscale, items 8 and 16 are reverse scored and all other items are summed without reverse scoring. Thus, a lower score means that the diabetes impact is less and this is considered to reflect more positive QoL. The total score is calculated by summing the scores from each of the items for satisfaction and items for impact. The raw scale score is converted into a 100-point scale where 0 represents the lowest possible QoL, and 100 represents the highest possible QoL. The respondents’ raw score minus the lowest possible score on each scale is divided by the possible score range and multiplied by 100 (Jacobson & DCCT, 1994). (Raw score − lowest possible score Transformed scale = * 100 Raw score range
For example, on the satisfaction subscale, the lowest possible score of 15 is subtracted from the total raw score of 61, giving a score of 46. This is then divided by the raw score range which is 60 (highest possible score, 75, minus the lowest possible score, 15). The result would be 0.77, which is then multiplied by 100 to yield a transformed satisfaction score of 77. Data Screening Before any statistical analyses, data were screened for missing values and outliers. Frequency analyses were used to identify missing values in each variable. Participants’ responses were re-checked to confirm whether the missing data was caused by data entry error or simply missing responses by the
123
participants. In the Malaysia-based and Australia-based data for the DQoL scale, impact subscale item-20 has a greater number of missing values than other items. Within this group more than 40% of the cases did not have scores for subscale impact items-20. Item-20 is “How often do you hide from others the fact that you are having an insulin reaction?” This happened because a number of participants were not on insulin treatment, so they would not be able to answer this question. Item-20 in subscale impact was deleted from the data set. There are possibilities that the patterns of missing data are characterised as MCAR (missing completely at random), MAR (missing at random), and MNAR (missing not at random or non-ignorable) types of missing data. Systematic patterns of missing data will impact the later results, so they may not be able to be generalized to the whole population (Kline, 2005). The patterns of missing data for the variables were examined using the missing value analysis in SPSS 17.0. Tabachnick and Fidell (2007) indicated that if there is less than 5% missing data in a random pattern for a large data set, it is unlikely to be problematic in the later results interpretation. They stated that any standard procedure for handling missing values yields similar results. The Australia-based sample had 14 (4.6%) cases and the Malaysia-based sample had 9 (3.2%) cases with one or more missing value in their responses. The percentage of the missing values in the data set was low. Thus, these cases were deleted from further analysis and the final total sample size is 291 for the Australia-based sample, and 276 for the Malaysia-based sample research. To identify any outliers, as suggested by Hair, Black, Babin, and Anderson, (2010), the distribution of the observations for each variable in the
124
analysis was examined. Outliers are those cases falling at the outer ranges (high or low) of the distribution. The graphical method of histograms was used to identify potential outliers that are seen to be trailing away or unattached to the rest of the distribution. Following inspection of outlier cases using a histogram, no potential outliers were observed in this data set. Analytical Procedures All statistical analyses were conducted using the computer packages SPSS 17.0 and AMOS 17.0. Differences were considered significant at p < 0.05 with 95% confidence intervals. In this study multiple statistical techniques were used. A combination of exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) was used for scale assessment and on model measurement. When a final set of measures was determined, the descriptive analysis on each variable was conducted. The final total score from each measure served as the observed variable in path analysis. In order to test the hypotheses, path analysis with observed variables was used to assess the hypothesised model for Australia-based and Malaysia-based samples. Then, in cross-cultural comparison, differences between the countries were identified. Exploratory Factor Analyses (EFA) EFA was used to identify items that belong to a factor in a multifactor structure. The principal components extraction method was used to extract the factors and their associated items. Then, the factors were rotated with Varimax rotation. Items with factor loadings lower than or equal to 0.3 were excluded from consideration. Then, the factor structure was examined and a suitable name for each factor was given based on its theoretical structure. Then, the
125
measurement model was specified by relating each observed variable (i.e., the measurement item) to its corresponding latent variable (i.e., the theoretical factor), using CFA. The internal consistency reliability of each factor identified in EFA was assessed with Cronbach’s alpha coefficient. There are different numbers of items in various SDSCA subscales, for example, the diet subscale has four items and the exercise subscale has two items. Cronbach’s alpha coefficient is influenced by both the number of items and the relationships between items (Toobert et al., 2000). For this reason, the inter-item correlations for SDSCA subscales were reported as well as the global correlation. EFA was conducted on measures SDSCA and DQoL. SDSCA is a multidimensional measure of diabetes self-management and EFA was used to assess whether the associated items fell within their designated factors (i.e., diet, exercise, blood glucose testing, and foot care) for the Australia-based and Malaysia-based samples. The SDSCA subscales are formed with small number of items (i.e., four items for diet, and two items for exercise, blood glucose testing, and foot care). Researchers now agree that different aspects of self-management should be assessed separately because of their multidimensionality. Therefore, a parcelling method was used to obtain one final score for each subscale (i.e., diet, exercise, blood glucose testing, and foot care). These parcels were then used as indicators (or served as observed variables) for hypotheses tested in path analysis. The DQoL measure consists of 15 items for subscale satisfaction and 20 items for subscale impact. The items in each subscale were explored using EFA to obtain a meaningful factor structure. Due to the large number of items per factor (or subscale) for the DQoL measure, CFA was used to confirm the items that defined
126
each factor. Problematic items were identified and discarded from further analysis. Analysis of CFA is explained in the following section. This study is not intended to examine the psychometric properties of questionnaires DKN and ATT19. The DKN scale is a self-administered multiplechoice questionnaire measuring diabetes knowledge. There are 15 questions that cover the general principles of diabetes management. The total score of the correct responses on DKN reflects the level of diabetes knowledge as suggested by Beeney et al. (1994). Researchers have used rigorous psychometric criteria to clarify the replicable factor structure of the measure, and to ensure the scale is reliable and psychologically meaningful for both T1DM and T2DM people. Their results showed that the most replicable interpretation was found for a singlefactor for both T1DM and T2DM people (Welch et al., 1996). The 19 items were found to be meaningful to address the attitudes of people with T2DM and have been used by researchers in different countries. Therefore, analysis of EFA and CFA were not conducted on DKN and ATT19. The total score of each scale was used as an indicator for the construct in the later path model assessment. Confirmatory Factor Analyses (CFA) Mulaik (1987) recognized that exploratory techniques can never answer definitively questions regarding the latent structure of a set of variables. He argued that EFA can only suggest structures and these require confirmation by CFA. CFA is also known as congeneric factor analysis when the scale is unidimensional. The variables that load on a factor become the descriptors of the underlying dimension. Therefore, examination of these variable loadings on the
127
factors gives understanding of the underlying dimension (Hair, Anderson, Tatham, & Black, 1998). Before testing the relationships between the variables, the appropriateness of the structure of the questionnaires was evaluated, using CFA analysed with the SPSS version of AMOS (Arbuckle, 2004). AMOS is a statistical program designed to perform structural equation modelling (SEM), path analysis, a very general, very powerful multivariate analysis technique (StatSoft, 2012). CFAs were performed using AMOS to test for the goodness-of-fit of the obtained data for each measure. CFA using Robust Maximum Likelihood estimation was conducted on the data set. In AMOS, several indices are available to express the fit of the underlying data. The most commonly-used indices evaluating goodness-of-fit were presented, namely the normed chi-square (χ2/ df), root mean square error of approximation (RMSEA), standardized root mean square (SRMR), goodness-offit index (GFI), the comparative fit index (CFI), and the Tucker and Lewis index (TLI). Following the CFA, the problematic items or any evidence that the models are over-parameterized (i.e., comprising too many factors) in the questionnaires were identified. Further revisions were determined, including exclusion of items, according to the problematic items identified. Following this, a new data set was derived and used in the latent analysis of hypothesis testing. In the next section, details of the statistics for model assessment used in this study are explained.
128
Model Assessment After the models had been specified and estimated, assessment was done of how well the observed data (i.e., items of DQoL) fitted to the model. This is one of the primary goals in CFA. Model assessment using CFA was not conducted on SDSCA because it consists of a small number of items per subscale (i.e., four items for the diet subscale and two items for the exercise, blood glucose testing, and foot care subscales). There are many model fit indices used in the SEM literature and different fit indices are reported in different articles. It is important to consider multiple criteria and fit indices in evaluating model fitness. Several goodness of fit indices are, thus, reported in this study and are explained in the following section. Absolute fit indices. Absolute fit indices determine how well a model fits the sample data (McDonald & Ho, 2002) and demonstrate which proposed model has the most superior fit (Hooper, Coughlan, & Mullen, 2008). The fit indices measure how well the model fits in comparison to no model at all, and do not rely on comparison with a baseline model (Joreskog & Sorbom, 1993). Fit indices included in this category that are used in this thesis are Normed chi-square, RMSEA, SRMR, GFI. Normed chi-square (χ2/ df). In order to reduce the sensitivity of χ2 to sample size, the χ2 value is divided by the degrees of freedom (χ2/ df). This lower value is called normed chi-square (NC). Bollen (1989) recommended that values of 2.0, 3.0, or even as high as 5.0 can be considered reasonable fit. The other fit indices described next are less affected by sample size.
129
Root mean square error of approximation (RMSEA). The RMSEA, developed by Browne and Cudeck (1993), is an estimation of lack of fit in a model compared to a perfect or saturated model. The model is a perfect fit when the value of RMSEA is zero. However, there is no model that will exactly fit a population. It is only possible to expect a close approximation to reality (Browne & Cudeck, 1993). It is suggested that an RMSEA value of .05 or less reflects a model of close fit, and values between .05 and .08 indicate reasonable fit. Standardized root mean square residual (SRMR). The SRMR and root mean square residual (RMR) are the square root of the difference between the residuals of the sample covariance matrix and the hypothesised covariance model. The RMR is calculated based on the scales of each indicator. When a questionnaire contains items with varying levels, for example, some items may range from 1 – 5 and others range from 1- 7, the RMR becomes difficult to interpret (Kline, 2005). The standardised RMR (SRMR) resolves this problem and is much more meaningful to interpret. The SRMR value can range from 0 to 1, but values less than .05 are considered reasonable fit (L. Hu & Bentler, 1999). Therefore, the SRMR was used in this study as variables are measured in varying levels. Goodness of fit index (GFI). The goodness-of-fit index, GFI is a ratio of the sum of the weighted variances from the estimated model co-variance matrix to the sum of the squared weighted variances from the sample co-variance (Tabachnick & Fidell, 2007). This index compares the hypothesized model with no model at all, so they are referred to as absolute indices of fit. GFI index should be between 0 and 1. The model is a good fit when the value is close to 1
130
(Joreskog & Sorbom, 1993). Values greater than .95 are recommended for GFI, so .95 is considered to be the acceptable threshold level for GFI and AGFI (Hooper et al., 2008). Incremental fit indices. Incremental fit indices are also known as comparative fit indices. They are a group of indices that do not use the chi-square in its raw form, but compare the chi-square value to a baseline model (Hooper et al., 2008). Fit indices included in this category and reported in this thesis are CFI and TLI. Comparative fit index (CFI). The CFI assesses fit relative to other models with a different approach. The CFI uses the non-central χ2 distribution with non-centrality parameters. The larger the value of the non-centrality parameters, the greater is the model misspecification (Bentler, 1990). CFI values vary between 0 and 1. The CFI index is relatively independent of sample size and it works well in estimating model fit even in small samples (L. Hu & Bentler, 1999). Values greater than .95 are recommended for CFI as acceptable, so .95 is considered to be the threshold level (Hooper et al., 2008). The Tucker Lewis Index (TLI). The TLI is also known as the Non Normed Fit index (NNFI). It was developed by Tucker and Lewis (1973) and it estimates the relative improvement per degree of freedom of the target model over an independence model (L. Hu & Bentler, 1998). The TLI value normally varies between 0 and 1 and values closer to 1 indicate the model fits well. The TLI index does exceed the value of 1 for an over-fitting model. The TLI index has been found to be consistently independent from sample size (J. C. Anderson & Gerbing, 1984). Hutchinson and Olmos (1998) reported that the TLI index is
131
more sensitive to the presence of model misspecification than other fit indices. Values greater than .95 are considered to reflect good fit, so .95 is recommended for TLI as an acceptable threshold level (Hooper et al., 2008). Summary of fit indices. There are no golden rules for assessment of model fit, therefore, reporting various fit indices is necessary (Crowley & Fan, 1997) because different indices reflect a different aspect of model fit as mentioned earlier. Although the model of chi-square is reported in some studies, it is affected by sample size and many problems are associated with it. For example, when the sample size is large, which is required in SEM and path analysis, it is more likely that the p-value will appear to be significant for the comparison between the model and the data (Kline, 2005). When the distributions of the data are severely non-normal, the value of chi-square tends to be high with a significant p-value. Consequently, true models will be rejected when chi-square is interpreted as a test fit statistic even when the model is properly specified (Mclntosh, 2006). Kline (2005) suggested a number of indices to include are RMSEA, SRMR, and CFI. For the purposes of analysis in the present thesis, if improvements have been made to the model and fit indices are improved, but are still not within their acceptable range for fit, the RMSEA, SRMR, and CFI will be prioritised in making the decision whether to accept or reject the specified model. Table 3.1 summarises the various fit indices produced by AMOS and the commended fit value of each index.
132
Table 4.1 Criteria for Model Fit Assessment Fit Indices
Recommended values for acceptable fit
Normed chi-square(χ2/df)
Between 1.0 and 5.0
Root Mean Square Error of Approximation
Lower than .08
(RMSEA) Standardised Root Mean Square (SRMR)
Lower than .05
Goodness-of-Fit Index (GFI)
Over .95
Comparative fit index (CFI)
Over .95
Tucker-Lewis Index (TLI)
Over .95
Descriptive Statistics The study variables consisting of diabetes knowledge, attitudes to T2DM, self-management of T2DM, and QoL of T2DM are described before any further analysis. Descriptive statistics for these variables are presented after model assessment where problematic items were deleted from the measures. Other demographic and health characteristics variables are also described and are presented at the beginning for each Results chapter (i.e., Chapter 5 and Chapter 6). Descriptive statistics, including mean, standard deviation, frequency, and charts are used to describe the data. Assessment of Path Model Assessment of the path model is where the hypotheses were tested. It was conducted after the confirmation of the measurement part of the model. When items were confirmed to fit adequately into the measurement model, an item parcelling method was used that summed a number of item responses into one
133
score. The procedure for summing a number of item responses for each variable was discussed in a previous section in this chapter on “Data Management”. The parcelled items are then used as indicator variables of a construct. A parcel is a total score across a set of items and is treated as a continuous indicator. The advantages of using parcels are the score reliability of parcels tends to be greater than those of the individual items, and the distribution of all parcels are roughly normal, therefore, a normal theory method, such as maximum likelihood may be used to estimate the model (Coffman & MacCallum, 2005; Cunningham, 2008). Path model assessment for hypothesis testing was first conducted separately for the Australia-based and Malaysia-based samples. Using the AMOS program, the path model was specified and estimated. The fit indices, which are described in “Model Assessment”, were used to identify the model fitness. The significant pathways of relationships between variables from the final specified model are illustrated in diagrams presented as Figures, and, along with the standardised regression weights and the p-value of the final models, they are presented in the Results chapters (Chapter 5 and 6). Cross-Cultural Comparison Cross-cultural study has become increasingly important in psychology research because it highlights the importance of similarities and variations across cultures (van de Vijver & Matsumoto, 2010). The findings of cross-cultural study promote international and intercultural exchange and contribute to a broader understanding of human behaviour and mind (van de Vijver & Matsumoto, 2010). Cross-cultural studies on diabetes QoL are still limited and rare. The challenge in cross-cultural studies that involve QoL is to identify a QoL
134
instrument that is reliable and valid across cultures. In response to this challenge, the DQoL measure used in this study was tested using multiple-group CFA in an effort to ensure this measure is equivalent for both cultural groups. The DQoL measure was examined to test for cross-cultural differences or invariance of a structural model. Measurement invariance was conducted using AMOS and they are explained in the following sections. Then, the differences between the cultural groups were compared in terms of demographic and health measures, and the study variables. Measurement invariance test. Measurement invariance is used when multiple-group CFA is involved. It is used to test whether the relations of latent variables with their indicators are identical for different groups (e.g., Australia, Malaysia) or whether the groups are different in particular ways across groups on the constructs represented (Byrne, Shavelson, & Muthen, 1989). The invariance test is important because constructs may have different meanings for different groups and this would result in invalid mean comparisons between them. In this study, measurement invariance using AMOS was conducted on the DQoL scale to test whether the relations of its subscales (satisfaction and impact) with their items are identical for the Australia-based and Malaysia-based data sets. Metric invariance or the measurement weight model was identified to determine whether the factor coefficients from each item were invariant for both groups (Cunningham, 2008). Metric invariance is a test of whether the factor coefficients are the same across groups. If the chi-square test is significant (p < .05), the factor coefficients are not equal across the groups. When metric invariance is not supported, at least one of the factor coefficients for the
135
Australia-based and Malaysia-based data sets cannot be assumed to be equal across the groups. In order to distinguish where the difference lies, the unstandardized factor coefficients for the two groups with the largest difference were first identified. The analysis is repeated by freeing the constraints on the factor coefficient, where the largest difference lies, in each of the groups. Then, the chi-square of the measurement weight was identified. If the chi-square test was not significant (p > .05), the factor coefficients were considered to be equal across the groups. Differences between two cultural groups. A relevant question to ask is whether there are any cultural difference in the diabetes knowledge, attitudes, self-management, and QoL between the Australia-based and the Malaysia-based samples. A cross-cultural comparison on diabetes variables is important to understand cultural differences between samples from the two countries. This study focused on similarity in patterns of relationships among various participants’ characteristics and studied variables (e.g., diabetes knowledge, attitudes to T2DM, self-management of T2DM, QoL of T2DM) and their differences in both cultural groups. Crosstabs with chi-square statistic was used to identify any differences between the Australia-based and Malaysia-based samples on categorical variables (e.g., type of treatment, education level, working status). The independent samples t-test was used to identify the difference between the Australia-based and Malaysia-based samples on numerical variables (e.g., diabetes, knowledge, attitudes, self-management, and QoL). The differences between gender and insulin treatment of both groups were tested using One-way ANOVA. The
136
differences between each pair of means were then determined by using Post-hoc Tukey test. The test statistics and p-values are presented in tables and discussed in the results of the cross-cultural comparison chapter (Chapter 7). These analyses are based on mean differences, therefore I do not consider that any differences between the samples confounded the questionnaire differences examined.
137
CHAPTER 5 RESULTS AND DISCUSSION FOR AUSTRALIA-BASED SAMPLE In this chapter, I present the results of the analyses for the Australia-based sample, people with T2DM living in Australia. This chapter begins with the sample overview of the Australia-based sample. In the second section of the chapter, results of the measurement model assessments for the questionnaires SDSCA and DQoL are reported. This section is divided into two sub-sections. In the first part, I address the results of EFAs on both questionnaires. That is followed by reporting of results of the CFA on DQoL, in the second part. In the third section of the chapter, I report on the descriptive analysis of diabetes knowledge, attitudes to T2DM, self-management of T2DM, and QoL of T2DM of the participants with T2DM in Australia. The fourth section of the chapter provides the correlations between the major variables. In the fifth section of the chapter, analyses related to hypotheses, research questions posed in this thesis and path analysis on the main and extraneous variables are presented. This section presents the final path model that provides the best fit for the Australiabased sample. This is followed by the sections of discussion of hypothesis testing for the Australia-based sample, comments on methodological issues, and suggestions for future research based on the findings reported for the Australiabased sample. This chapter ends with a summary of the research findings for the Australia-based sample.
138
Sample Overview This section begins with reports on the demographic and health characteristics of the participants. A total of 291 people with T2DM completed the questionnaires. The gender of participants in this study for the Australia-based sample included more males (n=192, 66%) than females (n=99, 34%). Participants’ ages ranged from 21 to 70 years, with mean age of 55.84 years and standard deviation of 11.10 years. Participants time since diagnosis with T2DM (duration of diabetes since diagnosis) ranged from 1 to 39 years, with a mean of 11.91 years and standard deviation of 9.01 years. Table 5.1 presents the mean and standard deviation of age and years since diagnosis in each gender. The gender of participants in this study was approximately even in the mean of age and duration of diabetes since diagnosis. Table 5.1 Mean (M) and Standard Deviation (SD) of Age and Duration of Diabetes since Diagnosis (Australia-Based Sample) Age
Duration of diabetes since diagnosis
Gender M
SD
Range
M
SD
Range
Female
55.95
11.33
24 - 70
12.67
9.34
1 – 39
Male
55.78
11.01
21 - 70
11.51
8.83
1 - 39
Overall
55.84
11.10
21 - 70
11.91
9.01
1 – 39
It was found that approximately half (53.6%) of the participants, had been treated with insulin at the time of data collection. As illustrated in Figure 5.1, the largest proportion of participants, 37.8% in this study are treated with diet and
139
tablets. Thus, over 90% of the participants in the Australia-based sample were prescribed with medication to help control their blood sugar level.
Figure 5.1. Number and percentage of types of treatment (Australia-based sample). The majority of the participants, in this study, 74.6%, had completed at least high school education level, of which 20.6% were university graduates and 15.8% were college graduates. Figure 5.2 illustrates the education background of the participants.
140
Figure 5.2. Participants’ education background (Australia-based sample).
Measurement Model Assessment Assessment of the measurement model involved EFA and CFA (Bollen, 1989). The next sub-sections include the results of the EFA followed by results of the CFA. Exploratory Factor Analysis (EFA) EFA is employed to explore whether the factor structure of the observed variables was the same as that in the proposed measurement model. It is also used to explore whether the proposed latent variable and observed variable relations were supported empirically. Principal components extraction methods were used to extract factors. Then the factors were rotated with varimax rotation to achieve interpretable results. The rotated factor loadings were examined for their ability to produce subscales that have items with loadings higher than .30. Items that had factor loadings lower than or equal to 0.3 were excluded from consideration. According to Hair, Black, Babin, Anderson, & Tatham (2006), factor loadings in
141
the range of .30 to .40 are considered to meet the minimal level for interpretation of structure, loadings greater than .50 are considered practically significant, and loadings that exceed .70 are considered to have a well-defined structure. In the present case, the reliability of the items that fell under the same factor was assessed with Cronbach’s alpha, which is the most widely used measure of internal consistency. Then the factor loadings and the Cronbach’s alpha of the items and construct were investigated. Items that did not meet the requirement of the cut-off point for factor loadings and Cronbach’s alpha were treated as problematic items and were excluded from the measurement model. The percentage of variance explained of each construct (or factor) is presented. Diabetes self-management (SDSCA). Table 5.2 presents the factor loadings for 10 items in the SDSCA scale for the Australia-based sample. It shows that all the items were loaded on their hypothesized factors, which were the four diabetes self-management aspects: diet, exercise, blood glucose testing, and foot care. All the items show factor loadings above the lower cut-off value, .30. Eigenvalues of each of the components extracted is more than the value of 1.0, which is considered to represent identifiable factors (Tabachnick & Fidell, 2007). According to the results, the total variance explained by all factors was 76.00. The Cronbach’s alpha varied from .74 to .89, which was considered acceptable. The inter-item correlations were high in each subscale except for items measuring specific diet (.22 for items S3 and S4,). This was also reported by Toobert et al. (2000) in their review of seven different studies. From the EFA analysis, the SDSCA scale revealed four constructs (i.e., diet, exercise, blood glucose testing, and foot care) distinct from each other in terms of factor loadings
142
and theoretically important. Thus, it was determined that four aspects of diabetes self-management, namely diet, exercise, blood glucose testing, and foot care, would be assessed separately, rather than examining a total score for selfmanagement. This result was in line with other researchers who agree that different aspects of self-management should be assessed separately because of the multidimensional nature of self-management (Toobert & Glasgow, 1994; Toobert et al., 2000; Weijman et al., 2005b).
143
Table 5.2 Results of Exploratory Factor Analyses for SDSCA Scale (Australia-Based Sample)
Item Diet: S1 How many of the last SEVEN DAYS have you followed a healthful eating plan S2 On average, over the past month, how many DAYS PER WEEK have you followed your eating plan? S3 On how many of the last SEVEN DAYS did you eat five or more servings of fruits and vegetables? S4 On how many of the last SEVEN DAYS did you eat high fat foods such as fried food or full-fat dairy products? Exercise: S5 On how many of the last SEVEN DAYS did you participate in at least 30 minutes of physical activity? (Total minutes of continuous activity, including walking). S6 On how many of the last SEVEN DAYS did you participate in a specific exercise session (such as swimming, walking, biking) other than what you do around the house or as part of your work? Blood sugar testing: S7 On how many of the last SEVEN DAYS did you test your blood sugar? S8 On how many of the last SEVEN DAYS did you test your blood sugar the number of times recommended by your health care provider? Foot care: S9 On how many of the last SEVEN DAYS did you check your feet?
1
Factor loading, λ 2 3
4
.84
.81
.56
.74
.91
.87
.93
.92
.86
144
S10
On how many of the last SEVEN DAYS did you inspect the inside of your shoes?
Cronbach’s alpha, α
.89
.77
.80
.89
.74
.22 to .84
.67
.81
.60
% of variance explained
23.22
18.91
17.42
16.45
% of cumulative variance explained
23.22
42.14
59.55
76.00
Inter-item correlations
Diabetes quality of life (DQoL)-satisfaction. Table 5.3 presents the factor loadings for the 15 items in the satisfaction scale for the Australia-based sample. It shows that the satisfaction scale consists of two constructs named as Factor 1S and Factor 2S. The items in each construct were further examined. In general, items in Factor 1S measured satisfaction in diabetes self-management, medical check-ups and treatment, and burden placed on family. Thus, Factor 1S was named “treatment, management, and burden to family”. For Factor 2S, items measured general healthy life in terms of physical activity, social life, and psychological well-being. Factor 2S was named “general life health”. All the items show factor loadings above the lower cut-off value, .30 and total variance explained was 49.59. The reliability test revealed that Cronbach’s alpha was .82 and .83 for Factor 1S and Factor 2S respectively, which were considered acceptable.
145
Table 5.3 Results of Exploratory Factor Analyses for DQoL-Satisfaction Scale (AustraliaBased Sample) Item Factor 1S: Treatment, management, burden to family QS1 QS2 QS3 QS4 QS5 QS6
How satisfied are you with the amount of time it takes to manage your diabetes? How satisfied are you with the amount of time you spend getting checkups? How satisfied are you with the time it takes to determine your sugar level? How satisfied are you with your current treatment? How satisfied are you with the flexibility you have in your diet? How satisfied are you with the burden your diabetes is placing on your family?
Factor loading 1 2
.80 .72 .76 .69 .61 .63
Factor 2S: General life health QS7 QS8 QS9 QS10 QS11 QS12 QS13 QS14 QS15
How satisfied are you with your knowledge about your diabetes? How satisfied are you with your sleep? How satisfied are you with your social relationships and friendships? How satisfied are you with your sex life? How satisfied are you with your work, school, and household activities? How satisfied are you with the appearance of your body? How satisfied are you with the time you spend exercising? How satisfied are you with your leisure time? How satisfied are you with life in general?
Cronbach’s alpha, α
.41 .61 .72 .71 .70 .68 .58 .57 .72
.82
.83
% of variance explained
26.18
23.41
% of cumulative variance explained
26.18
49.59
146
Diabetes quality of life (DQoL)-impact. Table 5.4 presents the factor loadings for the 19 items in the impact scale of the DQoL for the Australia-based sample. It shows that the impact scale consists of two constructs named as Factor 1I and Factor 2I. Items in each factor were further examined. Items in Factor 1I measured the impact of diabetes on physical health, psychological well-being, social life, and self-esteem. Thus, Factor 1I was named “general life health”. Items in Factor 2I measured self-efficacy related to physical activities. Thus, Factor 2I was named “self-efficacy”. All the items show factor loadings above the lower cut-off value, .30, except for item QI3 and item QI16.Total variance explained was 42.65 per cent after excluding QI3 and QI16. The reliability test revealed that Cronbach’s alpha was .81 and .82 for Factor 1I and Factor 2I respectively, which were considered acceptable levels of internal consistency for the two factors.
147
Table 5.4 Results of Exploratory Factor Analyses for DQoL-Impact Scale (Australia-Based Sample) Item Factor 1I: General life health QI1 QI2 QI3 QI4 QI5 QI6 QI7 QI8 QI9 QI10 QI16 QI17 QI18 QI19
How often do you feel pain associated with the treatment for your diabetes? How often are you embarrassed by having to deal with your diabetes in public? How often do you have low blood sugar? How often do you feel physically ill? How often does your diabetes interfere with your family life? How often do you have a bad night’s sleep? How often do you find your diabetes limiting your social relationships and friendships? How often do you feel good about yourself? How often do you feel restricted by your diet? How often does your diabetes interfere with your sex life? How often do you tell others about your diabetes? How often have you been teased because you have diabetes? How often do you feel that because of your diabetes you go to the bathroom more than others? How often do you find that you eat something you shouldn’t rather than tell someone that you have diabetes?
Factor loading 1 2
.54 .58 a .60 .60 .64 .63 .57 .41 .52 a .40 .46 .50
Factor 2I: Self-efficacy QI11 QI12 QI13 QI14 QI15
How often does your diabetes keep you from driving a car or using a machine (e.g. a typewriter)? How often does your diabetes interfere with your exercising? How often do you miss work, school, or household duties because of your diabetes? How often do you find yourself explaining what it means to have diabetes? How often do you find that your diabetes interrupts your leisure-time activities?
Cronbach’s alpha, α
.74 .69 .82 .59 .76 .81
.82
% of variance explained
21.85
20.80
% of cumulative variance explained
21.85
42.65
Note. a = Items with factor loading < 0.3 were deleted from the analysis. Then, EFA was rerun and the final results are presented in this table
148
Confirmatory Factor Analysis (CFA) Confirmatory factor analyses were conducted to further investigate the validity and reliability of the DQoL, which measures the satisfaction and impact experienced by T2DM people in this study. Confirmatory factor analyses were not conducted on SDSCA due to the small number of items per factor (i.e., four items for diet, two items for exercise, two items for blood glucose testing, and two items for foot care). The EFA has showed that the SDSCA consists of four meaningful factors with high inter-correlation, thus the four factors (variables) were used in the subsequent analyses in the present thesis. Various fit indices were used as discussed in the method chapter to examine the overall fit of the scales. L. Hu and Bentler (1999) suggested reporting two types of indices. Other researchers (Hoyle & Panter, 1995; Raykov, Tomer, & Nesselroade, 1991) suggested reporting more than two fit indices. In this study, models were empirically estimated with six fit indices. There were ratio of chi-squre to degrees of freedom (χ2/df), standardised root mean square residual (SRMR), root mean square error of approximation (RMSEA), goodnessof-fit index (GFI), comparative fit index (CFI), and Tucker-Lewis Index (TLI). In the higher-order factor analysis, the structure coefficient of the factors was inspected for discriminant validity. Then, construct reliability of the improved model and the variance extracted were estimated. Diabetes quality of life (DQoL)-satisfaction. In the EFA, I identified two factors for the satisfaction scale named “treatment, management, and burden to family” (Factor 1S) and “general life health” (Factor 2S). The first confirmatory models were for a one-factor model for Factor 1S and Factor 2S
149
from the satisfaction scale. The initial measurement model for Factor 1S is illustrated in a path diagram (Figure 5.3). All the fit indices for this initial measurement model were within the acceptable fit range.
Figure 5.3. DQoL-satisfaction Factor 1S measurement model (Australia-based sample). The initial measurement model for satisfaction on the Factor 2S was first specified using a graphic. Figure 5.4 shows the initial measurement model of the Factor 2S. The model fit indices in Figure 5.4 indicate that the initial measurement model of Factor 2S did not fit well. The majority of the fit indices were not within the acceptable fit range. Therefore, the problematic items were identified via the analysis output to improve the model. The problematic items were examined closely in terms of the wording and their importance in the questionnaire. Problematic items QS10 and QS12 were omitted one at a time from the path and the model was retested.
150
Figure 5.4. DQoL-satisfaction Factor 2S measurement model (Australia-based sample).
After removing the problematic items, the measurement model was improved. The fit indices of the improved model are shown in Figure 5.5. All the fit indices were within the range of recommended values for acceptable fit.
151
Figure 5.5. Improved DQoL-satisfaction Factor 2S measurement model (without QS10, QS12, Australia-based sample). Theoretically, the satisfaction scale consists of items QS1 to QS15 as discussed earlier. Thus, a second CFA test was conducted to confirm that Factor 1S and Factor 2S are correlated and are both significant indicators of satisfaction. The second confirmatory models were for a two-factor model of satisfaction dimension. When the latent variables (i.e., Factor 1S and Factor 2S) from the second confirmatory models are correlated at a noteworthy level, a higher-order factor is hypothesized as an explanation of the correlations that exists amongst the first-order factors. The second-order model illustrated in Figure 5.6 was an extension of the first-order CFA model for Factors 1S and Factor 2S (see figure 5.3 and 5.5). In this model, the two lower-order Factor 1S and Factor 2S were specified as indicators of a single higher-order construct of satisfaction. Figure 5.6 presents the standardized parameter estimates of the second-order satisfaction model for
152
the overall sample. The analysis in Figure 5.6 revealed that the majority of the fit indices were not within the range of recommended values for acceptable fit.
Figure 5.6. DQoL-satisfaction second order level measurement model (Australiabased sample). Figure 5.7 shows the second-order model of satisfaction dimensions after improvements had been carried out that were suggested by the CFA analysis and had adequate theoretical support. These improvements involved deleting the problematic item QS7 from the path diagram and co-varying two pairs of measurement error variances. Figure 5.7 shows that all the fit indices of the improved measurement model were within the recommended acceptable fit. The results of making these changes were that these 13 items showed that the empirical data fitted the improved second order level satisfaction measurement model.
153
Figure 5.7. Improved DQoL-satisfaction second order measurement model (Australia-based sample). An inspection of the structure coefficient in Table 5.5 indicates that the two factors display discriminant validity. Fornell and Larker’s (1981) formulas were used to calculate the construct reliability and variance extracted estimate for the second-order level satisfaction measurement model (refer to Appendix O for the calculation). The construct reliability and variance extracted estimates for the second-order level satisfaction measurement model were 0.78 and 0.64 respectively. The construct reliabilities and variance extracted estimates were in the acceptable range as recommended by Fornell and Larker (1981; construct reliabilities > .60, variance extracted estimates > .50). Therefore, this measure is
154
considered valid and reliable and it is considered that it provides reasonable basis for theoretical testing. Table 5.5 Structure Coefficients for DQoL-Satisfaction Factor 1S and Factor 2S (AustraliaBased Sample) Items
QS1 QS2 QS3 QS4 QS5 QS6 QS8 QS9 QS11 QS13 QS14 QS15
Satisfaction Factor 1S Factor Pattern .75 .71 .61 .63 .65 .56 0a 0a 0a 0a 0a 0a
Structure coefficient .75 .71 .61 .63 .65 .56 .32 .47 .40 .28 .43 .55
Satisfaction Factor 2S Factor Pattern 0a 0a 0a 0a 0a 0a .50 .75 .64 .45 .67 .87
Structure coefficient .47 .45 .39 .40 .41 .35 .50 .75 .64 .45 .67 .87
Note. a = Parameters fixed at reported levels in specifying the model. Structural coefficients are in boldface.
Thus, the final measurement model for satisfaction consists of items QS1, QS2, QS3, QS4, QS5, QS6, QS8, QS9, QS11, QS13, QS14, and QS15. The majority of the items were retained and the items were considered a good fit for the satisfaction measurement model. Diabetes quality of life (DQoL)-impact. EFA identified two constructs for impact subscale named as general life health (Factor 1I) and self-efficacy (Factor 2I). Figure 5.8 shows the dimensions of the Factor 1I model and also the model fit indices. Half of the fit indices for Factor 1I model did not meet the criteria. The items were further investigated to seek improvement of the model.
155
As suggested by CFA analysis, QI17and QI18 were identified as problematic items. Thus both items were omitted one at a time from the path and the model was retested. The measurement model was improved.
Figure 5.8. DQoL-impact Factor 1I measurement model (Australia-based sample). Figure 5.9 shows the construct dimension of impact for Factor 1I after the improvement was made. Figure 5.9 reveals that all the fit indices were improved and the majority of the fit indices were within the range of recommended values for acceptable fit.
156
Figure 5.9. Improved DQoL-impact Factor 1I measurement model (without QS17 and QS18, Australia-based sample). The initial measurement model of impact for Factor 2I is illustrated in Figure 5.10. All the fit indices reached the recommended acceptable fit value.
Figure 5.10. DQoL-impact Factor 2I measurement model (Australia-based sample). The two separate factor models of impact were further investigated for combination in a one-factor model representing impact, using a second-order model, illustrated in Figure 5.11. In this model, the two lower-order impact
157
factors, namely impact Factor 1I and impact Factor 2I were specified as indicators of a single higher-order construct of impact. Figure 5.11 shows the standardized parameter estimates of the second-order impact model for the Australia-based sample. The fit indices in Figure 5.11 reveal the majority of the fit indices did not reach the recommended fit value. Items were further investigated to seek improvement of this second-order level model.
Figure 5.11. DQoL-impact second order level measurement model (Australiabased sample). Figure 5.12 shows the diagram of impact for the second-order level dimension after improvement suggested by CFA analysis in AMOS, which had adequate theoretical support. This change involved co-varying the measurement error variance for items QI11 and QI13. All the fit indices in Figure 5.12 were improved and χ2/df , RMSEA, SRMR, and CFI reached recommended fit value.
158
The results of making these changes were that these 15 items showed that the empirical data fitted the improved second-order level impact measurement model.
Figure 5.12. Improved DQoL-impact second order level measurement model (Australia-based sample). An inspection of the structure coefficients in Table 5.6 indicates that the two factors display discriminant validity. Fornell and Larker’s (1981) formulas were used to calculate the construct reliability and variance extracted estimate for the second-order level impact measurement model (refer to Appendix O for the calculation). The construct reliability and variance extracted estimates for second-order level impact measurement model were 0.88 and 0.78 respectively. The construct reliabilities and variance extracted estimates were in the acceptable range as recommended by Fornell and Larker (1981; construct reliabilities > .60,
159
variance extracted estimates > .50). Therefore, this measure is considered valid and reliable and it is proposed that it provides a reasonable basis for theoretical testing. Table 5.6 Structure Coefficients for DQoL-Impact Factor 1I and Factor 2I (AustraliaBased Sample) Items
QI1 QI2 QI4 QI5 QI6 QI7 QI8 QI9 QI10 QI19 QI11 QI12 QI13 QI14 QI15
Impact Factor 1I Factor Structure Pattern coefficient .52 .52 .60 .60 .64 .64 .76 .76 .40 .40 .77 .77 .45 .45 .49 .49 .44 .44 .41 .41 a 0 .38 0a .53 a 0 .59 0a .40 0a .68
Impact Factor 2I Factor Structure Pattern coefficient 0a .39 a 0 .46 0a .49 a 0 .58 0a .30 0a .59 a 0 .34 0a .37 a 0 .33 0a .31 .50 .50 .69 .69 .77 .77 .52 .52 .89 .89
Note. a = Parameters fixed at reported levels in specifying the model. Structural coefficients are in boldface.
Thus, the final measurement model for impact consists of items QI1, QI2, QI4, QI5, QI6, QI7, QI8, QI9, QI10, QI11, QI12, QI13, QI14, QI15, and QI19. The majority of the items were retained and items were considered fit for the impact measurement model.
160
Descriptive Analysis The means and standard deviations of diabetes knowledge, attitude to T2DM, self-management of T2DM, and QoL of T2DM among participants from Australia-based sample are presented in Table 5.7, Table 5.8, and Table 5.9. For subscales of the SDSCA, which measures self-management of T2DM, and subscales of the DQoL, which measures QoL of T2DM, all mean and SD values were derived from items that were in the final models derived from the measurement model assessment previously described. Total scores of the final items identified in the measurement model assessment for self-management of T2DM and subscales of QoL of T2DM were computed based on the suggested formula discussed in the Method chapter (Chapter 4). These values are used in the following sections in this chapter examining the Australia-based sample. Diabetes Knowledge and Attitudes to Type 2 Diabetes Mellitus The participants based in Australia had a mean score of 61.65 out of 100 (SD=19.61) on the diabetes knowledge scale (DKN). Table 5.7 illustrates the mean score of diabetes knowledge by gender in the Australia-based sample. Australia’s male participants had lower average scores of diabetes knowledge compared to female participants. The attitude to T2DM total score for Australia-based participants was 63.39 (SD=11.68). Table 5.7 shows that the total score of attitude to T2DM of females was slightly higher than the score of males.
161
Table 5.7 Percentage Scores for Diabetes Knowledge and Attitudes to T2DM (AustraliaBased Sample) Diabetes Knowledge
Attitudes to T2DM
Gender
M
SD
Range
M
SD
Range
Female
66.20
17.62
13 - 93
64.16
11.25
34 – 88
Male
59.31
20.21
7 - 93
62.99
11.90
36 - 92
Overall
61.65
19.61
7 - 93
63.39
11.68
34 – 92
Self-Management of Type 2 Diabetes Mellitus The self-management of T2DM practices in a 7-day scale among participants in Australia are presented in Table 5.8. The self-management scores ranged from 0 days (without practices in any day of the week) to 7 days (practices every day of the week). Table 5.8 shows that participants from Australia based sample generally scored higher in diet and blood glucose testing compared to exercise and foot care. Both genders were similar in their number of days of practising self-management of T2DM in all four areas, but males selfmanaged little less across all four areas. Table 5.8. Self-Management of T2DM Practice Score Measured out of 7-days a Week (Australia-Based Sample) Gender
Self-management (M, SD) diet
Exercise
Blood glucose testing
Foot care
Female
4.91 (1.41)
3.50 (2.10)
4.95 (2.48)
3.67 (2.50)
Male
4.77 (1.64)
3.56 (2.34)
4.80 (2.50)
3.23 (2.61)
Overall
4.82 (1.56)
3.54 (2.26)
4.85 (2.49)
3.38 (2.58)
162
Quality of Life of Type 2 Diabetes Mellitus Individuals with T2DM with a higher level of QoL of T2DM will have a higher score on satisfaction and will have a lower score on impact. It was found that the mean score of satisfaction and impact was approximately equal for both genders. The participants in the Australia-based sample had a mean score of 67.99 on a scale with a maximum of 100 (SD = 17.94) for satisfaction and 28.85 per 100 (SD= 16.26) for impact. This indicates that participants generally experienced high satisfaction for and low impact of QoL of T2DM. Table 5.9 demonstrates the mean scores of satisfaction and impact for both genders. There was very little difference in satisfaction levels between females and males. However, the male participants experienced slightly higher impact with mean score of 30.00 (SD = 16.29) than female participants with mean score of 26.63 (SD = 16.06). Table 5.9 QoL of T2DM Score in Percentage Measured by Satisfaction Scale and Impact Scale of DQoL (Australia-Based Sample) Gender
Satisfaction %
Impact %
M
SD
Range
M
SD
Range
Female
67.82
17.92
14.58 - 100.00
26.63
16.06
1.67 - 71.67
Male
68.08
17.99
10.42 - 100.00
30.00
16.29
1.67 - 73.33
Overall
67.99
17.94
10.42 - 100.00
28.85
16.26
1.67 - 73.33
163
Correlation Analysis The correlation analysis explores the relationship between the five factors of the QoL scale presented by satisfaction and impact. These factors were duration of diabetes since diagnosis, age of participants, diabetes knowledge, attitudes to T2DM, and the four self-management of T2DM practices, which are diet, exercise, blood glucose testing, and foot care. Pearson’s product-moment correlation coefficients were calculated to determine the strength of the relationship between these variables. Table 5.10 presents the results of the correlation analysis in the overall Australia-based sample. The significant correlation coefficients range from r = -.12 to r = -.61. The strongest correlations were shown between satisfaction and impact which measure QoL of T2DM. The higher was the levels of satisfaction of individuals in regard to their QoL of T2DM, the lower was the impact of QoL of T2DM. Therefore, the correlation between satisfaction and impact is negative. Strong correlations were also shown between attitudes to T2DM and impact of QoL of T2DM and between attitudes to T2DM and satisfaction with QoL of T2DM. A moderate correlation was also shown between self-management of T2DM of diet and satisfaction with QoL of T2DM. There were significant negative correlations between age and diabetes knowledge, and duration of diabetes since diagnosis and self-management of T2DM of exercise. Rafique et al. (2006) also reported a negative relationship between age and diabetes knowledge, where younger participants scored higher in diabetes knowledge than those who were older. This information will be used as a guide to determine the initial path model, especially paths associated with diet, exercise, blood glucose testing, and
164
foot care. In general, studies have shown the existence of a relationship between diabetes self-management and knowledge (Kamel et al., 1999) and attitude (Greene et al., 1991). The methods for assessing self-management are diverse and some studies have assessed diabetes self-management as a total score, rather than individual diabetes self-management practices (Brownlee-Duffeck et al., 1987; Cerkoney & Hart, 1980). However, the EFA result in this study has shown SDSCA is a multidimensional scale that measures four important aspect of diabetes self-management. Therefore, each aspect is assessed separately in this study. Although, some intervention studies reported that exercise is related to QoL (Glasgow, Ruggiero, et al., 1997; D. W. Smith & McFall, 2005), research on the inter-relationship between other aspects of diabetes self-management with other variables (e.g., diabetes knowledge, attitude, QoL) is still limited. Research examining the relationship between the duration of diabetes since diagnosis and other variables in this study is also limited. Therefore, correlation analysis is useful to determine the initial path relationships for variables, such as diet, exercise, blood glucose testing, foot care, and duration of diabetes since diagnosis, with other variables in that are included in the path analysis.
165
Table 5.10 Descriptive Statistics and Correlations of Variables (Australia-Based Sample) Variable
Mean 55.84
SD 11.10
since diagnosis
11.91
9.01
3. Diabetes knowledge
61.65
19.61
4. Attitude to T2DM
63.39
11.68
5. SM-Diet
4.82
1.56
6. SM-Exercise
3.54
2.26
4.85
2.49
8. SM-Foot care
3.38
2.58
9. QoL-Satisfaction
67.99
17.94
10. QoL-Impact
28.86
16.26
1. Age
1 1
2 .294**
3 -.155**
4 .083
5 .156**
6 -.064
7 .153**
8 .170**
9 .209**
10 -.055
1
.205**
-.055
.048
-.119*
.205**
.173**
.009
.094
1
.252**
.083
.090
.154**
.095
.022
-.015
1
.233**
.091
.107
.020
.522**
-.525**
1
.290**
.380**
.248**
.406**
-.246**
1
.023
.111
.285**
-.168**
1
.211**
.206**
-.014
1
.136*
.023
1
-.608**
2. Duration of diabetes
7. SM-Blood glucose testing
Note. * p < .05, ** p < .01; SM = self-management of T2DM
1
166
Hypotheses for Path Analysis In path analysis, the components of self-management of T2DM, which are diet, exercise, blood glucose testing and foot care, and the components of QoL of T2DM, which are satisfaction and impact, serve as individual variables in the path model. In order to answer the main hypotheses and research questions proposed in Chapter 3, the sub-hypotheses are formulated to address each path relationship, including path arrows pointing toward each component of selfmanagement of T2DM and QoL of T2DM, reflecting the predicted direction of each relationship. In order to achieve a parsimonious model, rather than a complex model with all possible path arrows pointing toward each component of self-management of T2DM and QoL of T2DM, the result of correlation analysis was used to screen out the not significant path relationships. Hypotheses stated in Chapter 3 are expanded into their associate sub-hypotheses. All research questions in Chapter 3 are posited into hypotheses for path analysis. The following section explains the sub-hypotheses derived from the main hypotheses and research questions, which involve each component of self-management of T2DM and QoL of T2DM. Extension of Hypotheses and Research Questions Hypothesis 1 is: “H1: Greater knowledge about diabetes leads to more positive attitudes to T2DM.” This hypothesis is not posited into sub-hypotheses as it involves two main variables only which are diabetes knowledge and attitudes to T2DM. Hypothesis 2 is: “H2: Greater knowledge about diabetes leads to more regular self-management of T2DM practices.” From the correlation analysis,
167
diabetes knowledge is significantly correlated with blood glucose testing but not other component of self-management of T2DM. This hypothesis is posited into a hypothesis that links diabetes knowledge and blood glucose testing, which will be treated as individual main variables in the path model. Therefore, only one hypothesis is posited and named as H2. : H2: Greater knowledge about diabetes leads to more regular self-management of T2DM in terms of blood glucose testing. Hypothesis 3 is: “H3: More positive attitudes to T2DM lead to more regular self-management of T2DM practices.” Among the four components of self-management of T2DM, only component of diet in self-management of T2DM was significantly correlated with attitudes to T2DM. Therefore, H3 is posited into a hypothesis that links attitudes and diet which are treated as individual main variables. Thus, only one hypothesis is posited and named as H3: hypothesis H3: More positive attitudes to T2DM lead to more regular selfmanagement of T2DM in terms of diet. Hypotheses 4 is: “H4: More regular self-management of T2DM practices lead to higher levels of QoL of T2DM.” The components of diet and exercise in self-management of T2DM were significantly correlated with both satisfaction and impact in QoL of T2DM. However, the component of blood glucose testing and foot care in self-management of T2DM was significantly correlated with satisfaction only in QoL of T2DM. Therefore, H4 is posited into sub-hypotheses that link each component of self-management of T2DM and QoL of T2DM. The sub-hypotheses associated with H4 are:
168
Sub-hypothesis H4a: More regular self-management of T2DM in terms of diet leads to higher levels of satisfaction related to QoL of T2DM. Sub-hypothesis H4b: More regular self-management of T2DM in terms of diet leads to lower levels of impact related to QoL of T2DM. Sub-hypothesis H4c: More regular self-management of T2DM in terms of exercise leads to higher levels of satisfaction related to QoL of T2DM. Sub-hypothesis H4d: More regular self-management of T2DM in terms of exercise leads to lower levels of impact related to QoL of T2DM. Sub-hypothesis H4e: More regular self-management of T2DM in terms of blood glucose testing leads to higher levels of satisfaction related to QoL of T2DM. Sub-hypothesis H4f: More regular self-management of T2DM in terms of foot care leads to higher levels of satisfaction related to QoL of T2DM. Hypotheses 5 is: “H5: Greater knowledge about diabetes leads to higher levels of QoL of T2DM.” Diabetes knowledge was not significantly correlated with either satisfaction or impact in QoL of T2DM. Therefore, this hypothesis was excluded from the initial path model. Hypothesis 6 is “H6: More positive attitudes to T2DM lead to higher levels of QoL of T2DM.” Attitudes to T2DM was significantly correlated with both satisfaction and impact in QoL of T2DM. Therefore, H6 is posited into subhypotheses that link attitudes to both components of QoL of T2DM, which are satisfaction and impact. The sub-hypotheses associated with H6 are: H6a: More positive attitudes to T2DM lead to higher levels of satisfaction related to QoL of T2DM.
169
H6b: More positive attitudes to T2DM lead to lower levels of impact related to QoL of T2DM. The research questions that link the extraneous variables to main variables were also reviewed and hypotheses were proposed to answer the research questions. The first research question was, “R1: Is longer duration of diabetes since diagnosis associated with greater knowledge about diabetes?” In order to answer this research question, the associated hypothesis was formulated to provide a statistical assessment that is required for answering the research question. The first research question is assessed by testing the hypothesis that makes the link between duration of diabetes since diagnosis and diabetes knowledge. The hypothesis associated with R1 is: HRI: Longer duration of diabetes since diagnosis is associated with greater knowledge about diabetes. The second research question was, “R2: Is longer duration of diabetes since diagnosis associated with more regular self-management of T2DM practices?” In order to answer this research question, the associated subhypotheses were formulated. The components of exercise, blood glucose testing, and foot care in self-management of T2DM were significantly correlated with duration of diabetes since diagnosis. Therefore, R2 is posited into sub-hypotheses that link diabetes knowledge and the components of exercise, blood glucose testing, and foot care in self-management of T2DM. There is a negative correlation between duration of diabetes since diagnosis and exercise.The relationship between duration of diabetes since diagnosis and exercise for the
170
sub-hypothesis (HR2a) is stated to predict that these variables are negatively associated. The sub-hypotheses associated with R2 are namely: HR2a: Longer duration of diabetes since diagnosis is associated with less regular self-management of T2DM in terms of exercise. HR2b: Longer duration of diabetes since diagnosis is associated with more regular self-management of T2DM in terms of blood glucose testing. HR2c: Longer duration of diabetes since diagnosis is associated with more regular self-management of T2DM in terms of foot care. The third research question was, “R3: Do younger people with T2DM display greater knowledge about diabetes?” In order to answer this research question, the associated hypothesis was formulated. The research question R3 is assessed by testing the hypothesis that makes the link between age and diabetes knowledge. The hypothesis associated with R3 is: HR3: Younger people with T2DM display greater knowledge about diabetes. The fourth research question was, “R4: Do older people with T2DM display greater regularity of self-management of T2DM practices?” In order to answer this research question, the associated sub-hypotheses were formulated. The components of diet, blood glucose testing, and foot care in self-management of T2DM were significantly correlated with age. Therefore, R4 is posited into sub-hypotheses that link age and the components of diet, blood glucose testing, and foot care in self-management of T2DM. The sub-hypotheses associated with R4 are: HR4a: Older people with T2DM display greater regularity of self-management of T2DM in terms of diet.
171
HR4b: Older people with T2DM display greater regularity of self-management of T2DM in terms of blood glucose testing. HR4c: Older people with T2DM display greater regularity of self-management of T2DM in terms of foot care. Results of Assessment of the Path Models This section illustrates the relationships between the studied variables in a path diagram and presents the results of the overall path model assessment for the Australia-based sample. All the research hypotheses were tested using path analysis. A Parsimonious model is aimed at and preferred in path analysis (Byrne, 2001). Variables with significant correlation were identified. The specific hypotheses for each path arrow in Figure 5.13 are listed in Table 5.11. The specific hypotheses were tested in path analysis using SPSS and AMOS statistical packages.
172
Table 5.11 Hypotheses (Initial Structural Model/ Model 1, Australia-Based Sample) H1 H2 H3 H4a H4b H4c H4d H4e H4f H6a H6b HRI HR2a HR2b HR2c HR3 HR4a HR4b HR4c
Greater knowledge about diabetes leads to more positive attitudes to T2DM. Greater knowledge about diabetes leads to more regular selfmanagement of T2DM in terms of blood glucose testing. More positive attitudes to T2DM lead to more regular selfmanagement of T2DM in terms of diet. More regular self-management of T2DM in terms of diet leads to higher levels of satisfaction related to QoL of T2DM. More regular self-management of T2DM in terms of diet leads to lower levels of impact related to QoL of T2DM. More regular self-management of T2DM in terms of exercise leads to higher levels of satisfaction related to QoL of T2DM. More regular self-management of T2DM in terms of exercise leads to lower levels of impact related to QoL of T2DM. More regular self-management of T2DM in terms of blood glucose testing leads to higher levels of satisfaction related to QoL of T2DM. More regular self-management of T2DM in terms of foot care leads to higher levels of satisfaction related to QoL of T2DM. More positive attitudes to T2DM lead to higher levels of satisfaction related to QoL of T2DM. More positive attitudes to T2DM lead to lower levels of impact related to QoL of T2DM. Longer duration of diabetes since diagnosis is associated with greater knowledge about diabetes. Longer duration of diabetes since diagnosis is associated with less regular self-management of T2DM in terms of exercise. Longer duration of diabetes since diagnosis is associated with more regular self-management of T2DM in terms of blood glucose testing. Longer duration of diabetes since diagnosis is associated with more regular self-management of T2DM in terms of foot care. Younger people with T2DM display greater knowledge about diabetes. Older people with T2DM display greater regularity of selfmanagement of T2DM in terms of diet. Older people with T2DM display greater regularity of selfmanagement of T2DM in terms of blood glucose testing. Older people with T2DM display greater regularity of selfmanagement of T2DM in terms of foot care. Figure 5.13 presents the hypothesised path model which shows the path
relationships and their direction between the variables: age, duration of diabetes
173
since diagnosis, diabetes knowledge, attitude to T2DM, self-management of T2DM components: diet, exercise, blood glucose testing, foot care, and QoL of T2DM: satisfaction, and impact. The symbols of z1 to z8 are the error terms for each dependent variable (with arrow pointing toward), which are defined for path analysis purposes.
174
Figure 5.13. Initial hypothesised structural model (Australia-based sample).
175
The hypothesized model (Model 1) was tested using path analysis. The fit indices were examined to find out the goodness-of-fit of Model 1. Table 5.12 provides a comparison of the most commonly reported goodness-of-fit statistics in research. Table 5.12 Hypothesised Structural Model 1 Fit Indices (Australia-Based Sample) Fit indices
Result of fitness
χ2/df = 8.698
Not acceptable fit
RMSEA = .163
Not acceptable fit
SRMR = .115
Not acceptable fit
GFI = .870
Not acceptable fit
CFI = .627
Not acceptable fit
TLI = .355
Not acceptable fit
In evaluation of each of the 19 path relationships of Model 1, which are illustrated in Figure 5.14, I found that some path relationships were not significant. Path analysis of the output information for Model 1 revealed that the majority of the fit indices were not within the range of recommended values for acceptable fit. The model estimates were re-examined. Non-significant relationships between variables (or path estimated parameters) that did not contribute to the model were removed and the model was re-tested. The nonsignificant paths which were removed were pathways linking exercise to impact, blood glucose testing to satisfaction, and foot care to satisfaction in QoL of T2DM.
176
Figure 5.14. Hypothesised structural Model 1 with standardised regression weights (Australia-based sample).
177
The estimated parameters for path relationships were reduced to 16 and the revised model (Model 2) was tested for goodness-of-fit (see Figure 5.15). Path analysis of the output information for Model 2 (see Table 5.13) revealed that the majority of the fit indices were not within the range of recommended values for acceptable fit. Table 5.13 Hypothesised Structural Model 2 Fit Indices (Australia-Based Sample) Fit indices
Result of fitness
χ2/df = 8.029
Not acceptable fit
RMSEA = .156
Not acceptable fit
SRMR = .119
Not acceptable fit
GFI = .868
Not acceptable fit
CFI = .620
Not acceptable fit
TLI = .411
Not acceptable fit
Model 2 was re-examined. Some modification indices from path analysis output were suggested as part of the analysis investigation.
178
Figure 5.15. Hypothesised structural Model 2 with standardised regression weights (Australia-based sample).
179
The first modification index suggested some path relationships should be added into the model to improve the fit indices. Adequate theoretical support was identified to investigate the new path relationships suggested by the modification index. These paths were added into the model one at a time and the model was retested each time a new path was added. Any variable that did not contribute as a significant predictor was also removed from Model 2. Thus, foot care was removed from the model as it did not contribute as a predictor for DQoL variables. Arrows that represent correlations between duration of diabetes since diagnosis and age, satisfaction and impact, diet and exercise, diet and blood glucose testing were introduced to the final path model. Table 5.14 reveals that Model 3 fit indices were within the range of recommended values for acceptable fit. Table 5.14 Respecified Structural Model 3 Fit Indices (Australia-Based Sample) Fit indices
Result of fitness
χ2/df = 1.471
Acceptable fit
RMSEA = .040
Acceptable fit
SRMR = .042
Acceptable fit
GFI = .985
Acceptable fit
CFI = .987
Acceptable fit
TLI = .966
Acceptable fit
Figure 5.16 illustrates that a total of 18 paths (relationships between variables) were significant and theoretically important.
180
Figure 5.16. Respecified structural Model 3 with standardised regression weight (Australia-based sample).
181
Table 5.15 below presents the level of support for each of the hypotheses tested within the Australia-based sample. As illustrated in Table 5.15, 14 hypotheses were supported and five were not supported. Table 5.15 Level of Support for the Hypotheses (Australia-Based Sample) Hypothesis H1 H2 H3 H4a H4b H4c H4d H4e H4f H6a H6b HR1 HR2a
Greater knowledge about diabetes leads to more positive attitudes to T2DM. Greater knowledge about diabetes leads to more regular self-management of T2DM in terms of blood glucose testing. More positive attitudes to T2DM lead to more regular selfmanagement of T2DM in terms of diet. More regular self-management of T2DM in terms of diet leads to higher levels of satisfaction related to QoL of T2DM. More regular self-management of T2DM in terms of diet leads to lower levels of impact related to QoL of T2DM. More regular self-management of T2DM in terms of exercise leads to higher levels of satisfaction related to QoL of T2DM. More regular self-management of T2DM in terms of exercise leads to lower levels of impact related to QoL of T2DM. More regular self-management of T2DM in terms of blood glucose testing leads to higher levels of satisfaction related to QoL of T2DM. More regular self-management of T2DM in terms of foot care leads to higher levels of satisfaction related to QoL of T2DM. More positive attitudes to T2DM lead to higher levels of satisfaction related to QoL of T2DM. More positive attitudes to T2DM lead to lower levels of impact related to QoL of T2DM. Longer duration of diabetes since diagnosis is associated with greater knowledge about diabetes. Longer duration of diabetes since diagnosis is associated with less regular self-management of T2DM in terms of exercise.
Level of support Supported
Supported Supported
Supported Supported
Supported Not supported Not supported Not supported Supported Supported Supported
Supported
182
HR2b HR2c HR3 HR4a HR4b HR4c
Longer duration of diabetes since diagnosis is associated with more regular self-management of T2DM in terms of blood glucose testing. Longer duration of diabetes since diagnosis is associated with more regular self-management of T2DM in terms of foot care. Younger people with T2DM display greater knowledge about diabetes. Older people with T2DM display greater regularity of selfmanagement of T2DM in terms of diet. Older people with T2DM display greater regularity of selfmanagement of T2DM in terms of blood glucose testing. Older people with T2DM display greater regularity of selfmanagement of T2DM in terms of foot care.
Supported Not supported Supported Supported Supported Not supported
Table 5.16 presents the standardised regression weights for each path relationship tested in Model 3, including the estimates, standard errors, critical ratios, and the p value for the critical ratio for the 18 significant paths in Model 3. Significant associations are indicated by asterisks in Figure 5.16. Evaluation of output information revealed an overall acceptable fit of the framework to the data. Review of regression weights in Table 5.16 shows significant and important theoretical relationships. This result identified relationships among the variables as discussed earlier in Chapter 3. It also identified additional relationships that were statistically significant and theoretically important.
183
Table 5.16 Final Path Model Hypotheses (Australia-Based Sample) H
Pathways
H1
Attitude
H2
Blood glucose Testing Diet QoL-satisfaction QoL-impact QoL-satisfaction QoL-satisfaction QoL-impact Diabetes knowledge Exercise
H3 H4a H4b H4c H6a H6b HR1 HR2a HR2b
A
Blood glucose testing Diabetes knowledge Diet Blood glucose testing QoL-impact
A
Attitude
A A
Attitude QoL-satisfaction
HR3a HR4a HR4b
Diabetes knowledge diabetes knowledge Attitude Diet Diet Exercise Attitude Attitude Duration of diabetes Duration of diabetes Duration of diabetes Age Age Age Blood glucose testing Duration of diabetes Age Age
β
SE
CR
p
.317
.035
5.447