Paper 1650-2014
Risk factors and outcome of spinal epidural abscess from incident hemodialysis patients from the United States Renal ); 2
model sea = accesstype/dist = binomial link=log; estimate 'Catheter' accesstype -1 1 0 / exp; estimate 'Graft' accesstype -1 0 1 / exp; run; Notice the slight difference in adjusted relative risk for access type when all other covariates are included in the calculation from unadjusted relative risks. We are also interested in finding out whether there is a survival advantage for SEA HD patients. To obtain the survival curve and median residual lifetime for SEA HD patients, the following codes are used. proc lifetest data=usrds.sea_analysis method=lifetable plots=survival; time surv_time*death(0); strata sea; run; Figure 1. Survival curve for HD patients stratified by SEA diagnosis
Figure 1 shows that more than half of the HD patients without SEA are still living at the end of study, whereas the median residual lifetime for SEA HD patients is roughly 2.5 years. Table 4. Sample hazard ratios of co-morbidities regarding death in SEA HD patients Crude HR 95% HR CI Adjusted HR 95% HR CI SEA 1.368 1.237 1.513 1.179 1.064 1.305 Age of Patient ≥ 65 years 2.541 2.513 2.569 2.254 2.228 2.28 < 65 years Ref Ref Here is an example of obtaining crude hazard ratios for two age groups. proc phreg data=usrds.sea_analysis; class age65 (ref=first); model surv_time*death(0) = age65 / rl ties = efron; run; When all significant crude hazard ratios have been included in the full model, we then used backward elimination to reduce the full model into a reduced model where all covariates are statistically significant. Notice the drop in hazard ratio for SEA. When calculating crude HR, SEA patients are 36.8% more likely to die than non-SEA patients. However, when adjusted for all other demographic characteristics and clinical characteristics, hazard ratio for SEA is no longer as big. The new adjusted hazard ratio means that SEA patients are 17.9% more likely to die than non-SEA patients.
CONCLUSION SEA is an uncommon but serious complication of HD. Previously such condition has been difficult to study due to its rarity. We were able to conduct the largest cohort study to date and find risk factors and comorbidities of SEA on HD patients using various procedure steps in SAS® 9.3.
REFERENCES
Wong, S., Daka, S., Pastewski, A., Kyaw, W., Chapnick, E., Sepkowitz, D. Spinal Epidural Abscess in 3
Hemodialysis Patients: A Case Series and Review. Clinical Journal of the American Society of Nephrology 6 (2011): 1495-1500. Darouiche, R. Spinal Epidural Abscess. The New England Journal of Medicine 335 (2006): 2012-20. Atwater S, Colombo R, Baer S, Kheda M, Chebrolu P, Baulkmon E, Kintziger K, Nahman Jr NS. Risk factors for bacteremia in incident hemodialysis patients. J Amer Soc Nephrol, 23:983A, 2012. US Renal Data System, USRDS 2009 Annual Data Report: Atlas of Chronic Kidney Disease and End-Stage Renal Disease in the United States. Bethesda, MD: National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases; 2009.
CONTACT INFORMATION Your comments and questions are valued and encouraged. Contact the author at: Chan Jin Department of Biostatistics and Epidemiology Georgia Regents University th 1120 15 Street Augusta GA 30912
[email protected] SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration. Other brand and product names are trademarks of their respective companies.
4