Effects of Methylphenidate on Attention Deficits After Traumatic Brain [PDF]

Authors: John Whyte, MD, PhD Tessa Hart, PhD Monica Vaccaro, MS Patricia Grieb-Neff, MA Anthony Risser, PhD Marcia Polansky, ScD H. Branch Coslett, MD

Brain Injury

Research Article

Affiliations: From the Moss Rehabilitation Research Institute, Albert Einstein Healthcare Network, Philadelphia, Pennsylvania (JW, TH, MV, PGN, AR, HBC); the Department of Rehabilitation Medicine, Thomas Jefferson University, Philadelphia, Pennsylvania (JW, TH); the Department of Symptom Research, University of Texas, M. D. Anderson Cancer Center, Houston, Texas (AR); the Department of Biostatistics, Drexel University, Philadelphia, Pennsylvania (MP); and the Department of Neurology, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania (HBC).

Disclosures: Supported, in part, by grant R01NS39163 from the National Institute on Neurological Diseases and Stroke, National Institutes of Health, and grant R24HD39621 from the National Center for Medical Rehabilitation Research, National Institute on Child Health and Human Development, National Institutes of Health.

Correspondence: All correspondence and requests for reprints should be addressed to John Whyte, MD, PhD, Moss Rehabilitation Research Institute, 1200 West Tabor Road, Suite 213, Philadelphia, PA 19141. 0894-9115/04/8306-0401/0 American Journal of Physical Medicine & Rehabilitation Copyright © 2004 by Lippincott Williams & Wilkins DOI: 10.1097/01.PHM.0000128789.75375.D3

Effects of Methylphenidate on Attention Deficits After Traumatic Brain Injury A Multidimensional, Randomized, Controlled Trial ABSTRACT Whyte J, Hart T, Vaccaro M, Grieb-Neff P, Risser A, Polansky M, Coslett HB: Effects of methylphenidate on attention deficits after traumatic brain injury: A multidimensional, randomized, controlled trial. Am J Phys Med Rehabil 2004;83:401– 420. Objective: To evaluate the effects of methylphenidate on a variety of aspects of attention, ranging from laboratory-based impairment measures to caregiver ratings and work productivity, in individuals after traumatic brain injury. Design: A total of 34 adults with moderate to severe traumatic brain injury and attention complaints in the postacute phase of recovery were enrolled in a 6-wk, double-blind, placebo-controlled, repeated crossover study of methylphenidate, administered in a dose of 0.3 mg/kg/dose, twice a day. A wide range of attentional measures was gathered weekly, including computerized and paper-and-pencil tests of attention, videotaped records of individual work in a distracting environment, real-time observational scoring of attentiveness in a classroom environment, and caregiver and clinician rating scales of attentiveness. Participants also attempted to guess their drug condition each week. Data from the first ten participants were used for pilot purposes, to develop attentional factors for composite scoring, and to identify attentional dimensions suggestive of a treatment effect for independent replication. The remaining 24 participants’ results were used to confirm potential treatment effects seen in the pilot sample, using Wilcoxon’s signed-ranks test on composite factor scores and individual variables. Results: A total of 54 dependent variables were reduced to 13 composite factors and 13 remaining individual variables. Of the 13 attentional factors, five showed suggestive treatment effects in the pilot sample. Of these, three showed statistically significant treatment effects in the replication sample: speed of information processing (effect sizes, ⫺0.06 to 0.48; P ⬍ 0.001), attentiveness during individual work tasks (effect sizes, 0.15– 0.62; P ⫽ 0.01), and caregiver ratings of attention (effect sizes, 0.44 – 0.50; P ⫽ 0.01). Of the individual variables, four showed suggestive treatment effects in the pilot sample, but only one showed significant treatment effects in the replication sample: reaction time before errors in the Sustained Attention to Response Task (effect size, 0.20; P ⫽ 0.03). No treatment-related improvement was seen in divided attention, sustained attention, or susceptibility to distraction. None of the variables showed suggestive or definite negative treatment effects. Effect sizes for those performance measures positively affected by methylphenidate were in the small to medium range and included both impairment and activity level measures. Improvements in processing speed did not seem to come at the expense of accuracy. Conclusions: Methylphenidate, at 0.3 mg/kg/dose, given twice a day to individuals with attentional complaints after traumatic brain injury, seems to have clinically significant positive effects on speed of processing, caregiver ratings of attention, and some aspects of on-task behavior in naturalistic tasks. Further research is needed to identify the optimal dose and to extend these findings to less carefully selected individuals. Key Words:

June 2004

Traumatic Brain Injury, Methylphenidate, Attention Deficits

Effects of Methylphenidate After TBI

401

A

ttentional impairments are among the most common cognitive deficits observed after traumatic brain injury (TBI)1. Like attention itself, these impairments are multifaceted in nature; they can include lowered vigilance, increased distractibility, slowed processing time, and impaired ability to direct or allocate attention across different facets of the environment2. Attentional impairments may also contribute to memory deficits and impairments of executive functions, the two other predominant neuropsychological disorders observed in moderate-to-severe TBI3. The importance of attention in everyday life, its diversity in normal and defective expression, and the typical mechanisms of injury sustained during the closed head injuries that result in moderate-to-severe TBI all create the need for comprehensive examinations to specify the types of attentional problems that are present at any given time for an individual.4 Although moderate-to-severe TBI accounts for only roughly 20% of all documented TBIs,5 survivors require significant acute medical and rehabilitation inpatient care and postacute rehabilitation services. These survivors often present with changes in independent daily functioning, psychosocial relationships, and academic and vocational status.6 Of course, attentional impairments are not alone in determining functional outcome, but addressing them is essential to successful rehabilitation efforts. Attempts to improve acquired attentional problems have included both cognitive-behavioral remediation efforts and psychopharmacologic intervention, but neither has been conclusively shown to be of therapeutic benefit. Thus, there is a critical need for carefully controlled evaluations of treatment efficacy to support evidence-based approaches to intervention,7 for both cognitive-remediation8 and drug-based9 clinical-trial research. The lack of evidence-based treatment recommendations stems from a

402

Whyte et al.

number of methodologic challenges. Although clinicians, caregivers, and sometimes TBI survivors themselves cite attention deficits as key cognitive problems, it has been difficult for researchers to reach consensus on the nature of these problems.10 Although it is widely agreed that TBI results in slowed information processing,11 it is not clear that this, alone, can account for the range of the attention complaints reported.12 Other problems, including susceptibility to distraction,13 inefficient attentional selection,14 impaired sustained attention,12 impaired divided attention,15 and increased rates of off-task behavior in naturalistic tasks,16 have been reported in some but not all studies. This difficulty is not confined to the clinical study of attention after TBI. Indeed, controversy remains in cognitive psychology about how to define attention, how to subdivide it, and how best to measure its constituents17. When considering attention deficits from the perspective of rehabilitation, it also becomes important to consider the level of conceptual analysis of the attention measures of interest.18 That is, some research tasks focus primarily at the impairment level, others assess manifestations of attention in real-world tasks (activity level), and still others may address the effect of attention deficits on an individual’s ability to resume premorbid roles (participation level). Related to this point, pharmacologic treatments target the impairment level, whereas approaches to cognitive remediation may target impairment, activity levels, or a combination. Thus, measuring the effect of such treatments may involve a tradeoff between measurement sensitivity (which may be maximal when using impairment-level outcome measures) and clinical relevance (which is more related to activity and participation measures).18 Finally, shortcomings in experimental design have limited the interpretation of the results of a number of studies. Pre–

post designs without an untreated control group are rarely interpretable because of spontaneous recovery or practice effects.19 Small parallel group studies run the risk of both type II error, due to lack of statistical power, and type I error, due to between-group differences in patient characteristics, incompletely controlled by randomization. Methylphenidate (MP) and other psychostimulants have been used to treat attention deficit hyperactivity disorder (ADHD) at least since 1938, when it was discovered that benzedrine led to increased interest and effort in school for affected children20. A recent review and metaanalysis covering 40 yrs of research on MP treatment in ADHD concluded that despite great variability across methodology and quality of studies, MP consistently affects the inattention, excessive behavior, and impulsivity that constitute the core clinical features of the disorder21. Stimulants such as MP decrease hyperactivity and increase on-task behavior in ADHD22. Compared with the amphetamines to which it is chemically related, MP may affect mental functions more prominently than motor activity23. Recent studies have sought to isolate the cognitive functions most affected by MP in children with attention disorders. For example, on the Stroop Color Word Task, MP does not seem to improve interference control (e.g., ability to suppress prepotent word naming responses in the interference condition)24,25, but the drug does affect speed of “effortful” processing (e.g., color-naming alone) in a highly positive way.25 MP has also been used to treat ADHD in adults, although the research on this use is far less extensive. Variations in treatment response among adults may relate to differences in diagnostic criteria used, presence of co-morbidities, and doses. However, a rigorously designed crossover study showed highly

Am. J. Phys. Med. Rehabil.

●

Vol. 83, No. 6

significant treatment effects among adults with ADHD.26 The success of MP in children with attention and behavior disorders has prompted researchers and clinicians to use the drug for similar disorders seen in TBI and other acquired brain injuries. Other types of posttraumatic problems also have been addressed with MP. For example, the drug has been reported to be effective for the treatment of posttraumatic narcolepsy27 and difficulties with anger control28. Less conclusive effects have been suggested for overall adjustment29 and behavior30. However, the majority of published studies have tested the effects of MP on cognition. Recently, we published a detailed review and critique of studies on MP and other psychostimulants in TBI,9 including a previous randomized, controlled trial from our laboratory.31 As discussed in more detail in that review, studies are difficult to compare because they vary widely in degree of experimental rigor, statistical power, acuity and severity of TBI in the sample, age of the sample, number and type of dependent measures, and other key factors that affect the interpretation of results. The weight of the evidence thus far, however, seems to indicate that MP does not have a very strong effect on aspects of attention such as maintaining vigilance over time (sustained attention)29,31,32 or resistance to distraction.29,31 MP does seem to have a beneficial effect on processing speed, as measured in tasks of reaction time and speeded decision-making.31–34 In our previous controlled investigation,31 we demonstrated that this effect of MP was due to a specific effect on cognitive speed rather than speed of motor performance. The purpose of this study was to extend previous research findings on the effect of MP on attention deficits resulting from TBI. Because of the multifaceted nature of attention, as discussed above, we chose to study June 2004

the effects of MP on a wide range of attention measures, including those at impairment and activity levels, assessed over a 6-wk period of intensive data collection. Recognizing that the need to conduct statistical testing of the drug’s effect in each of many areas increases the risk of type I error, we sought to develop composite scores reflecting several different variables. We also designed the study to include a pilot phase, followed by an independent replication in a larger sample, of any drug effects that seemed promising. This complex design was undertaken to achieve two broad research objectives. One was to examine the clinical efficacy of MP for attention-related deficits associated with TBI in a more systematic and controlled fashion than has been undertaken in previous research. The other, equally important, objective was to clarify which attentional domains are responsive to MP treatment in this population so that future clinical trials can focus on a smaller number of empirically validated primary outcomes.

METHODS Participants Participants were recruited from a variety of sources, with most referrals coming from the outpatient clinical programs at the Drucker Brain Injury Center at MossRehab Hospital and the research registry maintained at MossRehab, of inpatients from MossRehab Hospital, Magee Rehabilitation Hospital, and Bryn Mawr Rehabilitation Hospital, all located in the Philadelphia area. We also advertised the study through the local public transportation system, a free newspaper distributed in a variety of places in the area, and with the brain injury associations of Pennsylvania and New Jersey. Participants who were interested in the study were interviewed and their medical records reviewed. To be

included, participants had to be between the ages of 16 and 60 yrs and to have a history of nonpenetrating TBI of at least moderate severity at least 3 mos before enrollment. Severity level was defined by significant and well documented loss or alteration of consciousness after injury (i.e., lowest Glasgow Coma Scale score of ⬍12 [Although many of the Glasgow Coma Scale scores used to determine eligibility were postresuscitation scores, in some instances, the initial Glasgow Coma Scale score was obtained from later secondary records in the absence of the acute care flow sheets, such that the exact timing of the score was unclear.] or prospectively documented posttraumatic amnesia of ⬎1 hr) or focal abnormality on a neuroimaging study that was attributable to traumatic injury. Participants needed to be able to perform tasks for 10 –15 mins semi-independently and be available to attend the project five full days per week for 6 wks. A subjective complaint of attention difficulties by the participant, treating clinician, or caregiver was also required. Potential participants were excluded if they were currently hospitalized; if they were pregnant or likely to become pregnant; if they had a history of premorbid neurologic disease, psychosis, major affective disorder, mental retardation, or ADHD; if they were taking psychotropic medications other than anticonvulsants; if they were currently abusing alcohol or recreational drugs, or if their history of using these substances was significant enough to place them at risk of long-term neurologic effects. Individuals were also excluded for behavioral problems or for impairments in vision, hearing, or motor function that were severe enough to preclude participation in the research tasks. Participants or their involved caregivers (depending on the participant’s cognitive capacity) provided informed consent. During a period of several years,

Effects of Methylphenidate After TBI

403

we reviewed ⬎1,500 individuals for possible participation in this study, as shown in Figure 1. Possible participants who were involved in clinical programs could often be excluded based on the available records. Those who were encountered as a result of public advertising were generally interviewed first to obtain releases for their medical records if they seemed eligible through interview. However, ascertainment of some of the eligibility criteria (e.g., history of ADHD or substance abuse, availability for 6 wks) generally required an interview, and this, in turn, required explaining the basics of the study as an introduction. Thus, there were a number of potential participants whose final eligibility could not be ascertained either because they did not return telephone calls or because they declined after hearing about the study but before answering the interview questions. A total of 39 participants consented to the study and were randomized to a drug order (see below). Two subsequently withdrew due to subjective adverse effects (one while receiving placebo), and one was dropped because of possible exacerbation of baseline hypertension. These three participants had insufficient data for analysis. Two additional participants were dropped because of suspected substance abuse during data collection. Although one of these was dropped near the end of data collection, the data were not included in the analysis because of the potential confounding effects of the substance abuse. Thus, the sample of 34 individuals who completed the study was clearly a highly selected subgroup of those who were theoretically eligible. Participants who met all eligibility criteria and provided informed consent underwent additional data collection to characterize the sample and establish baseline levels of function. Duration of posttraumatic amnesia, as an additional indicator of injury severity, was assessed retrospectively using the method developed by McMillan et al.35

404

Whyte et al.

Figure 1: Flowchart detailing the 1,500 individuals reviewed for possible participation in the study. MP, methylphenidate; P, placebo.

In brief, this involved comparing participants’ memory for events after trauma against temporal milestones abstracted from the medical record. Six participants were unable to provide relevant information, either because their injury was so long before the assessment that they could not adequately recall relevant events or because they had communication impairments that precluded participating effectively in the interview. For all participants, the Disability Rating Scale36 was scored by project staff based on participant and family interview at the time of enrollment as a measure of the level of current disability. Premorbid intellectual

status was estimated using demographic and occupational history data.37 Before starting the trial, participants also underwent a battery of neuropsychological tests focused primarily on attention and frontal/executive function; these results will be presented separately. Table 1 summarizes the demographic, injury and disability characteristics of the 34 participants. As is typical of research on moderate to severe TBI, the majority were male. Except for the exclusion of elderly participants, the distributions of age, education level, and ethnicity were typical of TBI samples in metropolitan ar-

Am. J. Phys. Med. Rehabil.

●

Vol. 83, No. 6

TABLE 1 Characteristics of participants (n ⫽ 34) Demographic characteristics Age, yrs Mean Range Sex, n (%) Male Female Ethnicity, n (%) White African American Hispanic Education, yrs Mean Median Range Estimated premorbid intelligence quotient Mean Range Injury and disability characteristics Participants with medical record documentation of, n (%): Loss/alteration of consciousness Abnormality on neuroimaging Estimated duration of posttraumatic amnesia (n ⫽ 28), n (%) ⬍2 wks 2 wks to ⬍1 mo 1 mo to ⬍3 mos 3 mos to ⬍6 mos ⬎6 mos Disability Rating Scale Mean Range

37 20–55 29 (85) 5 (15) 22 (65) 10 (29) 2 (6) 12.7 12 9–18

101 85–119

22 (65) 23 (68)

1 (4) 4 (14) 7 (25) 10 (36) 6 (21) 4 (moderate) 1 (mild) to 8 (moderate to severe)

Time postinjury Median Range

eas. In keeping with the exclusion for premorbid cognitive limitations, estimated premorbid intellectual statuses all fell within the average to high-average range. It may be seen in Table 1 that the typical participant was several years postinjury, exhibited a moderate level of functional disability, and reported a significant degree of posttraumatic amnesia. Table 1 also shows the proportion of participants for whom objective records verified at least moderate neurologic severity by significant alteration of consciousness or neuroimaging abnormalities. For two participants, medical records were not availJune 2004

3.2 yrs 4 mos to 34.2 yrs

able for documentation of severity. However, their histories, levels of disability, and motor impairment (one had spastic quadriparesis, one a dense hemiparesis) left no doubt as to the severity of injury.

Procedure The data collection for this study was integrated into a day activities program infrastructure, provided free of charge to participants. A rehabilitation clinician ran a “research classroom” for three to four participants at a time, from approximately 9:30 a.m. to 3:30 p.m., Monday through

Friday. Figure 2 shows a schematic diagram of the weekly schedule, including daily/weekly data collection elements described below. This classroom environment served several purposes. It provided a positive group environment to help maintain participation for the required 6 wks, which was an incentive for both participants lacking other productive activity and for their caregivers, who were frequently seeking structured activity for their family members. It also provided a standardized environment for behavioral ratings of attentiveness in individual work and group activities. Finally, it provided a milieu for participants when they were not being tested individually in the laboratory and allowed assessment of vital signs, side effects, and behavioral self-ratings. The classroom was a single, large room with a small office for the classroom clinician attached. The room was filled with large work tables, a whiteboard, and materials for educational and recreational activities. The clinician designed activities that could be tailored to the functional levels of individual participants and that required attention for their performance. These activities were not designed to meet individual clinical goals but were intended to be engaging and functionally relevant. The clinician was trained not to intervene to increase attentiveness through cueing or reinforcement because the intent of this study was specifically to assess drug effects. Participants were involved in the study for six consecutive weeks. They received a participant fee of $50/wk plus transportation expenses. During half of this time, they received MP, the study medication; during the other half, they took an identical-appearing placebo, with the two conditions alternating weekly. Half of the participants started with MP followed by placebo and the other started with placebo followed by MP (i.e., MP, placebo, MP, placebo, MP, placebo or placebo, MP, placebo, MP, placebo,

Effects of Methylphenidate After TBI

405

Figure 2: Weekly schedule for research classroom. CFQ/RSAB, Cognitive Failures Questionnaire/Rating Scale of Attentional Behavior. MP). A pharmacist who did not have day-to-day involvement in the study prepared the medications weekly and randomly determined the order of administration so that the conditions were balanced and concealed from the research staff, participants, and caregivers. Randomization occurred in blocks of four participants (two to each treatment order). The dose of MP was 0.3 mg/kg/dose, rounded to the nearest 2.5 mg, administered twice a day at approximately 8:30 a.m. and 12 noon, Monday through Saturday. The average dose of MP (given twice a day) was 24.4 mg (range, 15– 40). Sunday was a washout day before the crossover to the opposite condition (Fig. 2). Participants were screened for changes in vital signs for the first 2 wks of the study and for the occurrence of adverse events throughout the study. The screening procedure and the effects of the study drug on vital signs and adverse events have been reported separately.38 Once a week, each participant also guessed what drug he or she was receiving and provided a certainty rating for

406

Whyte et al.

this judgment to help ascertain the success of the double-blinding procedure. Before receiving the medication, participants were interviewed about their use of tobacco, caffeine, alcohol, prescription and over-the-counter drugs, and recreational drugs. Because of the length of the project, it was not realistic to expect participants to refrain from caffeine, tobacco, or alcohol, so they were instructed to maintain a consistent pattern of use of these substances. While in the study, participants filled out daily forms asking about their use to ensure regularity. They were also asked about sleep, pain, illness, medications taken, and any stressful events they might have experienced that day. The intent was to determine that the participants’ state was consistent enough from day to day so that variations in performance were most likely caused by the medication and not some other event or state. Research staff called participants in the early morning, at home, to remind them to take their medication. At arrival, they were asked

whether and when they had taken it. If a participant missed a dose, the tasks for that morning were skipped and made up at a later date under the correct medication condition. Midday doses were self-administered by participants under the observation of research staff. As depicted in Figure 2, participants attended four 1-hr sessions in the classroom each day, during which time observational data on on-task behavior were collected in both individual and group tasks. A 90-min period of unconstrained time plus a lunch break occurred each day, during which no classroom data were collected. Unconstrained time allowed the participants to work on activities of personal interest as a supplement to scheduled activities. Participants were taken from the classroom for an hour each day to do a variety of individual attention assessment tasks. During each of the 6 wks, the same tasks and assessments were performed, providing three samples of performance on each measure in the MP and placebo conditions and allowing a direct compari-

Am. J. Phys. Med. Rehabil.

●

Vol. 83, No. 6

son of the participants’ performance in the two conditions. Data were collected using laboratory measures of information processing, controlled simulations of attentional demands of everyday tasks, and ratings of attentiveness in everyday life. The order of tasks was the same each week, and tasks were performed at the same time each day for a given participant. Morning sessions began 90 mins after the morning dose, and afternoon sessions began 90 mins after the noon dose. When participants were unable to attend on a given day, the individual testing session for that day was skipped and made up at the end of the 6-wk period. The pharmacist supplied replacement doses of the appropriate medication for that date. Missing classroom data were not made up.

Attention Tasks Several of the following tasks, marked with an asterisk (*), were presented on an Apple PowerMac G3. All but one of these were programmed in PsyScope. For these tasks, participants were seated approximately 30 inches from the screen. Responses were recorded and timed using a PsyScope button box. For those tasks with two asterisks (**), the duration of stimulus presentation was individualized before study onset to minimize ceiling and floor effects. Other tasks described below used paper-and-pencil responding, rating scales, or observational data. **Sustained Arousal and Attention Task 50/50. The Sustained Arousal and Attention Task 50/5012 is a simple go/no-go visual reaction time task lasting approximately 20 mins. Participants were presented with 161 stimuli with an average interstimulus interval of 6 secs and instructed to respond as quickly and accurately as possible to the targets and to ignore the foils. The target rate for this task was 50%. The data derived from this task were divided into five blocks. The initial block of performance defined June 2004

the vigilance level, and robust regression slopes were calculated across all of the blocks to determine the vigilance decrements in reaction time, response bias, and accuracy. **Sustained Arousal and Attention Task 20/80. This modification of the above-described task used a 20% target rate, based on the hypothesis that individuals’ attentional lapses might be more frequent if fewer responses were required. Scoring on this task was identical to the original sustained arousal and attention task. *Speed/Accuracy Tradeoff Task. This task differed from the two described above in that the presented stimuli were visually degraded so that it was more difficult to distinguish between targets and foils, but they remained exposed for much longer (2000 msecs), allowing participants to prioritize speed or accuracy of responding. Specifically, we were interested in learning whether drug-induced improvements in speed came at the expense of accuracy or whether both speed and accuracy were positively affected by treatment. This task had the same number of trials and interstimulus interval as the two listed above and was scored in the same fashion. Unfortunately, a technical problem was identified after data collection was complete. This task was sufficiently more difficult than the sustained attention tasks described above, so response times were much slower for most subjects, but response times of ⬎2000 msecs were not recorded by the computer. Thus, some subjects seemed to have very low response rates, and their response time distributions were truncated at 2000 msecs. Consequently, this task was not incorporated into the overall analyses described below. However, those subjects with useable data were selected to specifically assess whether any effects of the drug

on speed occurred at the expense of accuracy. **Distraction Task. This task was a variation on one previously published.13 The same stimuli were presented with a 50% target rate with one addition: brightly colored shapes moved rapidly up and down either above or below the stimuli. These distracters occurred at various intervals just before or after stimulus presentation. There were seven intervals ranging from 200 msecs before the presentation of the stimulus to 200 msecs after stimulus presentation. In addition, a no-distracter condition was included, which allowed direct comparison of distracted and undistracted performance within the same task. There were 192 trials in this task, with an average interstimulus interval of 6 secs, organized into 12 blocks of 16 trials each (one target and one foil at each distracter interval, in random order). Variables from this task were derived by subtracting the scores in the no-distracter condition from the participant’s mean scores achieved at five of the distracter intervals at which a performance decrement was noted. This yielded distracter effect scores for accuracy, response bias, and reaction time. *Choice Reaction Time Task. The participant was required to press a number key on the numeric keypad of the computer keyboard in response to presentation of a digit on the screen. There were three blocks in this experiment, in which there were two, four, or six digits displayed. Response keys not being used during a given block were covered to make clear to the participants that the number of choices varied. The number stimulus remained on the screen until the participant responded. After each response, there was a 1000-msec blank interval, followed by an 800msec presentation of a central fixation cross. This was followed by the

Effects of Methylphenidate After TBI

407

presentation of the potential digit set in a random array for 1000 msecs. Finally, an auditory tone signaled the simultaneous brightening of one of the digits to identify it as the target. A regression slope of mean reaction time for correct responses vs. the natural logarithm of the number of choices available served as the index of processing speed, based on evidence that the steepness of this slope is an index of speed of mental processing (i.e., steeper ⫽ slower).39 *Dual Task. This program, written in C by Daniel Kimberg,15 was designed to assess the ability to divide attention across two concurrent tasks. The participant was first asked to respond as quickly as possible, by pressing the spacebar on the computer keyboard, to dots that appeared in pseudo-random locations on the computer monitor. The mean response time for these trials defined the participant’s baseline speed. The participant was then asked to listen to and repeat strings of digits read to them by the research assistant while responding to the dots as before. Digit string length was determined for each participant by initial digit span testing. The dual task decrement score is the difference between mean response time in the single vs. dual task versions. *Sustained Attention to Response Task. In this task,40 225 digits (numbers 1–9) are presented randomly, one at a time, in the center of the computer monitor, during a 4.5-min interval. The participant must press a response key as quickly as possible for all but one of the digits (3) and must withhold a response for that digit. This is also a go/no-go task, but it differs from those previously described in that targets are common and foils are rare (1:9). Measures in this task are the number of omission errors, the number of commission errors, and the reaction time of responses before errors.

408

Whyte et al.

Test of Everyday Attention. This task,41 not administered by computer, includes eight subtests available in three equivalent forms (given as forms A, B, C, A, B, C on successive testing weeks). The subtests are designed to assess selective attention, sustained attention, and attentional switching by simulating naturalistic tasks, such as locating items on a map, looking for phone numbers in a directory, determining which floor to get off an elevator, and checking winning lottery numbers. Inattentive Behavior Task. In this task,42 the participant was asked to perform three tasks at a worktable, including making a collage, sorting items into their correct bowls, and working on a complex jigsaw puzzle, while being videotaped. During the session, a research assistant carried out a series of naturalistic distractions (e.g., making a phone call, playing a noisy computer game, dropping a book) on cue from taped messages delivered through a concealed earphone. The videotapes were coded at a later time on a coding workstation, which consisted of a Dell, Windowscompatible personal computer with a Pentium II processor and a Panasonic video cassette recorder. Observational Coding Software, Version 3.2, developed by Triangle Research Collaborative, Inc. (Research Triangle Park, NC), was installed on the personal computer. This software allowed the beginning, end, and duration of each off-task event, external distraction, and period of the task during which the research assistant gave directions to be identified. Offtask events were defined with respect to direction of eye gaze, as described previously.16,42 Rates of off-task behavior were calculated separately for each task as events per minute of task time. The average duration of off-task events in each session was also calculated by dividing the total off-task time by the number of off-task events. The number of items used in

the collage and number of items correctly sorted were also recorded. Interrater reliability was assessed using Cohen’s Kappa for the beginnings and endings of off-task events, external distractions, and time-out events (e.g., direction giving). Two research assistants coded videotaped records of inattentive behavior. Initially, all videotaped records were double coded, until adequate agreement, as measured by an average Kappa of ⱖ0.8, was achieved. This process was maintained for 68 records, after which records subjected to interrater agreement assessment was reduced to 33% of all records. At this point, two research assistants coded one randomly selected record for each drug condition for each participant, without knowledge of drug condition. In all, 111 of all videotaped records of inattentive behavior (54%) were double coded. Whenever the agreement between coders on a given record fell below the standard (Kappa of 0.8) on any of the six variables included in the analysis, the coders met to reconcile the differences. The reconciled data were used for the final analysis of drug effects. To minimize the potential for bias (because reconciliation typically increases, rather than decreases, the number of events noted), whenever a record was subjected to reconciliation, another session in the opposite drug condition was also reconciled, regardless of the level of agreement between the coders on that second record. Mean Kappas for videotaped records of inattentive behavior during the posttraining phase were: 0.83, 0.82, 1.0, 98, 0.91, and 0.89 for offtask begin, off-task end, distracter begin, distracter end, time-out begin, and time-out end, respectively. Classroom Attentiveness. In addition to the laboratory testing, participants were observed as they participated in classroom activities. Data collection occurred during four 1-hr sessions each day: a group activity and an in-

Am. J. Phys. Med. Rehabil.

●

Vol. 83, No. 6

dividual activity in both the morning and the afternoon. During individual sessions, the participant was asked to carry out independent work (e.g., reading a book, working on crossword puzzles, or individual craft projects) in the same room with other participants. During group sessions, activities such as playing board games, discussing current events, and participating in lectures, which required the interaction and involvement of all participants, were conducted. For both types of session, a set of rules that defined the appropriate targets of attention was laid out by the classroom therapist before the activity began. For example, in a group activity with structured turntaking, responding during someone else’s turn or failing to respond during one’s own turn were both coded as off-task behavior. During each session, a research assistant sat in the classroom to observe and code offtask behavior. The research assistant wore a vibrating watch that provided a silent cue once per minute. A random sequence of participant identification codes was preprinted on a data collection sheet. Each minute, when the watch cued, the research assistant looked at the appropriate participant and coded whether, at that moment, he or she was on task or off task according to operationalized written behavioral criteria. Data were collected on off-task behavior as measured by eye gaze, speaking, and being out of seat. However, only the eye-gaze data are reported here because scores in the other two domains were frequently at ceiling. Ontask eye gaze was defined based on whether the research participant was looking at the appropriate task materials (for individual tasks) or the taskrelevant speaker or materials (for group tasks). Two research assistants independently observed 27 classroom sessions (approximately 2% of all data collection classroom sessions) using the same vibrating watch cue-driven June 2004

time sampling and the same random order of participant observation. Interrater agreement assessment, using both Cohen’s Kappa and percentage of agreement, was conducted on the presence or absence of off-task events as indicated by direction of eye gaze. Agreement between pairs of raters coding attentiveness in classroom activities was high. Perfect agreement was achieved in 23 of 27 sessions. Considering percentage of agreement across all double-coded sessions, average agreement was 99%. Kappas were calculated for those sessions in which both on-task and off-task events were observed (n ⫽ 15) and ranged from 0.65 to 1.0, with a mean of 0.95 and median of 1.0. There were 12 sessions during which both coders agreed that no off-task events occurred, and therefore, Kappa could not be calculated. Attention Ratings. Two different ratings of the participants’ attentiveness in everyday life were gathered on a weekly basis. Different versions of the Cognitive Failures Questionnaire43 were completed weekly by the participants themselves, their chosen caregiver, and the research staff (by consensus). This measure assesses the frequency of everyday lapses in attention using a 5-point Likert-type scale, ranging from “very often” to “never.” The participants completed a 25-item version, which includes more subjective questions about perceived attentional difficulties, whereas the caregivers and research staff filled out an eight-item version. The Rating Scale of Attentional Behavior11 is a 14item, 5-point Likert-type scale designed to assess observable attentionrelated behaviors, such as restlessness and distractibility. This was completed by the participant’s caregiver and the research staff.

Data Analysis Initial Data Reduction. Performance on most of the computerized infor-

mation processing tasks was assessed in several dimensions: D' (a signal detection measure of accuracy, or ability to discriminate the target vs. the foil), yes rate (a measure of response bias, or tendency to press the response key, irrespective of accuracy), and median response time (a measure of speed of processing and responding, less sensitive to outliers than the mean, and calculated only on correct key presses). For the tasks assessing sustained attention, the above three scores were calculated on the first block (first 20% of trials) to determine initial performance, and the slope of a robust regression across the scores of all five blocks was used to represent the ability to sustain performance over the duration of the task. The relevant scores from each task were averaged across the three (or in a few cases, due to missing data, two) sessions from each drug condition, yielding a pair of scores representing the drug conditions for each participant. Assessment of the Effects of MP. As discussed above, it was thought important to assess the effect of MP on numerous aspects of attentional function to avoid missing clinically important effects on unmeasured attentional dimensions, to evaluate the drug’s effect at impairment and activity levels of analysis, and to be able to compare the drug’s profile of action with those of other drugs. This produced a total of 54 separate scores for analysis (Table 2 contains descriptions of the scores used in the final analysis). These variables assessed initial performance in a variety of tasks, as assessed by accuracy, speed, and overall response rate; deterioration in performance over time, based on these same three performance indices; impulsive responding; distraction by competing stimuli; the ability to divide attention; attentiveness as judged by ratings scales; and attentiveness as judged by direct observation of off-task behavior.

Effects of Methylphenidate After TBI

409

410

Whyte et al.

Am. J. Phys. Med. Rehabil.

●

Vol. 83, No. 6

Impulsive errors

Test of Everyday Attention, elevator counting with distraction Test of Everyday Attention, telephone search while counting Test of Everyday Attention, visual elevator Test of Everyday Attention, elevator with reversal Sustained Attention to Response Task Errors of commission Errors of omission

Composite Factors Initial accuracy Perceptually simple visual go/no-go (50% targets) Perceptually simple visual go/no-go (20% targets) Accuracy decline Perceptually simple visual go/no-go (50% targets) Perceptually simple visual go/no-go (20% targets) Initial speed Perceptually simple visual go/no-go (50% targets) Perceptually simple visual go/no-go (20% targets) Visual go/no-go task (50% targets) with salient moving distracters Choice RT Dual task Test of Everyday Attention map search Test of Everyday Attention telephone search Inattentive behavior task: work productivity Speed decline Perceptually simple visual go/no-go (50% targets) Perceptually simple visual go/no-go (20% targets) Initial response bias Perceptually simple visual go/no-go (50% targets) Perceptually simple visual go/no-go (20% targets) Visual go/no-go task (50% targets) with salient moving distracters Decline in responding Perceptually simple visual go/no-go (50% targets) Perceptually simple visual go/no-go (20% targets) Divided attention Dual task

Task

0.04

Response rate, first 32 trials Response rate, first 32 trials Response rate, nondistraction trials

No. key presses to “3” No. of misses to non-“3” No. key presses with RT RT ⬍150 msecs

0.91

0.88

0.35

Robust regression slope, median RT vs. task block Robust regression slope, median RT vs. task block

RT decrement (mean RT with digit span–mean RT during baseline) No. correct Time/target Time/switch No. correct

0.02

Median RT, first 32 trials Median RT, first 32 trials Median RT, nondistraction trials Robust regression slope: mean RT vs. log (no. of choices) Mean RT, baseline condition No. of symbols circled (2 mins) Time per correctly circled target No. of items sorted

0.85

1.00

Robust regression slope, D' vs. task block Robust regression slope, D' vs. task block

Robust regression slope, response rate vs. task block Robust regression slope, response rate vs. task block

0.07

—

—

—

0.88

—

⬍0.001

—

0.29

Replication Sample

P Value Pilot Sample

D', first block D', first block

Score

TABLE 2 Significance testing of methylphenidate effects on performance factors and individual variables

⫺0.06 0.07 0.48 0.32 0.12 0.18 0.40 0.48

Effecta Size

June 2004

Effects of Methylphenidate After TBI

411

a

RT, response time. Effect size as measured by Cohen’s D.

Visual go/no-go task (50% targets) with salient moving distracters Dual task Dual task Dual task Cognitive Failures Questionnaire—self ratings Sustained attention to response task Test of Everyday Attention—visual elevator accuracy Test of Everyday Attention—lottery Inattentive behavior

Family ratings Cognitive Failures Questionnaire Rating Scale of Attentional Behavior Staff ratings Cognitive Failures Questionnaire Rating Scale of Attentional Behavior Inattention—individual Morning classroom—individual activities Afternoon classroom—individual activities Inattentive behavior—task 1 Inattentive behavior—task 2 Inattentive behavior—task 3 Inattention—individual rate Inattentive behavior—task 1 Inattentive behavior—task 2 Inattentive behavior—task 3 Inattention—group Morning classroom—group activities Afternoon classroom—group activities Individual scores Visual go/no-go task (50% targets) with salient moving distracters Visual go/no-go task (50% targets) with salient moving distracters Visual go/no-go task (50% targets) with salient moving distracters

Task

TABLE 2 Continued

D ⫺ nondistraction trials D' with distraction ⫺ D' without distraction Response rate with distraction ⫺ response rate without distraction Median RT with distraction ⫺ median RT without distraction Measured digit span Digit strings administered Digit strings correct, % Total score Mean RT before error of commission No. correct No. correct Collage paper used

0.49 0.19 0.56 0.63 0.18 0.11 0.12 0.56 0.65

0.70 0.62 0.92

0.92

% On-task eye gaze codes % On-task eye gaze codes

0.06

% On-task eye gaze % On-task eye gaze Average duration of Average duration of Average duration of

0.26

0.85

Total score Total score

Rate of off-task episodes Rate of off-task episodes Rate of off-task episodes

0.04

Pilot Sample

Total score Total score

codes codes off-task episodes off-task episodes off-task episodes

Score

— 0.70 — — 0.66 0.03 0.33 — —

— — —

—

—

0.01

—

0.01

Replication Sample

P Value

0.20

0.28 0.62 0.22 0.15 0.47

0.44 0.50

Effecta Size

As mentioned previously, to reduce the risk of type I error, a twostage analytic strategy was employed. Data gathered from the first ten participants were used only for pilot purposes, not in the final analysis, which served as an independent replication on 24 additional participants. Data from the pilot sample were used in two ways. First, the three sessions of placebo data from these ten participants were used to develop methods for collapsing scores from multiple empirically or conceptually related tasks into fewer composite scores. Second, data from the three placebo and three MP scores for these ten participants were analyzed for possible drug effects, using a relaxed Pvalue cutoff of 0.20. Only those variables that seemed promising, based on meeting this cutoff, were subjected to confirmatory analysis, using data from the next 24 participants (the replication sample). Thus, to be defined as responsive to MP, a particular variable had to show a drug effect in the same direction in the pilot sample (with a P value of ⱕ0.20) and in the replication sample (with a P value of ⬍0.05). Each of these analyses is described in greater detail below. Development of Composite Scores. The small sample size precluded formal factor analysis as a method to arrive at composite scores. Therefore, a combination of a priori reasoning and Spearman’s correlation was used. Scores from different tasks were grouped together in a preliminary fashion, based on similarities in task demands and scoring systems. For example, a number of different tasks emphasized speeded performance, suggesting that processing speed might be a single factor to which scores from multiple tasks could contribute. Spearman’s correlation matrices were then constructed among the potential members of a given factor (based on the average of three placebo scores for each of the ten

412

Whyte et al.

pilot participants). The two scores showing the highest pair-wise correlation were then assigned to the factor. Another score that had high correlations with each of the two previous scores was then identified and added to the factor if its average correlation with each of the previous scores was at least 0.50. This process continued until no other score satisfied this criterion. Individual scores that failed to “join” a factor were retained for individual analysis. Once scores were assembled into factors, composite factor scores were developed. Again, a pair of scores for each variable was computed for each participant by averaging the three sessions for each drug condition. Then, the MP and placebo scores were ranked together for each variable in the factor. Because there were occasions in which data were missing for a particular task for an individual participant, the raw ranks were divided by the maximal possible rank for each variable (i.e., in most instances, the raw rank was divided by 20 [2 drugs ⫻ 10 participants] for the pilot analysis, and 48 [2 drugs ⫻ 24 participants] for the replication analysis) so that the modified ranks ranged from 0 to 1.0 in all instances. The composite score was then computed by averaging all the placebo ranks for each factor variable for each participant and averaging all the MP ranks for each factor variable for each participant, resulting in a pair of scores (MP, placebo) for each participant for a factor. A total of 13 such factors were created and assessed for drug effects. Because average ranks were used in the final analyses of drug effects, the raw scoring system used in any given task had no effect, and this data analytic approach implicitly weighted each contributing score equally in the determination of the overall rank. Participants who were missing scores for a given task were always missing scores for both drug conditions. The correction mentioned above, dividing

by the ranks by the number of participants, ensured that tasks with fewer participants’ data would not have excessive weighting (by assuming higher ranks simply because of fewer possible ranks). Preliminary Assessment of Drug Effects from the Pilot Sample. Wilcoxon signed rank test was performed separately on the pairs of composite factor scores and on the remaining individual scores from the initial ten pilot participants. These analyses were done in their exact form (as opposed to their asymptotic approximations) because of the small sample size. Those factor and individual scores in which the calculated P value was ⱕ0.20 were used to derive a priori hypotheses for the larger replication sample. Final Analysis of Drug Effects Using the Replication Sample. Composite factor scores were computed as described above on the remaining 24 participants. Drug effects were then assessed as above, using Wilcoxon signed rank test, limited to those factors and individual scores that met the P-value cutoff from the pilot sample. Asymptotic approximations were used for the replication analysis because of the larger sample size. In instances in which the results were statistically significant at an alpha level of 0.05, the data from all 34 participants were combined to calculate descriptive statistics and effect sizes to provide the most precise estimates of the magnitude of the drug effect. This combining was done only after the effect in the pilot sample was confirmed independently in the replication sample. Note that these effect sizes were computed parametrically (Cohen’s D for within-participant treatment effects ⫽ mean of individual differences between treatment conditions/standard deviation of individual differences between treatment conditions),44 despite our reliance on nonparametric inferential

Am. J. Phys. Med. Rehabil.

●

Vol. 83, No. 6

statistical testing. Thus, the effect sizes may not correspond precisely to the P values, despite identical sample sizes, because data outliers have different effects on the two calculations. Effect of Missing Data. Missing data from the individual laboratory sessions were always because of impairments that prevented a participant from performing a particular task. For example, one participant was unable to learn to perform go/ no-go tasks, despite performing many other study tasks, including the choice reaction time task, adequately. These impairments led to the participant’s omission from both drug conditions. Missing data from classroom observation occurred more frequently because most participants performed their individual laboratory sessions during one of the four daily classroom data collection sessions. Again, these participants’ data were missing from both drug conditions for a given observation time period. Some participants missed additional scattered days of classroom observation because of conflicting appointments, illness, or other conflicts, but all participants had data from each study week. These absences seemed to have been random in nature. Thus, although missing data reduced the effective sample size and, hence, statistical power, for certain variables, it should not have biased the results. Whenever the sample for analysis was incomplete, we assessed qualitatively whether the obtained P value was close to the designated threshold and, thus, whether we might have committed a type II error for that variable. This was not a significant issue.

RESULTS Assessment of the Effects of MP. The procedure described above resulted in the creation of 13 factor scores and June 2004

left 13 individual scores that were not part of any of the calculated factors. One of the individual scores, the elevator counting task from the Test of Everyday Attention, was deleted because of ceiling effects in both drug conditions. The scores that were part of each factor, the results of the pilot data analysis, and where appropriate, the results of the replication analysis and calculated effect sizes are shown in Table 2. Descriptive statistics for those variables that showed significant drug effects are shown in Table 3. As shown in Table 2, a number of factors showed no apparent drug effect in the pilot analysis and were not evaluated further. In addition, two factors that seemed promising from the pilot analysis failed to replicate. These included the factor reflecting initial accuracy on two of the computerized tasks and the factor reflecting initial response rates on three of the computerized tasks. Three factors did, however, show statistically significant drug effects in both the pilot analysis and the replication. An initial speed factor, composed of the speed scores from eight different tasks, showed a statistically significant positive effect of MP in both the pilot and replication analyses, and this was the most robust drug effect detected. The variables contributing to this factor included not only computerized reaction time tasks, but also speeded paper-and-pencil tasks, a measure of work productivity, and a measure of mental processing speed, suggesting that the effects on speed are seen across a wide range of tasks and range from impairment to activity levels. The effect sizes ranged from negligible to medium, depending on the task. Only one of the eight tasks showed an effect size in the negative direction (slower performance while receiving MP), and this effect size was very close to zero. Caregiver ratings were also significantly improved by MP, despite

the fact that caregivers’ only opportunity to observe participants during peak drug effects occurred on Saturdays. Whereas the Rating Scale of Attentional Behavior items emphasize speed, the items on the Cognitive Failures Questionnaire include a variety of types of attentional lapses, and the drug effect sizes were quite comparable for both measures. Of interest, differences in the ratings on these same measures by the project staff did not achieve statistical significance. The third factor that was positively affected by MP was on-task behavior, in both the classroom environment’s individual work sessions, and the inattentive behavior task’s videotape coding. Interestingly, this factor combines the frequency of time-sampled classroom ratings that were coded as off task with the average duration of individual off-task episodes recorded on videotape. The effect sizes again were in the small to medium range. Note that MP seemed to have a larger effect on off-task behavior in the afternoon than in the morning. This result did not seem to be caused by ceiling effects in the morning on placebo because the afternoon effect was larger even in a subset of patients whose morning and afternoon off-task behavior on placebo was matched (data not shown). The frequency of off-task episodes from the videotaped records contributed to a separate factor that was not significantly affected by the study drug. Off-task behavior in the classroom during group activities, similarly, did not show a significant drug effect. Of the 12 individual scores that did not contribute to the above factors, four met the screening cutoff for the pilot sample. However, only the reaction time before errors of commission from the Sustained Attention to Response Task showed a significant drug effect in the replication sample. On the one hand, this is consistent with the general effect of MP

Effects of Methylphenidate After TBI

413

TABLE 3 Descriptive statistics for performance variables showing significant drug effects Minimum/Mean/Median/Maximum Factor Name Initial speed

a

Task

n

Perceptually simple visual go/no-go (50% targets), response time in msecs

Perceptually simple visual go/no-go (20% targets), response time in msecs

Visual go/no-go task (50% targets) with salient moving distracters, response time in msecs no distractor condition

Family ratings

Placebo

416

390

33

804 695 1922 355

797 713 1660 421

33

775 696 1760 438

789 778 1517 455

745 663 1695 33 387 230 2769 265 449 391 1111 12.0

791 697 1739 76 510 321 2012 287 457 396 1129 10.7

Choice, response time (slope coefficient)

34

Dual task, baseline response time in msecs

33

Test of Everyday Attention map search, no. of symbols circled in 2 mins

33

Test of Everyday Attention, telephone search, time per target in secs

Methylphenidate

33

33

Inattentive behavior task, no. of items sorted

34

Cognitive Failures Questionnaire

32

Rating Scale of Attentional Behavior

32

45.8 49.3 77.7 1.87

44.8 42.3 77.0 1.94

5.73 4.82 15.25 122 484 478 916 0.00 10.89 12.33 23.00 0.00 13.48 13.67 32.33

6.31 4.75 19.84 104 458 433 856 0.00 11.99 12.83 22.00 0.00 16.13 17.5 31.67

a

Sample sizes vary from the full sample of 34 because of individual participants’ inability to perform certain tasks and the scheduling of laboratory tasks for some participants during classroom time.

on speed of performance. However, in general, accelerating response times during the Sustained Attention to Re-

414

Whyte et al.

sponse Task are interpreted as indicating an automatic, or inattentive, mode of responding.

Does MP Improve Speed at the Expense of Quality? Because of the technical problem with the speed/ac-

Am. J. Phys. Med. Rehabil.

●

Vol. 83, No. 6

TABLE 3 continued Minimum/Mean/Median/Maximum Factor Name Inattention– individual

Task Morning classroom–individual activities, % on task

Afternoon classroom–individual activities, % on task

Inattentive behavior–task 1, average duration of off-task episodes in secs

Inattentive behavior–task 2, average duration of off-task episodes in secs

Inattentive behavior–task 3, average duration of off-task episodes in secs

Individual score

Sustained Attention to Response Task, response time before commission error in msecs

curacy tradeoff task, described above, a subset of ten participants was identified whose raw response time distributions had no more than 15% of trials slower than 1750 msecs in either drug condition. This criterion was based on preliminary visual inspection of response time distributions that seemed to be complete (i.e., not abruptly truncated at 2000 msecs). Thus, these tended to be the faster participants overall. For these individuals, the drug-effect sizes on speed and accuracy were both small (0.26 and 0.21, respectively) but were both in the direction of improvement with MP. Considering the flawed data on the whole sample, the effect on speed was small (0.21) and on accuJune 2004

a

n

Methylphenidate

Placebo

26

65.3

73.5

24

93.6 97.0 99.5 79.1

92.0 96.4 100.0 62.6

34

96.5 98.3 100.0 0.42

90.8 94.6 100.0 0.47

32

2.72 1.49 14.01 0.26

3.69 1.83 26.63 0.41

30

1.25 1.16 3.73 0.25

1.47 1.19 8.59 0.72

34

1.52 1.06 4.74 259

3.48 1.90 17.14 259

400 380 680

410 390 713

racy was medium (0.52), but again, both effects are in the direction of improvement with MP. Thus, there is no evidence from this task that the faster processing associated with MP is associated with lowered accuracy or quality. Similarly, in the Sustained Attention to Response Task, individuals tend to adopt an automatic or inattentive mode of processing. As they do so, their response times on targets tend to become faster and faster until they ultimately make an error of commission to a nontarget.40 Control participants tend to slow down on their next trial, as they recognize their error. Our participants with TBI did have faster response times before their errors of commis-

sion with MP than placebo, but they did not make more errors of commission, suggesting that this more rapid responding was not evidence of lessened attentional monitoring. Success of Blinding. Participants’ average ratings of their drug condition assessments during placebo and MP weeks were compared via Wilcoxon signed rank test. The median ratings did differ between drug conditions (MP: median ⫽ 2.5, range ⫽ 1– 4; placebo: median ⫽ 3.0, range ⫽ 1– 4.67; P ⫽ 0.03, where 1 ⫽ certainty they were taking the active study drug and 5 ⫽ certainty they were taking the placebo). Although the difference between the mean MP

Effects of Methylphenidate After TBI

415

and placebo ratings is statistically significant, the actual differences are quite small, and in both conditions, the average ratings were in the direction of thinking that they were taking the active study drug. Inspection of individual data indicated that there was only one participant who was consistently accurate in his drug self-assessments.

DISCUSSION The results of this study are largely consistent with previous research from our laboratory31 and from research in other populations,25,45– 47 demonstrating that MP has reliable positive effects on performance speed across a range of tasks in a postacute adult TBI population. This effect is probably exerted primarily on cognitive speed, as judged by the improvement on the choice reaction time task, for which the motor response is identical across blocks with different numbers of choices. The magnitude of this effect varied from task to task, with small to medium effect sizes. In a naturalistic work task (i.e., object sorting), approximately 6% more work was accomplished with MP in a 15-min interval. Although there were technical problems with one of the tasks designed to assess speed/accuracy tradeoffs, it did not seem, from this task or from the Sustained Attention to Response Task, that the improvement in speed occurred at the expense of accuracy. Given that slowed processing is one of the most widely reported and disabling impairments resulting from TBI, a positive effect of MP in this domain may have considerable clinical significance. The clinical relevance of this aspect of MP is further supported by the fact that performance was improved on a wide range of tasks that shared demands on cognitive speed, from computerized laboratory tasks to simulated work activities. Improved speed of processing with MP also may have

416

Whyte et al.

contributed to the higher attentiveness ratings of caregivers. In children with ADHD, MP is reported to improve on-task behavior. No such effect was identified in our previous TBI pilot study,31 and this study provided somewhat mixed results. To interpret these effects, one must first consider the breakdown of the factors that pertain to attentiveness in the classroom and inattentive behavior task. As mentioned previously, variables that pertain to the duration of off-task episodes in the videotaped records did not correlate with the frequency of such events in the same records, but they did correlate with the proportion of off-task observations in the classroom. The videotape coding attempted to capture each off-task event that occurred, regardless of its duration, whereas the time-sampling method used in the classroom merely provided “spot checks” of whether participants were on task or off task at random moments. Thus, it is possible that longer off-task events occurring in the classroom were more likely to be captured by this sampling method. If so, this factor might more accurately be thought of as assessing the duration of these off-task episodes rather than their frequency. Under this account, it is quite possible that MP’s effects are primarily on how rapidly participants return to task, once off, rather than on the number of such events, per se. Thus, the fact that no drug effect was seen on offtask event frequency on the videotaped records would not be surprising. Although MP did seem to decrease the counts (perhaps by decreasing their duration) of off-task behavior in individual classroom sessions, no such effect was seen in group sessions. It is possible that the duration of off-task behaviors during individual work tasks is determined primarily by the participant’s internal goal state, whereas those same behaviors during group sessions may be more related to the nature and sa-

lience of distracting behaviors produced by other group members. If so, it would not be surprising for MP to have more robust effects on the former than the latter. It is also of interest that MP affected on-task classroom behaviors more strongly in the afternoon than in the morning sessions. In children with ADHD, MP seems to improve cognitive speed particularly on effortful tasks.25 Possibly, maintaining one’s attention during a mild diurnal “slump” is particularly enhanced by MP because it is more effortful than paying attention to the same activity in the morning. Of particular importance, MP produced substantial improvements in caregiver ratings of attentiveness on two different measures, an aspect of the drug not assessed in our previous research. This is somewhat surprising because one would expect the effects of MP to be waning by the time participants returned home on weekdays, and caregivers only spent 1 day/wk (Saturdays) with the participants when they were at peak drug levels. It is difficult to know precisely what changes the caregivers observed to compose these ratings. The results of the laboratory studies, however, may suggest that improved processing speed contributes to improvements in a variety of domains. Alternatively, drug effects not measured in this study, such as improved mood or decreased irritability, may have contributed to caregivers’ perceptions of positive drug benefits, although the assessment questions did not pertain to these dimensions. Alternatively, if the study was not successfully blinded, and participants and caregivers communicated about the perceived drug condition, this could have contributed to bias in the caregiver ratings. However, most participants were close to chance in guessing the correct drug condition, making this explanation less likely. The lack of drug effects on self-ratings is likely attributable to some

Am. J. Phys. Med. Rehabil.

●

Vol. 83, No. 6

combination of impairment in participants’ memory for their performance over days and weeks and to their reduced self-awareness. Regardless of its source, however, it does suggest that participants remained reasonably well blinded to their drug condition. The fact that staff ratings did not show effects of MP but that caregiver ratings did may relate to the more intimate effect on caregivers of the participants’ usual behaviors and, hence, their greater sensitivity to behavioral changes. Differences in the perception of deficits and their effect by professional staff and caregivers are often noted in the literature. The current data suggest that caregiver ratings may be particularly important in assessing real-world treatment effectiveness. Our previous study31 also found that the decline in responding seen during a sustained attention task was blunted by MP, an effect that was not replicated in this study. This difference may simply have occurred by chance, given the number of performance measures assessed in the previous study, or may be attributable to broader impairment across a range of aspects of attention at a more acute stage of recovery. The previous study also found a reduction of the effect of external distracters on measures of accuracy. However, it seemed that that result relied on faster processing of the target, before the onset of the distracter (i.e., another manifestation of faster processing with MP). In the current study, modifications in distracter timing may have eliminated that effect. Although MP’s effects were statistically and clinically significant in a number of performance domains, the effect sizes ranged from small to medium, with absolute improvements ranging from 5% to 25%. This is in contrast to a study of adults with ADHD by Spencer et al.26 in which rating scale improvements ranged from 17% to 55%. There are a numJune 2004

ber of differences between studies that might account for the differences in magnitude of treatment effects. It is certainly possible that adults with ADHD respond more dramatically to MP than do those with TBI, particularly because improvements in hyperactivity (55%) and impulsivity (54%), symptom clusters that are not prominent in TBI, were greater in the ADHD study than were improvements in inattentiveness (17%). It is also possible that the outcome measures used in the Spencer et al.26 study are more sensitive to drug effects than those used in this research. However, because of the different presentation of ADHD vs. TBI-related inattentiveness, the scales used by Spencer et al.26 would not be appropriate in this population. Finally, the difference may be attributable to differences in dosage. Spencer et al. used 1.0 mg/kg/day, divided into three doses, whereas we used 0.6 mg/kg/day (0.3 mg/kg/dose), divided into two doses. Although the daily dose administered in the Spencer et al.26 study was dramatically larger than ours, individual doses were similar (0.3 mg/kg/dose vs. 0.33 mg/kg/ dose). Given the short duration of action of MP, one would expect these different dosing regimens to affect duration of the effect more than magnitude, particularly for those shortduration tasks that were assessed near the time of peak drug levels. Nevertheless, it is possible that larger effects would have been seen in the rating scale data with three-timesper-day dosing. Given the absence of serious adverse events, it would be of interest to assess whether larger effect sizes could be obtained in patients with TBI by escalating the doses of MP used. A number of limitations in this study must be kept in mind in interpreting its findings. The study sample might be most accurately termed an “inconvenience sample” in the sense that a very large number of potential participants had to be screened to iden-

tify the 34 participants who met the inclusion and exclusion criteria and who were willing to commit 6 wks of their time. Thus, individuals who had returned to employment or were too impaired to travel, or those with premorbid neurologic deficits or current psychoactive drug treatment, were underrepresented in the sample. Moreover, because of the time elapsed between injury and enrollment for many participants, the details of their initial neuropathology were often unavailable, making it difficult to assess variations in hypoxic/ischemic injury and other factors that might have affected treatment response. On the other hand, in routine clinical practice, one is often called on to treat patients in the chronic stage whose initial histories are similarly unobtainable, requiring the physician to tailor treatment to the current functional status rather than the original injury. Although generalization to a broader population of individuals with TBI is limited by this selection and data collection process, we believe that the types of drug effects seen here can be plausibly expected in other individuals with TBI. In addition, the goal of this study was less to make definitive recommendations regarding clinical use of MP than to help understand the mechanisms by which this drug might modify cognitive function after TBI. It seems likely that the fundamental effects of MP on attentional processes are likely to be fairly general. Moreover, by identifying a smaller subset of domains in which MP effects are likely to be seen, this study should facilitate the conduct of a more streamlined clinical trial in a broader sample, with data collection limited to a few variables likely to be responsive. A larger clinical trial of this type will address more definitively the issue of generality of treatment response and whether individual patient characteristics substantially modify this response. Another limitation is the use of a single weight-adjusted MP dose for all participants rather than conducting individualized dosage titration to

Effects of Methylphenidate After TBI

417

the optimal dose for each participant or testing each participant fully on several different doses. This decision was made largely for practical reasons. In our experience, individual dosage titration is very difficult because of the substantial variability in performance at any particular dose and the presence of confounding practice effects, not to mention the uncertainty regarding the measure against which to optimize the dose. Testing each participant at two or three different doses would have lengthened the testing protocol to an infeasible degree. Thus, our goal here was to establish a restricted set of variables that can be used to assess MP response. These variables can then be used more efficiently to study the response to different drug doses. This approach cannot, however, completely rule out the possibility that other behavioral domains that showed no treatment effect at this dose might show such an effect at a different dose. Finally, the complex statistical method used may have resulted in either type I or type II error. Certainly, the fact that a given score or factor does not meet a probability level cutoff of 0.20 in a sample of ten participants does not mean that it could not show a drug effect in a larger sample. Thus, we may have sacrificed the opportunity to detect some treatment effects (particularly small ones) by failing to further assess those variables that did not meet the pilot screening cutoff. In addition, that different factors contained different numbers of raw scores may have caused variations in sensitivity to detect drug effects. To the extent that the data are variable or “noisy,” collapsing more scores into a single factor tends to improve the signal-tonoise ratio. Thus, one might expect this method to be slightly less sensitive in detecting drug effects on those factors that have fewer score “members.” Whereas it is true that the speed factor had the largest number

418

Whyte et al.

of contributing variables and the most robust treatment effect, the caregiver rating factor had only two members, yet it proved highly significant. With respect to type I error, this two-stage method of analysis makes is more difficult to determine precisely the odds that a variable could reach statistical significance by chance. However, it is worth noting that the direction of effect for all of the variables that met the screening cutoff was in the direction of MP benefit (with the exception of one speed factor that had essentially a zero effect size), an occurrence that would be extremely unlikely by chance. With these limitations in mind, the current findings do support the notion that MP has a beneficial effect on cognitive processing speed after moderate to severe TBI. This effect, and possibly other specific benefits of MP, translates not only into faster performance on laboratory measures of attention, but also the ability to attend to naturalistic tasks or return more quickly to them after lapses of attention. Although MP did not affect several facets of attention measured in the current study, there was a highly significant effect on everyday attentional behavior as assessed by relatives very familiar with the dayto-day effect of attention deficits on our participants with chronic TBI.

CONCLUSION MP, in a dose of 0.3 mg/kg twice a day, seems to have clear and consistent positive effects on speed of processing and caregiver ratings of attentiveness in a highly selected sample of individuals with moderate to severe TBI. The effects on inattentiveness are more complex and may be traceable to shorter duration of off-task behaviors without a reduction in their frequency. Effects on sustained attention were not evident, in contrast to our previous pilot study, nor did MP improve dual task performance, in contrast to a previ-

ous study of bromocriptine.48 Positive drug effects covered a range from impairment to activity and participation measures. Further clinical trials on larger and more representative samples of individuals with TBI are warranted and should benefit from these results in selecting their primary outcomes. Similarly, studies comparing different doses of MP could help shed light on the optimal dose for specific target goals.

ACKNOWLEDGMENTS We thank the many individuals who assisted in the research: Joseph Alban, Natosha Bailey, Christopher Gantz, Monica Hopson, and Vonetta Drakes managed participant recruitment, testing, and behavioral observation; Adelyn Brecher, Anne Hammond, Karen Hawkey, Andrea Laborde, Walter Lewis, Nathaniel Mayer, Jeanne Pelensky, and Rosadele Plumari all helped to identify and refer eligible participants; Kelly Card and Vivian Ly facilitated the research classroom; Thouron Smith provided volunteer assistance in the laboratory and classroom; Stephen Hinshaw suggested the model for the research classroom and provided methodological consultation; Guylaine Gagnon, Mary Kennedy, and Ronald Charno managed the blinded randomization code and the preparation of study medications; Antonella Pavese and Tracy Veramonti programmed the computerized laboratory tasks; Fei Sha and Gemma Baldon programmed the research database; and Mary Czerniak assisted in the preparation of the manuscript; and most importantly, we thank the participants and their families for investing their time and energy in this study.

REFERENCES 1. Capruso DX, Levin HS: Neurobehavioral sequelae of head injury, in Cooper PR, Golfinos JG (eds): Head Injury. New York, McGraw-Hill, 2000, pp 525–53 2. van Zomeren AH, Brouwer WH: Clin-

Am. J. Phys. Med. Rehabil.

●

Vol. 83, No. 6

ical Neuropsychology of Attention. New York, Oxford University Press, 1994 3. Mateer CM, Mapou RL: Understanding, evaluating, and managing attention disorders after traumatic brain injury. J Head Trauma Rehabil 1996;11:1–16 4. Lezak MD: Neuropsychological Assessment, ed 3. New York, Oxford University Press, 1995 5. Kraus JF, Sorenson SB: Epidemiology, in Silver JM, Yudofsky SC, Hales RE (eds): Neuropsychiatry of Traumatic Brain Injury. Washington, DC, American Psychiatric Press, 1994 6. Prigatano GP: Principles of Neuropsychological Rehabilitation. New York, Oxford University Press, 1999 7. Rehabilitation of persons with traumatic brain injury. NIH Consensus Statements 1998;16:1– 41 8. Cicerone KD, Dahlberg C, Kalmar K, et al: Evidence-based cognitive rehabilitation: Recommendations for clinical practice. Arch Phys Med Rehabil 2000;81: 1596 –1615

behavior after traumatic brain injury: Effects of distraction, task and practice. J Int Neuropsychol Soc 2000;6:1–11 17. Mirsky AF, Anthony BJ, Duncan CC, et al: Analysis of the elements of attention: A neuropsychological approach. Neuropsychol Rev 1991;2: 109 – 45 18. Whyte J: Distinctive methodologic challenges, in Fuhrer M (ed): Assessing Medical Rehabilitation Practices: The Promise of Outcomes Research. Baltimore, Paul H. Brookes, 1997, pp 43–59 19. Park NW, Ingles JL: Effectiveness of attention rehabilitation after an acquired brain injury: A meta-analysis. Neuropsychology 2001;15:199 –210 20. Pliszka S: Child and adolescent psychopharmacology: The use of psychostimulants in the pediatric patient. Pediatric Clin North Am 1998;45:1085–99 21. Conners CK: Forty years of methylphenidate treatment in Attention-Deficit/ Hyperactivity Disorder. J Atten Disord 2002;6(Suppl 1):S17–30

9. Whyte J, Vaccaro M, Grieb-Neff P, et al: Psychostimulant use in the rehabilitation of individuals with traumatic brain injury. J Head Trauma Rehabil 2002;17: 284 –99

22. Birmaher B, Greenhill LL, Cooper MA, et al: Sustained release methylphenidate: Pharmacokinetic studies in ADDH males. J Am Acad Child Adolesc Psychiatry 1989;28:768 –72

10. Whyte J, Hart T, Laborde A, et al: Rehabilitation of the patient with traumatic brain injury, in DeLisa J (ed): Rehabilitation Medicine: Principles and Practice. Philadelphia, Lippincott, 1998, pp 1191–1239

23. Hoffman BB: Catecholamines, sympathomimetic drugs, and adrenergic receptor antagonists, in Hardman JG, Limbird LE (eds): The Pharmacological Basis of Therapeutics. New York, McGraw-Hill, 2001

11. Ponsford J, Kinsella G: The use of a rating scale of attentional behavior. Neuropsychol Rehabil 1991;1:241–57

24. Scheres A, Oosterlaan J, Swanson J, et al: The effect of methylphenidate on three forms of response inhibition in boys with AD/HD. J Abnorm Child Psychol 2003;31:105–20

12. Whyte J, Polansky M, Fleming M, et al: Sustained arousal and attention after traumatic brain injury. Neuropsychologia 1995;33:797– 813 13. Whyte J, Fleming M, Polansky M, et al: The effects of visual distraction following traumatic brain injury. J Int Neuropsychol Soc 1998;4:126 –36 14. Stuss DT, Stethem LL, Hugenholtz H, et al: Reaction time after head injury: Fatigue, divided and focused attention, and consistency of performance. J Neurol Neurosurg Psychiatry 1989;52:742– 8

25. Bedard AC, Ickowicz A, Tannock R: Methylphenidate improves Stroop naming speed, but not responsive interference in children with attention deficit hyperactivity disorder. J Child Adolesc Psychopharmacol 2002;12:301–9 26. Spencer T, Wilens T, Biederman J, et al: A double-blind crossover comparison of methylphenidate and placebo in adults with childhood-onset attention-deficit hyperactivity disorder. Arch Gen Psychiatry 1995;52:434 – 43

ger. Arch Phys Med Rehabil 1993;74: 153– 60 29. Speech TJ, Rao SM, Osmon DC, et al: A double-blind controlled study of methylphenidate treatment in closed head injury. Brain Inj 1993;7:333– 8 30. Williams SE, Ris MD, Ayyanger R, et al: Recovery in pediatric brain injury: Is psychostimulant medication beneficial? J Head Trauma Rehabil 1998;13:73– 81 31. Whyte J, Hart T, Schuster K, et al: The effects of methylphenidate on attentional function after traumatic brain injury: A randomized, placebo-controlled trial. Am J Phys Med Rehabil 1997;76: 440 –50 32. Gualtieri C, Evans R: Stimulant treatment for the neurobehavioural sequelae of traumatic brain injury. Brain Inj 1988;2:273–90 33. Mahalick DM, Carmel PW, Greenberg JP, et al: Psychopharmacologic treatment of acquired attention disorders in children with brain injury. Pediatr Neurosurg 1998;29:121– 6 34. Evans RW, Gualtieri CT, Patterson D: Treatment of chronic closed head injury with psychostimulant drugs: A controlled case study and an appropriate evaluation procedure. J Nerv Ment Dis 1987;175: 106 –9 35. McMillan TM, Jongen ELMM, Greenwood RJ: Assessment of post-traumatic amnesia after severe closed head injury: Retrospective or prospective? J Neurol Neurosurg Psychiatry 1996;60:422–7 36. Rappaport M, Hall KM, Hopkins K, et al: Disability rating scale for severe head trauma: Coma to community. Arch Phys Med Rehabil 1982;63:118 –23 37. Barona A, Reynolds CR, Chastain R: A demographically based index of pre-morbid intelligence for the WAIS-R. J Consult Clin Psychol 1984;52:885–7 38. Alban JP, Hopson MM, Ly V, et al: Effect of methylphenidate on vital signs and adverse effects in adults with traumatic brain injury. Am J Phys Med Rehabil 2004;83:131–7 39. Posner MI: Chronometric Explorations of Mind. New York, Oxford University Press, 1986

15. McDowell SK, Whyte J, D’Esposito M: Working memory impairments in traumatic brain injury: Evidence from a dualtask paradigm. Neuropsychologia 1997; 35:1341–53

27. Francisco G, Ivanhoe C: Successful treatment of post-traumatic narcolepsy with methylphenidate: A case report. Am J Phys Med Rehabil 1996;75:63–5

40. Robertson IH, Manly T, Andrade J, et al: “Oops!”: Performance correlates of everyday attentional failures in traumatic brain injured and normal subjects. Neuropsychologia 1997;35:747–58

16. Whyte J, Schuster K, Polansky M, et al: Frequency and duration of inattentive

28. Mooney GF, Haas LJ: Effect of methylphenidate on brain injury-related an-

41. Robertson IH, Ward T, Ridgeway V, et al: The structure of normal human

June 2004

Effects of Methylphenidate After TBI

419

attention: The test of everyday attention. J Int Neuropsychol Soc 1996;2:525–34 42. Whyte J, Polansky M, Cavallucci C, et al: Inattentive behavior after traumatic brain injury. J Int Neuropsychol Soc 1996;2:274 – 81

Practice: Appendix C. Power Analysis and Determination of Sample Size. New York, Appleton and Lange, 1993, pp 651–7

43. Broadbent DE, Cooper PF, FitzGerald P, et al: The cognitive failures questionnaire (CFQ) and its correlates. Br J Clin Psychol 1982;21:1–16

45. Carlson CL, Pelham WE, Swanson JM, et al: A divided attention analysis of the effects of methylphenidate on the arithmetic performance of children with attention-deficit hyperactivity disorder. J Child Psychol Psychiatry 1991;32: 463–71

44. Portney LG, Watkins MP: Foundations of clinical research, in: Applications to

46. Reid MK, Borkowski JG: Effects of methylphenidate (Ritalin) on information

processing in hyperactive children. Abnorm Child Psychol 1984;12:169 – 85 47. Coons HW, Peloquin LJ, Klorman R, et al: Effect of methylphenidate on young adult’s vigilance and event-related potentials. Electroencephalogr Clin Neurophysiol 1981;51:373– 87 48. McDowell S, Whyte J, D’Esposito M: Differential effect of a dopaminergic agonist on prefrontal function in traumatic brain injury patients. Brain 1998;121: 1155– 64

Letters to the Editor The American Journal of Physical Medicine & Rehabilitation welcomes letters to the Editor that objectively critique information appearing in the Journal. Letters should be limited to a maximum of 500 words (approximately two double-spaced pages). No unsigned letters will be accepted for publication. If a letter is accepted for publication, a copy will be sent to the principal author of the material being discussed, who will have an opportunity to submit a response for publication. All letters are subject to editing by the Editor.

420

Am. J. Phys. Med. Rehabil.

●

Vol. 83, No. 6