PHQ-9 indicates Patient Health Questionnaire 9; SCI, spinal cord injury; SCID, Structured Clinical Interview for DSM-IV-TR; and XR, extended release.
A, Intent-to-treat adjusted mixed-effects models revealed no significant differences between groups in improvement on the Hamilton Depression Rating Scale 17-item version from baseline to 12 weeks. B, There was a significant Bonferroni-corrected effect on the Maier subscale. XR indicates extended release.
Fann JR, Bombardier CH, Richards JS, Wilson CS, Heinemann AW, Warren AM, Brooks L, McCullumsmith CB, Temkin NR, Warms C, Tate DG, for the PRISMS Investigators. Venlafaxine Extended-Release for Depression Following Spinal Cord InjuryA Randomized Clinical Trial. JAMA Psychiatry. 2015;72(3):247-258. doi:10.1001/jamapsychiatry.2014.2482
Depression is prevalent and associated with negative outcomes in individuals with spinal cord injury (SCI). Antidepressants are used routinely to treat depression, yet no placebo-controlled trials have been published in this population to our knowledge.
To determine the efficacy and tolerability of venlafaxine hydrochloride extended-release (XR) for major depressive disorder (MDD) or dysthymic disorder in persons with chronic SCI.
Design, Setting, and Participants
Multisite, randomized (1:1), double-blind, placebo-controlled Project to Improve Symptoms and Mood After SCI (PRISMS) trial. All research staff conducting screening, intervention, and outcome procedures were blinded to randomization status. We screened 2536 patients from outpatient clinics at 6 SCI treatment centers in the United States and randomized 133 participants into the trial. Participants were 18 to 64 years old and at least 1 month after SCI, with MDD or dysthymic disorder. Seventy-four percent of participants were male, and participants were on average 40 years old and 11 years after SCI. Forty-seven percent had cervical injuries, 53.4% had American Spinal Injury Association injury severity A (complete injury) SCI, 24.1% had at least 2 prior MDD episodes, and 99.2% had current MDD. Common comorbidities included chronic pain (93.9%), significant anxiety (57.1%), and history of substance dependence (44.4%).
Twelve-week trial of venlafaxine XR vs placebo using a flexible-dose algorithm.
Main Outcomes and Measures
The Hamilton Depression Rating Scale (HAM-D 17-item version and Maier subscale, which focuses on core depression symptoms and excludes somatic symptoms) over 12 weeks.
Mixed-effects models revealed a significant difference between the venlafaxine XR and placebo groups in improvement on the Maier subscale from baseline to 12 weeks (treatment effect, 1.6; 95% CI, 0.3-2.9; P = .02) but not on the HAM-D 17-item version (treatment effect, 1.0; 95% CI, −1.4 to 3.4; P = .42). Participants receiving venlafaxine XR reported significantly less SCI-related disability on the Sheehan Disability Scale at 12 weeks compared with placebo (treatment effect, 4.7; 95% CI, 1.5-7.8; P = .005). Blurred vision was the only significantly more common new or worsening adverse effect in the venlafaxine XR group compared with the placebo group over 12 weeks.
Conclusions and Relevance
Venlafaxine XR was well tolerated by most patients and an effective antidepressant for decreasing core symptoms of depression and improving SCI-related disability. Further research is needed to determine the optimal treatment and measurement approaches for depression in chronic SCI.
clinicaltrials.gov Identifier: NCT00592384
Spinal cord injury (SCI) is an often catastrophic injury, estimated to affect 276 281 persons in the United States.1 From 2000 to 2009, there were 5928 SCIs among military personnel.2 Depression is the most prevalent and disabling psychiatric comorbidity after SCI. Major depressive disorder (MDD) occurs in 25% to 30% of community-residing persons with SCI.3 In this population, depression tends to be persistent4- 6 and is associated with the development of pressure ulcers and urinary tract infections,7 poorer self-appraised health,8 less community mobility and social and leisure activities,9- 11 greater unemployment,12 and higher all-cause mortality.13 Nevertheless, only 29% of persons with probable MDD are treated with an antidepressant (11% at guideline levels), and 11% receive psychotherapy (6% at guideline levels) within the past 3 months.14 Undertreatment may be attributable to residual beliefs that depression is a “normal” response to catastrophic injury15 but also to the absence of definitive depression treatment trials.16
Research on the efficacy and tolerability of antidepressant medications for MDD in persons with SCI is long overdue.16 Antidepressant efficacy remains uncertain because persons with SCI have characteristics associated with poor response to treatment such as high rates of comorbid chronic medical conditions,17,18 chronic pain,19,20 history of traumatic brain injury,21,22 dependence for activities of daily living,23,24 and unemployment and impoverishment.25,26 Poor tolerability is suspected due to reported exacerbation of SCI-related spasticity by selective serotonin reuptake inhibitor antidepressants27 and early discontinuation of antidepressants by veterans with SCI.28
We conducted a multisite, randomized, double-blind, placebo-controlled Project to Improve Symptoms and Mood After SCI (PRISMS) trial of venlafaxine hydrochloride extended-release (XR) for MDD or dysthymic disorder (DD) in persons with SCI. The aims of the study were to determine the efficacy and tolerability of venlafaxine XR vs placebo for MDD or DD that occurs more than 1 month following traumatic SCI in persons younger than 65 years. The primary outcome measure was the Hamilton Depression Rating Scale (HAM-D). Secondary outcomes were self-rated depression severity, global improvement, and SCI-related disability.
Details about the methods of this trial are reported elsewhere.29 The institutional review boards of all 6 participating sites (Table 1) approved the study protocols. Written informed consent was obtained from all study participants. Participants with traumatic SCI were recruited from community and clinical settings and referrals from clinicians. Additional inclusion criteria were (1) age 18 to 64 years, (2) SCI of any level (American Spinal Injury Association injury severity A-D), (3) at least 1 month after SCI, (4) moderate depression severity on the Patient Health Questionnaire 9 (PHQ-9)30 (score of ≥10, with ≥1 cardinal symptom) administered twice at least 10 days apart, and (5) satisfaction of DSM-IV criteria for MDD or DD on the Structured Clinical Interview for DSM-IV-TR (SCID).31 Exclusion criteria were (1) current suicidal ideation with intent or plan, (2) history of more than 1 suicide attempt or any suicide attempt within the past 5 years, (3) history of schizophrenia or bipolar disorder, (4) current evidence-based depression treatment, (5) current alcohol or drug dependence (within the past month), (6) history of nonresponse to venlafaxine XR, (7) unstable dosage of psychoactive medications within the past 2 weeks, (8) unstable medical condition within the past 2 weeks, (9) pending surgery within 3 months, (10) pregnancy, (11) breastfeeding, (12) cognitive impairment significant enough to invalidate self-reported outcomes, and (13) nonfluency in English.
Randomization was computer generated by the study biostatisticians (led by N.R.T.) and stratified by site, history of MDD, and history of substance dependence, using permuted blocks of 2. Randomization was 1:1 to venlafaxine XR vs inactive placebo and occurred immediately after baseline assessment (Figure 1).
Demographic and injury characteristics, including level and severity of injury, were obtained from participant interviews with trained research staff and medical record reviews. Current and past MDD, DD, and substance dependence, as well as exclusion diagnoses of bipolar disorder or psychosis, were based on the relevant SCID modules.31,32 Research staff trained by and checked against a single expert clinician (C.W.) administered the SCID. At least 90% interrater item-level agreement with the expert was achieved and maintained each year. History of posttraumatic stress disorder (PTSD),33 current anxiety,34 and current alcohol abuse35 and drug use36 were assessed. Baseline antidepressant treatment history, medical comorbidity, pain, community participation, environmental factors, social support, and health-related quality of life were also assessed.29
Study assessments and dose titrations occurred in person at baseline and at 3, 6, 10, and 12 weeks and by telephone at 1 and 8 weeks. The primary depression outcomes were the HAM-D 17-item version37 and 6-item Maier subscale.38 We included the Maier subscale because it is a unidimensional scale with superior sensitivity to change.39- 41 It excludes somatic items and is especially appropriate for individuals who have substantial physical impairment and medical comorbidity.39 We used a structured version of the HAM-D42 for improved interrater reliability. Interviewer training and interrater reliability standards were the same for the HAM-D 17-item as for the SCID. All research staff conducting screening, intervention, and outcome procedures were blinded to randomization status.
Secondary outcomes were self-rated depression severity (on the Symptom Checklist 20)43 and the Patient Global Impression,44 as well as an additional measure of treatment response,45 self-reported SCI-related disability (on the modified Sheehan Disability Scale).46 We used an adverse effects checklist interview that captured the onset, severity, and resolution of 28 common antidepressant-related adverse effects plus spasticity. Objective spasticity was also assessed using the Ashworth Spasticity Scale.47
Study medication (identical-appearing venlafaxine XR or inert placebo) was administered per a flexible-dose algorithm.29 Pill counts were used to monitor adherence to study drug. Staff counted pills during each in-person visit, and participants counted pills and reported to staff during each telephone visit. Tolerability was assessed at each follow-up.
All data were electronically entered into a common database. Data were inspected for missing information, consistency, and accuracy. Mixed-effects linear regression models were used to compare treatment efficacy of venlafaxine XR with placebo in part to make use of all the collected data and maximize statistical power. This method also accounts for missing data, including dropouts.48
Primary analyses included observations at baseline and at 1, 3, 6, 8, 10, and 12 weeks, analyzed according to the intent-to-treat principle. Random effects were included for participants’ intercepts. Fixed effects were time (linear), treatment (venlafaxine XR vs placebo), interaction of time with treatment, stratification factors, and potential confounders and variables to improve sensitivity. The primary indicator of treatment effect was the interaction of time with treatment using a Bonferroni-corrected significance level of 0.025. We calculated the mean treatment effect for the depression outcomes as the interaction of time with treatment evaluated over 12 weeks (ie, the estimated difference between treatment groups over 12 weeks in excess of the difference at baseline). Exact logistic regression was used to compare dichotomous outcomes. Variables examined as potential confounders included stratification variables, age and sex, education, medical comorbidity, number of prior MDD episodes, baseline scores on alcohol use, generalized anxiety, PTSD, nonneuropathic pain, community participation, environmental factors, and social support.29 We examined potential effect modification by the following domains: demographic characteristics (age, sex, race/ethnicity, and education), depression (severity, number of prior MDD episodes, and preinjury onset), injury characteristics (SCI severity and time since injury), other psychiatric history (history of substance dependence, current alcohol abuse, and current anxiety), nonneuropathic pain, and study site. We examined nonneuropathic pain because separate analyses indicated that venlafaxine XR reduced this pain type compared with placebo.49
Severity of adverse effects (none, mild, moderate, or severe) was recorded at baseline and at each follow-up, allowing reporting of new or worsening adverse effects. The difference in the mean Ashworth Spasticity Scale scores from week 1 to week 12 was analyzed using a linear regression model, adjusting for the corresponding baseline value.
From June 7, 2007, to May 18, 2012, six sites screened 2536 persons, performed baseline assessments on 174 of them, and randomized 133 participants to the intervention trial (Figure 1). Table 1 and Table 2 list the demographic, injury, and clinical characteristics of the study sample. Groups were equivalent at baseline on these characteristics. Sample demographics were consistent with the epidemiology of SCI.1 Fifty-six percent had at least 1 and 24.1% had at least 2 prior MDD episodes, 99.2% had current MDD, and 15.0% had failed a prior antidepressant trial for their current depressive episode. Common medical comorbidities included chronic pain (93.9%), significant anxiety (57.1%), history of substance dependence (44.4%), and current alcohol abuse (21.1%).
The HAM-D 17-item and Maier subscale scores over time are shown in Figure 2. Intent-to-treat adjusted mixed-effects models, summarized in Table 3, revealed no significant differences between the venlafaxine XR and placebo groups in improvement on the HAM-D 17-item from baseline to 12 weeks (treatment effect, 1.0; 95% CI, −1.4 to 3.4; P = .42). There was a significant Bonferroni-corrected effect on the Maier subscale (treatment effect, 1.6; 95% CI, 0.3-2.9; P = .02). That is, at 12 weeks, those receiving venlafaxine XR had an estimated decrease from baseline on the HAM-D 17-item of 1.0 point (5.2%) more than those receiving placebo. On the Maier subscale, those receiving venlafaxine XR had an estimated decrease of 1.6 points (17.6%) more than those receiving placebo.
Mixed-effects model results from the subgroups of completers (those prescribed study medication >9 weeks after randomization) or those who achieved therapeutic doses (≥150 mg/d for at least 6 weeks) did not differ significantly from the overall study sample. There were also no significant differences between groups in improvement on the self-reported Symptom Checklist 20 (treatment effect, 0.22; 95% CI, −0.07 to 0.51; P = .14) or for the percentage of responders or remitters on the HAM-D 17-item or Maier subscale. However, participants in the venlafaxine XR group reported significantly greater improvement in SCI-related disability on the Sheehan Disability Scale at 12 weeks compared with those in the placebo group (treatment effect, 4.7; 95% CI, 1.5-7.8; P = .005). There was a nonsignificant trend for participants in the venlafaxine XR group to rate their depression severity as “much” or “very much” improved compared with those in the placebo group (53.6% vs 40.3%, P = .09).
We found no effect modifiers for the HAM-D 17-item. Study site was the only significant modifier of the Maier subscale treatment effect (P = .006). Three sites showed individually significant effects in favor of venlafaxine XR, one showed a nonsignificant effect in the same direction, and 2 showed nonsignificant effects in the opposite direction. The overall effect was not driven by any particular site.
Only 7 participants (5.3%) did not provide 12-week primary outcome data. Data on treatment dosages attained are summarized in Table 4. Significantly fewer participants in the venlafaxine XR group (81.2%) than in the placebo group (95.3%) achieved a dosage of at least 150 mg/d (P = .02). There were no group differences in number of participants who had at least 80% study drug adherence (as determined by pill counts) or number of assessments via telephone.
Data on blood pressure, spasticity, and adverse effects are summarized in Table 5. Groups were equivalent in terms of their mean new or worsened adverse effects over 12 weeks except that blurred vision was worse in the venlafaxine XR group and constipation was worse in the placebo group. Changes in blood pressure, subjective spasticity, and objective spasticity on the Ashworth Spasticity Scale were similar between groups. The only serious adverse effect was a suicide attempt in the venlafaxine XR group.
The results of this multisite, randomized, double-blind, placebo-controlled trial showed that venlafaxine XR is well tolerated by most patients and is associated with significantly greater improvement in core depressive symptoms compared with placebo controls on one of our primary depression outcome measures, the Maier subscale. The threshold for clinically meaningful differences in improvement between antidepressant and placebo conditions is 3 to 4 points on the HAM-D 17-item, or 15%.50,51 While improvement in the venlafaxine XR group on the HAM-D 17-item and Symptom Checklist 20 was not statistically significant compared with placebo, improvement in the venlafaxine XR group was estimated to be 17.6% greater than that in the placebo group on the Maier subscale, a difference that can be interpreted as clinically meaningful yet only partial remission. Our findings are consistent with results from prior nonrandomized studies52,53 in populations with SCI that showed antidepressants to be well tolerated and potentially efficacious.
The Maier subscale consists of 6 core items measuring depressed mood, anhedonia, guilt, psychomotor agitation, psychomotor retardation, and psychic anxiety. It has been found to be more sensitive than the HAM-D 17-item to change40,41 and excludes somatic items such as problems with sleep, energy, appetite, headache, sexual dysfunction, weight loss, and bodily concern,39 which are causally linked to SCI as well as being manifestations of depression.52- 54 Therefore, we speculate that the larger antidepressant treatment effect on the Maier subscale may be due to the stronger psychometric properties of this measure as well as the absence of somatic symptoms that may be less responsive to antidepressant treatment in the context of SCI. An important direction for future research is the identification of measures of depression severity that are sensitive to change in persons with chronic SCI.
Compared with placebo controls, those treated with venlafaxine XR reported significantly greater improvement in self-reported SCI-related disability after 12 weeks, particularly in the areas of social life and family life or home responsibilities. Rapaport et al45 have argued that quality-of-life and functioning measures, such as the Sheehan Disability Scale, are important outcomes in depression treatment trials and may be more sensitive to treatment effects than standardized depression scales when measuring treatment response.45,55 The use of venlafaxine XR also was associated with a trend toward greater self-appraised global improvement compared with placebo controls.
Despite the presence of risk factors for nonresponse such as medical comorbidities,17 chronic pain,20 and impoverishment,26 the magnitude of the venlafaxine XR HAM-D 17-item response rate was robust (55.1%) compared with the response rate of venlafaxine in a recent systematic review (57%)56 and compared with the antidepressant response rate in trials among persons with physical illness (45%).57 Instead, a high placebo response rate (50.0%) probably attenuated the magnitude of the drug response effect. Placebo response rates of 30% or more are associated with investigations not demonstrating the superiority of antidepressants.58 Moderate initial depression severity,59,60 frequent clinical assessments,61 and longer trial duration62 may have inflated the placebo response rate in our trial. Future researchers should consider design features that minimize placebo responding as well as procedures, such as placebo run-in, that exclude placebo responders from the trial.63
Our sample exhibited high rates of pain and psychiatric comorbidities. Prior investigations in medical populations have also shown a smaller overall difference between active medication and placebo compared with healthy populations.57 Contrary to some prior evidence,59 possibly also due to the high level of comorbidity and insufficient statistical power, we did not find that higher baseline depression or a greater history of prior depression was associated with a larger treatment effect of antidepressant compared with placebo. Our effect size is noteworthy considering the low proportion (81.2%) of patients who achieved recommended therapeutic dosing of venlafaxine XR of 150 mg/d and the lower antidepressant response rates seen in randomized clinical trials vs comparator trials.64
Blurred vision was the only adverse effect that was significantly worse in the venlafaxine XR group. However, some adverse effects, such as anorexia, spasticity, and erectile dysfunction, were nonsignificantly more common in the venlafaxine XR group vs the placebo group, which may have contributed to a lower rate of adherence and therapeutic dosing. While spasticity has been noted in a case report27 as a concern with selective serotonin reuptake inhibitor antidepressants in populations with SCI, this adverse effect was not significantly more severe or frequent in our sample. The suicide attempt in one participant receiving venlafaxine XR underscores the importance of monitoring suicidal risk in this population, which has a risk of completed suicide that is 2 to 6 times greater than in the general population.65
Our findings suggest several important points related to clinical practice. First, education about potential adverse effects and engagement strategies are important to maximize adherence and chances of achieving a therapeutic antidepressant dose and duration. In addition, given the high level of medical and psychosocial comorbidity in many patients with SCI, adjunctive and alternative treatment options, such as anxiolytics or psychotherapy, should be considered, particularly if there is evidence of treatment resistance or intolerance.
Several study limitations should be noted. Our sample demonstrated higher levels of medical and psychiatric comorbidity compared with most antidepressant investigations, as well as the general population with SCI and depression.14 Populations with high levels of psychiatric66,67 and medical18 comorbidity tend to have a lower antidepressant treatment effect in randomized placebo-controlled trials, likely affecting our ability to demonstrate significant differences in treatment outcomes between groups on some measures. Hispanic/Latino persons and Asians were underrepresented; therefore, generalizability to these groups is uncertain. Our findings cannot be generalized to the etiologically diverse category of nontraumatic spinal cord diseases. Achievement of a therapeutic dose and duration was significantly lower for venlafaxine XR than for placebo. While this finding may be in part a reflection of the clinical tolerability of venlafaxine XR, study participants may have been less likely to adhere to the medication regimen within the context of a randomized, double-blind, placebo-controlled trial compared with routine clinical care. Future research should examine the potential role that genetic polymorphisms might have in explaining antidepressant response in this population.68,69 Finally, although the multisite design of this study was advantageous in facilitating geographic, socioeconomic, and racial/ethnic diversity in the study sample, there was a significant difference in the Maier subscale treatment effect between study sites, suggesting the possibility of latent confounding by study site or staff-related factors.
Future research regarding treatment of depression in persons with chronic SCI should consider other antidepressant medications, including adjunctive strategies and treatment algorithms, as well as combining pharmacotherapy with nonpharmacological approaches, such as behavioral activation.70 Research is needed to evaluate sensitivity to change of common depression measures and develop new or optimized outcome assessment tools. A forthcoming article will compare the measurement properties of the HAM-D 17-item, PHQ-9, and Symptom Checklist 20 after SCI. Nevertheless, treatment of somatic symptoms, such as sleep disturbance, is integral to overall depression treatment and should not be ignored in future studies. Depression treatment approaches should also take advantage of alternative delivery strategies, such as telephone, Internet, and mobile technologies, to improve reach, adherence, and follow-up. Studies are needed to determine ways to maintain treatment gains and prevent relapse. Following the findings from this tightly controlled efficacy study, larger studies in real-world settings, with greater statistical power and ecological validity, are needed to determine how best to deliver specific treatment approaches to patients with particular demographic and clinical characteristics. Patient-centered, multifaceted, stepped-care approaches71- 73 that address common comorbid conditions (eg, pain, anxiety, and inactivity), that take into account patient preferences and personality characteristics,74,75 and that are integrated into routine health care may hold the most promise but require further investigation.
Submitted for Publication: May 21, 2014; final revision received August 15, 2014; accepted September 22, 2014.
Corresponding Author: Jesse R. Fann, MD, MPH, Department of Psychiatry and Behavioral Sciences, University of Washington, 1959 NE Pacific St, PO Box 356560, Seattle, WA 98195 (firstname.lastname@example.org).
Published Online: January 21, 2015. doi:10.1001/jamapsychiatry.2014.2482.
Author Contributions: Drs Fann and Temkin had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Fann, Bombardier, Richards, Heinemann, Temkin, Tate.
Acquisition, analysis, or interpretation of data: Bombardier, Richards, Wilson, Heinemann, Warren, Brooks, McCullumsmith, Temkin, Warms, Tate.
Drafting of the manuscript: Fann, Bombardier, Richards, Temkin, Warms.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Fann, Bombardier, Temkin.
Obtained funding: Fann, Bombardier, Richards, Wilson, Heinemann, Temkin, Tate.
Administrative, technical, or material support: Fann, Bombardier, Richards, Wilson, Heinemann, Warren, Brooks, Tate.
Study supervision: Fann, Bombardier, Heinemann.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported by grant H133A060107 from the Department of Education, National Institute on Disability and Rehabilitation Research. Pfizer Pharmaceuticals provided the study drug.
Role of the Funder/Sponsor: The Department of Education, National Institute on Disability and Rehabilitation Research made some recommendations in the design and conduct of the study but had no role in the collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication. Pfizer Pharmaceuticals had no part in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.
Group Information: The PRISMS Investigators were Charles H. Bombardier, PhD, Department of Rehabilitation Medicine, University of Washington; Jesse R. Fann, MD, MPH, Departments of Psychiatry and Behavioral Sciences, University of Washington; Nancy R. Temkin, PhD, Department of Neurological Surgery, University of Washington; Denise G. Tate, PhD, Department of Physical Medicine and Rehabilitation, University of Michigan; Anthony E. Chiodo, MD, Department of Physical Medicine and Rehabilitation, University of Michigan; Claire Z. Kalpakjian, PhD, Department of Physical Medicine and Rehabilitation, University of Michigan; Martin Forchheimer, MPP, Department of Physical Medicine and Rehabilitation, University of Michigan; J. Scott Richards, PhD, Department of Physical Medicine and Rehabilitation, University of Alabama; Cheryl B. McCullumsmith, MD, PhD, Department of Psychiatry and Behavioral Neuroscience, University of Cincinnati; Allen W. Heinemann, PhD, Rehabilitation Institute of Chicago, Northwestern University; Catherine S. Wilson, PsyD, Rehabilitation Institute of Chicago, Northwestern University; Ki (Alex) Kim, MD, Rehabilitation Institute of Chicago, Northwestern University; Susan E. Pearlson, MD, Rehabilitation Institute of Chicago (deceased); Ann Marie Warren, PhD, Division of Trauma, Baylor Institute for Rehabilitation, Baylor University Medical Center; R. Lance Bruce, MD, Division of Trauma, Baylor Institute for Rehabilitation, Baylor University Medical Center; Larry Brooks, PhD, Department of Rehabilitation Medicine, University of Miami; and Diana Cardenas, MD, Department of Rehabilitation Medicine, University of Miami.
Additional Contributions: Jason Barber, MS (Department of Neurological Surgery, University of Washington, Seattle), assisted with data management. Gerald van Belle, PhD (Department of Biostatistics, University of Washington, Seattle), chaired the Data and Safety Monitoring Board. Christian Buhagiar, MS (Department of Rehabilitation Medicine, University of Washington, Seattle; currently at Seattle Children’s Hospital), and Holly Demark, MA (Rehabilitation Institute of Chicago, Northwestern University), assisted with study coordination.