If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Ashman TA, Cantor JB, Gordon WA, Spielman L, Flanagan S, Ginsberg A, Engmann C, Egan M, Ambrose F, Greenwald B. A randomized controlled trial of sertraline for the treatment of depression in persons with traumatic brain injury.
To examine the efficacy of sertraline in the treatment of depression after traumatic brain injury (TBI).
Double-blind, randomized controlled trial.
Research center at a major urban medical center.
Subjects were a referred and volunteer sample of 52 participants with TBI, a diagnosis of major depression disorder (MDD), and a score on the Hamilton Rating Scale for Depression (HAM-D) of 18 or greater. The majority of the sample was male (58%), had less than 14 years of education (73%), had incomes below $20,000 (82%), and were from minority backgrounds (75%). Approximately one third of the sample had mild brain injuries, and two thirds had moderate to severe brain injuries. The mean age was 47±11, and the mean time since injury was 17±14 years. One participant withdrew from the study because of side effects.
Daily oral sertraline in doses starting at 25mg and increasing to therapeutic levels (up to 200mg) or placebo for 10 weeks.
Main Outcome Measures
The HAM-D, the Beck Anxiety Inventory, and the Life-3 quality of life (QOL).
No statistically significant differences were found at baseline between drug and placebo groups on baseline measures of depression (24.8±7.3 vs 27.7±7.0), anxiety (16.4±12.3 vs 24.0±14.9), or QOL (2.96±1.0 vs 2.9±0.9). The income level of those receiving placebo was significantly lower than those participants receiving medication. Analyses of covariance revealed significant changes from preintervention to posttreatment for all 3 outcome measures (P<.001) but no group effects. Random-effects modeling did not find any significant difference in patterns of scores of the outcome measures between the placebo and medication groups.
Both groups showed improvements in mood, anxiety, and QOL, with 59% of the experimental group and 32% of the placebo group responding to the treatment, defined as a reduction of a person's HAM-D score by 50%.
The large variation in rates of post-TBI depression reported in the literature can be attributed to inconsistencies in diagnostic and/or sampling methods used to categorize a person as depressed. Although the majority of studies since the mid-1990s have used more rigorous diagnostic criteria as defined by the DSM-IV,
there continue to be significant methodologic differences between studies investigating prevalence (eg, whether the sample assesses only a current diagnosis of depression or includes lifetime rates of depression, both before and after injury). Even when confining the literature review to studies that diagnose MDD based on DSM-IV criteria determined through clinical interview, the research
Post-TBI depression has been associated with numerous negative outcomes including greater functional disability, reduced participation in activities of daily living, less social and recreational activity, less employment potential, increased caregiver burden, greater sexual dysfunction, lower ratings of health, poor subjective well-being, poorer QOL, and increased rates of suicidal ideation.
Guideline Development Group of the British Society of Rehabilitation Medicine, British Geriatrics Society, Royal College of Physicians London Use of antidepressant medication following acquired brain injury: concise guidance.
found only 2 class II studies and 5 class III studies on the utility of medications in the treatment of depression after TBI, consisting largely of case studies or open trial studies. Sertraline has been the most frequently studied medication for the treatment of depression in persons with TBI and stroke.
There has been no published research on the effectiveness of pharmacologic treatments of anxiety after TBI, especially when anxiety is comorbid with MDD. However, selective serotonin reuptake inhibitors have been approved by the Food and Drug Administration for the treatment of the 3 most common post-TBI anxiety disorders, posttraumatic stress disorder, obsessive-compulsive disorder, and panic disorder,
The purpose of the present study was to test the hypothesis that, compared with placebo, sertraline would be significantly more effective in reducing post-TBI depression and anxiety symptoms and that these changes would be associated with significantly improved QOL. Hypothesized outcomes were measured by using the HAM-D,
This 10-week controlled, double-blind randomized controlled trial of sertraline (Zoloft) versus placebo was designed to determine the efficacy of sertraline in the treatment of post-TBI depression and examine the impact of sertraline on any comorbid anxiety disorders. The severity of depressive and anxiety symptoms was assessed at baseline, biweekly throughout the treatment, and immediately at the end of treatment. Participants were accessed into the study on a rolling admission basis and random assignment to treatment conditions (sertraline or placebo) was balanced by using a block design (blocks of 8) with computer-generated random number sequences in individual presealed envelopes.
Participants were recruited through physician referrals and flyers posted in outpatient rehabilitation treatment areas and distributed at TBI-survivor support groups. In addition, participants from past research projects who had agreed to be contacted for future studies were contacted. The study was approved by the Institutional Review Board of the Mount Sinai School of Medicine, and the conduct of the investigation conformed to the protocol and ethical principles governing research with human beings. When a potential participant arrived at the research center for a screening evaluation, the nature and purpose of the study was explained. After written informed consent was obtained, all potential subjects were evaluated for eligibility. To be included in the study, participants had to be 18 years or older, have a history of TBI with a documented loss of consciousness or other evidence of a TBI (ie, pathology on neuroimaging), be at least 6 months postinjury, meet diagnostic criteria for a major depressive episode using DSM-IV criteria, have a score 18 or greater on the HAM-D, and be able to comprehend or answer verbal or written questionnaires. All participants also had to abstain from seeking new psychosocial or pharmacologic treatments during the course of the intervention, use daily reminders to take the medication, document daily medication compliance and side effects in a written diary; and come to the study facility every other week for the duration of the study for physician visits and a brief evaluation of mood and anxiety.
Participants were excluded from the study if they were known or suspected (1) to be using antidepressant medication, (2) to be in psychotherapy currently, (3) to have active suicidal plans or depression severe enough to require hospitalization, (4) to have a serious medical illness, (5) to be pregnant or breast-feeding, (6) to have mass brain lesions or other neurologic diagnoses other than TBI, (7) to have a history of current or past psychosis or mania using DSM-IV criteria, (8) to have a current substance abuse disorder using DSM-IV criteria, or (9) to have a history of clinically significant liver or renal disease.
Participants who met eligibility criteria were then administered the BAI
Additional demographic data and TBI-related history (ie, date of injury, duration of altered mental state, time since injury) were also obtained. Medical eligibility was assessed by one of the study physicians (SF, BG, AA) who determined if the person's medical history or current medical conditions and medication regimen contraindicated the use of sertraline. Once medically cleared, all participants were randomized into the medication or placebo arm of the study by the pharmacy. Thus, both study staff and participants were blinded to the type of intervention received (ie, whether the participant was prescribed sertraline or placebo).
Participants were randomly assigned to placebo or drug (sertraline) condition for a 10-week trial. Following the protocol used by Fann et al,
participants started at doses of 25mg daily of either sertraline or placebo for weeks 1 and 2, with dosages adjusted during the physician visit at the end of week 2. Further adjustments of dosage were determined by the physician during subsequent office visits occurring every 2 weeks for the duration of the intervention. During the week 2 visit, physicians increased the dosage to between 25 and 100mg/d depending on the participant's self-report of symptoms of depression as reported on the HAM-D and the physician's clinical assessment of the severity of these symptoms. During weeks 4, 6, and 8, dosage of medications continued to be adjusted as clinically indicated, with dosages falling in the range of 25 to 200mg/d depending on clinical response and tolerance by the end of the clinical trial.
The pharmacy implemented masking and random assignment of participants to sertraline or placebo conditions and was responsible for dispensing medications with identical markings to the project physicians for distribution. All medications were packaged in 25-mg quantities and had the same markings so that increases in the number of pills taken in both drug conditions were identical. Medication side effects were noted at each assessment point. All adverse events during the intervention period were recorded, and their relationship to the medication was assessed.
After the initial screening and baseline assessment, all participants were seen by a study physician every 2 weeks for the duration of the clinical trial to monitor drug dosage, reaction to medications, and mood. In addition, a postdoctoral fellow or licensed psychologist rated the clinical severity of the participant's depressive (HAM-D) and anxiety (BAI) symptoms independently from the physician at each biweekly visit. In an attempt to control for psychologic support, research staff followed a standard protocol, and the duration of each contact was kept constant. In addition, participants were not necessarily seen by the same staff member (ie, physician or psychologist) at each visit. At the end of the intervention, all participants completed a posttreatment assessment. These assessments were administered by a different staff member than the person who administered the initial assessment. Participants underwent clinical interviews by using the SCID and the HAM-D. Additionally, the BAI and Life-3 were administered.
Structured Clinical Interview for DSM-IV Axis I Disorders
were used to establish psychiatric diagnoses at the baseline and posttreatment assessments. This interview resulted in a dichotomous score indicating the presence or absence of current and lifetime depressive disorders, lifetime psychotic disorders, and current anxiety disorders and substance abuse disorders in accordance with DSM-IV criteria. Reliability scores for the SCID are high, with the kappa value between .70 and 1.00,
to address unidimensional aspects of the measure, with effect sizes ranging from .40 to .45. The total score and these 4 subscales were tested to detect changes in depression between the treatment arms.
is a 21-item self-report measure that assesses subjective, somatic, or panic-related symptoms associated with anxiety. The instrument has good internal consistency (.92), good test-retest reliability (.75), and acceptable validity.
It is typically administered twice during an evaluation, and the mean of the 2 obtained scores is used. Higher scores on this measure indicate higher levels of subjective QOL. The Life-3 has correlated well with other measures of global or domain-specific life satisfaction. Prior research
reported that an 8-week treatment, with medication individually adjusted up to 200mg, resulted in a response to medication for 87% of study participants (as measured by a reduction of the HAM-D score by at least 50%). In estimating study power, the following assumptions were made, which are more conservative than those reported in the Fann study
: (1) in the treatment group, 75% would respond to sertraline (defined as a reduction of HAM-D score by more than 50%), and (2) in the placebo group, 35% would experience a similar response (the placebo response in general population clinical trials of antidepressant medication is typically 30%–35%).
Sample-size estimates were based on the ability to test these proportions (75% vs 35% response rate). Setting power requirements at .80, a significance level of .05, and a 1-sided test of difference of proportions indicated that 38 cases were needed to complete the study.
A second power analysis was conducted with regard to the repeated administrations of the HAM-D as a continuous outcome measure. These sample-size estimates were based on the formula outlined by Diggle et al
for longitudinal studies. Thus, in addition to estimating the size of the effect that would be clinically interesting and establishing desired significance level and power, rho, the within-subject correlation over time also was estimated. Assuming a baseline HAM-D score ± SD of 25±10, a 5-point drop (ie, half of a SD) would be the smallest effect that would still indicate a clinically viable treatment. The calculations were based on the assumption of a moderate correlation within subjects for the 6 assessments over time, ρ equal to .30 and α equal to .05. A sample size of 21 per group was found to be sufficient at 80% power to detect the effect.
Thus, the primary outcome measure in this study was the HAM-D, and it was the basis of the power analysis. Additional analyses of other measures (eg, BDI-II, Life-3) informed interpretation of the effects of the intervention on additional symptoms but were not the main goal of the study, and, therefore, they were not included in the power calculations.
Between-group comparisons of demographic and other data at baseline were computed by using chi-square analyses and t tests. An intent-to-treat analysis compared the outcomes of all study participants, including participants who received minimal treatment (eg, ≤6wks), to determine if noncompletion was based on treatment failure or some other variable. A last-observation-carried-forward procedure combined the 2 participants who completed 8 weeks with those who completed all 10 weeks of treatment. Random-effects models for repeated measures with 6 time points were conducted to assess within- and between-group differences over time between the medication and placebo groups. The dependent variable in both the intent-to-treat and last-observation-carried-forward subsamples came from the last assessment of each participant. Random-effects models are preferred over traditional repeated-measures analysis of variance because these approaches more appropriately model the within-subject correlation of observations over time and for their ability to tolerate different numbers of observations (eg, a few subjects were missing 1 of the 6 assessments).
Of the 91 respondents screened, 52 were eligible and agreed to participate in the study. Reasons for ineligibility included no diagnosis of depression (n=16), current substance abuse diagnosis (n=9), other exclusionary psychiatric diagnosis (ie, bipolar disorder or psychosis) (n=10), history of stroke (n=2), and current antidepressant use (n=2). Out of the 52 participants who qualified for the study, 41 completed the clinical trial. Of the 11 who did not complete the study, 3 were removed for clinical reasons: 1 participant because of increased anxiety after the baseline assessment, a second because of increased depression after a death in the family 4 weeks into the treatment, and a third because of a mild allergic reaction 4 weeks into the treatment (the latter 2 were receiving placebo). The 8 other noncompleters were noncompliant with the protocol by missing multiple follow-up appointments, not picking up medications, or failing to respond to follow-up contacts.
As shown in table 1, the sample represented a middle-aged, urban, lower socioeconomic status group with a wide range of times since injury. Compared with national estimates,
there was underrepresentation of mild TBI, nonminorities, youths, and older adults. No significant differences were found between drug and placebo groups on sex; age; education; race; income; etiology of injury; time since injury; injury severity; or baseline measures of depression, anxiety, or QOL.
Table 1Sample Demographics
Age (mean ± SD)
Time since injury (mean ± SD)
High school/GED or less
Trade school/some college
Bachelors degree or more
Etiology of injury
GSW to head
Hit by falling object
MVC or pedestrian hit by MV
NOTE. Values are percentages unless otherwise noted.
Abbreviations: GSW, gun shot wound; MV, motor vehicle; MVC, motor vehicle collision.
Eleven participants dropped out of the study before completing 8 weeks of treatment, 5 dropped out after the initial assessment, 3 after 4 weeks, and 3 after 6 weeks. The intent-to-treat analyses found differences between the last assessment HAM-D score for those who dropped out of treatment as compared with the assessment HAM-D score of those participants who remained in treatment (F=4.0, P=.052), with the dropouts tending to have higher HAM-D scores.
Of the 11 participants who continued to have a depressive disorder diagnosis at the end of the intervention according to the SCID, 7 were in the placebo condition (reflecting a nonresponse rate of 37% for this group) and 4 were in the medication condition (reflecting a nonresponse rate of 18% for this group). The difference in nonresponse rates between the groups was not statistically significant (χ2=1.81, NS). Those who continued to have a depressive diagnosis had significantly lower scores on the initial Life-3 than did participants who no longer had a depression diagnosis at the end of treatment (t39=2.2, P=.03; Mnonresp=2.3±1.1, Mresp=3.0±0.8) and a trend toward higher scores on the initial BAI score (t39=2.0, P=.06; Mnonresp=29.3±17.5, Mresp=19.3±13.2). They also had significantly higher postintervention scores on the HAM-D (t39=2.3, P=.03; Mnonresp=20.3±10.0, Mresp=12.9±8.8) and BAI (t39=2.2, P=.03; Mnonresp=18.2±12.7, Mresp=9.6±10.5) than those without a posttreatment diagnosis. All means and SDs for all measures can be found in table 2.
Table 2Outcome Variables and Diagnosis by Time and Group
Participants were considered treatment responders if their initial HAM-D decreased by 50% or dropped below a score of 10 at the end of the intervention. Fifty-nine percent of the participants in the medication condition (13 out of 22) and 32% of those in the placebo condition (6 out of 19) were treatment responders (χ2=3.1, P=.08; with continuity correction P=.15).
Random-effects models for repeated measures, calculated for the 6 assessments, were performed on the HAM-D scores, with group as the fixed effect. The group effect was NS (FHAM-D(1 222.93)=.481, NS). A similar model was examined for the Life-3 and BAI measures; as with the HAM-D, the group effects were not statistically significant, although the results for the Life-3 showed a trend toward significance (FLife1,227.44=2.94, P=.09; FBAI1,215.17=.008, NS). Repeated-measures analysis of variances was also performed on the baseline and endpoint HAM-D, BAI, and Life-3 scores. Findings indicated significant time effects for depression (F1,39=66.3, P<.001; Mpre=26.4±7.5, Mpost=14.9±9.6), anxiety (F1,39=25.1, P<.001; Mpre=21.9±14.9, Mpost=11.9±11.6), and QOL (F1,39=8.1, P<.01; Mpre=2.8±0.9, Mpost=5.7±6.7) but no group effects (Fig 1, Fig 2, Fig 3).
Because the HAM-D is a multidimensional measure of depressive symptoms, some researchers
have indicated that unidimensional subscales may be more effective in detecting differences on core depressive symptoms, particularly in medication trials. Therefore, the analyses described previously were conducted by using these unidimensional subscales. Results were the same as with the HAM-D total score, indicating significant time effects but no group effects.
Finally, the biweekly unidimensional subscales of the HAM-D were used in a random-effects model to determine whether the pattern of changes in depression and anxiety symptoms was different between the medication and placebo group. Two of the subscales, HamG and HamFive, did show a significant fixed effect for group (FHamG[1,226.71]=4.26, P=.04; FHamFive[1,225.74]=3.79, P=.05). In both cases, the placebo group scored significantly higher than the medication group (HamG: meanpla ± SE, 9.62±0.52, meanmed ± SE, 8.16±0.48; HamFive: meanpla ± SE, 6.64±0.37, meanmed, 5.65±0.35), suggesting that 2 scales were more sensitive to changes in depression symptoms related to medication use.
In this sample, both groups showed significant improvements in depressive symptoms, as measured by the HAM-D. Even those participants (37% in the placebo group and 18% in the medication group) who continued to meet diagnostic criteria for a depressive disorder (eg, MDD, depression not otherwise specified, or dysthymia) had significantly lower HAM-D scores at the end of treatment than they did at the beginning of treatment. Additionally, all participants reported significantly lower anxiety symptoms as measured by the BAI with no differences between drug and placebo group.
It is interesting that the findings of this study do not support those of Fann et al.
It is possible that this different set of findings could be related to the fact that the participants in their study were restricted to those with mild TBI and those included herein were more heterogeneous with regard to injury severity. In addition, all participants in Fann's study received placebo for the first week of the trial followed by 8 weeks of treatment with sertraline, so the long-term effects of placebo were not examined.
It is important to review the sample characteristics when considering the findings of the current study. For example, their average time since injury was longer than is often found in the literature; however, this ruled out the recovery of emotional function as playing a role in the findings. In addition, although the participants who entered this study were representative of people with TBI, there was a high proportion of participants with low incomes and from minority backgrounds who are often underrepresented in TBI research. Although the literature indicates that post-TBI depression is not associated with either ethnicity or income, epidemiologic surveys have found that the relationship between race and MDD is complex.
Therefore, it is possible that the lower socioeconomic status of the participants in this study may have resulted in poorer prestudy access to regular medical care, including treatment for depression than in other samples of persons with TBI.
The remission of depression in the majority of participants, regardless of whether they were receiving active or inactive medication, suggests that engagement with the research treatment team by this largely underserved group could have been an important component in the study outcomes. The therapeutic effect of connecting with a medical center and receiving regularly scheduled assessments and treatment by a caring research team may have played a role in reducing depressive symptoms in both groups of participants.
However, this sample may have unique needs and barriers to health care that are not representative of the larger population of persons with TBI, including the etiology of injury, and may therefore underestimate the efficacy of sertraline on other samples of persons with TBI. Although this is a potential limitation of this study, it is also a strength because persons from minority groups are infrequently included in randomized controlled trials.
Because the sample size was small, the participants who dropped out of the study could have reduced the differential effect between the medication and placebo groups. There was a difference between those who dropped out of treatment and those who remained in treatment, with the dropouts having higher depression scores at the time they left treatment; however, the number of dropouts was similar for both conditions. Additionally, among those who completed the study, there were more responders in the treatment group as compared with the placebo group (59% and 32%) and fewer nonresponders (17% and 37%), although these percentages were not statistically significant.
There are other limitations to the study. First, the sample was heterogeneous in terms of the severity of injury and the time since injury. Thus, the medication may indeed be helpful for persons with mild TBI and not helpful for those with moderate or severe TBI. Second, the high representation of persons from minority backgrounds of lower socioeconomic status limits the generalizability of the findings to the larger TBI population. Future research on more homogeneous samples may better define which persons with TBI are most likely to experience an improvement in their mood after treatment with sertraline.
The findings of this study highlight the importance of identifying high-risk subgroups in the TBI population and the continuing need to target effective interventions to them. Finally, although the sample size is small, even without the intent-to-treat analysis, the remaining participants provide sufficient power (70%) to detect a medium effect. Of the 11 who did not complete the study, 5 dropped out immediately; however, the remaining 6 had some treatment, and half of the 6 had 6 weeks of treatment. This would suggest that the findings are robust despite the small sample.
Given the high prevalence of post-TBI depression (and anxiety), more randomized controlled studies are needed to examine the efficacy of various types of both pharmaceutical and behavioral treatments. This study is just one of many that need to be completed to develop an empirical basis for treatment.
The authors gratefully acknowledge all of the conscientious clinicians who conducted the structured interviews (Joshua Masino, PhD, Giselle Braganza, PhD, Charles Filanosky, PhD, Theodore Tsaousides, PhD, Adam Warshowsky, PhD, Amanda Sacks, PhD, Guido Mascialino, PhD). The authors also thank Jonathan Silver, MD, for his advice on study design.
Psychiatric challenges in the first 6 years after traumatic brain injury: cross-sequential analyses of Axis I disorders.
Supported by the National Institute of Disability and Rehabilitation Research, United States Department of Education (grant no. H133A020501), and Pfizer Pharmaceutical Company (ClinicalTrials.gov identifier: NCT00233103).
No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit on the authors or on any organization with which the authors are associated.