The American Psychiatric Association (APA) has updated its Privacy Policy and Terms of Use, including with new information specifically addressed to individuals in the European Economic Area. As described in the Privacy Policy and Terms of Use, this website utilizes cookies, including for the purpose of offering an optimal online experience and services tailored to your preferences.

Please read the entire Privacy Policy and Terms of Use. By closing this message, browsing this website, continuing the navigation, or otherwise continuing to use the APA's websites, you confirm that you understand and accept the terms of the Privacy Policy and Terms of Use, including the utilization of cookies.

×

Abstract

Objective:

The authors examined the effect of patient treatment preference on the differential effectiveness of prolonged exposure and sertraline for the treatment of posttraumatic stress disorder (PTSD).

Method:

In a doubly randomized preference trial, 200 patients with PTSD viewed standardized treatment rationales prior to randomization. Patients were first randomized to choice of treatment or no choice. Those assigned to no choice were then randomized to prolonged exposure or sertraline. Acute treatment was 10 weeks, with 24-month follow-up. Interviewer-rated PTSD symptom severity was the main outcome measure, and depression, anxiety, and functioning were assessed as additional outcomes.

Results:

Patients preferred prolonged exposure over sertraline (number needed to benefit [NNTB]=4.5). Using intent-to-treat analyses (N=200), both prolonged exposure and sertraline showed large gains that were maintained over 24 months. Although no differential effect was observed on interviewer-rated PTSD severity, there was a significant benefit of prolonged exposure over sertraline on interview-rated loss of PTSD diagnosis (NNTB=7.0), responder status (NNTB=5.7), and self-reported PTSD, depression, and anxiety symptoms and functioning (effect sizes, 0.35–0.44). Patients who received their preferred treatment were more likely to be adherent, lose their PTSD diagnosis (NNTB=3.4), achieve responder status (NNTB=3.4), and have lower self-reported PTSD, depression, and anxiety symptoms (effect sizes, 0.40–0.72).

Conclusions:

Prolonged exposure and sertraline confer significant benefits for PTSD, with some evidence of an advantage for prolonged exposure. Giving patients with PTSD their preferred treatment also confers important benefits, including enhancing adherence.

Posttraumatic stress disorder (PTSD) is prevalent and debilitating (1, 2), especially among returning veterans, creating an unprecedented need for empirically supported treatments. Providing evidence-based treatment for those in need could reduce the estimated $4.0–$6.2 billion 2-year costs of the disorder by as much as 27% (2).

Recent PTSD reports (3, 4) highlight the need for clinical trials that compare psychotherapy and pharmacotherapy and take into account patient preference. Cognitive-behavioral therapies (CBTs), including prolonged exposure, and selective serotonin reuptake inhibitors (SSRIs) are considered first-line interventions for chronic PTSD (5). Across randomized controlled trials, SSRIs generally produce improvement but have relatively low response rates in contrast to prolonged exposure.

No large-scale trials have directly compared a CBT to an SSRI for PTSD. Only four small-scale trials (N<100) have been published (69), generally showing comparable initial efficacy and a potential benefit for psychotherapy at follow-up (7, 8). Two smaller augmentation studies showed that adding prolonged exposure improved outcomes for poor SSRI responders (10, 11).

PTSD studies have not systematically examined whether patient preference for CBT or SSRIs affects outcome. It is unclear whether providers should accommodate patient preferences or try to persuade them to accept one treatment over another (12). In depression, treatment preference has been shown to have an impact on both pharmacotherapy adherence (13) and psychotherapy outcomes (14). Notably, when presented with options, patients with PTSD have clear treatment preferences (6, 15).

Doubly randomized preference trials allow for estimation of the causal effect of choosing a treatment and assessment of effectiveness under real-world preference conditions (16). Prolonged exposure and sertraline are very different treatment options. Given such differences, patients may have clear treatment preferences, and such preferences may influence treatment outcome and adherence. In this doubly randomized preference trial, we examined the relative effectiveness of prolonged exposure and sertraline in the treatment of chronic PTSD. Patients with chronic PTSD were first randomized to “choice” or “no choice” of prolonged exposure or sertraline. Those who were randomized to no choice were randomized again to prolonged exposure or sertraline. Given the lack of large-scale comparison trials, we did not have an a priori hypothesis as to whether prolonged exposure or sertraline would be more effective in reducing PTSD and related symptoms. We hypothesized that there would be a preference for prolonged exposure over sertraline (6, 15). We hypothesized further that the group of patients who did not receive their preferred treatment would have a higher treatment dropout rate, lower adherence rates, and less reduction in PTSD and related symptoms at the posttreatment and follow-up assessments than those who received their preferred treatment.

Method

Study Design and Participants

In this doubly randomized preference trial (Optimizing PTSD Treatments), patients were recruited from two outpatient clinics in Seattle and Cleveland. Participants were recruited through referrals from community agencies (e.g., rape crisis centers) and treatment providers (e.g., private clinicians, psychiatry departments), fliers and brochures, and newspaper, bus, and magazine advertisements. Recruitment materials included information on the nature of the study as a clinical trial and the treatment options of either medication or therapy.

Eligible participants were adult patients (ages 18–65 years) with a primary DSM-IV diagnosis of chronic PTSD, determined using standard clinical interviews (the Structured Clinical Interview for DSM-IV Axis I Disorders [SCID] [17] and the PTSD Symptom Scale–Interview Version [PSS-I] [18]). Exclusion criteria included a current diagnosis of schizophrenia or delusional disorder; medically unstable bipolar disorder; depression with psychotic features or severe enough to require immediate psychiatric treatment (e.g., actively suicidal); severe self-injurious behavior or suicide attempt within the past 3 months; no clear trauma memory or trauma before age 3; a current diagnosis of alcohol or substance dependence within the past 3 months; ongoing intimate relationship with assailant (in assault cases); unwillingness or medical inadvisability to stop current trauma-focused psychotherapy or antidepressant, based on condition assignment; previous nonresponse to an adequate trial of prolonged exposure (eight sessions or more) or sertraline (150 mg/day for 8 weeks); or a medical contraindication for sertraline (e.g., pregnancy).

Measures

The primary outcome measure was interviewer-rated PTSD symptoms after 10 weeks of acute treatment and 3, 6, 12, and 24 months after treatment, as measured by the PSS-I, a 17-item interview that uses DSM-IV symptom criteria and yields a severity rating (range, 0–51) and a diagnosis (yes/no). Both loss of PTSD diagnosis and responder status (a PSS-I score <24 and a Clinical Global Impressions improvement score <4) were derived. The SCID was also used to assess comorbidity and exclusion diagnoses. Approximately 10% of cases were rerated to assess diagnostic reliability.

Additional outcome measures included scores on the PTSD Symptom Scale–Self-Report (PSS-SR) (18), the Beck Depression Inventory (BDI) (19)), the Hamilton Rating Scale for Depression (HAM-D) (20), the Spielberger State-Trait Anxiety Inventory (STAI) (21), the Sheehan Disability Scale (22), and the CGI improvement scale (23).

Procedures

The study was approved by the University of Washington and Case Western Reserve University/University Hospitals institutional review boards. All patients provided written informed consent. All treatments and procedures were manualized. Between July 1, 2004, and January 6, 2009, we recruited 426 potential participants and randomly allocated 200 participants to the choice condition (N=97) or the no-choice condition (N=103), with 24-month follow-ups through September 20, 2011. Within the choice condition, 61 participants (63%) chose prolonged exposure and 36 participants (37%) chose sertraline (Figure 1). Baseline characteristics were similar between groups (Table 1).

FIGURE 1.

FIGURE 1. CONSORT Diagram for a Doubly Randomized Preference Trial of Prolonged Exposure Compared With Sertraline for PTSD

TABLE 1. Demographic and Clinical Characteristics of Participants in a Doubly Randomized Preference Trial of Prolonged Exposure Compared With Sertraline for PTSD

Characteristic
MeanSD
Age (years)37.4111.30
Time since trauma exposure (years)11.9712.69
N%
Gender
 Female15175.5
 Male4924.5
Ethnicity
 Caucasian13165.5
 African American4321.5
 Other2613.0
Not college educated14070.0
Income <$20,000/year9748.5
Trauma type
 Adult sexual assault6231.0
 Childhood assault (e.g., childhood sexual abuse)4824.0
 Adult assault (nonsexual)4522.5
 Accident (e.g., motor vehicle crash, natural disaster)2713.5
 Death or violence to loved one136.5
 Combat or war52.5
MeanSD
Number of other DSM-IV criterion A events9.056.23
N%
Axis I co-occurrence
 Current13467.0
 Lifetime19296.0
Previous psychiatric treatment
 Any treatment17085.0
 Inpatient treatment3819.0

TABLE 1. Demographic and Clinical Characteristics of Participants in a Doubly Randomized Preference Trial of Prolonged Exposure Compared With Sertraline for PTSD

Enlarge table

Pretreatment assessment.

After patients provided informed consent, independent evaluators conducted diagnostic interviews. Patients then underwent a physical examination and a urine drug screen, and female patients underwent a pregnancy test. Self-report measures were completed prior to randomization. Eligible participants watched videotaped treatment rationales and indicated treatment preference.

Videotaped treatment rationales.

Prior to treatment assignment, participants watched videotaped treatment rationales and then indicated a preference for either sertraline or prolonged exposure. Rationales for sertraline and prolonged exposure included efficacy information, an analogy about mechanism of action, procedures, and possible side effects (6, 24). Rationales were reviewed, revised, and approved by experts in pharmacotherapy (Randall D. Marshall, M.M., and P.P.R.-B.) and in psychotherapy (Edna B. Foa, N.C.F, and L.A.Z.). The order of rationales and the background of the clinician (psychiatrist or psychologist) delivering the rationale were counterbalanced.

Randomization and masking.

After eligibility was confirmed, consent obtained, baseline data collected (diagnostic interviews and self-report measures), and treatment preference recorded, the study coordinator at each site randomly assigned participants in a 1:1 ratio to choice or no choice of treatment, using a computer-generated urn randomization sequence. Participants in the no-choice condition were then randomly assigned in a 1:1 ratio to prolonged exposure therapy or sertraline. Randomization was stratified by PTSD severity according to the PSS-I (scores <35 and scores ≥35) and current antidepressant status (yes: N=58 [29%]; no: N=142 [71%]). The study coordinator at each site, who was responsible for randomization, placed the assignment in a sealed opaque envelope that was then opened with the patient by a Ph.D.-level clinician blind to assignment. All assessors of outcomes were blind to randomization status.

Pretreatment, posttreatment, and follow-up assessments.

Independent evaluators conducted interviewer-rated assessments (PSS-I, HAM-D, CGI improvement scale), and patients completed self-report measures (PSS-SR, BDI, STAI, Sheehan Disability Scale) prior to treatment, after 10 weeks of acute treatment, and 3, 6, 12, and 24 months after treatment. Responders were offered continued sertraline up to 24 months free of charge (which 42 of 46 responders utilized) or up to two prolonged exposure booster sessions (which 13 of 84 responders utilized). Nonresponders were offered the other treatment.

Interventions

Sertraline.

Sertraline was chosen as one of two SSRIs approved by the U.S. Food and Drug Administration for treatment of PTSD at the time the project began. Using a treatment manual, a psychiatrist met with participants who were randomized to or chose sertraline for 45 minutes during the first session and then for no more than 30 minutes for the remainder of the 10 weekly sessions (25). The dosage was started at 25 mg/day, with the goal of 200 mg/day if indicated and tolerated. Final dosages ranged from 12.5 mg/day to 300 mg/day, with an average final dosage of 115 mg/day (SD=78.00). Pill counts and medication diaries were used to document medication use and management.

Prolonged exposure.

Prolonged exposure was chosen because it was one of the best validated psychotherapies for PTSD. For patients who chose or were randomized to prolonged exposure, the psychotherapy was delivered in 10 weekly 90- to 120-minute sessions according to a manual (26). Procedures included education, repeated recounting of trauma memories during the session (imaginal exposure), in vivo exposure, and discussion of the exposure exercises. Homework was assigned at each session.

Supervision and treatment integrity.

The medical director at each site oversaw the administration of sertraline, and all psychiatrists were board-certified and had experience in the treatment of anxiety disorders. Master’s- or Ph.D.-level therapists with CBT experience were trained to deliver prolonged exposure. All sessions were recorded, and prolonged exposure supervision occurred weekly at each site. Independent raters reviewed 10% of videotaped exposures for fidelity, based on Foa et al. (26) and Marshall et al. (25). Prolonged exposure providers completed 90% of essential components, and sertraline providers 96%. No protocol violations were observed.

Adverse events.

Adverse events were defined as any medical symptom accompanied by functional or clinical impairment (e.g., missed work, emergency department visit). Serious adverse events were defined as any inpatient hospitalization, death, suicide attempt, or other life-threatening experience. Forty-one adverse events were reported (15 in the prolonged exposure condition, 25 in the sertraline condition, and one prior to randomization). Twenty adverse events were possibly study related (e.g., headaches, constipation/diarrhea, increased PTSD symptoms), and 27 were serious adverse events, with one possibly related to sertraline (suicidal ideation, intent, and plan).

Data Analysis

All analyses were intent-to-treat, using PROC MIXED in SAS (SAS Institute, Cary, N.C.). Random coefficient regression models were used to examine between-condition differences in the trajectories of change in outcomes across time with predicted values at each assessment phase in active treatment (before treatment, at sessions 1–10 for self-report measures, and after treatment) or follow-up (after treatment and at 3, 6, 12, and 24 months), using maximum likelihood estimation for handling missing data. Random coefficient regression models allow for random intercept and random slopes for the trajectories of individuals included in each condition. The trajectory of change across time includes baseline as time 0 of the trajectory. This approach compares pattern and rate of change across time. A Little’s test confirmed that missing data were missing completely at random. We also compared the number of completed assessments for each treatment preference and did not find significant differences in the preference conditions, which suggests that the impact of missing data on parameter estimates across the trajectory may be less than originally anticipated.

The effects of treatment (prolonged exposure or sertraline) were examined in one model, and preference (match or mismatch) in a separate model. For examination of treatment modality, fixed effects were condition (treatment [prolonged exposure or sertraline]), site, time, and the condition-by-time interaction, and random effects were participant and the participant-by-time interaction. For examination of preference, fixed effects were condition (preference [match or mismatch]), site, time, and the condition-by-time interaction, and random effects were participant and the participant-by-time interaction. The effects of time, redundant across the two models, were reported first in both models, with the treatment models providing the parameter estimates and effect sizes and the preference models simply noting the presence of the effect. Then, to emphasize the main findings, specific condition-by-time effects were reported. Models were fitted for linear and quadratic time and condition-by-time interaction terms, using random intercepts and slopes and an unstructured covariance matrix, except for the acute analyses with interview measures, where compound symmetry was used. Between-group sizes (d) were based on Feingold (27, equation 9), incorporating the quadratic effect for self-reported PTSD symptom scores and scores on the BDI, the STAI, and the Sheehan Disability Scale. The Kenward-Roger method was used to compute the denominator degrees of freedom for the tests of fixed effects resulting from the models with an unstructured covariance matrix specified. For the preference models, education was tested as a fixed effect but failed to alter the findings. Removing site from the analysis did not change the pattern of findings. Post hoc analyses, which were not a priori hypothesized or powered in the study design, examined the relationship of treatment and preference over time, using the same analytic methods as described above.

Results

Prolonged Exposure Compared With Sertraline

Acute outcome: pretreatment to posttreatment assessments.

On interviewer-rated PTSD severity (PSS-I), scores improved from the pretreatment to the posttreatment assessment (F=347.58, df=1, 264, p<0.001); the treatment-by-time interaction was not significant. However, patients in the prolonged exposure condition were significantly more likely than those in the sertraline condition to achieve loss of PTSD diagnosis (number needed to benefit [NNTB]=7.0; χ2=4.22, df=1, N=200, p=0.04) and responder status (NNTB=5.7; χ2=6.67, df=1, N=200, p=0.009) (Table 2).

TABLE 2. Outcome Measures in a Doubly Randomized Preference Trial of Prolonged Exposure Compared With Sertraline for PTSD (Intent-to-Treat Sample, N=200)a

Treatment Condition and MeasurePretreatment AssessmentPosttreatment Assessment24-Month Follow-Up Assessment
Mean95% CIMean95% CIMean95% CI
Prolonged exposure
 PTSD Symptom Scale–Interview29.429.1–29.710.59.5–11.58.06.6–9.3
 Reexperiencing7.67.5–7.62.52.3–2.61.41.3–1.6
 Avoidance12.312.2–12.34.13.7–4.53.32.7–4.0
 Arousal9.69.5–9.74.03.8–4.33.02.6–3.5
 PTSD Symptom Scale–Self Report33.932.6–35.213.811.8–15.712.110.5–13.8
N%N%N%
 PTSD diagnosis116100.03631.03631.0
 Responder status00.08472.48573.3
Mean95% CIMean95% CIMean95% CI
 Hamilton Depression Rating Scale (24-item)22.822.4–23.211.511.1–12.09.78.5–10.9
 Beck Depression Inventory24.222.5–25.99.37.8–10.99.27.8–10.6
 Spielberger State Anxiety Inventory–State54.853.0–56.538.436.4–40.439.137.1–41.1
 Sheehan Disability Scale18.016.8–19.110.18.9–11.48.26.9–9.4
Sertraline
 PTSD Symptom Scale–Interview29.829.5–30.113.311.9–14.711.29.4–13.0
 Reexperiencing7.57.5–7.62.92.7–3.11.61.4–1.8
 Avoidance12.112.0–12.25.34.8–5.84.53.7–5.3
 Arousal10.210.1–10.35.14.7–5.54.94.2–5.7
 PTSD Symptom Scale–Self Report35.834.4–37.317.014.4–19.515.913.9–17.9
N%N%N%
 PTSD diagnosis84100.03845.23845.2
 Responder status00.04654.85059.5
Mean95% CIMean95% CIMean95% CI
 Hamilton Depression Rating Scale (24-item)25.024.5–25.413.713.1–14.312.510.8–14.2
 Beck Depression Inventory25.823.9–27.813.110.7–15.613.110.9–15.3
 Spielberger State Anxiety Inventory–State57.255.3–59.143.540.6–46.342.540.0–45.0
 Sheehan Disability Scale19.318.0–20.712.510.7–14.410.28.5–11.9

aResponder status was defined as having a score ≤24 on the PTSD Symptom Scale–Interview Version and a score <4 on the Clinical Global Impressions improvement scale. Means and 95% confidence intervals were generated from the mixed-effect models. Data for outcome measures are estimated using the intent-to-treat analysis.

TABLE 2. Outcome Measures in a Doubly Randomized Preference Trial of Prolonged Exposure Compared With Sertraline for PTSD (Intent-to-Treat Sample, N=200)a

Enlarge table

Self-reported PTSD severity, depression symptoms (HAM-D, BDI), anxiety symptoms, and disability showed similar patterns of improvement, with effects of time across all measures (PSS-SR: F=104.76, df=1, 175, p<0.001; HAM-D: F=99.95, df=1, 333, p<0.001; BDI: F=50.43, df=1, 173, p<0.001; STAI: F=16.97, df=1, 164, p<0.001; Sheehan Disability Scale: F=18.32, df=1, 166, p<0.001). There were also significant treatment-by-time effects for self-reported PTSD severity (PSS-SR: linear F=20.10, df=1, 175, p<0.001; quadratic F=28.22, df=1, 175, p<0.001), self-reported depression (BDI: linear F=5.78, df=1, 173, p=0.02; quadratic F=10.39, df=1, 158, p=0.002), state anxiety (STAI-S: linear F=14.14, df=1, 164, p<0.001; quadratic F=22.57, df=1, 157, p<0.001), and disability (Sheehan Disability Scale: linear F=7.68, df=1, 166, p=0.006; quadratic F=11.69, df=1, 159, p<0.001), indicating that symptoms declined over time in both treatment conditions, with the rate of decline slowing more over time among patients in the sertraline condition than among those in the prolonged exposure condition (see Figure S1 in the online supplement), with final differences between conditions moderate in size (PSS-SR: d=0.39, mean difference=3.2, 95% CI=2.09–4.31; BDI: d=0.37, mean difference=3.8, 95% CI=2.44–5.16; STAI-S: d=0.44, mean difference=5.1, 95% CI=3.47–6.73; Sheehan Disability Scale: d=0.35, mean difference=2.4, 95% CI=1.46–3.34).

Maintenance of gains: posttreatment to 24-month follow-up assessment.

For interviewer-rated PTSD severity (PSS-I), gains were maintained from posttreatment assessment to 24-month follow-up assessment; there was a time effect (F=4.19, df=1, 117, p=0.043), indicating continued improvement over time (d=0.22), and a differential treatment effect, with patients in the prolonged exposure condition showing more improvement over the 24-month follow-up (F=4.00, df=1, 138, p=0.047; d=0.47). For both loss of diagnosis (NNTB=7.0; χ2=4.18, df=1, N=200, p=0.04) and responder status (NNTB=7.3; χ2=8.23, df=1, N=200, p=0.004), more patients in the prolonged exposure condition maintained their loss of PTSD diagnosis and meaningful gains than in the sertraline condition at 24-month follow-up.

Self-reported PTSD severity, depression and anxiety symptoms, and disability showed similar patterns of maintenance of gains over time, with continued improvement in functioning on the Sheehan Disability Scale from the posttreatment to the 24-month follow-up assessment (F=10.81, df=1, 123, p=0.001; d=0.28) and differential effects favoring prolonged exposure over sertraline at the 24-month follow-up assessment for self-reported depressive symptoms (F=4.45, df=1, 148, p=0.04; d=0.40; mean difference=3.9, 95% CI=2.54–5.26).

Patient Treatment Preference

Overall, patients preferred prolonged exposure (61%) over sertraline (39%) (NNT=4.5, χ2=9.68, df=1, N=200, p<0.01). Those who had a college education were more likely to prefer prolonged exposure (77%) than sertraline (23%) (NNTB=4.5). Those who preferred prolonged exposure reported slightly lower depression symptom severity (BDI score, mean=23.81, SD=9.50) than those who preferred sertraline (mean=26.95, SD=9.96; d=0.32; mean difference=3.14, 95% CI=1.78–4.50). Age, gender, trauma type, time since trauma, prior psychotherapy or pharmacotherapy, or any other baseline psychopathology measures were not significantly associated with preference (p values >0.05).

Acute outcome: pretreatment to posttreatment assessment.

When examining the effect of whether patients got their preferred treatment (match) or not (mismatch) on interviewer-rated PTSD severity (PSS-I), the analysis showed that scores improved from pretreatment to posttreatment, although the preference-by-time interaction was not significant. However, patients who received their preferred treatment were significantly more likely that those who did not to achieve loss of PTSD diagnosis (NNTB=3.4; χ2=13.99, df=1, N=200, p<0.001) and responder status (NNTB=3.4; χ2=14.38, df=1, N=200, p<0.001) (Table 3).

TABLE 3. Outcome Measures and Match or Mismatch With Patient Preference in a Doubly Randomized Preference Trial of Prolonged Exposure Compared With Sertraline for PTSD (Intent-to-Treat Sample, N=200)a

MeasurePretreatment AssessmentPosttreatment Assessment24-Month Follow-Up Assessment
Mean95% CIMean95% CIMean95% CI
Match with preference
 PTSD Symptom Scale–Interview29.529.3–29.710.89.9–11.79.48.1–10.6
 Reexperiencing7.57.5–7.62.42.3–2.51.61.4–1.7
 Avoidance12.312.2–12.44.44.0–4.73.93.3–4.5
 Arousal9.69.5–9.74.13.9–4.33.83.3–4.3
 PTSD Symptom Scale–Self Report34.533.3–35.713.912.2–15.613.311.9–14.8
N%N%N%
 PTSD diagnosis149100.04429.54832.2
 Responder status00.010872.510973.2
Mean95% CIMean95% CIMean95% CI
 Hamilton Depression Rating Scale (24-item)23.723.4–24.011.811.4–12.211.210.1–12.4
 Beck Depression Inventory24.823.4–26.210.08.6–11.410.49.0–11.7
 Spielberger State Anxiety Inventory–State55.954.4–57.439.637.8–41.440.738.9–42.5
 Sheehan Disability Scale18.717.7–19.711.310.0–12.69.17.9–10.3
Mismatch with preference
 PTSD Symptom Scale–Interview29.729.3–30.114.712.8–16.68.46.3–10.6
 Reexperiencing7.57.4–7.63.63.4–3.81.20.9–1.4
 Avoidance12.111.9–12.35.44.7–6.13.42.4–4.3
 Arousal10.410.3–10.55.75.2–6.23.72.8–4.6
 PTSD Symptom Scale–Self Report35.433.7–37.119.616.2–23.014.011.3–16.7
N%N%N%
 PTSD diagnosis51100.03058.82651.0
 Responder status00.02243.12651.0
Mean95% CIMean95% CIMean95% CI
 Hamilton Depression Rating Scale (24-item)23.923.3–24.514.413.6–15.29.57.6–11.4
 Beck Depression Inventory25.322.5–28.114.310.9–17.712.39.4–15.3
 Spielberger State Anxiety Inventory–State55.452.6–58.244.340.5–48.238.835.7–42.0
 Sheehan Disability Scale18.116.4–19.810.48.4–12.48.66.8–10.4

aResponder status was defined as having a score ≤24 on the PTSD Symptom Scale–Interview Version and a score <4 on the Clinical Global Impressions improvement scale. Means and 95% confidence intervals were generated from the mixed-effect models. Data for outcome measures are estimated using the intent-to-treat analysis.

TABLE 3. Outcome Measures and Match or Mismatch With Patient Preference in a Doubly Randomized Preference Trial of Prolonged Exposure Compared With Sertraline for PTSD (Intent-to-Treat Sample, N=200)a

Enlarge table

Self-reported PTSD severity, depression (HAM-D, BDI) and anxiety symptoms, and disability showed similar patterns of improvement, with effects of time across all measures. There were no linear preference-by-time effects for these measures. However, patients who did not receive their preferred treatment showed a nonlinear, sawtooth pattern of gradual decline of symptoms, with final differences emerging between those who received their preferred treatment and those who did not in PTSD severity (PSS-SR: p=0.03, d=0.72; mean difference=5.7, 95% CI=4.59–6.81), depression (BDI: p=0.05, d=0.44; mean difference=4.3, 95% CI=2.94–5.66), and state anxiety (STAI-S: p=0.09, d=0.40; mean difference=3.2, 95% CI=1.56–4.48). This nonlinear pattern was also seen in individual patients.

Maintenance of gains: posttreatment to 24-month follow-up assessment.

For interviewer-rated PTSD severity (PSS-I), gains were maintained from the posttreatment to the 24-month follow-up assessment, with an effect of time, but no preference-by-time effect. For both loss of diagnosis (NNTB=5.3, χ2=5.61, df=1, N=200, p=0.02) and responder status (NNTB=4.5, χ2=8.23, df=1, N=200, p=0.004), those who received their preferred treatment were more likely to maintain their loss of diagnosis and meaningful gains at 24-month follow-up than those who did not.

Self-reported PTSD severity, depression and anxiety symptoms, and disability showed maintenance of gains over time, with continued improvement on interviewer-rated depression and disability from the posttreatment to the 24-month follow-up assessment, but there was no preference-by-time effect.

Treatment Adherence

Number of sessions attended ranged from 0 to 10 (mean=7.04, SD=3.65). There was no difference in dropout rates between the prolonged exposure (N=39, 34%) and sertraline (N=27, 32%) conditions (χ2=0.00, df=1, N=200, p=0.95), and final sertraline dosage was not related to PTSD symptom change (r=0.08, p=0.55). However, there was an effect of treatment preference, with those who received their preferred treatment being more likely to complete treatment (73.2%) than those who did not (49%) (Table 4). Finally, as seen in Table 4, across adherence measures, those who received their preferred treatment were more adherent than those who did not.

TABLE 4. Treatment Adherence, by Patient Preference, in a Doubly Randomized Preference Trial of Prolonged Exposure Compared With Sertraline for PTSDa

MeasureMatch With PreferenceMismatch With Preferencepd or NNTMean Difference or RR95% CI
N%N%
Completed treatment10973.22549.00.0024.10.530.36–0.77
Dropped out after randomization64.01325.5<0.0014.70.160.06–0.39
Dropped out after <4 sessions2315.42345.1<0.0013.40.340.21–0.55
MeanSDMeanSD
Sessions completed7.693.195.144.26<0.0010.672.551.44–3.67
Sertraline dosage (mg/day)144.2066.4662.0369.63<0.0011.2182.1751.55–112.79
Adherence scoresb
 Sertraline21.9812.8412.6713.580.0020.709.323.38–15.25
 Prolonged exposure in vivo exposure homework19.5311.7113.5714.560.0460.455.950.98–11.81
 Prolonged exposure imaginal exposure homework15.219.7110.1910.730.0370.495.020.30–9.75

ad=Cohen’s d; NNT=number needed to treat; RR=relative risk.

bSertraline adherence was assessed with the Medication Adherence Form, and prolonged exposure adherence was assessed with the Usefulness of Techniques Form.

TABLE 4. Treatment Adherence, by Patient Preference, in a Doubly Randomized Preference Trial of Prolonged Exposure Compared With Sertraline for PTSDa

Enlarge table

Post Hoc Analysis: Treatment Modality and Patient Preference

Post hoc analyses examined the relationship of treatment and preference (condition: preferred prolonged exposure, received prolonged exposure [N=95]; preferred sertraline, received prolonged exposure [N=21]; preferred sertraline, received sertraline [N=54]; preferred prolonged exposure, received sertraline [N=30]) over acute treatment. For PTSD diagnosis at the posttreatment assessment, 73.7% of those who preferred and received prolonged exposure were diagnosis free, whereas only 36.7% of those who preferred prolonged exposure but received sertraline were diagnosis free (NNTB=2.7, χ2=15.78, df=3, N=200, p=0.001). The results were similar for responder status, with 76.8% of those who preferred prolonged exposure and received prolonged exposure responding, compared with 36.7% who preferred prolonged exposure but received sertraline (NNTB=2.5, χ2=17.91, df=3, N=200, p<0.001). For PTSD severity, the interaction was not significant, although the pattern was in the same direction (d=0.93).

For self-report measures, there were also significant condition-by-time interactions for self-reported PTSD severity (PSS-SR: linear F=7.63, df=3, 179, p<0.001; quadratic F=10.25, df=3, 164, p<0.001), self-reported depression (BDI: linear F=3.14, df=3, 175, p=0.03; quadratic F=3.96, df=3, 158, p=0.009), state anxiety (STAI-S: linear F=5.26, df=3, 168, p=0.002; quadratic F=7.54, df=3, 158, p<0.001), and disability (Sheehan Disability Scale: linear F=2.70, df=3, 169, p=0.047; quadratic F=3.90, df=3, 161, p=0.01). As seen previously, these interactions were driven by lower scores for those who preferred prolonged exposure and received prolonged exposure than for those who preferred prolonged exposure and received sertraline (PSS-SR: d=0.90; BDI: d=0.80; STAI: d=0.90; Sheehan Disability Scale: d=0.28).

Finally, in terms of patient adherence, there were significant effects of condition across dropout after randomization (χ2=22.32, df=3, N=200, p<0.001), dropout after trying (χ2=20.17, df=3, N=200, p<0.001), completion (χ2=10.86, df=3, N=200, p=0.013), and sessions completed (F=7.58, df=3, 196, p<0.001), with better adherence for those who preferred and received prolonged exposure than for those who preferred prolonged exposure and received sertraline (dropout after randomization: NNTB=1.5; dropout after trying: NNTB=3.0; completion: NNTB=4.9; sessions completed: d=0.83).

Table S1 in the online supplement lists means, 95% CIs, effect sizes, and analyses for randomization to choice compared with no choice, showing that randomization to choice conferred benefits on loss of PTSD diagnosis (NNTB=6.3) and responder status (NNTB=6.3).

Discussion

When given a choice, men and women with chronic PTSD typically preferred exposure therapy (61%) to pharmacotherapy (39%). In a methodologically rigorous doubly randomized preference trial, two important main findings emerged: prolonged exposure and sertraline showed good short- and long-term efficacy across a range of outcomes, with some advantage of prolonged exposure over sertraline; and receiving one’s preferred treatment was associated with moderately better outcomes on some measures and substantially better adherence.

Notably, both prolonged exposure and sertraline produced large pretreatment to posttreatment effect sizes across PTSD severity, depression symptoms, anxiety symptoms, and general functioning. These results are generally consistent with the broader literature. For a prolonged exposure benchmark, an effect size of 2.83 for pretreatment to posttreatment change in PTSD severity is similar to results of multiple randomized controlled trials of prolonged exposure (28, 29). For a sertraline benchmark, an effect size of 2.47 for pretreatment to posttreatment change in PTSD severity is larger than seen in previous studies (30). The absence of placebo and a protocol encouraging medication maintenance through follow-up may in part explain these positive results.

Although both treatments were efficacious, there was some evidence of an advantage for prolonged exposure in magnitude of change, likelihood of losing PTSD diagnosis, and responder status at the end of acute treatment. For sertraline, the rate of improvement slowed more over time than it did with prolonged exposure, with moderate differences at the posttreatment assessment across multiple indices. At the 24-month follow-up, participants in the prolonged exposure condition were still more likely not to have a PTSD diagnosis and to be responders than those in the sertraline condition.

This is, to our knowledge, the first randomized controlled trial examining the impact of patient treatment preference on adherence and outcome of CBT and SSRIs for PTSD. To better personalize PTSD care, we must understand the impact of preferences on empirically supported interventions. The majority of patients in this study preferred prolonged exposure, which is consistent with previous work (6, 24, 31) suggesting a typical preference for psychotherapy over pharmacotherapy. In PTSD, in particular, a preference for prolonged exposure may align with patients’ sense that they need to talk about what happened to them (6, 24). Although few predictors of preference emerged, participants who preferred prolonged exposure were more likely to have a college education and to report slightly lower self-reported depressive symptom severity than those who preferred sertraline. Thus, prolonged exposure may be a better match for patients who are more psychologically minded and have more motivation to engage in a behavioral treatment.

Receiving one’s preferred treatment was associated with a greater likelihood of losing a PTSD diagnosis, a greater likelihood of treatment response, and lower self-reported PTSD, depression, and anxiety symptoms. Patients who did not get their preferred treatment were more likely to drop out of treatment and less likely to be adherent (e.g., doing homework, taking all medication doses). Merely being initially randomized to choice of a treatment conferred a benefit on PTSD diagnosis and responder status, which suggests that simply giving patients a choice matters, although it does not have as great an impact as receiving a preferred treatment. Indeed, the mere act of choosing may have therapeutic powers in and of itself that enhance outcome (32). Furthermore, receiving prolonged exposure when it was preferred was associated with better outcomes and better adherence than receiving sertraline when prolonged exposure was preferred. This likely underestimates the impact of preference, as patients had to be willing to be randomized to either treatment to enter this trial, and degree of preference was not included. Notably, preference effects were as strong as, if not stronger than, treatment modality effects, which suggests that accommodating patients’ preferences in treating PTSD is as important as, if not more so, than the specific choice of an empirically supported treatment. This is consistent with an emerging literature indicating that matched patients are likely to show greater improvement, are less likely to drop out of treatment, and have greater satisfaction compared with those who are not matched with their preferred treatment (33).

Several limitations of the study should be noted. There was no placebo condition; however, meta-analyses suggest that chronic PTSD symptoms often do not substantially improve in control conditions, and both prolonged exposure and sertraline show effects above waiting list or placebo (5). There was also no combination condition, and therefore it is not known whether patients would prefer prolonged exposure plus sertraline, or whether the combination treatment would be more effective than monotherapy. Furthermore, the clinical trial was not designed or powered to test the full interactional model, which limited the power to fully examine specific therapy-by-preference effects. A small percentage of the sample (29%) was stabilized on an antidepressant prior to being randomized, potentially offering either PTSD treatment resistance or augmentation effects for those who were randomized or chose prolonged exposure. However, in prolonged exposure, the observed mean change scores from the pretreatment to the posttreatment assessment were higher for those without a current antidepressant (19.65, SD=10.13) than those on a current antidepressant (17.52, SD=10.21), which argues against an augmentation hypothesis. The assumption of data missing at random may have inflated the effect size estimates; however, this approach should not have differentially affected the results for one condition or the other (treatment or preference), since the degree and pattern of missingness did not differ between conditions. The amount of time with providers was quite different for the two treatment conditions. However, this mirrors real-world practice. Finally, this study used the DSM-IV criteria for PTSD, which could limit generalizability. It is likely, though, that the findings hold for diagnosis with DSM-5 criteria, given that 94% (N=188) of the sample met DSM-5 PTSD diagnostic criteria without including the three new DSM-5 PTSD symptoms.

In the first large head-to-head comparison of prolonged exposure and sertraline, both interventions showed strong and sustained effects for PTSD, related psychopathology, and functioning. These findings have broad real-world applicability, as this trial included patients with common comorbidities, extensive trauma histories, and complicated treatment histories. Accordingly, patients have both trauma-focused psychotherapy and serotonergic antidepressant options for effective PTSD treatment. Accommodating patient preferences, within the context of empirically supported treatments, and developing strategies for enhancing patient buy-in are important next steps in facilitating access, initiation, adherence, and completion of empirically supported treatment for PTSD.

From the Department of Psychology and the Department of Psychiatry and Behavioral Sciences, University of Washington, Seattle; and the Department of Psychiatry and the Department of Psychology, Case Western Reserve University, Cleveland.
Address correspondence to Dr. Zoellner ().

Supported by NIMH grants R01MH066347 and R01MH066348 and the William T. Dahms, M.D., Clinical Research Unit, funded under the Cleveland Clinical and Translational Science Award (UL1 RR024989). Pfizer supplied the medication at no cost but had no input in the trial development, conduct, analysis, or interpretation.

ClinicalTrials.gov identifier: NCT00127673.

Dr. Roy-Byrne has served as editor-in-chief for UpToDate Psychiatry, Journal Watch Psychiatry, and Depression and Anxiety, and he has stock options in Valent Medical Solutions. The other authors report no financial relationships with commercial interests.

The authors thank the participants, therapists, and psychiatrists involved in the study and acknowledge the vital contributions of study researchers and administrators in Seattle and Cleveland. Specifically, the authors thank the investigative team on the grants: Jason Doctor, Ph.D., Joshua McDavid, M.D., Alice S. Friedman, M.S.N., A.R.N.P., and Nora McNamara, M.D. Afsoon Eftekhari, Ph.D., and Lisa Stines Doane, Ph.D., were integral in the implementation of this study. Edna B. Foa, Ph.D., and her team provided prolonged exposure therapy integrity ratings. The authors acknowledge Susan Silva, Ph.D., who was primarily responsible for the statistical analyses, and Eric Youngstrom, Ph.D., and Kevin King, Ph.D., for their statistical consultation.

References

1 Kessler RC, Berglund P, Demler O, et al.: Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the National Comorbidity Survey Replication. Arch Gen Psychiatry 2005; 62:593–602Crossref, MedlineGoogle Scholar

2 Eibner C, Ringel JS, Kilmer B, et al.: The cost of post-deployment mental health and cognitive conditions, in Invisible Wounds of War: Psychological and Cognitive Injuries, Their Consequences, and Services to Assist Recovery. Edited by Tanielian T, Jaycox LH. Santa Monica, Calif, RAND Corporation, 2008, pp 169–241Google Scholar

3 Institute of Medicine: Treatment of Posttraumatic Stress Disorder: An Assessment of the Evidence. Washington, DC, National Academies Press, 2007Google Scholar

4 US Department of Veterans Affairs Office of Research and Development; National Institute of Mental Health; and Department of Defense: Advancing Research Standards for PTSD Interventions: Suggested Approaches for Designing and Evaluating Clinical Trials: A Meeting Summary (Rockville, Md, January 22–23, 2008). https://www.research.va.gov/services/csrd/ptsd-standards.pdfGoogle Scholar

5 Watts BV, Schnurr PP, Mayo L, et al.: Meta-analysis of the efficacy of treatments for posttraumatic stress disorder. J Clin Psychiatry 2013; 74:e541–e550Crossref, MedlineGoogle Scholar

6 Feeny NC, Zoellner LA, Mavissakalian MR, et al.: What would you choose? Sertraline or prolonged exposure in community and PTSD treatment seeking women. Depress Anxiety 2009; 26:724–731Crossref, MedlineGoogle Scholar

7 Frommberger U, Stieglitz R-D, Nyberg E, et al.: Comparison between paroxetine and behaviour therapy in patients with posttraumatic stress disorder (PTSD): a pilot study. Int J Psychiatry Clin Pract 2004; 8:19–23Crossref, MedlineGoogle Scholar

8 van der Kolk BA, Spinazzola J, Blaustein ME, et al.: A randomized clinical trial of eye movement desensitization and reprocessing (EMDR), fluoxetine, and pill placebo in the treatment of posttraumatic stress disorder: treatment effects and long-term maintenance. J Clin Psychiatry 2007; 68:37–46Crossref, MedlineGoogle Scholar

9 Schneier FR, Neria Y, Pavlicova M, et al.: Combined prolonged exposure therapy and paroxetine for PTSD related to the World Trade Center attack: a randomized controlled trial. Am J Psychiatry 2012; 169:80–88LinkGoogle Scholar

10 Rothbaum BO, Cahill SP, Foa EB, et al.: Augmentation of sertraline with prolonged exposure in the treatment of posttraumatic stress disorder. J Trauma Stress 2006; 19:625–638Crossref, MedlineGoogle Scholar

11 Simon NM, Connor KM, Lang AJ, et al.: Paroxetine CR augmentation for posttraumatic stress disorder refractory to prolonged exposure therapy. J Clin Psychiatry 2008; 69:400–405Crossref, MedlineGoogle Scholar

12 Street LL, Niederehe G, Lebowitz BD: Toward greater public health relevance for psychotherapeutic intervention research: an NIMH workshop report. Clin Psychol Sci Pract 2000; 7:127–137CrossrefGoogle Scholar

13 Pyne JM, Rost KM, Farahati F, et al.: One size fits some: the impact of patient treatment attitudes on the cost-effectiveness of a depression primary-care intervention. Psychol Med 2005; 35:839–854Crossref, MedlineGoogle Scholar

14 Chilvers C, Dewey M, Fielding K, et al.: Antidepressant drugs and generic counselling for treatment of major depression in primary care: randomised trial with patient preference arms. BMJ 2001; 322:772–775Crossref, MedlineGoogle Scholar

15 van Minnen A, Hendriks L, Olff M: When do trauma experts choose exposure therapy for PTSD patients? A controlled study of therapist and patient factors. Behav Res Ther 2010; 48:312–320Crossref, MedlineGoogle Scholar

16 Long Q, Little RJ, Lin X: Causal inference in hybrid intervention trials involving treatment choice. J Am Stat Assoc 2008; 103:474–484CrossrefGoogle Scholar

17 First MB, Spitzer RL, Gibbon M, et al.: Structured Clinical Interview for DSM-IV Axis I Disorders, Patient Edition (SCID-I/P). New York, New York State Psychiatric Institute, Biometrics Research, 1995Google Scholar

18 Foa EB, Riggs DS, Dancu CV, et al.: Reliability and validity of a brief instrument for assessing post-traumatic stress disorder. J Trauma Stress 1993; 6:459–474CrossrefGoogle Scholar

19 Beck AT, Ward CH, Mendelson M, et al.: An inventory for measuring depression. Arch Gen Psychiatry 1961; 4:561–571Crossref, MedlineGoogle Scholar

20 Hamilton M: A rating scale for depression. J Neurol Neurosurg Psychiatry 1960; 23:56–62Crossref, MedlineGoogle Scholar

21 Spielberger CD, Gorsuch RL, Lushene R, et al.: Manual for the State-Trait Anxiety Inventory. Palo Alto, Calif, Consulting Psychologists Press, 1983Google Scholar

22 Sheehan DV: The Anxiety Disease. New York, Charles Scribner and Sons, 1983Google Scholar

23 Guy W (ed): ECDEU Assessment Manual for Psychopharmacology. Rockville, Md, US Department of Health, Education, and Welfare, 1976Google Scholar

24 Zoellner LA, Feeny NC, Cochran B, et al.: Treatment choice for PTSD. Behav Res Ther 2003; 41:879–886Crossref, MedlineGoogle Scholar

25 Marshall RD, Beebe KL, Oldham M, et al.: Efficacy and safety of paroxetine treatment for chronic PTSD: a fixed-dose, placebo-controlled study. Am J Psychiatry 2001; 158:1982–1988LinkGoogle Scholar

26 Foa EB, Hembree EA, Dancu CV: Prolonged Exposure Manual: Revised Version. Philadelphia, University of Pennsylvania, 2002Google Scholar

27 Feingold A: A regression framework for effect size assessments in longitudinal modeling of group differences. Rev Gen Psychol 2013; 17:111–121Crossref, MedlineGoogle Scholar

28 Resick PA, Nishith P, Weaver TL, et al.: A comparison of cognitive-processing therapy with prolonged exposure and a waiting condition for the treatment of chronic posttraumatic stress disorder in female rape victims. J Consult Clin Psychol 2002; 70:867–879Crossref, MedlineGoogle Scholar

29 Foa EB, Hembree EA, Cahill SP, et al.: Randomized trial of prolonged exposure for posttraumatic stress disorder with and without cognitive restructuring: outcome at academic and community clinics. J Consult Clin Psychol 2005; 73:953–964Crossref, MedlineGoogle Scholar

30 Brady K, Pearlstein T, Asnis GM, et al.: Efficacy and safety of sertraline treatment of posttraumatic stress disorder: a randomized controlled trial. JAMA 2000; 283:1837–1844Crossref, MedlineGoogle Scholar

31 Barlow DH: Psychological treatments. Am Psychol 2004; 59:869–878Crossref, MedlineGoogle Scholar

32 Seligman MEP. The effectiveness of psychotherapy: the Consumer Reports study. Am Psychol 1995; 50:965–974Crossref, MedlineGoogle Scholar

33 Lindhiem O, Bennett CB, Trentacosta CJ, et al.: Client preferences affect treatment satisfaction, completion, and clinical outcome: a meta-analysis. Clin Psychol Rev 2014; 34:506–517Crossref, MedlineGoogle Scholar