Consistent with our previous work, there was a strong preference for PE in comparison to SER.
In Study 1, a community sample of trauma-exposed women, 81% chose PE; similarly, in Study 2, an open treatment trial for women with chronic PTSD, 72% chose PE, highlighting a likely link between theoretical and actual treatment choices. Consistent with these results, previous work also highlights the preference for psychotherapy over medication in recent assault survivors.
Our work has shown this general preference persists regardless of information presented (i.e., side effects and treatment mechanism) in the treatment description (Feeny et al., submitted). Further, this may not just reflect a general preference for psychotherapy, as even when other therapies for PTSD are described in detail, exposure-based therapies are ranked among the most preferred.
Consistent with other preference work,
in Study 2, higher levels of psychopathology and comorbidity were associated with choosing pharmacotherapy. While this was not the case in Study 1, this may reflect a reduced range of psychopathology in this sample. Notably, women who chose pharmacotherapy also reported being less likely to be employed full time (Study 1) or college educated (Study 2), suggesting a possible association between lower socioeconomic status and preference for pharmacotherapy. Indeed, these findings are also consistent with an association between higher education and the receipt of nonmedical treatment or psychotherapy (Feeny et al., submitted).
Both SER and PE yielded medium to large unadjusted effects on PTSD severity over time, with evidence of an advantage of PE in propensity-adjusted analyses at posttreatment. It should be noted that effect sizes reported on are adjusted to account for differential preference rates for SER and PE and accordingly do not reflect overall improvement. Notably, the unadjusted effect sizes obtained in this nonrandomized trial are, from a benchmarking perspective, comparable to outcome seen in randomized trials (SER[2,31]
). Continued improvement over time on SER is also consistent with the literature.[3,33]
Our treatment completion rates are similar to those seen in randomized trials as well (38% for PE;
30.9% for SER
). Interestingly, those with co-occurring MDD and PTSD were more likely to choose SER than those without this co-occurrence; perhaps those with MDD do not have the energy or motivation to undergo an intense psychotherapy such as PE, or they believe that pharmacotherapy is more necessary for more severe symptoms. However, among those with MDD, PE was particularly effective, suggesting a further dissociation between treatment choice and efficacy.
Several limitations should be noted. First, both Studies 1 and 2 used a forced-choice scenario, which does not reflect the full range of treatment options available and inappropriately implies disinterest in the other option.
Thus, it is possible that the preference seen for PE as opposed to SER may reflect a preference for therapy over medication and not a preference for PE specifically. Similarly, the videotaped treatment descriptions do not capture the dynamic nature of a collaborative discussion between clinician and patient about possible treatment options. Second, the fact that our samples are limited to females who are primarily assault survivors may limit the generalizability of our findings. However, given that women are consistently found to be twice as likely as men to develop PTSD
and that assault survivors are among the most likely group exposed to trauma to develop PTSD, we believe the focus on this group is quite warranted. Third, the end dosage of SER (105 mg/day) was lower than in some published trials (e.g., 133.3 mg/day
). However, end dosage was not strongly associated with symptom improvement (PTSD: r
= –.13; BDI: r
= .02; STAI: r
= –.25; ns
). Fourth, we used an open choice design and thus could not directly address the impact of choice or treatment modality on outcome. While propensity scores were used, unmeasured factors cannot be accounted for. Finally, rates of choice (3:1, PE:SER) impacted our power, making treatment effect sizes potentially unstable. Accordingly, the present results should be interpreted with caution and await future studies utilizing randomization to “choice” vs. “no choice” and PE vs. SER.
This study is the first to examine treatment preferences and outcome in individuals with chronic PTSD, highlighting the presence of clear preferences and their potential impact on outcome. Notably, while those with MDD preferred SER, in this subsample, PE was particularly effective. While novel in PTSD, there is growing evidence in the treatment of depression that patient preferences may significantly impact outcome,
compliance, and cost effectiveness.
Taken together, these studies and this study underscore the importance of systematic study of patient preferences and encourage a rethinking of one-size fits all approaches to treatment for mental disorders.