Interpersonal Psychotherapy for Posttraumatic Stress Disorder

Randomized Controlled Trial.
 With funding from the National Institute 
of Mental Health (NIMH), we next conducted a randomized controlled 
trial, comparing non- exposure IPT to both Prolonged Exposure, the gold 
standard and best tested exposure- based treatment, developed by Edna Foa 
(Foa & Rothbaum, 1998); and to Relaxation Therapy (Jacobsen, 1938), which 
had previously been compared to Prolonged Exposure and functioned as an 
active control condition. Relaxation Therapy controlled for therapist time, 
attention, and empathy, and it offered a different mechanism for treatment 
improvement: namely, progressive muscle relaxation. We designed this trial 
as a 
non- inferiority
 study, to test whether IPT would produce improvements 
within 15 CAPS points of Prolonged Exposure and would fare better than 
Relaxation Therapy. According to the developers of the CAPS (Weathers 
et al., 2001), a 15- point score on the CAPS constitutes a meaningful clinical 
difference. Based on power analyses, we therefore chose 12.5 points as a 
cutoff: a mean difference in CAPS change of 12.5 points or more would indicate 
at least minimal inferiority, and less than 12.5 points would indicate less than 
minimal inferiority (Markowitz et al., 2015).
Data analyses followed the intention- to- treat principle. Some participants 
who discontinued treatment were later assessed at the specified assessment 
times, whereas other participants who completed treatment missed the mid- 
treatment assessment. We compared patients with missing post- randomization 
data to those without missing data on their baseline characteristics. No com-
parisons between subjects with and without post- randomization assessment 
overall, or within treatment groups, were statistically significantly different, 
and no differences approached clinically meaningful magnitude. Efficacy of 
the three treatments with respect to symptom severity was estimated based 
on longitudinal mixed- effects models, using multiple imputation for the miss-
ing values. For each variable (score on the CAPS, the PSS- SR, the Hamilton 
Depression Rating Scale (HAM- D), the Social Adjustment Scale– Self- Report, 
 

---

Is Exposure Therapy Necessary to Treat PTSD? 
11
the Quality of Life Measure, and the IIP), we used the Markov chain Monte 
Carlo technique to obtain a monotone missing data pattern. We then applied 
a predictive mean- matching regression method separately for the three treat-
ment groups. To increase the likelihood that the missing- at- random assump-
tion was valid, in addition to the previous values of the variable being imputed, 
we used all other symptom variables and baseline major depression status as 
predictors in predictive mean- matching regression. Fifty imputed data sets 
were generated.
We modeled the post- randomization values as functions of treatment, time, 
and their interaction, controlling for baseline values of the outcome and major 
depression status. If the time- by- treatment interaction reached statistical sig-
nificance, differences between treatments were estimated separately at mid- 
treatment (week 7) and at end of treatment (week 14); otherwise, the model 
was refitted with only main effects for treatment and time, and the differences 
were assessed from a model postulating similar relationships between the 
treatments at all times.
Response and remission rates were estimated based on the observed data 
using pre- specified criteria: response was defined as a decrease of > 30% from 
baseline CAPS score, and remission was defined as a CAPS score of < 20 (Blake 
et al., 1995; Weathers et al., 2001; Davidson et al., 2002). Participants for whom 
these data were missing were categorized as “nonresponders” and “nonremit-
ters.” Statistical significance was assessed throughout at an alpha of 0.05 (two- 
sided). We reported 
p
 values without adjustment for multiple testing, as the  
reported results pertain to pre- specified hypotheses and tests. All analyses 
used Statistical Analysis Software (SAS/ STAT), version 9.2.
We randomly assigned 110 patients who were not taking any psychotropic 
medication to 14 weeks during which they received either 14 50- minute ses-
sions of IPT (Weissman et al., 2007), ten 90- minute sessions of Prolonged 
Exposure (Foa & Rothbaum, 1998), or nine 90- minute and one 30- minute 
session of Relaxation Therapy (Jacobsen, 1938). Treatments were run at their 
standard lengths, amounting to 700 minutes of IPT, 900 minutes of Prolonged 
Exposure, and 840 minutes of Relaxation Therapy. Figure 1. 1 is a Consolidated 
Standards of Reporting Trials (CONSORT) diagram illustrating the flow of 
the study.
Therapists were trained by expert supervisors, followed treatment manu-
als, and conducted pilot cases to develop expertise before beginning the study. 
The Prolonged Exposure supervisor, Elizabeth Hembree had been involved in 
numerous randomized trials of Prolonged Exposure and was perhaps the pri-
mary trainer of therapists, in collaboration with Edna Foa, who invented the 
treatment. Karina Lovell, who had supervised therapists in two prior stud-
ies comparing Relaxation Therapy to Prolonged Exposure, supervised our 

---

12 
I P T   F O R   P T S D
Relaxation Therapy therapists. I  supervised the IPT therapists. All sessions 
were recorded on audiotape, and reliable, trained raters blind to treatment and 
session number rated a sub- sample of treatment sessions using instruments 
that included the best available discriminator of IPT and cognitive behavioral 
treatments, the Collaborative Study Psychotherapy Rating Scale (CSPRS- 6; 
Hollon, 1984).
Study therapists were Ph.D./ Psy.D. psychologists or psychiatrists, who each 
treated a minimum of two pilot cases to ensure their competence and adher-
ence to treatment. The study included two Prolonged Exposure therapists ex-
perienced in a previous PTSD study (Schneier et al., 2012); four IPT therapists; 
and four Relaxation Therapy therapists. Therapists reported primary allegiance 
to their study therapy, an important guard against bias in a treatment trial 
(Falkenström et al., 2013). Therapy teams did not differ significantly in mean 
age (Prolonged Exposure 47.5 years [SD = 10.6], IPT 41.0 years [SD = 9.1], and 
Relaxation Therapy 34.8 years [SD = 5.1]) or in years of modality- specific psy-
chotherapy experience (Prolonged Exposure 7.5 years [SD = 0.7], IPT 9.0 years