Article Text
PDF
Statistics from Altmetric.com
Question
Question: What is the efficacy of different treatments for post-traumatic stress disorder (PTSD)?
Outcomes: Change in PTSD symptoms as measured by the individual trials. When available, preference was given to assessments made using the clinician-administered PTSD Scale or PTSD Symptom Scale-Interview version; if this was not available, participant self-report measures were used.
Methods
Design: Systematic review and meta-analysis.
Data sources: PubMed, MEDLINE, PILOTS, PsycINFO, Cochrane Library were searched for articles published between January 1980 and April 2012, supplemented by hand search of reference lists of included studies and previous meta-analyses.
Study selection and analysis: Randomised clinical trials (RCTs) comparing at least one active treatment with a control in adults aged 18 or over with a Diagnostic and Statistic Manual-confirmed diagnosis of PTSD. Studies were grouped into top-level treatment categories (psychotherapy, somatic therapy and pharmacotherapy) then further subdivided, allowing authors to compare effects across and within therapy categories. Effect sizes were calculated for the between-group difference in pretreatment and post-treatment score changes using Hedges g correction for small sample sizes. Effect sizes were pooled using the random-effects model and heterogeneity was measured using the Q and I2 statistics. Publication bias was assessed using funnel plot analysis.
Main results
A total 112 studies (n=9256) were included in the meta-analysis. All treatments were significantly more effective than control with effect sizes for psychotherapy g=1.14 (95% CI 0.97 to 1.3), somatic therapy g=1.24 (95% CI 0.35 to 2.13) and pharmacotherapy g=0.42 (95% CI 0.31 to 0.53). The effect size for psychotherapy was significantly larger than for pharmacotherapy (p<0.001), but the effect size for somatic therapies did not differ from psychotherapy (p<0.26) or pharmacotherapy (p<0.41). Cognitive behavioural therapy was the most common psychotherapy examined, and was the most effective (g=1.26, 95% CI 1.09 to 1.44). The next most common psychotherapy was eye movement desensitisation and reprocessing (g=1.01, 95% CI 0.42 to 1.62). Resilience therapy was demonstrated to have large effect in a single trial (g=1.26, 95% CI 0.58 to 1.93). Only two somatic therapies were studied. Acupuncture demonstrated a large effect in a single trial (g=1.28, 95% CI 0.67 to 1.89). Four trials (n=80) demonstrated a non-significant effect of repetitive transcranial magnetic stimulation (g=1.23, 95% CI 0.00 to 2.53). Examining broad categories of pharmacotherapy, antidepressants (g=0.43, 95% CI 0.31 to 0.55) and atypical antipsychotics (g=0.36, 95% CI 0.05 to 0.66) both showed moderate effects. Effect sizes differed between the subcategories of both of these medications. The other pharmacotherapy classes did not have significant effects compared with the placebo. Regression analyses showed that, in psychotherapy and pharmacotherapy studies, effect sizes were larger in studies including more women, whereas studies with more veterans had smaller effect sizes.
Conclusions
Psychotherapies, somatic therapies and pharmacotherapies are all more effective than control in the treatment of PTSD. However, there is variation in the effectiveness of individual therapies within these categories which makes it difficult to identify a best treatment approach.
Abstracted from
Commentary
In the past 20 years, studies of potential treatment strategies for post-traumatic stress disorder (PTSD) have proliferated, notably in the last decade, in the wake of combat, terrorist attacks and natural disasters. Numerous review articles and a smaller number of meta-analyses have attempted to summarise findings, and treatment guidelines have been proposed. Despite these studies reviewing essentially the same evidence; however, diverging conclusions remain.
Watts and colleagues seek to address the question of treatment efficacy with this exhaustive meta-analysis of randomised clinical trials (RCTs) of PTSD, consisting of 112 articles. Especially noteworthy is the authors’ detailed treatment categorisation system, which begins with modality and continues with further delineation of types and subcategories. The specificity of this breakdown allows the authors to truly characterise the nature of treatments to which the findings belong and is an extremely valuable addition to the PTSD literature.
Results show large effect sizes for all categories. Among psychotherapies, CBT had the largest effect size, while among medications, paroxetine, fluoxetine, sertraline, venlaxafine, risperidone and topiramate were found to be effective. Notably, effect sizes for psychotherapy were significantly greater than those for medication, though the authors caution that potential positive publication bias and other factors may contribute to these findings, and suggest further direct comparisons are needed.
It is essential that future reviews specify the standards used to define treatment success. While Watts and colleagues rigorously include only RCTs and describe moderating variables, an analysis of outcome measures is needed to determine if statistically significant differences translate into clinically meaningful change. Indeed, some studies report a 6-point decrease as meaningful while others require a 50% change in PTSD outcome scores. Equally important is an understanding of the length of symptom remission. Clear findings on long-term outcomes and symptom relapse after medication and therapy termination are especially meaningful to clinicians when choosing treatment options, but remain elusive and must be further explored.
Nonetheless, this rigorous meta-analysis will serve as a valuable resource for researchers and clinicians as the quest to optimally treat individuals with PTSD continues.
Footnotes
-
Sources of funding: US Department of Veterans Affairs and US Department of Defense.
Footnotes
-
Competing interests None.