Publication Bias in Meta-Analyses of Posttraumatic Stress Disorder Interventions: A Comparison of Methods
Date created: | Last Updated:
: DOI | ARK
Creating DOI. Please wait...
Description: Meta-analyses are susceptible to publication bias, the selective publication of studies with statistically significant results. If publication bias is present in psychotherapy research, the efficacy of interventions will likely be overestimated. This study has two aims: to investigate the degree and impact of publication bias in meta-analyses of the efficacy of psychotherapeutic treatment for posttraumatic stress disorder (PTSD) and to compare the performance of methods to assess publication bias. A comprehensive literature search was conducted and 93 data sets, reported in 24 meta-analyses, were eligible for bias assessment. The rank-correlation test, Egger's regression test, the test of excess significance, trim and fill, PET-PEESE, and p-uniform were applied and found rather few indications for the presence of publication bias. Aside from the explanation that there is hardly publication bias in meta-analyses on PTSD, an additional explanation is that characteristics of the data sets that were unfavorable for these publication bias methods (i.e., the small number of primary studies in each data set) resulting in publication bias tests with low statistical power, and effect size estimates corrected for publication bias that are imprecise. Our results show that methods for detecting publication bias often disagree on the evidence for publication bias in a particular data set. Hence, we recommend including a variety of publication bias methods in each meta-analysis rather than focusing on a single method.