Do the seven indicators of risk of bias by Cochrane have construct validity when applied on randomized clinical trials about psychological therapies for post-traumatic stress disorder (PTSD) in adults?

Abstract

Frequently developed after traumatizing events like homicides, accidents and sexual violence, Post-Traumatic Stress Disorder (PTSD) affects individuals at any age. According to the Diagnostic and Statistical Manual of Mental Disorders (DSM-V), PTSD has 3,5% of prevalence in 12 months in adults in North America and around 0,5 to 1,0% in Europe, most of Asian countries, Africa and Latin America when grouped. That way, the study on this subject has shown itself necessary so to comprehend and better deal with the development of the disorder, offering better life quality to those who live with it. Several randomized clinical trials and nature of interventions have been proposed to reduce PTSD. Psychological intervention, for example, showed a reduction in the severity of PTSD symptoms in adults on the systematic review entitled "Psychological therapies for chronic post-traumatic stress disorder (PTSD) in adults". Primary clinical trials that are included in systematic reviews related to psychological interventions for the treatment of PTSD in adults can contain besides the effect of interventions, risk of bias. Cochrane, an institution created with the intention of producing health related information through systematic reviews uses 7 indicators to evaluate risk of biases of randomized clinical trials included to generate a meta-analysis. These 7 indicators are observable entities that have, underlying to them, a factor (latent) called risk of bias. These 7 indicators, like any test, scale, quiz or battery need, also, to pass through a validation process providing validity evidences in different scopes; up to the present moment, record of quality evidence of construct to this group of items was not found. This process allows us to understand if the function to which the observable indicators were created is, in fact, being conducted. This work proposes to, starting from the utilization of Confirmatory Factorial Analysis (CFA), a tool that allows the confirmation or rejection of a theory, validate or not, the utilization of the 7 Cochrane indicators in the evaluation of studies related to PTSD in adults. Starting from the review entitled "Psychological therapies for chronic post-traumatic stress disorder (PTSD) in adults", published in 2013 by Jonathan I. Bisson, Neil P. Roberts, Martin Andrew, Rosalind Cooper and Catrin Lewis, we are going to use the 70 clinical trials contained in it as observations (research subjects). The 7 indicators of risk of bias structured in Likert scale (low risk, unclear, high risk) are the observable indicators and underlying them we have as a hypothesis to be tested aunidimensional model (called risk of bias). We will use Mplus (a specific software designed to work with latent variables) and the following fit indexes to the Cochrane model of evaluation of risk of bias will be considered: X² (Chi-Square), RMSEA (Root Mean Square Error of Aproximation), CFI (Comparative Fit Index), TLI (Tucker-Lewis Index) e WRMR (Weighted Root Mean Residuals). Through them it will be evaluated if there exists good adjustment between the theoric model of hypothesis (7 observable indicator and a latent measure) and the data collected and described by the authors of the systematic review. (AU)