Population attributable fraction (PAF) in repeated measures design: Childhood traumas as predictors of psychotherapy outcomes

Abstract

Introduction

Population attributable fraction (PAF) represents the proportion of treatment failure, which could be avoided, if the individuals at high risk were similar to the individuals at low risk. The PAF, however, has not been available for repeated measures designs.

A relatively prevalent and strong risk factor for many adulthood disorders, such as depression and anxiety, are adversities and traumas experienced in childhood. Little is, however, known of their implications for common treatments such as psychotherapy.

Objectives

To develop PAF for repeated measures, and to provide a useful tool in various research fields to provide decision-makers results, which are easier to interpret.

This study will examine the relative importance of different childhood traumas as predictors of psychotherapy outcome in a patient population with depressive and anxiety disorders.

Methods

PAF was calculated using generalized linear mixed models and Bayesian predictive distributions.

The data is based on 326 outpatients, randomized in one long-term and two short-term psychotherapies by the Helsinki Psychotherapy Study. Patients were assessed up to 10 times during a 5-year follow-up. A combination of psychiatric symptoms measured, is used as the outcome measure.

Results

The repeated measures PAF will provide a useful aggregate measure over the follow-up time and over the patient population.

Conclusions

The repeated measures PAF will provide insight on the relative importance of the different domains of childhood traumas on therapy outcome. Associations of individual-level risk factors do not provide guidelines for policy decisions, which should acknowledge also prevalences of the risk factors in the patient population.

Disclosure of interest

The authors have not supplied their declaration of competing interest.