A review of spin and bias use in the early intervention in psychosis literature.

Abstract

The early intervention in psychosis literature has recently appropriated clinical terms with etiologic implications such as staging and pluripotent from the oncology literature without adopting the methodological rigor of oncology research. Oncology research maintains this rigor, among other methods, by examining the literature for evidence of bias and spin, which obscures negative trials. This study was designed to detect possible use of reporting bias and spin in the early intervention in psychosis literature. Articles were selected from PubMed searches for early intervention in psychosis, duration of untreated psychosis, first-episode psychosis, ultra-high risk, and at risk mental state between January 1, 2000, and May 31, 2013. 38 RCT and quasi-experimental articles reporting results from early intervention in psychosis paradigms were selected for inclusion. Articles were examined for evidence of inappropriate reporting of primary and secondary end points in the abstract (reporting bias) and presentation as positive despite negative primary end points (spin). While only 13% of early intervention articles reported positive primary end points, abstracts implied that 76% of articles were positive. There was evidence of bias in 58% of articles and spin in 66% of articles. There was a high prevalence of spin and bias in the early intervention in psychosis literature compared to previous findings in the oncological literature. The most common techniques were changing the primary end point or focusing on secondary end points when the primary end point was negative and reporting analyses using only a subset of the data. There appears to be a need for greater scrutiny of the early intervention in psychosis literature by editors, peer reviewers, and critical readers of the literature.