Abstract

Recently, I read the article by Hayes et al. (1) in Acta Psychiatrica Scandinavica. The authors conducted a systematic review and subsequent meta-analysis of studies that estimated standardized mortality ratios (SMRs) in people diagnosed with bipolar affective disorder (BPAD) for all-cause and cause-specific mortalities. Their findings have focused attention on the major issue of premature mortality among sufferers of mental disorders like BPAD, especially because meta-analyses are considered some of the strongest forms of epidemiological evidence. However, while their review attempted to address concerns about statistical heterogeneity and included a highly comprehensive literature search, it still suffered from some important weaknesses. I was concerned with the authors’ treatment of BPAD as if it were seemingly one mental disorder when it, in fact, is conceptualized by mental health professionals as a broad spectrum consisting of multiple disorders. Both the DSM-IV and ICD-10 describe several subtypes of BPAD (2, 3). These various subtypes differ notably in regard to diagnostic criteria, symptom presentation, standard treatment, and prognosis. Despite performing several subgroup analyses, the authors did not address any concerns of high heterogeneity in their metaanalysis resulting from different diagnoses among subjects or even the different criteria used to diagnose these subjects. Obviously, information on BPAD subtype diagnoses may not have been available to the authors; however, this issue should have been addressed or at least mentioned along with their other concerns. Perhaps even more importantly, the authors failed to assess the methodological quality of the studies included in the review. In large part, the internal validity of the results of a meta-analysis depends on the statistical rigor of these included studies. Grading the strength of the evidence provided by each study and reporting the results of this assessment allows the readers to be able to assess the quality of the review’s findings. Although this problem appears to be somewhat common in systematic reviews, there are numerous tools that can be used to assess the scientific quality of included studies. For example, the Newcastle–Ottowa Scale, which is used to assess the quality of observational (non-randomized) studies, would have been an invaluable asset to the authors when they were conducting this review (4). Although the authors examined an important, often overlooked public health problem, and adhered to most of the guidelines listed in the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) (5), their review suffered from methodological flaws that could not be ignored. Because of these issues, the conclusions of this meta-analysis and its potential as a tool for assisting decision-making in public health policy and other areas should be interpreted with caution.