Cardiovascular risk with antipsychotics: case–control study or survey?

Mackin et al (2007) highlighted the significantly higher risk of cardiovascular and metabolic diseases in people with severe mental illness. Rates of metabolic syndrome and cardiovascular risk similar to those in schizophrenia have been reported in bipolar disorders, and atypical antipsychotics have been approved for the treatment of the latter (Fagiolini et al, 2005; Birkenaes et al, 2007). This implies that all such populations should be studied for putative long-term adverse outcomes, as in the timely study of Mackin et al (2007).

However, some methodological issues need clarification. Mackin et al state that their study is a case-control study, but by definition a case-control study starts with an outcome and investigates exposure to putative risk factors in groups with and without the outcome (Lewallen & Courtright, 1998), generating a measure of relative risk with regard to a given risk factor. Mackin et al started with a group with mental illness on antipsychotics and studied the prevalence of metabolic disease and cardiovascular risk compared with controls. Thus the study has really used a survey design with a control group. The use of a control group alone does not justify the label of a ‘case-control study’.

As an important corollary of this distinction, the sample size is rather low for a community-based survey. Mackin et al mention that comparative data for physical comorbidity in people with diagnoses other than schizophrenia are sparse; unfortunately, the study fails to generate such data owing to the inadequate sample size. We feel that Mackin et al have gone beyond their brief to analyse the effect of individual factors such as diagnoses, type of antipsychotic and smoking; not surprisingly, they failed to emerge with convincing findings as the sample was underpowered to generate such data.

Finally, we wonder whether the inclusion of several patients with depression and anxiety is appropriate for a study on ‘severe mental illness’, a term traditionally reserved for psychotic and bipolar disorders. The common denominator seems to be ‘treated with antipsychotics’ rather than ‘severe mental illness’. It is interesting to note that the type of antipsychotics had no impact on the outcome measures (except serum insulin). If replicated in a much larger community sample, this so-called negative finding could have far-reaching implications regarding choice of treatment. Another important factor in the secondary analysis could have been the duration of treatment with antipsychotics and the dosages used. In a recent study, higher doses of medication were associated with increased cardiovascular risk scores (Osborn et al, 2006). Including the dosage and duration of antipsychotics in the analysis could provide important insights regarding the true impact of antipsychotics on the outcome measures.

This study, like several others, reiterates that patients treated with antipsychotics are at heightened risk for cardiovascular events and metabolic syndrome. Longitudinal studies are needed to explore the relative contribution of putative aetiological factors to physical comorbidity in severe mental illness.