I commend the authors of the consensus statement in Diabetes Care for focusing our attention on the physical health of people with schizophrenia and serious mental illnesses and on the need for glucose monitoring (1). There is little consensus about the risk of diabetes in people with schizophrenia and the role of atypical antipsychotic drugs (2–4). Recently I was a member of a group of international psychiatrists and diabetologists who reviewed the evidence surrounding this issue (proceedings have been published in a supplement to the April 2004 issue of the British Journal of Psychiatry). I was, therefore, interested to see that the conclusions published in Diabetes Care differed in some respects from our deliberations and those of the U.S. Food and Drug Administration (FDA). In September 2003, the FDA wrote to the manufacturers of all atypical antipsychotic drugs to ask that their respective labels be changed to recommend regular glucose testing for all schizophrenia patients at risk of diabetes. The FDA went on to explain that “[p]recise risk estimates for hyperglycemia-related adverse events in patients treated with atypical antipsychotics are not available.”
The Diabetes Care consensus statement suggests that the risk of hyperglycemia for some “atypicals,” such as clozapine and olanzapine, is greater than for others (1). It is unclear how this opinion was reached, because although a bibliography was given, the work was unreferenced. One may speculate that the weight given to the 35 retrospective studies was greater than that for the prospective trials (4); the placebo-controlled studies (5–7) and the longest prospective study (8), which assessed glucose over 1 year and compared clozapine and chlorpromazine, are not mentioned in the bibliography.
None of the retrospective studies can state how many patients given each drug received blood tests. This is a crucial confounder, as patients receiving typical antipsychotics have less blood monitoring than those receiving “atypicals” (9). Any study that introduces glucose screening will undoubtedly find new previously undiagnosed diabetes because of the high prevalence of undiagnosed type 2 diabetes.
I am unaware of any prospective trial showing any difference among “atypicals” in terms of emergent glucose abnormalities. The best trial data comparing aripiprazole with olanzapine over 6 months found that emergent glucose abnormalities were identical (4.7 vs. 4.5%) (5,7). The use of placebo cohorts is critical to understanding what part of the risk of glucose abnormalities is attributable to drugs. There are two such datasets, and the incidence of diabetes in the placebo cohorts does not differ from that in the active-drug group (5,6).
Interestingly, during the randomized control trials, weight gain was not associated with the development of diabetes and the incidence of diabetes did not differ among the various “atypicals,” despite differing propensity for weight gain. Linking short-term weight gain to the risk of diabetes ignores the many other genetic and environmental reasons why people with schizophrenia develop diabetes (10,11).
Antipsychotic medication is essential for people with schizophrenia, and effectiveness should be the most important consideration when selecting treatment. The FDA was nearer to the mark in its judgement, and choosing an antipsychotic drug on the basis of its potential to worsen glycemia is failing to understand the available data.
R.I.G.H. has been a member of an advisory panel for, has received honoraria from, and has received grant support from Eli Lilly.