Abstract
Background. Retrospective large patient database studies have reported conflicting findings regarding diabetes risks associated with antipsychotics. This study compared two study designs to assess antipsychotic-related diabetes risk.
Methods. Claims data were analyzed for over 60,000 patients with psychosis, both treated and untreated with antipsychotics, between January 1999 and April 2002. Diabetes odds ratios for patients treated with antipsychotics versus untreated patients were estimated. All patients and patients stratified by low, medium, and high antipsychotic dose were analyzed. Logistic regression controlled for age, sex, type of psychosis, length of observation/treatment, preexisting excess weight, and use of other drugs.
Results. Under a less rigorous study design, diabetes risk was statistically significant with all antipsychotics versus no treatment. Under a more rigorous design, relative odds for quetiapine and risperidone declined and became statistically nonsignificant, whereas those for olanzapine and conventional antipsychotics increased and remained significant. By dose, only quetiapine showed a lack of statistical significance at all dose levels.
Conclusions. In database studies estimated risks of antipsychotic-related diabetes are affected by study design. With a more rigorous design, the risk associated with quetiapine and risperidone was not significantly different from that in untreated patients. These findings may explain inconsistent findings in pharmacoepidemiologic database studies. Presented at the following meetings: American Psychiatric Association, 56th Institute on Psychiatric Services, October 6–10, 2004, Atlanta, GA USA; New Clinical Drug Evaluation Unit, 44th Annual Meeting, June 1–4, 2004, Phoenix, AZ USA; International Society for Pharmacoeconomics & Outcomes Research, 9th Annual International Meeting, May 16–19, 2004, Arlington, VA USA; American Psychiatric Association, 157th Annual Meeting, May 1–6, 2004, New York, NY USA.