A real-world data analysis of dose effect of second-generation antipsychotic therapy on hemoglobin A1C level

Previous studies have demonstrated an association between certain second-generation antipsychotics (SGAs) and diabetes mellitus. The study assessed the impact of SGA dose on hemoglobin A1C (HbA(1c) >6.0) levels in a real-world setting. Patients aged >= 18 years during 2002-2006 in Ingenix LabRx claims database were included. The database collects medical and prescription claims and a subset of laboratory results for an employed, commercially insured population distributed throughout the United States. Patients with previously diagnosed diabetes, identified by the ICD-9-CM code of 250.x or use of antidiabetic agents, were excluded. The main exposure measure was the cumulative dose over a 30 day period before the HbA(1c) test, calculated as [sum of (number of pills per day x strength)]/100. A logistic regression was used to examine the relation with HbA(1c) >6.0 by tertile of the cumulative dose and average daily dose, adjusted for the covariates. The study included 391 patients on olanzapine, 467 on quetiapine, and 262 on risperidone. Patients treated with aripiprazole or ziprasidone (n = 212) were included as a secondary reference because of their minimal metabolic risk. Compared to lower (Tertiles 1 and 2) cumulative doses of risperidone, patients with a high cumulative dose of risperidone (Tertile 3) had a significantly higher odds ratio (OR) for HbA(1c) >6.0 (adjusted OR = 2.45; 95% confidence interval = 1.13-532; P = 0.023). A similar increase in OR was seen in patients with high cumulative dose of olanzapine (2.41; 1.19-4.89; P = 0.015). Analyses of average daily dose revealed that quetiapine >= 400 mg/day and risperidone >= 2 mg/day had an OR of 2.29 (1.04-5.06; P = 0.041) and 2.28 (1.08-4.83; P = 0.032), respectively, compared to aripiprazole/ziprasidone. Both olanzapine groups (>= 10 and <10 mg/day) were associated with a significantly increased OR. All results remained similar after further adjustment for the predicated probability of having an HbA(1c) test and additional medication covariates. In this claims data study, use of olanzapine was associated with elevated HbA(1c) and risperidone and quetiapine appeared to have dose-related association with elevated HbA(1c). One of the limitations of a claims data analysis is the lack of information on potential confounders such as ethnicity and weight. (C) 2011 Elsevier Inc. All rights reserved.