Antipsychotics and Risk of Hyperglycemic Emergencies

Overview

The purpose of this study is to determine whether the use of atypical antipsychotic medication increases the risk of hospitalization for a hyperglycemic emergency. The investigators will carry out separate population-based cohort studies using administrative health databases in eight jurisdictions in Canada and the UK. Cohort entry will be defined by the initiation of a new antipsychotic medication. Follow-up will continue until hospitalization for a hyperglycemic emergency or the end of 365 days. The results from the separate sites will be combined to provide an overall assessment of the risk of hyperglycemic emergencies among new users of various antipsychotic drugs.

The objective of this study is to determine whether the use of atypical antipsychotics is associated with an increased risk of hospitalization for hyperglycemic emergencies (hyperglycemia, diabetic ketoacidosis, and hyperosmolar hyperglycemic state). A common-protocol approach will be used to conduct retrospective cohort studies using administrative health care data from eight jurisdictions (the Canadian provinces of British Columbia, Alberta, Saskatchewan, Manitoba, Ontario, Quebec, and Nova Scotia, as well as the United Kingdom (UK) General Practice Research Database \[GPRD\]). Briefly, the Canadian databases include population-level data on physician billing, diagnoses and procedures from hospital discharge abstracts, and dispensations for prescription drugs. The data in Nova Scotia, Ontario, and Alberta will be restricted to patients aged 65 years and older, as prescription data are not available for younger patients. The GPRD is a clinical database that is representative of the UK population and contains the records for patients seen at over 680 general practitioner practices in the UK. In each jurisdiction, the investigators will assemble a study cohort that includes all patients newly prescribed an antipsychotic medication (WHO Anatomical Therapeutic Chemical (ATC) code N05A) from April 1, 1998 (or the earliest date of data availability at each site) to March 31, 2010 (or the last date of data availability at each site if earlier). The date of study cohort entry will be defined by the date of the prescription (for GPRD) or dispensation (for all other sites) of the newly-prescribed antipsychotic. Patients in the study cohort will be followed from the date of study cohort entry until an event (defined below), censoring due to death or departure from the database, 365 days after cohort entry, or the end of the study period (March 31, 2011), whichever occurs first. The exposure categories will be separated by the type of antipsychotic medication prescribed upon cohort entry as: 1) olanzapine users; 2) other atypical antipsychotic users; 3) typical antipsychotic users; and 4) risperidone users. Exposure to olanzapine will be defined as a prescription for olanzapine on the date of cohort entry. Exposure to other atypical antipsychotics will be defined as a prescription for an atypical antipsychotic other than olanzapine on the date of cohort entry. Exposure to typical antipsychotics will be defined as a prescription for a typical antipsychotic on the date of cohort entry. Exposure to risperidone will be defined as a prescription for risperidone on the date of cohort entry. The investigators will use an analysis analogous to an intention-to-treat approach. The primary outcome will be defined as a first hospitalization for a hyperglycemic emergency within 365 days of the initiation of the cohort defining antipsychotic medication. The reference group throughout all analyses will be risperidone users. The primary comparator will be olanzapine users; other atypical antipsychotic users and typical antipsychotic users will both serve as secondary comparators. Inverse probability treatment weighting (IPTW) using propensity scores to estimate marginal treatment effects will be used. These propensity scores will be estimated using logistic regression models where treatment with the comparator antipsychotic will be regressed on a number of pre-identified covariates that influence the risk of a hyperglycemic emergency. The absolute difference in the probability of a hyperglycemic emergency will be estimated. The cause-specific hazard for the comparator vs. risperidone will be estimated using Cox-proportional hazards regression models, incorporating the IPT weights and using a robust variance estimator. Meta-analyses of the site-specific results will then be performed using fixed-effects models with inverse variance weighting, with the amount of between site heterogeneity estimated using the I-squared statistic. Secondary analyses will be performed by means of nested case-control analyses of the cohorts. Cases will be defined as subjects having a hyperglycemic event and will be matched to controls that did not experience a hyperglycemic event. The event date of the cases will serve as the index date for both controls and their matched cases. Exposure will be defined as the most recent antipsychotic medication prescribed prior to the index date. Conditional logistic regression will be used to estimate the odds ratio of a hyperglycemic event for the comparator vs. risperidone.