Influence of the drug exposure definition on the assessment of the antipsychotic metabolic impact in patients initially treated with mood-stabilizers

What is already known about this subject: • Metabolic disturbances represent a well-known side effect of second generation antipsychotics. However, studies comparing second generation antipsychotic drugs (SGAPs) and first generation antipsychotic drugs (FGAPs) through administrative databases have shown contrasting findings, which may be attributable to methodological differences.

What this paper adds: • The definition of antipsychotic exposure impacts on the association between antipsychotics and metabolic risk in studies carried out through administrative databases. • Considering cumulative exposure to antipsychotics or including patients exposed to an antipsychotic drug for months or years is likely to over-represent patients who tolerate the drug well with a depletion of susceptible effects. • Antipsychotic drug exposure is a time-varying determinant and episodes of no use, past use and current use should be distinguished over the study period to avoid any misclassification bias that might lead to misleading findings.

Aims: To assess the influence of three definitions of antipsychotic exposure on the comparison between first generation (FGAP) and second generation (SGAP) antipsychotic drugs and 'conventional' mood stabilizers towards the risk of metabolic events using (i) a dichotomous measure (exposed/non-exposed over the follow-up), (ii) a categorical measure taking into account the chronology of exposure at the time of the metabolic event (current, recent and no use) and (iii) a continuous measure (cumulative duration).

Methods: A historical fixed cohort was identified from the 2004-2006 claims database of the French health insurance programme for self-employed workers, including 3172 patients aged 18 years and over who used conventional mood stabilizers over a 3 month period. A metabolic event was defined as an incident dispensing of an anti-diabetic or lipid-lowering drug.

Results: A metabolic event occurred in 367 patients (11.6%). At least one FGAP had been prescribed in 29% of patients who did not develop a metabolic event and in 22% of patients who developed a metabolic event. In addition, at least one SGAP had been prescribed in 12% of patients who did not develop a metabolic event and in 7% of patients who developed a metabolic event. Compared with conventional mood stabilizers, the risk of a metabolic event was negatively associated with exposure to SGAPs over the follow-up period (HR 0.53, 95% CI 0.34, 0.82, P= 0.004), positively associated with recent, but not current, exposure to SGAPs (HR 2.1, 95% CI 1.2, 3.7, P= 0.006) and not associated with cumulative duration of SGAPs (HR 1.001, 95% CI 0.999, 1.003, P= 0.20).

Conclusions: The definition of exposure to antipsychotics in epidemiological studies exploring their metabolic impact is of paramount importance in understanding this association. Different definitions can lead to opposite and seemingly nonsensical results. Not taking into account past exposure, in order to minimize the depletion of susceptible effects, may lead to absurd results.