(P-257) The Use of Uncertain Exposure – a Method to Define switching and Add-on in Pharmacoepidemiology

Background: The definition of exposure is a central part of pharmacoepidemiology. When using observational data from prescription registers to define exposure to pharmacological treatments, misclassification may occur when an individual with an ongoing pharmacological treatment fills a prescription for an additional treatment.

Objectives: This study aims to investigate two methods for definition of switching and add-on therapies and their impact on exposure-outcome associations. The first method used conditioning on future exposure (FE) and the second used “uncertain exposure” (UE) which occurs during time when the data does not contain sufficient information to determine exposure.

Methods: Observational cohort study utilizing Swedish population-based health registers. To define exposure to antidepressant (AD) prescription fillings in Sweden from 2010 to 2014, the FE and the UE methods were applied. Hospitalizations for psychiatric reasons were used as an empirical outcome example to illustrate the difference between the two methods. Hazard ratios (HRs) and 95% confidence intervals were estimated using Cox proportional hazards regression adjusted for age and sex.

Results: When defining exposure using the UE method, a notable proportion of psychiatric hospitalizations (7.7%) occurred during the time of uncertain exposure, whilst when using the FE method these hospitalizations were distributed over the unexposed time and the AD use time. The HRs showed that the highest risk of hospitalizations occurred during UE time.

Conclusions: To avoid immortal time bias and selection biases, and to reduce exposure misclassification inherent to the FE method, the UE method could be used when identifying switching and add-on therapies. In addition, the UE method may be used to assess the amount of uncertain exposure time.