A Positive Event Dependence Model for Self‐Controlled Case Series with Applications in Postmarketing Surveillance

Abstract

A primary objective in the application of postmarketing drug safety surveillance is to ascertain the relationship between time-varying drug exposures and recurrent adverse events (AEs) related to health outcomes. The self-controlled case series (SCCS) method is one approach to analysis in this context. It is based on a conditional Poisson regression model, which assumes that events at different time points are conditionally independent given the covariate process. This requirement is problematic when the occurrence of an event can alter the future event risk. In a clinical setting, for example, patients who have a first myocardial infarction (MI) may be at higher subsequent risk for a second. In this work, we propose the positive dependence self-controlled case series (PD-SCCS) method: a generalization of SCCS that allows the occurrence of an event to increase the future event risk, yet maintains the advantages of the original model by controlling for fixed baseline covariates and relying solely on data from cases. As in the SCCS model, individual-level baseline parameters drop out of the PD-SCCS likelihood. Data sources used for postmarketing surveillance can contain tens of millions of people, so in this situation it is particularly advantageous that PD-SCCS avoids doing a costly estimation of individual parameters. We develop expressions for large sample inference and optimization for PD-SCCS and compare the results of our generalized model with the more restrictive SCCS approach.

Self controlled case series methods: an alternative to standard epidemiological study designs

“…16 If events are dependent, a simple solution is to study just first events (table 1 ). Alternatively, an extension that allows a first event to increase the future event risk may be used 17 …”

Self‐controlled case series studies: Just how rare does a rare non‐recurrent outcome need to be?

“…There are several potential sources of bias that researchers should be mindful of when conducting SCCS or SCRI studies, including time‐varying confounders, small sample estimation bias (Musonda et al., ), systematic bias resulting from outcomes that prohibit or precipitate subsequent exposure (Farrington et al., ; Kuhnert et al., ), systematic bias resulting from outcomes that censor subsequent observation (Farrington et al., ), and bias resulting from outcomes that reasonably do not arise according to a non‐homogeneous Poisson process (Simpson, ), which includes common unique outcomes. Outcomes that potentially censor the observation period, such as myocardial infarction or stroke that carry high mortality risk, have not been mentioned in the present paper, but similar to bias relating to unique common outcomes, the magnitude and direction of bias depends on the distribution of exposure risk windows over observation periods.…”

Self-controlled case series with multiple event types

“…The original SCCS method requires that event times (of a single event type) are independent within individuals. The only context in which the SCCS method has been extended to handle dependence between events is that in which the event rate depends on the previous number of events (Simpson, 2013). A common solution is to analyse only first events.…”

“…This is the case for myocardial infarction, for example. An SCCS model has been proposed to allow the event rate to depend on the previous number of events (Simpson, 2013). However, a simple alternative is to analyse just the first events.…”