Avoiding bias in self-controlled case series studies ofcoronavirus disease 2019.

Abstract

Many studies, including self‐controlled case series (SCCS) studies, are being undertaken to quantify the risks of complications following infection with severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2), the virus that causes coronavirus disease 2019 (COVID‐19). One such SCCS study, based on all COVID‐19 cases arising in Sweden over an 8‐month period, has shown that SARS‐CoV‐2 infection increases the risks of AMI and ischemic stroke. Some features of SARS‐CoV‐2 infection and COVID‐19, present in this study and likely in others, complicate the analysis and may introduce bias. In the present paper we describe these features, and explore the biases they may generate. Motivated by data‐based simulations, we propose methods to reduce or remove these biases.