Chapter 4: Examples of COVID-19 Real-World Evidence Studies

Self-Controlled Case Series Study

Study citation: Katsoularis I, Fonseca-Rodríguez O, Farrington P, Lindmark K, Fors Connolly AM. Risk of acute myocardial infarction and ischaemic stroke following COVID-19 in Sweden: a self-controlled case series and matched cohort study. Lancet. 2021 Aug 14;398(10300):599-607. doi: 10.1016/S0140-6736(21)00896-5. Epub 2021 Jul 29. PMID: 34332652; PMCID: PMC8321431.4

Study objective: As the science and understanding of COVID-19 has evolved, its effect on specific organ systems (e.g., vascular) requires study. This study sought to characterize the risk of both myocardial infarction and ischemic stroke for those infected with SARS-CoV-2.

PICO: This study used two methods for its analysis: a self-controlled case series (SCCS) and a matched cohort study of patients in Sweden who received a prior diagnosis of COVID-19, and later received diagnosis codes related to a myocardial infarction or ischemic stroke. At most time points of the study, prior laboratory confirmed SARS-CoV-2 infection was identified as a risk factor for these 2 vascular events.

Data source: Patient data were derived from SmiNet at the Swedish Public Health Agency (including all patients registered as having COVID-19 until September 14, 2020) and linked with registries administered by the Swedish National Board of Health and Welfare. Like many other public health databases, SmiNet has been used to help track COVID-19 in Sweden since February 2020, as COVID-19 has been a notifiable disease in Sweden, and diagnosed individuals are reported to SmiNet (Swedish Public Health Agency) daily. The source has suitable coverage for the population of interest.

Study period: February 1 to September 14, 2020

Key sources of error and how they were handled: All individuals in Sweden were eligible for inclusion, which is a particular strength of this nationally representative study, given the comprehensive population level data capture in this RWD source, and making it the largest study evaluating this association at the time of publication. Another strength was the use of 2 different designs (SCCS and conventional matched cohort), which rely on different sets of assumptions and address different types of biases. The use of the 2 designs contributed to the ability to control for bias using different assumptions about pathways for biases. This study also adjusted for income, education level, and country of origin, which are important demographic and socioeconomic factors to consider in real-world evidence studies of COVID-19 sequelae. Generalizability from a single country study may be limited, and RWD studies of COVID-19 outcomes will also benefit from understanding race and ethnicity when available. Additionally, the authors noted that there was a peak of vascular events on day 0 (date of vaccination). A sensitivity analysis that excluded patients with events on day 0 found similar results to the main analysis. This speaks to the robustness of the study results.