Control yourself: the self-controlled case series

Published by


Does influenza increase the risk of acute myocardial infarction (AMI)? Do bivalent mRNA COVID-19 vaccines cause acute ischemic stroke? Relevant questions, but difficult to answer. Incidences of both outcomes are low, precluding an answer from randomised trials (in case of vaccination) and analyses from observational studies will be flawed by confounding.

Two new studies used the self-controlled case series (SCCS) design to answer both questions. In SCCS only individuals who have experienced the outcome of interest are included and they serve as their own control, thereby controlling for confounders that do not change over time. The logic is simple: if influenza increases the risk of AMI, then the incidence of AMI will depend on the occurrence of influenza and will be higher in the week after influenza than in the weeks (or months) before influenza. If it does not, the relative incidence will be similar at any point in time. The same holds for rare events that might be caused by vaccination.

One problem: You need large patient-specific databases covering a time period in which the occurrence of both the outcome (AMI or ischemic stroke) and the time of exposure (influenza or vaccination) are known.

Annemarijn de Boer managed to get results from 158,777 PCR tests for influenza between 2008-2019 (26,221 positive) from 16 Dutch laboratories. Among those subjects, 406 AMI episodes were identified within one year before and one year after confirmed influenza infection. Twenty-five cases of AMI occurred during the first seven days after influenza diagnosis and 394 during the control period, yielding an adjusted relative incidence of AMI in the week after influenza of 6.16 (95% CI 4.11 to 9.24), and even of 16.60 (95% CI, 10.45 to 26.37) in individuals without prior hospitalisation for coronary artery disease. The first finding confirms previous work. In fact, this study was funded through the Replication Studies program of the Dutch Research Council.

The second SCCS study was more complex: the outcome of interest was ischemic stroke diagnosed between September 1, 2022, and March 31, 2023, and the exposures were Pfizer-BioNTech or Moderna bivalent COVID-19 vaccines. Risk intervals were 1-21 days and 1-42 days after vaccination, and subgroups included those that also received influenza vaccination, those that had had COVID-19, and those from 12-65 and above 65 years.

With 4,933 ischemic stroke events, there was no increased risk within the 21-day risk interval for the 2 vaccines. Yet the relative incidence was 2.13 (95% CI 1.01-4.46) within the 42-day risk interval among those <65 years with co-administration of Pfizer-BioNTech bivalent and influenza vaccines on the same day. A small risk, but important to follow-up and to compare with similar studies with different findings.

Both studies offer excellent reading for this powerful methodology that will be applied more frequently as access to large patient-specific databases increases.