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

Target Trial Emulation

Study citation: Dagan N, Barda N, Kepten E, Miron O, Perchik S, Katz MA, Hernán MA, Lipsitch M, Reis B, Balicer RD. BNT162b2 mRNA Covid-19 Vaccine in a Nationwide Mass Vaccination Setting. N Engl J Med. 2021 Apr 15;384(15):1412-1423. doi: 10.1056/NEJMoa2101765. Epub 2021 Feb 24. PMID: 33626250; PMCID: PMC7944975.1

Study objective: The objective of this study was to rapidly characterize and evaluate the effectiveness of the BNT162b2 mRNA vaccine in a real-world setting based on emulation of the target trial of the causal effect. As noted by the authors, COVID-19 vaccination is expanding globally, and understanding vaccine performance in a real-world setting is of paramount importance to patients, public health officials, researchers, and other stakeholders.

PICO: The study population was derived from Clalit Health Services (CHS) which includes 53% of the insured population in Israel. The outcomes of interest were occurrence of documented SARS-CoV-2 infections, symptomatic COVID-19, related hospitalizations, severe COVID-19, and death from COVID-19 among both vaccinated and unvaccinated populations. The intervention was receipt of the BNT162b2 mRNA vaccine, based on availability and individual agreement to receive a vaccination. Patients were matched 1:1 based on clinical and non-clinical characteristics.

Data source: CHS is a large health care organization in Israel, covering approximately four million citizens—just over half of Israel’s total population.

Study period: December 20, 2020, to February 1, 2021

Key sources of error and how they were handled: This study, using the target trial paradigm, is one of the largest population-based studies available and allows for insight into the real-world effectiveness of the BNT162b2 mRNA vaccine. COVID-19 diagnoses were confirmed with gold-standard polymerase chain reaction (PCR) testing, which is an important element of case definition for other large real-world studies as there can be significant clinical overlap between COVID-19 and other viral respiratory illnesses resulting in potential outcome misclassification. To mitigate the potential for confounding, the authors were able to balance a wide range of factors, including described risk factors for severe COVID-19 disease, by matching patient cohorts on demographic and clinical characteristics. A number of sensitivity analyses were performed, accounting for potential biases. Date of symptom onset was unavailable for this study and presents a minor limitation. There was limited race and ethnicity data appropriate to the region of the study population and the ability to assess treatment effect heterogeneity across other populations is limited. Additionally, several patient populations (those without documented addresses, health care workers, etc.) were excluded from the study for concern of skewing results in the case these individuals were exposed to a greater degree than the general population. Overall, 34% of the otherwise-eligible vaccinated population was excluded due to the rigorous matching performed. This is important to consider when extrapolating these data to other population groups.