Clalit Health Services (CHS) is the largest integrated health care service provider in Israel, with more than 4.7 million active members. CHS maintains a comprehensive health care data warehouse that combines hospital and community medical records, including laboratory and imaging data and data on medication use and health care utilization. Covid-19–related data are collected by the Israeli Ministry of Health for all members. During the study period, children were required to be tested when they had contact with an infected person or if they opted to take part in any activity that required a “green pass” — a certificate issued by the Israeli Ministry of Health that allowed a person who had been vaccinated against Covid-19, had recently been infected and recovered from Covid-19, or had tested negative for Covid-19 on a recent (within the previous 24 hours) antigen test or a recent (within the previous 72 hours) polymerase-chain-reaction (PCR) test to partake in activities such as cultural events or travel abroad.
Study Design and Population
We conducted an observational cohort study emulating a target trial. We enrolled children 5 to 11 years of age who were vaccinated on or after November 23, 2021, when vaccination became available for this age group in Israel. We matched each vaccinated child with an unvaccinated control on the date of vaccination (the recruitment date). The study period ended on January 7, 2022, when a new testing policy was implemented in Israel (Fig. S1 in the Supplementary Appendixavailable with the full text of this article at NEJM.org). This new policy required unvaccinated persons to be tested in an official setting supervised by the Ministry of Health, whereas vaccinated persons were permitted to conduct rapid antigen tests at home. We examined testing rates of both vaccinated and unvaccinated populations before and after the change in testing policy on January 7, 2022, to gauge the effect of this change.
We evaluated vaccine effectiveness against documented SARS-CoV-2 infection, as confirmed by a positive PCR test, and against symptomatic Covid-19, which was defined as a PCR-confirmed infection with a report of any Covid-19 symptom in the patient’s electronic health record. Details of outcome definitions are provided in Table S1. We estimated vaccine effectiveness for documented infection and symptomatic Covid-19 over two separate periods — from 14 to 27 days after the first dose and from 7 to 21 days after the second dose. To assess the changing dominance of omicron variant infections in Israel over the study period, we estimated the daily proportion of omicron cases on the basis of a sample of positive PCR test results that had been collected and sequenced by the Israeli Ministry of Health.
Children were eligible for the study if they were 5 to 11 years of age at the time of recruitment; had at least 12 months of continuous membership in CHS before recruitment; had no previous PCR, serology, or antigen test that was positive for SARS-CoV-2; had a valid residence location and assignment to a population sector; were not homebound because of medical reasons; and had no interaction with the health care system (physician appointment, hospitalization, or laboratory testing) in the 3 days preceding the recruitment date, since such interaction would potentially be an indication of a developing case of symptomatic Covid-19. On each day of recruitment, all newly vaccinated children who met the inclusion criteria were matched one to one with eligible unvaccinated children on the basis of age, sex, population sector, residential area, number of influenza vaccines received in the past 5 years, overweight status, and number of diagnosis codes in the patient’s medical record that were considered by the physician to represent chronic conditions. Thousands of diagnosis codes are included in the variable that captured background conditions; therefore, when describing the study population, we provided estimates of grouped diagnoses and key single conditions that were defined by the Centers for Disease Control and Prevention as risk factors for severe Covid-19. Definitions of population characteristic variables are provided in Table S2. This study was approved by the institutional review board at CHS, and an exemption from the requirement for informed consent from the participants was granted.
All the authors conceived of and designed the study. A subgroup of authors collected and analyzed the data and wrote the manuscript. All the authors critically reviewed the manuscript. The first and last authors supervised the study and vouch for the accuracy and completeness of the data and that the conduct of the analyses followed the study design. No one who is not an author participated in the writing of the manuscript. The funding institutions did not dictate the design of the study, have access to the data, or influence the decision to submit the manuscript for publication.
In the estimation of the per-protocol effect of vaccination, data from both members of a matched pair were censored if and when the control received a vaccination; such censoring was done in order to maintain the comparability of the two study groups with respect to the matching factors. The control for whom data were censored could be rerecruited as a vaccinated person if a new matched unvaccinated control was found. Follow-up also ended when a person had an outcome or died or the end of the study period was reached (in the latter two circumstances, the patient data were considered to have been censored).
The Kaplan–Meier estimator was used to estimate the cumulative incidence (risk) of each outcome in the vaccinated and unvaccinated groups. Risk ratios and risk differences were calculated by dividing and subtracting the period-specific risk estimates, respectively. Vaccine effectiveness was defined as 1 minus the risk ratio, and 95% confidence intervals were estimated with the use of the nonparametric percentile bootstrap method with 500 repetitions. Estimates in each period included only matched pairs in which both members of the pair were still undergoing follow-up at the beginning of the period relevant for that analysis. We also conducted subgroup analyses according to age (5 or 6 years, 7 to 9 years, and 10 or 11 years), examining the cumulative incidence of and comparing the vaccine effectiveness against documented and symptomatic infections 7 to 21 days after the second dose. All analyses were performed with the use of R software, version 220.127.116.11