Data Source
Data on all residents of Israel who had been fully vaccinated before June 1, 2021, and who had not been infected before the study period were extracted from the Israeli Ministry of Health database on September 2, 2021. We defined fully vaccinated persons as those for whom 7 days or more had passed since receipt of the second dose of the BNT162b2 vaccine. We used the Ministry of Health official database that contains all information regarding Covid-19 (see Supplementary Methods 1 in the Supplementary Appendix, available with the full text of this article at NEJM.org). We extracted from the database information on all documented SARS-CoV-2 infections (i.e., positive result on PCR assay) and on the severity of the disease after infection. We focused on infections that had been documented in the period from July 11 through 31, 2021 (study period), removing from the data all confirmed cases that had been documented before that period. The start date was selected as a time when the virus had already spread throughout the entire country and across population sectors. The end date was just after Israel had initiated a campaign regarding the use of a booster vaccine (third dose). The study period happened to coincide with the school summer vacation.
We omitted from all the analyses children and adolescents younger than 16 years of age (most of whom were unvaccinated or had been recently vaccinated). Only persons 40 years of age or older were included in the analysis of severe disease because severe disease was rare in the younger population. Severe disease was defined as a resting respiratory rate of more than 30 breaths per minute, oxygen saturation of less than 94% while the person was breathing ambient air, or a ratio of the partial pressure of arterial oxygen to the fraction of inspired oxygen of less than 300.14 Persons who died from Covid-19 during the follow-up period were included in the study and categorized as having had severe disease.
During the study period, approximately 10% of the detected infections were in residents of Israel returning from abroad. Most residents who traveled abroad had been vaccinated and were exposed to different populations, so their risk of infection differed from that in the rest of the study population. We therefore removed from the analysis all residents who had returned from abroad in July.
Vaccination Schedule
The official vaccination regimen in Israel involved the administration of the second dose 3 weeks after the first dose. All residents 60 years of age or older were eligible for vaccination starting on December 20, 2020, thus becoming fully vaccinated starting in mid-January 2021. At that time, younger persons were eligible for vaccination only if they belonged to designated groups (e.g., health care workers and severely immunocompromised adults). The eligibility age was reduced to 55 years on January 12, 2021, and to 40 years on January 19, 2021. On February 4, 2021, all persons 16 years of age or older became eligible for vaccination. Thus, if they did not belong to a designated group, persons 40 to 59 years of age received the second dose starting in mid-February, and those 16 to 39 years of age received the second dose starting in the beginning of March. On the basis of these dates, we defined our periods of interest in half months starting from January 16; vaccination periods for individual persons were determined according to the time that they had become fully vaccinated (i.e., 1 week after receipt of the second dose). All the analyses were stratified according to vaccination period and to age group (16 to 39 years, 40 to 59 years, and ≥60 years).
Statistical Analysis
The association between the rate of confirmed infections and the period of vaccination provides a measure of waning immunity. Without waning of immunity, one would expect to see no differences in infection rates among persons vaccinated at different times. To examine the effect of waning immunity during the period when the delta variant was predominant, we compared the rate of confirmed infections (per 1000 persons) during the study period (July 11 to 31, 2021) among persons who became fully vaccinated during various periods. The 95% confidence intervals for the rates were calculated by multiplying the standard confidence intervals for proportions by 1000. A similar analysis was performed to compare the association between the rate of severe Covid-19 and the vaccination period, but for this outcome we used periods of entire months because there were fewer cases of severe disease.
To account for possible confounders, we fitted Poisson regressions. The outcome variable was the number of documented SARS-CoV-2 infections or cases of severe Covid-19 during the study period. The period of vaccination, which was defined as 7 days after receipt of the second dose of the Covid-19 vaccine, was the primary exposure of interest. The models compared the rates per 1000 persons between different vaccination periods, in which the reference period for each age group was set according to the time at which all persons in that group first became eligible for vaccination. A differential effect of the vaccination period for each age group was allowed by the inclusion of an interaction term between age and vaccination period. Additional potential confounders were added as covariates, as described below, and the natural logarithm of the number of persons was added as an offset. For each vaccination period and age group, an adjusted rate was calculated as the expected number of weekly events per 100,000 persons if all the persons in that age group had been vaccinated in that period. All the analyses were performed with the use of the glm function in the R statistical software package.17
In addition to age and sex, the regression analysis included as covariates the following confounders. First, because the event rates were rising rapidly during the study period (Figure 1), we included the week in which the event was recorded. Second, although PCR testing is free in Israel for all residents, compliance with PCR-testing recommendations is variable and is a possible source of detection bias. To partially account for this, we stratified persons according to the number of PCR tests that had been performed during the period of March 1 to November 31, 2020, which was before the initiation of the vaccination campaign. We defined three levels of use: zero, one, and two or more PCR tests. Finally, the three major population groups in Israel (general Jewish, Arab, and ultra-Orthodox Jewish) have varying risk factors for infection. The proportion of vaccinated persons, as well as the level of exposure to the virus, differed among these groups.18 Although we restricted the study to dates when the virus was found throughout the country, we included population sector as a covariate to control for any residual confounding effect.
We conducted several secondary analyses to test the robustness of the results, including calculation of the rate of confirmed infection in a finer, 10-year age grouping and an analysis restricted to the general Jewish population (in which the delta outbreak began), which comprises the majority of persons in Israel. In addition, a model including a measure of socioeconomic status as a covariate was fitted to the data, because this was an important risk factor in a previous study.18 Since socioeconomic status was unknown for 5% of the persons in our study and the missingness of the data seemed to be informative, and also owing to concern regarding nondifferential misclassification (persons with unknown socioeconomic status may have had different rates of vaccination, infection, and severe disease), we did not include socioeconomic status in the main analysis. Finally, we compared the association between the number of PCR tests that had been conducted before the vaccination campaign (i.e., before December 2020) with the number that were conducted during the study period in order to evaluate the possible magnitude of detection bias in our analysis. A good correlation between past behavior regarding PCR testing and behavior during the study period would provide reassurance that the inclusion of past behavior as a covariate in the model would control, at least in part, for detection bias.