Vaccination in Estonia began in January 2021, with a cumulative vaccination uptake about 70% among adult population by June 2022. Within the time period of data underlying the present study, Estonia had three large pandemic waves: the first was from March to June 2020 (SARS-CoV-2 prevariant of concern era); the second was from November 2020 to May 2021 (first the Alpha variant, then the Delta variant); and the third was from December 2022 (Omicron variant)25. Our analysis used data derived from the nationwide and population-based universal tax-funded Estonian health care system.
We conducted a retrospective cohort study based on linking individual-level data on laboratory-confirmed COVID-19s, SARS-CoV-2 vaccination status, and health care utilization between 26 February 2020 and 23 February 2022 from the national e-health records26.
The Health and Welfare Information Systems Centre (TEHIK)
Data on COVID-19 vaccination (dates), SARS-CoV-2 testing (dates) and laboratory confirmed (real-time polymerase chain reaction (PCR) or antigen testing) cases of SARS-CoV-2 infections (dates) were retrieved from TEHIK26. According to law, all health-care providers and laboratories in Estonia are obligated to report to TEHIK, with an expected coverage of 100%.
The Estonian Health Insurance Fund (HIF)
By the end of 2021, universal public health insurance covered 95.2% of the Estonian population (1,328,889 people)27. The HIF maintains a complete record of the health care services provided. Diagnoses are defined according to the International Classification of Diseases, tenth revision (ICD-10). The HIF database records sex, age, health care utilization information (dates of service, diagnoses, treatment type: in- or outpatient), and date of death.
The Population Register is a unified database of Estonian citizens and foreign nationals living in Estonia based on right of residence or residence permits. Population Register data were used to identify the study subjects’ education and ethnicity.
The databases are linked using a unique personal code given to all persons living in Estonia.
Our population was drawn from 329,496 individuals aged 18 years or older. Based on various histories of immunity-conferring events (i.e., infection and/or vaccination) from 26 February 2020 to 25 June 2021, we determined four exposure states:
Individuals with no immunity (SARS-CoV-2 immune-naïve) were defined as those who were unvaccinated and did not have documented previous SARS-CoV-2 infections (n = 130 874);
Individuals with natural immunity (the recovered, unvaccinated cohort) were those with a documented previous infection but without previous vaccination (n = 47,491);
Individuals with vaccine-induced SARS-CoV-2 immunity (vaccinated-only cohort) were those without previously recorded infections who received a full vaccination course (BNT162b2; mRNA-1273; AZD1222; Ad26.COV2. S) (n = 127,460); and
Individuals with hybrid SARS-CoV-2 immunity (the recovered, vaccinated cohort) were defined as those with documented previous infections who received at least one vaccine dose (n = 23,671).
Construction of study cohorts
We constructed three mutually exclusive cohorts to assess the risk of SARS-CoV-2 (re)infection and COVID-19 hospital admission. Each cohort consisted of two subcohorts with different types of SARS-CoV-2 immunity (see Fig. 1).
Cohort 1 was formed to compare people with natural SARS-CoV-2 immunity to those without SARS-CoV-2 immunity. All individuals with natural immunity were randomly matched (1:1, without replacement) by birth year and sex to unvaccinated individuals with no immunity at baseline (for this cohort, the date of the positive SARS-CoV-2 test for individuals with natural immunity).
Cohort 2 was formed to compare hybrid SARS-CoV-2 immunity with natural immunity. Hybrid immunity was defined as having a documented previous infection and a single vaccine dose either before or after infection or having received two or more vaccine doses, with at least the second dose given after infection (One individual had received three vaccine doses).
All individuals with hybrid immunity were matched to those with natural immunity in a 1:1 ratio (with replacement) based on sex, birth year, and time. For this cohort, the baseline date was defined as (the date of the last immunity-conferring event for individuals with hybrid immunity. Matched subjects with natural immunity had to be alive, previous infected and unvaccinated on the baseline date. Matching was performed as an iterative process until all subjects with hybrid immunity had a suitable match of individuals with natural immunity only (n = 23,580 individuals). The follow-up started on the baseline date for both individuals in the matched pair.
Cohort 3 was formed to compare vaccine-induced immunity (vaccine only) to natural immunity. Those with vaccine-induced immunity had received at least one vaccine dose (30 individuals had received 3 or more vaccine doses). All individuals with vaccine-induced immunity were matched to those with natural immunity (n = 45,888 individuals) using the same principles as in the second cohort. The follow-up started at baseline (the date of the last immunity-conferring event of the individual with vaccine immunity) for both individuals in the matched pair.
The matching procedure resulted in 246,113 individuals being matched into the three cohorts.
The primary outcome was laboratory confirmed SARS-CoV-2 infection occurring after the baseline date: (i) at any time for individuals with no immunity; (ii) after 60 days of recovery from a previous infection for individuals with natural immunity (i.e., reinfection)23; (iii) after being vaccinated for at least 14 days for individuals with vaccine-induced immunity (SARS-CoV-2 vaccination only) (i.e., breakthrough infection); and (iii) after being vaccinated for at least 14 days or after 60 days28 of recovery from a previous infection, whichever came later, for individuals with hybrid immunity.
The second outcome was hospitalization with COVID-19 as the reason for admission. This was defined as SARS-CoV-2-related hospitalization occurring from 3 days before to 14 days after a positive SARS-CoV-2 test and the presence of at least one of the following diagnoses (ICD-10) in relation to hospitalization: U07.1, U07.2, acute respiratory tract infections (J00–J06, J12, J15-J18, J20-J22, J46) or severe complications of lower respiratory tract infections (J80–84, J85–J86)29.
The follow-up duration was counted in days until the date of an outcome (the date of the positive SARS-CoV-2 test), vaccination (for individuals with natural or no immunity), death, or end of the study period (23 February 2022), whichever occurred first.
Variables accounted for in the risk model
The number of SARS-CoV-2 tests a person received throughout the pandemic was accounted for by counting the number of tests that an individual underwent from baseline to the end of the study. We defined three individualized testing intensities (< 1, 1–1.99, ≥ 2 per 100 person-days).
The comorbidity status was computed based on health data within 12 months before the baseline date using the Charlson comorbidity index (CCI)30, and study subjects were divided into three groups comprising those with no (CCI score of 0), one or two (CCI score of 1 or 2) or at least three (CCI score ≥ 3) comorbid conditions.
The follow-up period from baseline was split into four segments: up to 2, 2–4, 4–6, and 6–8 months. The number of and the time since the last immunity-conferring event (since SARS-CoV-2 infection or vaccination) within these groups were also quantified.
We analysed data from two time periods: from the start of follow-up until 19 December 2021, when the Delta variant was the predominant circulating SARS-CoV-2 strain (proportion of sequenced strains 93%); and 20 December 2021 to 23 February 2022 (end of follow-up), when the Omicron variants (BA1, BA2 and their sublineages) were the predominant strains (proportion of sequenced strains 88%)25.
Frequencies and proportions for categorical variables, means and standard deviations (SD) for age, and median and range for baseline date were used to characterize the study cohorts (Table 1). The follow-up duration is presented in months. The number of confirmed infections and the crude incidence rates (IRs) per 100 person-years were counted for each cohort (Tables S1–S3). Cumulative Kaplan‒Meier curves are presented to describe SARS-CoV-2 infections in cohorts by different subcohorts (Fig. 2).
We performed Cox regression with the SARS-CoV-2 infection or COVID-19 hospitalization as the dependent variable and sex, age group (18–49, 50–64, 65–79, 80 + years), education (higher, < higher), nationality (Estonian, other), CCI score, time (in months) since the last conferring event and number of conferring events, and SARS-CoV-2 testing intensity as independent variables. Multivariable-adjusted hazard ratios (aHRs) and their 95% confidence intervals (CIs) are presented (Tables 2, 3, 4) (see Supplement for additional information on data analysis).
A p value of less than 0.05 was considered to indicate statistical significance in all analyses.
Data analysis was performed with the statistical software Stata 17.0.
The Research Ethics Committee of the University of Tartu approved our study and waived the requirement for informed consent. Whole research was performed in accordance with the relevant guidelines and regulations.