Age and product dependent vaccine effectiveness against SARS-CoV-2 infection and hospitalisation among adults in Norway: a national cohort study, July–November 2021

For this population-based cohort study, we linked data from the Norwegian National Preparedness Register for COVID-19 (Beredt C19) (Additional file 1: Table S1). This register is a “data lake” in which several different publicly owned data sources, like the central health registers and national administrative registers, are gathered and made available to the Norwegian Institute of Public Health. All residents in Norway have a unique personal identification number. This repository allows for the real-time surveillance and analysis of all individual-level data relating to the pandemic. For this study, we extract and link data from six different sources—the National Population Registry (NPR), The Norwegian Immunisation Register (SYSVAK), The Norwegian Intensive Care and Pandemic Registry (NIPaR), The Norwegian Surveillance System for Communicable Diseases (MSIS), Statistics Norway (SSB), and an internally created table of risk group membership. These registers include and cover the entire Norwegian population, and the reporting of testing, test results, vaccinations, hospitalisations, and mortality are mandatory by law and considered complete. We included all adults (≥ 18 years by the end of 2021) with a valid national identity number and registered in the NPR as living in Norway. To remove non-standard vaccination histories, we removed individuals with more than three doses before the end of the study period and excluded individuals for which the interval between first and second dose was shorter than the recommended minimum intervals and censored those with a third dose registered before the recommended 120 days of the second dose. The recommended minimum interval between first and second dose was based on the vaccine type given as the first dose; 19 days for Comirnaty, 22 days for Spikevax, and 21 days for Vaxzevria. We only included individuals who had received either of the three vaccines that are part of the Norwegian vaccination programme (Comirnaty, Spikevax or Vaxzevria). Finally, individuals admitted to hospital with COVID-19 without a corresponding match in the database for the time of positive test were excluded. Data were extracted from the registries on 15 February 2022.

SARS-CoV-2 infection is defined as a positive SARS-CoV-2 PCR test reported to the MSIS registry, which has been complete for all domestic laboratories since April 2020. Individuals who were hospitalised or admitted to an intensive care unit (ICU) with COVID-19 are registered in NIPaR, which covers mandatory reporting from all Norwegian hospitals [25]. We included all hospitalisation where COVID-19 was registered as the primary diagnosis for admission and ICU admission of individuals who tested positive for SARS-CoV-2 and were admitted to an ICU (length of stay ≥ 24 h), required mechanical ventilatory support (invasive or non-invasive), or persistent administration of vasoactive medication. All COVID-19-associated deaths are defined as anyone with a positive SARS-CoV-2 PCR test who died with COVID-19 reported on the death certificate in the Cause of Death Register (DÅR) or those notified directly to MSIS. We use testing date as time of infection (positive PCR test) and vaccination status is determined at time of infection for all outcomes. Individual vaccination histories were generated from SYSVAK and categorised into the following vaccination statuses:

The vaccine regimens included were Comirnaty, Spikevax, heterologous mRNA regimen, Vaxzevria, or Vaxzevria in combination with an mRNA vaccine, all with or without an mRNA booster. The period between seven days before the first vaccine dose until 21 days after the first dose was included as a separate status not reported, since vaccination was postponed if individuals showed signs of infection which could potentially bias infection rates for both unvaccinated and partially vaccinated if included in the adjacent vaccination statuses. Similarly, individuals were also included as a separate status and not reported for the first 7 days after receiving the third dose. Individuals with a SARS-CoV-2 infection registered prior to 1 June 2021 were included as a separate category (see below).

To adjust for confounding, several covariates were included in our analysis. When stratifying by age, we used the categories 18 to 44 years, 45 to 64 years, and 65 years or older. For adjustment in other models, we used 10-year age bands (NPR). County of residence (NPR) was included as both infection rates and speed of vaccine rollout has varied across Norway. We included country of birth (NPR; Norway, abroad or unknown) and crowded living conditions (SSB; crowded, not crowded, or unknown) since both are associated with vaccine coverage and risk of infection. Individuals with pre-existing medical conditions associated increased risk of severe COVID-19 illness were prioritised for vaccination and this covariate was also included in the adjusted model. Missing values were considered as a separate category for each of the variables where relevant. More details on the data sources and variables can be found in the supplementary information as well as the number of infections, hospitalisations, and vaccination status over time (Additional file 1: Table S1, Fig. S1 and Fig. S2).

We estimated the vaccine effectiveness using Cox proportional hazards models on an open cohort, using vaccine status as time-varying covariate for all individuals included in the statistical software R [26]. We included all SARS-CoV-2 infections reported from 15 July until 30 November 2021, the period in which the delta variant was dominating in Norway [10]. We right-censored individuals at the time of an event (SARS-CoV-2 infection, hospitalisation, ICU admission, or death associated with COVID-19), time of death (all cause), or end of follow-up period (30 November 2021). During the study period, the registries in Norway only report re-infections if 6 months or more since last positive test, individuals registered with an infection prior to 15 July 2021 entered the dataset 180 days since positive test with status ‘previously infected’.

Vaccine effectiveness is defined as 100*(1–β), where β represents the hazard ratio associated with a particular vaccine status. For crude vaccine effectiveness estimates, we only used vaccine status as a time-varying covariate (supplementary analyses; Additional file 1: Tables S2-S5). For adjusted estimates, we implemented stratified analyses using strata(variable) in the survival-package [27], i.e. that the impact of the adjustment variables can be non-proportional.

To estimate specific vaccine effectiveness for age groups and vaccine product regimens, we use independent Cox-models while still adjusting for the remaining covariates as strata. Vaccine status was factored either by combining all vaccine types, including Comirnaty, Spikevax, and Vaxzevria (thus assuming similar effectiveness across vaccines) to estimate a population level vaccine effectiveness of the vaccination programme in Norway or by implementing vaccine status as the combination of vaccine type and vaccine status. Due to the smaller numbers and partial exclusion of Vaxzevria from the vaccination programme in Norway, all individuals who received a dose of Vaxzevria were censored at time of the dose in the product-specific analyses. Models were also run excluding all unvaccinated individuals who have never had a recorded SARS-CoV-2 PCR test in Norway, results from these models can be found in the supplementary materials (Additional file 1: Tables S2-S6).

The adjusted vaccine effectiveness against infection was high in the first period (2 to 9 weeks) after the second dose (81.3%, CI: 80.7 to 81.9). Effectiveness waned with time since vaccination and more than 33 weeks after receiving the second dose the effectiveness against infection was 8.6% (CI: 4.0 to 13.4). Similarly, the effectiveness against hospitalisation was 98.6% (CI: 97.5 to 99.2) in the period right after receiving the second dose, decreasing to 66.6% (CI: 57.9 to 73.6) after more than 33 weeks. For admission to ICU and death, not enough events in the first period following the second dose had occurred to reliably estimate a vaccine effectiveness, but in the following period (10 to 17 weeks), vaccine effectiveness was 96.9% (CI: 94.7 to 98.1), and 93.4% (CI: 85.4 to 97.0) against ICU and death respectively, with a less pronounced reduction over time for ICU admissions (86.7%, CI: 73.9 to 93.2 after more than 33 weeks) than for death (68.6%, CI: 55.4 to 77.9). One dose provided little protection against infection (30.0%, CI: 28.2 to 31.9) but did protect against hospitalisation (79.4%, CI: 73.0 to 84.2) and ICU admission (92.4%, CI: 80.9 to 97.0). The estimates against death were should not be used for inference as they have wide confidence intervals (46.9%, CI: − 0.2 to 71.9), likely due to small sample. The vaccine effectiveness against infection after receiving a third dose (75.9%, CI: 73.4 to 78.1) was similar to the effectiveness in the initial two to nine weeks after the second dose (81.3%, CI: 80.7 to 81.9), albeit based on a relatively small sample (Additional file 1: Table S2) with short follow-up time (median: 13 days, Q1–Q3: 6–20 days). For those with a previous reported infection (> 6 months prior), the protection against infection was 93.1% (CI: 92.3 to 93.9), whereas too few events amongst those with a reported prior infection were reported to estimate effectiveness against hospitalisation. Vaccine effectiveness for all outcomes split by time is shown in Fig. 1 (details can be found in Additional file 1: Tables S2-S5). The vaccine effectiveness estimates were used to estimate the overall cohort-wide level of protection against SARS-CoV-2 infection over time, showing the overall impact of waning and booster doses (Additional file 1: Fig. S3).

Vaccine effectiveness against infection was highest in 2 to 9 weeks after the second dose amongst 18- to 44-year-olds (83.2%, CI: 82.6 to 83.8) compared to 45- to 64-year-olds (75.6%, CI: 74.1 to 77.0) and those over 64 years (74.9%, CI: 67.2 to 80.7). No significant protection was found more than 33 weeks after the second dose: the estimated effectiveness against infection was 5.2% (CI: − 1.9 to 11.8) for 18- to 44-year-olds, 0.5% (CI: − 9.4 to 9.5) amongst 45- to 64-year-olds, and 8.4% (CI: − 2.8 to 18.5) amongst those over 65 years (Fig. 2, Additional file 1: Table S6). Amongst those with a reported prior infection, protection against infection was 92.6% (CI: 91.5 to 93.5), 95.1% (CI: 93.6 to 96.3), and 89.0% (CI: 81.3 to 93.5) for 18- to 45-year-olds, 45- to 64-year-olds, and over 65-year-olds respectively (Fig. 2). The protection against hospitalisation was already significant after one dose in all age groups: 82.9% (CI: 73.0 to 89.2) amongst 18- to 44-year-olds, 71.4% (CI: 56.2 to 81.4) amongst 45- to 64-year-olds, and 56.5% (CI: 24.6 to 74.9) amongst those over 65 years. The effectiveness against hospitalisation decreased less with time than protection against infection (Fig. 2). For 45- to 64-year-olds the effectiveness was 99.1% (CI: 97.7 to 99.6) in 2 to 9 weeks after the second dose, compared to 65.3% (CI: 33.3 to 82.0) more than 33 weeks after the second dose. Similarly, amongst those above 65 years, the protection waned from 93.6% (CI: 89.9 to 96.0) to 61.9% (CI: 50.1 to 70.9) (Fig. 2, Additional file 1: Tables S6 - S7). Receiving a third dose increased vaccine effectiveness against hospitalisation to 85.4% (CI: 65.3 to 93.9) amongst 45- to 64-year-olds and 95.3% (CI: 92.6 to 97.0) for over 65-year-olds; the number of events were too small for 18- to 44-year-olds (Fig. 2, Additional file 1: Tables S6 - S7).

When stratifying by product regimen, individuals who received two doses of Spikevax (86.6%, CI: 85.6 to 87.6) or a heterologous mRNA regimen (84.1%, CI: 83.2 to 85.0) had a higher estimated vaccine effectiveness against infection than those who received two doses Comirnaty (77.7%, CI: 76.8 to 78.5) in 2 to 9 weeks after the second dose (Fig. 3, Additional file 1: Table S8). All product regimens showed waning of vaccine effectiveness against infection (Fig. 3). The vaccine effectiveness against infection after receiving the third dose was highest for those who received two doses of Spikevax followed by a booster with Comirnaty (87.1%; CI: 80.1 to 91.6) or Spikevax (84.9%; CI: 71.8 to 91.9), compared to 75.3% (CI: 72.5 to 77.8) and 68.2% (CI: 57.6 to 76.1) for those who received two doses of Comirnaty followed by a booster with Comirnaty or Spikevax respectively. The vaccine effectiveness against hospitalisation was high after one dose (77.1% and 75.3%) for both Spikevax and Comirnaty, as well as one to 32 weeks after the second dose (range 81.8 to 97.5%). There were only five hospital admissions amongst those who received heterologous mRNA vaccination during our study period and therefore vaccine effectiveness against hospitalisation since time of vaccination could not be estimated. Amongst those who received a booster dose, vaccine effectiveness against hospitalisation could only be estimated for those with primary regimen of Comirnaty and protection was high 95.6% (CI: 93.1 to 97.2) for those receiving a Comirnaty booster and slightly lower but more uncertain for those receiving Spikevax (73.5%, CI: 45.7 to 87.1) (Fig. 3, Additional file 1: Table S9). Supplementary analyses split by age and product regimen showed similar trends (Additional file 1: Fig. S4).