https://orcid.org/0000-0001-7274-9641
Figures
Abstract
Background
While multiple studies report that the BNT162b2 XBB.1.5-adapted mRNA COVID-19 vaccine (BNT162b2 XBB vaccine) is effective in preventing COVID-19 hospitalisation and death, effectiveness beyond six months remains unexplored.
Methods
We extended our previous study of BNT162b2 XBB vaccine effectiveness (VE) to evaluate durability against JN.1-related hospitalisation up to 46 weeks since dose using the id.DRIVE platform across Europe. This multi-country, multi-centre test-negative case-control study assessed the effectiveness and durability of the BNT162b2 XBB vaccine in preventing JN.1-associated hospitalisation among adults with severe acute respiratory infection between October 2023 and August 2024. Each case was matched with up to four controls by symptom onset date and study site. Multivariable analyses were adjusted for symptom onset date, age, sex, number of chronic conditions, and influenza vaccination receipt.
Results
Among 827 test-positive cases and 2232 test-negative controls, protection against hospitalisation was sustained from two to <30 weeks since dose, with evidence of significant waning thereafter. VE was 64.5% (95% CI: 56.6; 71.0) at two to <30 weeks, and 4.9% (95% CI: −30.3; 30.7) at 30 to <46 weeks.
Conclusion
Despite the vaccine target not matching the predominant subvariant, BNT162b2 XBB vaccine protected against JN.1-related hospitalisation for up to 30 weeks. Protection against hospitalisation was non-significant after 30 weeks since dose, potentially due to further shift in circulating SARS-CoV-2 strains and/or waning immunity. Given the high COVID-19 activity in Europe during summer 2024, an additional vaccination after six months is warranted for those at risk of COVID-19 hospitalisation to maintain year-round protection.
Citation: Volkman HR, de Munter L, Nguyen JL, Tran TMP, Mitratza M, Marques C, et al. (2026) Durability of the BNT162b2 XBB:1.5-adapted vaccine against JN.1 hospitalisation in Europe, October 2023 to August 2024: A test-negative case-control study using the id.DRIVE platform. PLoS One 21(2): e0342382. https://doi.org/10.1371/journal.pone.0342382
Editor: Haihai Jiang, Nanchang University, CHINA
Received: July 9, 2025; Accepted: January 21, 2026; Published: February 11, 2026
Copyright: © 2026 Volkman et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The data supporting the findings of this study are not publicly available due to the inclusion of identifiable protected health information. Privacy regulations and ethical obligations prevent the open sharing of individual-level data. Access to de-identified data that underlie the results reported in this article (text, tables, figures, and appendices) may be obtained in accordance with id.DRIVE data access policy, as outlined in the id.DRIVE Governance Charter, which is available upon request (info@iddrive.eu).
Funding: This study was conducted as a collaboration between Pfizer Inc. and P95. Pfizer Inc. funded this study. HRV, JLN, C Marques, LC, SV, JY, LJ, and JMML are employees of Pfizer Inc. HRV, JLN, LC, SV, JY, LJ, and JMML hold stocks or stock options in Pfizer Inc. and were provided support in the form of salaries from Pfizer Inc. These authors from Pfizer Inc. were involved in the study design, data collection, data interpretation, and writing of the manuscript, and their specific roles are articulated in the ‘author contributions’ section. P95 is the study sponsor. EC-S, KB, LD, LdM, MM, and TMPT are employees of P95 Clinical and Epidemiology Services, which received funding from Pfizer Inc. in connection with the development of this manuscript. Pfizer Inc., Novavax, and Valneva provided funding for the data collection of the study reported in this publication.”.
Competing interests: Pfizer Inc. funded this study. HRV, JLN, C Marques, LC, SV, JY, LJ, and JMML are employees of Pfizer Inc. HRV, JLN, LC, SV, JY, LJ, and JMML hold stocks or stock options in Pfizer Inc. EC-S, KB, LD, LdM, MM, and TMPT are employees of P95 Clinical and Epidemiology Services, which received funding from Pfizer in connection with the development of this manuscript. TMPT declares the possibility of owning stock or stock options in P95. KB has received consulting fees from Pfizer Inc. for conducting this study. AOS declares payment or honoraria for lectures, presentations, speaker bureaus, manuscript writing, or educational events from GSK, MSD, and Sanofi. AOS also reports support from GSK, MSD, and Sanofi to their institution for attending meetings and/or travel. IC reports support from Pfizer for congress attendance. AM-I reports being associated with the Fisabio foundation. Fisabio received funding from P95 via id.DRIVE to conduct the study. GR declares being the chairman of the executive board (unpaid) of the CAPNETZ foundation. Furthermore, GR reports personal fees from AstraZeneca, Atriva, Boehringer Ingelheim, GSK, Insmed, MSD, Sanofi, Novartis, and Pfizer for consultancy during advisory board meetings and personal fees from AstraZeneca, Berlin Chemie, BMS, Boehringer Ingelheim, Chiesi, Essex Pharma, Grifols, GSK, Insmed, MSD, Roche, Sanofi, Solvay, Takeda, Novartis, Pfizer, and Vertex for lectures. SO-R declares consulting fees from Moderna and Sanofi, and payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing, or educational events from GSK and AstraZeneca. Moreover, SO-R reports support from Pfizer for attending meetings and/or travel. AA, BG, C Martin, EC-S, GI, GLtK, LdM, LD, and MM declare no conflicts of interest. This does not alter authors’ adherence to PLOS ONE policies on sharing data and materials.
Introduction
Since its authorisation by the European Medicines Agency (EMA) on 30 August 2023 [1], numerous real-world studies have reported on the vaccine effectiveness (VE) of the BNT162b2 XBB.1.5-adapted mRNA COVID-19 vaccine (Pfizer/BioNTech 2023–2024 formulation, hereafter referred to as the BNT162b2 XBB vaccine) during the 2023–2024 season [2–20]. These studies collectively show that the BNT162b2 XBB vaccine is effective against severe endpoints such as hospitalisation [2–5,8,9,11,12,15–18], ICU admission [4], and death [3,17]. However, its effectiveness in preventing hospitalisation is likely lower against Omicron subvariant JN.1 than XBB (VE of 61%−76% during XBB predominance and VE of 35%−54% during JN.1 predominance) [2–5,9,11,12,17].
Despite this broad evidence base, the long-term durability of BNT162b2 XBB vaccine beyond 6 months since dose remains unknown. This gap in evidence on long-term effectiveness is especially relevant considering that from June–August 2024, many countries in Europe and around the globe experienced an increase in COVID-19 hospitalisations [21], long after Northern Hemisphere autumn/winter-focused COVID-19 vaccination campaigns had concluded. Changes in COVID-19 dynamics, including SARS-CoV-2 strain shifts [22], complex and evolving population immunity [23], and aseasonal/acyclical increases in hospitalisations [24], have continued since the conclusion of the World Health Organization (WHO)-declared global health emergency [21,25]. Therefore, it is critical for public health decision-makers to have timely information on the long-term durability of COVID-19 vaccines.
Therefore, we sought to extend the study period of our previously published real-world effectiveness analysis of the BNT162b2 XBB vaccine against JN.1-associated hospitalisation across four European countries (Belgium, Germany, Italy, and Spain) [12]. Our previous report indicated that the BNT162b2 XBB vaccine was 54.8% (95% CI: 39.7; 66.0) effective against JN.1-related hospitalisation after a median of 9 weeks (63 days) following vaccination, with no evidence of waning protection through 22 weeks [12]. In this analysis, we measured the long-term durability of the BNT162b2 XBB vaccine, both overall and in risk groups, up to 46 weeks post-dose using the id.DRIVE study platform during a period of JN.1 predominance.
Methods
Study design
As previously described in our analysis of BNT162b2 XBB VE through April 2024 [12], id.DRIVE is an ongoing multi-country, multi-centre test-negative case-control study that began in Europe in November 2020. The study aims to estimate VE against COVID-19-related hospitalisations with laboratory confirmation of SARS-CoV-2 infection among patients admitted with severe acute respiratory infections (SARI). The id.DRIVE Master protocol (EUPAS42328) was approved by independent ethics committees (IECs) at each study site. Patients with SARI were enrolled both prospectively and retrospectively across eight sites, including six hospitals and two hospital networks across Belgium, Germany, Italy, and Spain from 2 October 2023–31 August 2024 (S1 Table in S1 File).
Study objective
The primary objective was to determine the VE and long-term durability of the BNT162b2 XBB vaccine in preventing COVID-19 hospitalisations through 31 August 2024 (up to 46 weeks since dose).
Study participants
Patients included adults aged ≥18 years eligible for COVID-19 vaccination based on national or regional guidelines (S2 Table in S1 File). All participating countries initiated their vaccination campaigns for the 2023–2024 season in September 2023 and recommended the use of an updated vaccine targeting the XBB variant. Each country defined its at-risk population according to national criteria, although recommendations consistently prioritised older adults and individuals at increased risk of COVID-19 infection or severe disease due to underlying health conditions and occupational exposure. All countries permitted vaccination for the broader adult population beyond the defined risk groups.
Included patients must have been hospitalised for ≥1 night and diagnosed with SARI, based on the European Centre for Disease Prevention and Control (ECDC) case definition [26]. Qualifying SARI symptoms included cough, objective fever (≥38⁰C), shortness of breath, anosmia, ageusia, or dysgeusia. Symptom onset date must have occurred ≤14 days before hospital admission. Patients were excluded if admitted to the hospital for COVID-19 within the past 3 months, were incapable of providing a nasopharyngeal or oropharyngeal specimen, or had received a COVID-19 vaccine dose from a manufacturer not authorised by the EMA. Informed consent was obtained from patients or their legal representatives, except at study sites where IECs authorised waivers of consent.
SARS-CoV-2 laboratory testing and variant identification
All patients were tested for SARS-CoV-2 by reverse transcription polymerase chain reaction (RT-PCR) or multiplex PCR between 14 days prior to, and up to 72 hours following, hospital admission. The SARS-CoV-2 infecting strain was further characterised in positive specimens using whole genome sequencing or next-generation sequencing. When sequencing was not possible due to insufficient viral load or poor sample quality, the predominant subvariant present in the country of hospitalisation on symptom onset date was imputed from national reporting sources.
A predominance threshold of ≥70% of sequenced samples was established using country-level genomic reporting by calendar time. Sub-lineages of XBB or JN.1 (such as EG.5, KP.2, and KP.3) were collapsed into their parent lineages. When neither XBB nor JN.1 reached the predominance threshold of ≥70%, the time was classified as a transition period. Country-specific XBB and JN.1 predominance periods are defined in S1 Table in S1 File. Patients with a confirmed JN.1 sequencing result or who experienced symptom onset during the country-specific JN.1 predominance period were included. Accordingly, patients with symptom onset during the country-specific XBB predominance or the transition period, and/or with non-JN.1 sequencing results were excluded.
Exposure and outcome definition
Exposed patients received at least one dose of the BNT162b2 XBB vaccine from start of the country-specific vaccination campaigns up to the end of the study period, irrespective of prior COVID-19 vaccination status, at least 14 days before symptom onset and at least 12 weeks after receipt of any prior COVID-19 vaccine. The unexposed group included patients who did not receive any COVID-19 vaccine during the 2023–2024 season, including never vaccinated patients and those who received COVID-19 vaccines prior to the 2023–2024 season and at least 12 weeks before symptom onset, following national eligibility guidelines [27].
’Test-positive cases’ were patients who met the SARI case definition and had at least one positive SARS-CoV-2 result by PCR. ‘Test-negative controls’ were patients who also met the SARI case definition and received a negative result for all SARS-CoV-2 tests performed.
Data collection and management
COVID-19 vaccination data were collected from registries, medical records, or vaccination cards. Data were collected on age, sex, predefined chronic conditions associated with an increased risk of severe outcomes from SARS-CoV-2 infection (S3 Table in S1 File), body mass index (BMI), symptom onset date, and clinical course of the hospitalisation. Trained study staff entered data from medical records and patient interviews into an electronic Case Report Form (eCRF, Castor®). Pseudonymised data were transferred to a secure central research server hosted by P95 Clinical and Epidemiology Services (Leuven, Belgium) for analysis.
Statistical methods
Odds ratios (OR) comparing the odds of vaccination between test-positive cases and test-negative controls were estimated using multivariable Generalized Estimating Equation (GEE) logistic regression models. These models accounted for potential intra-cluster correlation among patients from the same study site [28]. The variances associated with ORs were calculated using a robust sandwich estimator, yielding a consistent and asymptotically normal estimator [29]. VE was calculated as (1–OR) × 100%. Models adjusted for symptom onset date, age, sex, number of chronic conditions, and receipt of influenza vaccination in the past 12 months. Each case was matched with up to four controls by study site and 14-day interval from the earliest onset date to minimise potential residual time-varying confounding. Matching was conducted independently for each study objective, using distinct datasets. Calendar time for symptom onset date was modelled using a cubic spline, with the number of knots aligned with the number of symptom onset months included in the analysis. The effect of age was captured using a cubic spline with two predefined knots at ages 50 and 65 years. When <10% of patients in the dataset were <50 years of age, a single knot at age 65 was used. The cutoff points used to define the age group categories determined the selection of knots for the age effect. To overcome multicollinearity while fitting GEE models, we applied penalisation for spline terms of symptom onset date utilising a modified Penalized GEE (PGEE) package with a ridge penalty term [30].
To estimate VE by time since the last BNT162b2 XBB vaccine dose, the logistic regression model was extended to include the interaction between exposure status and time since the last vaccine dose. The effect of time since the last vaccine dose was captured by a cubic regression spline with a knot at 182 days (or 26 weeks), based on prior evidence of waning VE after 26 weeks for any XBB vaccine [31]. The time-dependent OR was derived on a one-dimensional grid ranging from the minimum to the maximum values of time since the last dose, with a step size of 1 day. The sandwich estimator was used to calculate the variance of the VE by time since the last vaccine dose.
Several sensitivity analyses were performed to assess the robustness of the results obtained from the PGEE approach. These analyses considered different choices regarding the number and location of knots for both symptom onset date and age, the impact of prior COVID-19 infection, the impact of removing adjustment for influenza vaccination status, and excluding Belgian, Italian, and German sites.
Only patients with complete data on exposure status, SARS-CoV-2 test results, and relevant confounders were included in regression analyses [32]. As 1.6% (113/7223) of the patients with SARI were excluded from analysis due to missing chronic conditions or missing influenza vaccination history (S1 Fig in S1 File), the complete case analysis is considered unlikely to produce biased estimates.
Sample size calculations, model specifications, and fitting procedures are provided in the supporting information (Appendix A.1, A.2, and A.3). All analyses were conducted using R version 4.1.2 [33].
Results
Study population
From 2 October 2023–31 August 2024, 7223 patients hospitalised with SARI were enrolled. There were 801 patients with symptom onset during XBB predominance or the transition period, and/or with non-JN.1 sequencing results who were excluded. Overall, 6094 (84.4%) patients with a confirmed JN.1 sequencing result or a SARI onset date during the country-specific JN.1 predominance period were eligible for analysis. After matching, a total of 3059 patients were ultimately included (S1 Fig in S1 File).
Fig 1 indicates an increase in enrolment among patients with SARI overall and among test-positive cases in the winter months, which peaked in late December 2023. A subsequent increase among only test-positive cases peaked again in July 2024, corresponding to the summer increase of COVID-19 observed in Europe [34]. During JN.1 predominance, 77.4% (2368/3059) of all hospitalised patients with SARI were ≥65 years of age, 53.3% (1630/3059) were male, and 41.4% (1266/3059) had ≥ 3 chronic conditions (Table 1). The most prevalent chronic conditions included hypertension (62.4% of total patients), cardiovascular disease (41.9%), lung disease (35.6%), type 2 diabetes (30.4%), and immunodeficiency or cancer (21.2%). The median age of test-positive cases was 78 years (interquartile range [IQR] 66–86), and more than half were male (53.2%, n = 440), similar to test-negative controls.
Abbreviations: SARI, severe acute respiratory infections.
In total, 53.4% (1633/3059) of participants received the BNT162b2 XBB vaccine, including 45.7% (378/827) of test-positive cases and 56.2% (1255/2232) of test-negative controls (Table 1). Among the 1426 patients who did not receive a COVID-19 vaccine during the 2023–2024 season, 161 (11.3%) had never been vaccinated, 697 (48.9%) received ≥2 mRNA wildtype doses only, and 379 (26.6%) most recently received an original bivalent BA.4/5-adapted booster. Patients who were not vaccinated during the 2023–2024 season tended to be younger compared with those who received the BNT162b2 XBB vaccine (median age 71 years [IQR: 57–82] versus 80 years [IQR: 73–88]) and had fewer chronic conditions (35.3% with ≥3 chronic conditions compared to 46.7%).
The median length of hospital stay was 6 days (IQR: 4–9) (Table 1). In total, 5.9% of patients (181/3059) died in the hospital. One-quarter (24.9% [592/2379]) of patients did not require respiratory support, including 27.9% (167/598) of test-positive cases and 23.9% (425/1781) of test-negative controls. Overall, 62.9% (1497/2379) of patients received oxygen therapy, 11.3% (270/2379) received non-invasive ventilation, and 0.8% (20/2379) received invasive mechanical ventilation or extracorporeal membrane oxygenation (ECMO).
COVID-19 vaccine effectiveness
The overall effectiveness of the BNT162b2 XBB vaccine against JN.1-related hospitalisation was 47.9% (95% CI: 37.0; 56.9) at a median of 107 (IQR: 65–223) days since dose compared with no COVID-19 vaccination during the 2023–2024 season (Fig 2).
Abbreviations: ECMO, extracorporeal membrane oxygenation; ICU, intensive care unit; IH, in-hospital; IQR, interquartile range; LOS, length of stay; N, number; VE, vaccine effectiveness. 1 Vaccine effectiveness estimates are adjusted for date of symptom onset, age, sex, number of chronic conditions and receipt of influenza vaccine in the 12 months prior to current hospital admission. 2 ‘Never vaccinated’ study participants were excluded when calculating the median (IQR) of time since last vaccine dose in the unexposed group (patients who did not receive any dose of a COVID-19 vaccine in the 2023–2024 autumn/winter season). 3 The categories “Cancer” and “Immunodeficiency” among chronic conditions are counted as separate types of chronic conditions in all adjusted VE estimates. Here, the VE estimate among participants having “Immunodeficiency or cancer” is provided as a composite. 4 The categories “0” and “1” among the number of chronic conditions are counted as separate categories of number of chronic conditions in all adjusted VE estimates. Here, the VE estimate among participants having “0–1” chronic conditions is provided as a composite. 5 The categories “Hospital admission, no ICU, no IH death”, “ICU admission, no IH death”, and “In-hospital death” are mutually exclusive. An additional category is added to combine the latter two categories: “ICU admission or IH death”. 6 Respiratory support level, presented in ascending order of severity with the highest level used during hospital stay determine the category, are mutually exclusive: 1) none; 2) oxygen therapy (e.g., nasal cannula or mask); 3) non-invasive ventilation (support without endotracheal intubation such as high flow nasal oxygen, continuous or bi-level positive airway pressure); 4) invasive mechanical ventilation (support with endotracheal intubation); and 5) extracorporeal membrane oxygenation (ECMO).
Effectiveness was higher among patients 18–64 years of age than those ≥65 years (65.6% [95% CI: 55.5; 73.5] and 44.9% [95% CI: 39.5; 49.7], respectively). VE was similar when stratified across a broad range of chronic condition groups; however, when considering the number of chronic conditions, patients with two chronic conditions had a higher VE than patients with 0–1 or ≥3 chronic conditions. VE was higher against more severe in-hospital endpoints, such as ICU admission or death than hospital admission alone, although confidence intervals overlapped. VE against hospital admission without ICU admission or death was 46.5% (95% CI: 31.9; 58.0). VE against ICU admission or in-hospital death was 55.4% (95% CI: 21.3; 74.8). The sample size was insufficient to estimate VE for the more severe respiratory support endpoints, such as invasive mechanical ventilation or ECMO. Effectiveness appeared slightly higher against longer hospital stays, although confidence intervals overlapped, with VE against a JN.1-related hospitalisation of ≥2 days was 49.2% (95% CI: 40.7 to 56.4), while VE against a hospitalisation with a length of ≥7 days was 61.5% (95% CI: 51.8; 69.2) (Fig 2).
To determine BNT162b2 XBB VE according to prior COVID-19 vaccination status and to discern any potential residual protection from COVID-19 vaccines received before the 2023–2024 season, VE was stratified by distinct comparison groups (S2 Fig in S1 File). Compared with receiving ≥1 mRNA BA.4/5 original bivalent dose, the BNT162b2 XBB vaccine was 48.6% (95% CI: 38.9; 56.7) effective, similar to the overall estimate that used a comparison group of no vaccine in the 2023–2024 season. When compared with ≥2 mRNA wildtype doses only, VE was also similar to the overall VE estimate at 46.4% (95% CI: 37.1 to 54.4). VE of the BNT162b2 XBB vaccine was lower when only compared to patients never vaccinated against COVID-19 at 26.7% (95% CI: −38.8; 61.3), although wide confidence intervals suggested considerable uncertainty of the point estimate. Characteristics of the study participants with prior vaccination categories as reference groups are presented in S4 Table in S1 File.
Durability
VE stratified by time since BNT162b2 XBB vaccine dose was 64.5% (95% CI: 56.6; 71.0) at two to <30 weeks, with a VE of 68.6% (95% CI: 56.0; 77.6) at two to <12 weeks, 75.3% (95% CI: 68.4; 80.6) at 12 to <18 weeks, 59.2% (95% CI: 43.9; 70.3) at 18 to <24 weeks, and 50.5% (95% CI: 35.8; 61.9) at 24 to <30 weeks (Fig 2). VE at 30 to <46 weeks since BNT162b2 XBB vaccine dose was 4.9% (95% CI: −30.3; 30.7). These stratified results are consistent with estimates from the spline-based approach (Fig 3), which indicated a slight increase in VE from approximately two to 15 weeks since receipt of the BNT162b2 XBB vaccine, followed by relatively flat VE through 25 weeks, and lastly, a steeper decrease in VE after 25 weeks that reached a central estimate of 0% effectiveness at 33 weeks since dose.
Vaccine effectiveness estimates are adjusted for date of symptom onset, age, sex, number of chronic conditions, and influenza vaccination status for the last 12 months. The tick marks on the x-axis of the figure indicate the time since the BNT162b2 XBB vaccine received by the SARI patients included in the analysis. The grey-shaded area corresponds to the 95% CI.
When stratified by age groups and number of chronic conditions (S3 Fig and S4 Fig in S1 File), the durability of the BNT162b2 XBB vaccine followed a waning pattern consistent with the overall trend, wherein effectiveness begins decreasing in the 18 to <24 weeks post-dose stratum. Similar results were observed when stratified by level of SARI severity (S5 Fig in S1 File).
Sensitivity analyses
Results of sensitivity analyses that changed the number and location of knots for both symptom onset date and age were comparable to VE estimates obtained in the main analysis. When stratified by prior COVID-19 infection status, VE was 39.8% (95% CI: 12.4; 58.6) among patients with documented prior COVID-19 infection and 46.4% (95% CI: 36.0; 55.1) among patients with no documented prior COVID-19 infection (S10 Fig in S1 File). VE estimates were comparable after removing sites from Belgium, Germany, and Italy (S11 Fig in S1 File).
Estimates were lower after removing adjustment for influenza vaccination status; VE was 40.9% (95% CI: 24.3; 53.9) without influenza vaccination status adjustment (S12 Fig in S1 File) and 47.9% (95% CI: 37.0; 56.9) with influenza vaccination status adjustment (Fig 2).
Discussion
In this test-negative case-control study conducted in the id.DRIVE platform in Europe, the BNT162b2 XBB vaccine was 47.9% (95% CI: 37.0; 56.9) effective at a median of 107 days (IQR: 65; 223) since dose against JN.1-associated hospitalisation. VE peaked at 10–12 weeks since dose, with a decline observed after 15 weeks, and continuing to drop through 24 to <30 weeks since dose, irrespective of age group and comorbidity status. We found no evidence of protection against JN.1-associated hospitalisation beyond 30 weeks since dose.
The VE patterns observed in this study are consistent with the limited available evidence on the durability of XBB-adapted vaccines during the JN.1 predominant period. A study conducted in Europe by Nunes and colleagues of primarily BNT162b2 XBB vaccine during JN.1 predominance, found that VE against hospitalisation was 47% (95% CI: 32; 59) for adults aged 65–79 years and 39% (95% CI: 17; 54) for adults aged ≥80 years at 90–179 days (~13 to 25.5 weeks) since dose, with no evidence of waning through 25.5 weeks [17]. In our study, we observed a numerically higher VE of 56.4% (95% CI: 34.7; 70.9) at 18 to <24 weeks since dose among adults aged ≥65 years, albeit with confidence intervals overlapping that of Nunes et al.
Link-Gelles reported VE against hospitalisation up to 43 weeks since dose for any XBB vaccine in the VISION Network in the United States [31]. Among adults aged ≥18 years, VE against COVID-19 hospitalisation during the JN.1 predominant period was 14% (95% CI: 2; 24) at 17 to 25.5 weeks since dose. At 26–43 weeks, XBB-adapted vaccines no longer provided significant protection against hospitalisation (VE: −1% [95% CI: −10; 8]) [31]. We found a higher VE of 59.2% (95% CI: 43.9; 70.3) at a comparable intermediate timepoint since dose at 18 to <24 weeks. However, minimal to no remaining protection was apparent in either study at the latest time points, > 26 or >30 weeks since dose. Together, these findings suggest XBB vaccines may provide limited protection against COVID-19 hospitalisation in the latter 6 months of a once-yearly vaccination schedule.
As shown by Shrestha et al. (2024), Tartof et al. (2024), and Caffrey et al. (2024), the effectiveness of the BNT162b2 XBB vaccine was higher during XBB predominance compared to its effectiveness during JN.1 predominance [9–11]. This has also been shown in studies not limited to the BNT162b2 vaccine [35,36]. VE tends to decrease when the circulating SARS-CoV-2 strains exhibit significant divergence from the subvariant targeted in the vaccine, supporting the need for periodic updates to variant-adapted COVID-19 vaccines. Nevertheless, we observed that BNT162b2 XBB vaccine was effective against JN.1-associated hospitalisation for up to six months.
Our main findings remained consistent across various sensitivity analyses, including modifying the number and location of knots for symptom onset date and age, stratifying by prior COVID-19 infection status, and removing Belgian, German, and Italian sites. Similar to our first analysis [12], adjusting for influenza vaccination status in the last 12 months increased VE estimates. The inclusion of influenza vaccination status as a covariate in adjusted models can reduce potential residual bias related to the likelihood of enrolment as a test-negative control [37].
The inclusion of multiple sites across four European countries was a strength of this multi-centre study. This enabled broader surveillance to identify the JN.1 subvariant early, allowing follow-up for up to 46 weeks after XBB vaccination. We determined vaccination status using vaccine registries, vaccination cards, and health records, which likely reduced misclassification of exposure. Patients’ case status was verified using PCR, which reduced the likelihood of misclassifying cases and controls. Furthermore, utilising the test-negative approach and requiring both cases and controls to be hospitalised with a SARI illness helped to account for unmeasured confounding factors linked to healthcare-seeking behaviours and exposure to respiratory pathogens.
Our study is subject to several limitations. Although results from sensitivity analyses were similar to the main results, our study nonetheless could be subject to residual confounding, particularly related to factors that are difficult to accurately measure, such as history of COVID-19 and time since the last infection. While we did collect information on prior infection status, this information was frequently missing, and prior infections are likely underestimated, given known discrepancies in the underreporting of SARS-CoV-2 infections [38]. Overall, 71% of the patients with SARI included in our analysis did not have a documented prior SARS-CoV-2 infection. Second, limited sample size for some stratified analyses prevented VE estimation or reduced the precision of VE estimated for certain in-hospital endpoints, such as the use of respiratory support. There was a high proportion (505/2025; 25%) of missing data for in-hospital respiratory support, which reduced the statistical power. Missingness was most prevalent in patients who did not receive the BNT162b2 XBB vaccine and was equally distributed among unvaccinated test-positive cases and test-negative controls (both 34%). Third, the generalisability of our findings beyond European populations primarily comprised of older adults is unknown.
We found that the effectiveness of the BNT162b2 XBB vaccine waned over time, with no evidence of remaining protection against hospitalisation by about 30 weeks since dose. These findings have important implications for the frequency of COVID-19 vaccination for high-risk patients, such as those with comorbid conditions and older adults. As evidenced by the July–August 2024 summer increase in COVID-19 hospitalisations in Europe, COVID-19 activity remains unpredictable, unlike the seasonal patterns displayed by influenza and some other respiratory viruses. Our findings support the need for additional vaccination in the spring season to maintain year-round protection against constantly changing circulating SARS-CoV-2 strains and unexpected increases in COVID-19 activity.
Supporting information
S1 File. Supporting information: main supporting tables and figures – Tables S1–S4 and Figures S1–S12.
https://doi.org/10.1371/journal.pone.0342382.s001
(DOCX)
Acknowledgments
The authors thank Prof. Elizabeth Miller, a member of the id.DRIVE Independent Scientific Committee, for her valuable scientific advice to the id.DRIVE studies. The authors acknowledge the contribution of Sultan Abduljawad and Shanti Pather of BioNTech. The authors thank Hospital Universitario Vall d’Hebrón, Hospital Universitari Germans Trias i Pujol, University Hospital Ulm, Centro Interuniversitario per la Ricerca sull’Influenza e le altre Infezioni Trasmissibili, Universitair Ziekenhuis Antwerpen, Le Centre Hospitalier Universitaire St Pierre, Valencia Hospital Network for the Study of Infectious Diseases (VAHNSI)-Fundación para el Fomento de la Investigación Sanitaria y Biomédica de la Comunitat Valenciana (Fisabio – Public Health), and University Hospital Frankfurt for collecting data of the study reported in this publication. The authors wish to acknowledge the contribution of Griet Rebry (study operations), Juan Hernandez (statistician), Somsuvro Basu (medical writer) (all from P95 Clinical and Epidemiology Services), Brenda Marquina Sánchez (coordinator) of FISABIO, and all other members of the id.DRIVE Co-Coordination team.
References
- 1.
EMA. Comirnaty: EMA recommends approval of adapted COVID-19 vaccine targeting Omicron XBB.1.5. https://www.ema.europa.eu/en/news/comirnaty-ema-recommends-approval-adapted-covid-19-vaccine-targeting-omicron-xbb15. 2023. Accessed 2024 July 24.
- 2. Hansen CH, Moustsen-Helms IR, Rasmussen M, Søborg B, Ullum H, Valentiner-Branth P. Short-term effectiveness of the XBB.1.5 updated COVID-19 vaccine against hospitalisation in Denmark: a national cohort study. Lancet Infect Dis. 2024;24(2):e73–4. pmid:38190834
- 3. Monge S, Humphreys J, Nicolay N, Braeye T, Van Evercooren I, Holm Hansen C, et al. Effectiveness of XBB.1.5 monovalent COVID-19 vaccines during a period of XBB.1.5 dominance in EU/EEA countries, october to november 2023: A VEBIS-EHR Network Study. Influenza Other Respir Viruses. 2024;18(4):e13292. pmid:38654485
- 4. van Werkhoven CH, Valk A-W, Smagge B, de Melker HE, Knol MJ, Hahné SJ, et al. Early COVID-19 vaccine effectiveness of XBB.1.5 vaccine against hospitalisation and admission to intensive care, the Netherlands, 9 October to 5 December 2023. Euro Surveill. 2024;29(1):2300703. pmid:38179623
- 5. Tartof SY, Slezak JM, Frankland TB, Puzniak L, Hong V, Ackerson BK, et al. Estimated Effectiveness of the BNT162b2 XBB Vaccine Against COVID-19. JAMA Intern Med. 2024;184(8):932–40. pmid:38913355
- 6. Laniece Delaunay C, Melo A, Maurel M, Mazagatos C, Goerlitz L, O’Donnell J, et al. Effectiveness of COVID-19 vaccines administered in the 2023 autumnal campaigns in Europe: Results from the VEBIS primary care test-negative design study, September 2023-January 2024. Vaccine. 2024;42(19):3931–7. pmid:38839521
- 7. Huiberts AJ, Hoeve CE, de Gier B, Cremer J, van der Veer B, de Melker HE, et al. Effectiveness of Omicron XBB.1.5 vaccine against infection with SARS-CoV-2 Omicron XBB and JN.1 variants, prospective cohort study, the Netherlands, October 2023 to January 2024. Euro Surveill. 2024;29(10):2400109. pmid:38456217
- 8.
UKHSA. COVID-19 vaccine surveillance report (week 17). 2024. https://assets.publishing.service.gov.uk/media/662a43ed55e1582b6ca7e5ec/Vaccine_surveillance_report_2024_week_17.pdf
- 9. Caffrey AR, Appaneal HJ, Lopes VV, Puzniak L, Zasowski EJ, Jodar L, et al. Effectiveness of BNT162b2 XBB vaccine in the US veterans affairs healthcare system. Nat Commun. 2024;15(1):9490. pmid:39488521
- 10. Shrestha NK, Burke PC, Nowacki AS, Gordon SM. Effectiveness of the 2023-2024 Formulation of the COVID-19 Messenger RNA Vaccine. Clin Infect Dis. 2024;79(2):405–11. pmid:38465901
- 11. Tartof SY, Slezak JM, Puzniak L, Frankland TB, Ackerson BK, Jodar L, et al. Effectiveness of BNT162b2 XBB Vaccine Against XBB and JN.1 Sublineages. Open Forum Infect Dis. 2024;11(7):ofae370. pmid:39015348
- 12. Nguyen JL, Mitratza M, Volkman HR, de Munter L, Tran TMP, Marques C, et al. Effectiveness of the BNT162b2 XBB.1.5-adapted vaccine against COVID-19 hospitalization related to the JN.1 variant in Europe: a test-negative case-control study using the id.DRIVE platform. EClinicalMedicine. 2024;79:102995. pmid:39726669
- 13. Shrestha NK, Burke PC, Nowacki AS, Gordon SM. Effectiveness of the 2023-2024 Formulation of the Coronavirus Disease 2019 mRNA Vaccine against the JN.1 Variant. openRxiv. 2024.
- 14. Merdrignac L, Laniece Delaunay C, Verdasca N, Vega-Piris L, O’Donnell J, Sève N, et al. Effectiveness of XBB.1.5 Vaccines Against Symptomatic SARS-CoV-2 Infection in Older Adults During the JN.1 Lineage-Predominant Period, European VEBIS Primary Care Multicentre Study, 20 November 2023-1 March 2024. Influenza Other Respir Viruses. 2024;18(11):e70009. pmid:39523781
- 15. Tartof SY, Frankland TB, Puzniak L, Slezak JM, Ackerson BK, Hong V, et al. BNT162b2 XBB Vaccine for COVID-19 Among Children 5-17 Years of Age. JAMA Netw Open. 2024;7(12):e2449944. pmid:39666343
- 16. Andersen KM, Allen KE, Nepal RM, Mateus JS, Yu T, Zhou A, et al. Effectiveness of BNT162b2 XBB.1.5 vaccine in immunocompetent adults using tokenization in two U.S. states. Vaccine. 2025;52:126881. pmid:39987879
- 17. Nunes B, Humphreys J, Nicolay N, Braeye T, Van Evercooren I, Holm Hansen C, et al. Monovalent XBB.1.5 COVID-19 vaccine effectiveness against hospitalisations and deaths during the Omicron BA.2.86/JN.1 period among older adults in seven European countries: A VEBIS-EHR network study. Expert Review of Vaccines. 2024;23(1):1085–90.
- 18.
UKHSA. COVID-19 vaccine surveillance report (week 29). UKHSA. 2024. https://assets.publishing.service.gov.uk/media/669923b20808eaf43b50d1fd/Vaccine_surveillance_report_2024_week_29.pdf
- 19. Ward T, Paton RS, Overton CE, Mellor J, Aziz NA, Charlett A, et al. Understanding the effectiveness of the Comirnaty monovalent and bivalent vaccines during the Winter Coronavirus (COVID-19) Infection Study. Journal of Infection. 2025;90(5):106461.
- 20. Lin D-Y, Du Y, Xu Y, Paritala S, Donahue M, Maloney P. Effectiveness of XBB.1.5 Vaccines Against Omicron Subvariants. openRxiv. 2024.
- 21.
ERVISS Dashboard. https://erviss.org/. Accessed 2025 January 10.
- 22. Planas D, Staropoli I, Michel V, Lemoine F, Donati F, Prot M, et al. Distinct evolution of SARS-CoV-2 Omicron XBB and BA.2.86/JN.1 lineages combining increased fitness and antibody evasion. Nat Commun. 2024;15(1):2254. pmid:38480689
- 23. Jian F, Wang J, Yisimayi A, Song W, Xu Y, Chen X, et al. Evolving antibody response to SARS-CoV-2 antigenic shift from XBB to JN.1. Nature. 2025;637(8047):921–9. pmid:39510125
- 24.
CDC. COVID-19 can surge throughout the year 2024. https://www.cdc.gov/ncird/whats-new/covid-19-can-surge-throughout-the-year.html. Accessed 2025 January 10.
- 25.
WHO. Statement on the fifteenth meeting of the IHR (2005) Emergency Committee on the COVID-19 pandemic. https://www.who.int/news/item/05-05-2023-statement-on-the-fifteenth-meeting-of-the-international-health-regulations-(2005)-emergency-committee-regarding-the-coronavirus-disease-(covid-19)-pandemic. Accessed 2025 January 10.
- 26.
ECDC. ECDC technical report: Core protocol for ECDC studies of COVID-19 vaccine effectiveness against hospitalisation with severe acute respiratory infection laboratory-confirmed with SARS-CoV-2, version 1.0. 2021. https://www.ecdc.europa.eu/sites/default/files/documents/Core-protocol-for-ECDC-studies-of-COVID-19-vaccine-effectiveness-against-hospitalisation-with-SARI.pdf
- 27.
EMA and ECDC statement on updating COVID-19 vaccines to target new SARS-CoV-2 virus variants. https://www.ema.europa.eu/en/news/ema-and-ecdc-statement-updating-covid-19-vaccines-target-new-sars-cov-2-virus-variants. 2023. Accessed 2024 July 24.
- 28. LIANG K-Y, ZEGER SL. Longitudinal data analysis using generalized linear models. Biometrika. 1986;73(1):13–22.
- 29. Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes. Biometrics. 1986;42(1):121–30. pmid:3719049
- 30. Inan G, Wang L. PGEE: an R Package for analysis of longitudinal data with high-dimensional covariates. R J. 2017;9:393–402.
- 31. Link-Gelles R, Rowley EAK, Irving SA, Klein NP, Grannis SJ, Ong TC, et al. Estimated 2023-2024 COVID-19 Vaccine Effectiveness in Adults. JAMA Netw Open. 2025;8(6):e2517402. pmid:40560584
- 32.
Lee SY. Handbook of latent variable and related models. Elsevier. 2007.
- 33.
R-Core-Team. R: A Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical Computing. 2021.
- 34.
ECDC. Communicable disease threats report (Week 53; 28 December 2024 – 3 January 2025). 2025. https://www.ecdc.europa.eu/sites/default/files/documents/communicable-disease-threats-report-week-53-2024.pdf
- 35. Ma KC, Surie D, Lauring AS, Martin ET, Leis AM, Papalambros L, et al. Effectiveness of updated 2023-2024 (Monovalent XBB.1.5) COVID-19 vaccination against SARS-CoV-2 omicron XBB and BA.2.86/JN.1 lineage hospitalization and a comparison of clinical severity-IVY network, 26 Hospitals, October 18, 2023-March 9, 2024. Clin Infect Dis. 2024;:ciae405. pmid:39107255
- 36. Carazo S, Skowronski DM, Brousseau N, Guay C-A, Sauvageau C, Racine É, et al. Monovalent mRNA XBB.1.5 vaccine effectiveness against COVID-19 hospitalization in Quebec, Canada: Impact of variant replacement and waning protection during 10-month follow-up. PLoS One. 2025;20(6):e0325269. pmid:40460407
- 37. Doll MK, Pettigrew SM, Ma J, Verma A. Effects of confounding bias in coronavirus disease 2019 (COVID-19) and influenza vaccine effectiveness test-negative designs due to correlated influenza and COVID-19 vaccination behaviors. Clin Infect Dis. 2022;75(1):e564–71. pmid:35325923
- 38.
Mustapha M, Togo K, Volkman HR, Yang J, Yehoshua A, Fusco MD, et al. Under-reporting of SARS-CoV-2 infections in 27 countries, 2020–2022. 2024. https://dx.doi.org/10.2139/ssrn.5014145