RESEARCH Associations of BNT162b2 vaccination with SARS-CoV-2 ­infection and hospital admission and death with covid-19 in nursing homes and healthcare workers in Catalonia: prospective cohort study Carmen Cabezas,1 Ermengol Coma,2 Nuria Mora-Fernandez,2 Xintong Li,3 Montse Martinez-Marcos,1 Francesc Fina,2 Mireia Fabregas,2 Eduardo Hermosilla,5 Angel Jover,2 Juan Carlos Contel,6 Yolanda Lejardi,4 Belen Enfedaque,2 Josep Maria Argimon,7 Manuel Medina-Peralta,2 Daniel Prieto-Alhambra3,8 For numbered affiliations see ABSTRACT hospital with covid-19, and 450 died with covid-19 end of the article OBJECTIVE during the study period. In parallel, 1828 nursing Correspondence to: C Cabezas To determine associations of BNT162b2 vaccination home staff and 2968 healthcare workers were found carmen.cabezas@gencat.cat with SARS-CoV-2 infection and hospital admission and to have SARS-CoV-2 infection, but fewer than five (or @CarmenCabezas10 on Twitter ORCID 0000-0002-4291-0454) death with covid-19 among nursing home residents, were admitted or died with covid-19. The adjusted Additional material is published nursing home staff, and healthcare workers. hazard ratio for SARS-CoV-2 infection after two doses online only. To view please visit DESIGN of vaccine was 0.09 (95% confidence interval 0.08 the journal online. to 0.11) for nursing home residents, 0.20 (0.17 to Prospective cohort study. Cite this as: BMJ 2021;374:n1868 0.24) for nursing home staff, and 0.13 (0.11 to 0.16) http://dx.doi.org/10.1136/bmj.n1868 SETTING for healthcare workers. Adjusted hazard ratios for Nursing homes and linked electronic medical record, Accepted: 13 July 2021 hospital admission and mortality after two doses test, and mortality data in Catalonia on 27 December of vaccine were 0.05 (0.04 to 0.07) and 0.03 (0.02 2020. to 0.04), respectively, for nursing home residents. PARTICIPANTS Nursing home staff and healthcare workers recorded 28 456 nursing home residents, 26 170 nursing home insufficient events for mortality analysis. staff, and 61 791 healthcare workers. CONCLUSIONS MAIN OUTCOME MEASURES Vaccination was associated with 80-91% reduction in Participants were followed until the earliest outcome SARS-CoV-2 infection in all three cohorts and greater (confirmed SARS-CoV-2 infection, hospital admission reductions in hospital admissions and mortality or death with covid-19) or 26 May 2021. Vaccination among nursing home residents for up to five months. status was introduced as a time varying exposure, More data are needed on longer term effects of with a 14 day run-in after the first dose. Mixed effects covid-19 vaccines. Cox models were fitted to estimate hazard ratios with index month as a fixed effect and adjusted Introduction for confounders including sociodemographics, The global pandemic of covid-19 has caused more than comorbidity, and previous medicine use. 195 million confirmed cases and 4 million deaths to RESULTS date.1 Spain is one of the most affected countries in Among the nursing home residents, SARS-CoV-2 Europe, reporting more than 4 million cases by July infection was found in 2482, 411 were admitted to 2021, the second highest figure among the European Union and European Economic Area member states.2 Three vaccines had been approved by the European WHAT IS ALREADY KNOWN ON THIS TOPIC Medicines Agency at the time of writing—Pfizer- Four covid-19 vaccines have been approved for use in the UK and EU to date BioNTech BNT162b2 mRNA, Moderna mRNA-1273, A large US based trial found that the first vaccine to be approved, the BNT162b2 and Oxford-AstraZeneca ChAdOx1 nCoV-19.3 All mRNA vaccine, had >90% efficacy against symptomatic covid-19 three have shown high efficacy in clinical trials, with 95% efficacy against symptomatic covid-19 for Emerging evidence from observational studies have confirmed similar results in BNT162b2 mRNA, 94.1% for mRNA-1273, and 70.4% the UK and Israel for ChAdOx1 nCoV-19.4-6 However, the trials have WHAT THIS STUDY ADDS not included large enough sample sizes to provide Two dose BNT162b2 vaccination was associated with 80-91% reductions in reliable evidence of protection against severe disease symptomatic and asymptomatic SARS-CoV-2 infections among nursing home or mortality. residents, nursing home staff, and healthcare workers Certain population groups, such as nursing home residents, have been under-represented in existing Additionally, vaccination with two doses of BNT162b2 led to ≥95% reductions clinical trials of covid-19 vaccination, despite evidence in covid-19 related hospital admission and mortality among nursing home suggesting that nursing home residents and staff are residents disproportionately affected by covid-19. Nursing homes The effects of two dose vaccination with BNT162b2 in nursing home residents accounted for a large proportion of deaths globally and are equivalent to those shown in randomised controlled trial participants showed a disproportionately higher mortality than the bmj | BMJ 2021;374:n1868 | doi: 10.1136/bmj.n1868 1 RESEARCH the general population of the same age.7-9 Preventing We followed non-vaccinated participants from the outbreaks of infections and reducing related mortality beginning of the vaccination campaign until the earliest in nursing home settings is crucial for minimising the of first dose of vaccine plus 14 day run-in (they then impact of the pandemic. However, no trial has studied switched to the “single dose vaccinated” arm), an this population specifically. Determining how effective outcome (positive RT-PCR or lateral flow test for SARS- covid-19 vaccines are in these high risk populations is CoV-2 or hospital admission, intensive care admission, important. or death with covid-19) or the end of the study (26 Little is known about the effectiveness of the May 2021). We followed participants vaccinated with approved vaccines in actual practice conditions, a single dose from the day they received the first dose outside research settings. Differences in compliance of the vaccine plus 14 days of run-in until the earliest with vaccine doses and intervals, testing for SARS- of a second dose of vaccine (they then switched to the CoV-2 infection, and management of covid-19 in busy “two dose vaccinated” arm), an outcome, or the end of clinical settings may all affect the external validity of the study. We followed two dose vaccinated participants reported trial data. from the day they received their second dose until an Spain started its mass immunisation programme outcome or the end of the study. We treated exposure on 27 December 2020, soon after the first covid-19 as time varying, with a participant able to contribute vaccine (BNT162b2 mRNA) was approved earlier that person days of follow-up to all three arms. We did an month. No data have been reported yet on compliance additional analysis comparing time before first dose with with vaccination and its observable effects in the the 14 days after the date of the first dose as a measure Spanish population. We aimed to characterise the of residual confounding (see Statistical analysis). first three cohorts of vaccinated people (nursing home residents, nursing home staff, and healthcare Outcomes workers) and estimate the short term effectiveness of We studied SARS-CoV-2 infection, hospital admission the BNT162b2 mRNA vaccine in preventing infections, with covid-19, and covid-19 as cause of death. We hospital admissions, and deaths. defined SARS-CoV-2 infection by the date of the earliest of a positive RT-PCR or lateral flow test, Methods regardless of symptoms. Screening of all nursing Study design and setting home staff and residents using RT-PCR was conducted Our prospective cohort study included three after any one case was identified. In addition, RT- populations that were analysed separately: we PCR and lateral flow tests were recommended among identified nursing home residents and nursing home healthcare workers on a fortnightly basis and after staff from primary care records and administrative three or more weeks of absence (for example, after a data, and healthcare workers from a bespoke registry of holiday or leave). We considered hospital admission to healthcare workers. Vaccination status against SARS- be the date of admission for covid-19 as reported in a CoV-2 infection came from the Catalan Shared Clinical bespoke official covid-19 inpatient registry. Death due Records, a clinical database of electronic medical to covid-19 was based on the reported diagnosis in the records that links primary care and hospital diagnoses mortality registry. and treatments for the whole universal Catalan health system. We further linked data to the regional central Additional variables and potential confounders database of reverse transcriptase polymerase chain We assessed individual level sociodemographics and reaction (RT-PCR) and lateral flow tests for SARS- clinical features at the time of inclusion, as collected CoV-2, hospital admissions, mortality registries, and from primary care electronic health records: age (in primary care electronic health records. Ninety per cent years), sex, residence status (nursing home resident or of primary care practices in Catalonia, and 90% of the staff) or profession (healthcare worker); we assessed population, were included in the dataset. Information pre-existing comorbidities if present any time before on professional roles for the included healthcare the index date; and we identified long term use of workers came from the workforce census of the Institut medicines on the basis of primary care prescriptions if Catala de la Salut. Data from these databases have been active/ongoing on the index date. Lists of ICD-10-CM previously validated and used for epidemiological (international classification of diseases, 10th revision, research,10-12 including many studies of covid-19.7 13-15 clinical modification) codes for comorbidities and lists of medicines identified using Anatomical Therapeutic Participants and follow-up Chemical Classification System codes are provided in We included all people alive in Catalonia at the supplementary table A. beginning of the covid-19 vaccination campaign with BNT162b2 mRNA on 27 December 2020 who were Statistical analysis nursing home residents or staff eligible for vaccination For descriptive analysis, we expressed continuous or who were identified as healthcare workers. We variables as mean (standard deviation) or median excluded those with a previous SARS-CoV-2 infection (interquartile range) and summarised categorical identified by a positive RT-PCR or lateral flow test and variables as number (percentage). We analysed the those who were not assigned to one of the primary care existence of confounding by indication by using the practices contributing to our database. standardised mean difference of all confounders listed 2 doi: 10.1136/bmj.n1868 | BMJ 2021;374:n1868 | the bmj RESEARCH above to compare vaccinated and unvaccinated groups. healthcare workers were available for the study. We We considered a standardised mean difference >0.1 to excluded 10 462 (24.4%) nursing home residents, be equivalent to a relevant imbalance and adjusted 3839 (11.8%) nursing home staff, and 12 213 (14.7%) for it in multivariable analyses.16 Additionally, we healthcare workers as they had previously been did an analysis of effects in the first 14 days after the infected with SARS-CoV-2. We also excluded 3885 first dose to assess the likely presence of residual nursing home residents, 2487 nursing home staff, confounding due to participant (unrecorded) variables, and 9340 healthcare workers owing to a lack of linked cluster effects at nursing home level, or changes in primary care records or dose interval/s or date data (fig epidemiological parameters related to the covid-19 1). We therefore analysed data from 28 456 nursing pandemic at the community level. Any departure from home residents, 26 170 nursing home staff, and 61 791 the expected null effect (hazard ratio=1) in these first healthcare workers. Supplementary table B reports 14 days after first dose vaccination can be interpreted professional roles for the included healthcare workers, as a measure of residual confounding. and supplementary table C shows sociodemographics We analysed vaccination as a time varying exposure and vaccination status for the excluded populations with three follow-up intervals. (1) No vaccination: compared with the analysed populations. from 27 December 2020 until first dose vaccination By the end of the study period, 26 987 (94.8%) plus 14 day run-in (where applicable), outcome, or nursing home residents, 21 870 (83.6%) nursing end of study. (2) One dose vaccination: from date of home staff, and 55 790 (90.3%) healthcare workers first dose administration plus 14 day run-in to date of had been vaccinated with at least one dose. Figure 2 second dose, outcome, or end of study. (3) Two dose shows rapid uptake of vaccine in the three cohorts over vaccination: from date of second dose administration the study period, with >50% of participants having to outcome or end of study. For each of these periods, received at least one dose by mid-January 2021, and we calculated the rate of outcomes per 10 000 person earlier for nursing home residents. Second doses were days by dividing the number of observed events within administered within a median 21 (interquartile range a period by the number of days of exposure, multiplied 0) days after the first dose. Supplementary figure A by 10 000. depicts the average weekly incidence of covid-19 in We plotted Kaplan-Meier estimates for each nursing home residents and the general population for study outcome stratified by vaccination status for context. visualisation. We fitted random effects time varying Table 1 and supplementary figure B compare the Cox models to calculate hazard ratios and 95% vaccinated and unvaccinated groups of the three confidence intervals for each study outcome according cohorts. Few differences existed between vaccinated to vaccination status. All Cox models used the index and unvaccinated nursing home residents. Vaccinated month as a random effect and were adjusted for any residents had a mean age of 85.7 years and were confounders with a standardised mean difference 73.3% female, whereas unvaccinated residents had >0.1. Three models were conducted separately for a mean age of 85.6 years and were 70.1% female. each of the cohorts (nursing home residents, nursing They had a similar prevalence of comorbidities and home staff, and healthcare workers). We assessed use of most medicines except sedatives/hypnotics and proportionality of hazards in the Cox models by visual antidepressants. All other covariates were sufficiently inspection of scaled Schoenfeld residuals. We used R balanced without adjustment (standardised mean version 3.5.1 for all analyses. difference ≤0.1). Vaccinated and unvaccinated nursing home staff were also similar in terms of sex, Sensitivity analyses but differed in age, prevalence of some comorbidities We did a sensitivity analysis as suggested after peer (obesity, hypertension, osteoarthritis), and medicine review, in which we excluded participants who never use (lipid modifying drugs and angiotensin converting received a vaccine during the study period. These enzyme inhibitors/angiotensin receptor blockers), all analyses were therefore entirely focused on the time with standardised mean difference >0.1. Vaccinated varying exposure in vaccinated participants, potentially healthcare workers differed (standardised mean further reducing confounding by indication. difference >0.1) from unvaccinated ones in terms of age (43.7 v 39.9 years), sex (75.2% v 79.7% female), Patient and public involvement prevalence of hypertension (7.8% v 5.3%), and history No patients or members of the public were directly of use of lipid modifying agents (4.6% v 2.5%), with no involved in the design or analysis of the reported data. other relevant differences observed. Because of covid-19 related restrictions, interaction Supplementary figure C depicts the number of with relevant patients, and especially with nursing participants tested over time. By the end of the study home residents, has been difficult. Some of the period, a total of 25 834 (95.7%) vaccinated and 1223 co-authors are healthcare workers and therefore (83.3%) unvaccinated nursing home residents, 21 185 represented in some of our analyses. (96.9%) vaccinated and 3589 (83.5%) unvaccinated nursing home staff, and 35 699 (64.0%) vaccinated and Results 4158 (69.3%) unvaccinated healthcare workers had Before exclusions, data for 42 803 nursing home been tested at least once. A median of 3 (interquartile residents, 32 496 nursing home staff, and 83 344 range 1-5) tests were performed during the study the bmj | BMJ 2021;374:n1868 | doi: 10.1136/bmj.n1868 3 RESEARCH 42 803 32 496 83 344 Residents Staff Healthcare workers 32 341 28 657 71 131 No previous covid No previous covid No previous covid 28 613 26 405 62 673 EMR linkage EMR linkage EMR linkage 28 456 26 170 61 791 No erroneous dose intervals No erroneous dose intervals No erroneous dose intervals Fig 1 | Population flowchart. EMR=electronic medical records period in vaccinated and 3 (2-5) in unvaccinated not the second (251). The incidence rates of infection nursing home residents, 11 (8-16) and 9 (4-14) in staff, were 10.2/10 000 person days for unvaccinated staff, and 3 (1-5) and 3 (2-5) among healthcare workers. 8.8 for staff with one dose, and 1.3 for staff with two In total, 1335 SARS-CoV-2 infections occurred doses. Any vaccination led to an adjusted hazard ratio among unvaccinated nursing home residents, 620 in for infection of 0.22 (0.19 to 0.24). A single vaccine the first 14 days following the first dose, and 1147 dose led to an adjusted hazard ratio of 0.62 (0.55 to among vaccinated residents. Most post-vaccination 0.69), and a second dose led to an adjusted hazard infections were after the first vaccine dose (882), ratio of 0.20 (0.17 to 0.24) (table 2; fig 3). The adjusted with few after the second dose (265). The incidence hazard ratio associated with one dose of vaccine during rates of infection were 22.6/10 000 person days for the first 14 days was 0.89 (0.71 to 1.02), suggesting the unvaccinated residents, 14.3/10 000 for residents absence of relevant unresolved confounding. with one dose, and 1.0/10 000 for residents with two Finally, in the cohort of healthcare workers, 1961 doses. Any vaccination led to an adjusted hazard ratio unvaccinated and 1007 vaccinated staff tested for SARS-CoV-2 infection of 0.21 (95% confidence positive for SARS-CoV-2 (785 after one dose, 222 interval 0.19 to 0.24). A single vaccine dose led to after the second). The incidence rates of infection an adjusted hazard ratio of 0.53 (0.49 to 0.58) and a were 8.6/10 000 person days among unvaccinated second dose to an adjusted hazard ratio of 0.09 (0.08 healthcare workers, 5.6/10 000 after one dose, and to 0.11) (table 2). Figure 3 shows Kaplan-Meier plots 0.5/10 000 after two doses. Any vaccination led to stratified by vaccination status, in which a modest but an adjusted hazard ratio of 0.13 (0.11 to 0.14). A noticeable reduction in infections was apparent earlier single dose led to an adjusted hazard ratio of 0.60 than expected, already seen in the first 14 days among (0.55 to 0.66), and a second dose led to an adjusted the vaccinated versus unvaccinated residents. The hazard ratio of 0.13 (0.11 to 0.16) (table 2; fig 3). The resulting adjusted hazard ratio of 0.77 (0.69 to 0.86) is adjusted hazard ratio for the first 14 days after first a measure of residual confounding. dose vaccination was 0.97 (0.87 to 1.08), suggesting We found similar results for nursing home staff, the absence of residual confounding. with 1144 infections among unvaccinated staff and A sensitivity analysis excluding the “never 684 among vaccinated staff. Most infections among vaccinated” nursing home residents resulted in a vaccinated staff were after the first vaccine dose (433), hazard ratio of 0.98 (0.87 to 1.10), suggesting further reductions in confounding in these analyses. The corresponding adjusted hazard after a second dose of 100 vaccine was 0.11 (0.09 to 0.13). More detail is reported Vaccination coverage in supplementary table D. 80 Hospital admissions with covid-19 were recorded for 411 nursing home residents, with incidence rates 60 of 3.6/10 000 person days for unvaccinated residents, 2.1/10 000 after one dose, and 0.2/10 000 after two 40 Residents doses. Any vaccination led to an adjusted hazard for Staff admission of 0.35 (0.28 to 0.43). One vaccine dose 20 Healthcare workers led to an adjusted hazard ratio of 0.43 (0.34 to 0.54), 0 and a second dose led to a hazard ratio of 0.05 (0.04 Dec 1 Jan 1 Feb 1 Mar 1 Apr 1 May 1 Jun 1 to 0.07) (table 3). Figure 4 shows Kaplan-Meier plots Date for hospital admissions in nursing home residents by Fig 2 | Covid-19 vaccine uptake expressed as percentage with at least one dose (y vaccination status. axis) over time (x axis) in nursing home residents, nursing home staff, and healthcare We observed 450 deaths among nursing home workers residents: 272 before vaccination, 145 after one dose 4 doi: 10.1136/bmj.n1868 | BMJ 2021;374:n1868 | the bmj RESEARCH Table 1 | Baseline characteristics stratified by vaccination (none versus any) status. Values are numbers (percentages) unless stated otherwise Nursing home residents Nursing home staff Healthcare workers Ever vaccinated Never vaccinated Ever vaccinated Never vaccinated Ever vaccinated Never vaccinated Characteristics (n=26 987) (n=1469) (n=21 870) (n=4300) (n=55 790) (n=6001) Mean (SD) age, years 85.70 (9.08) 85.58 (11.33) 45.18 (12.6) 39.99 (12.83) 43.71 (12.27) 39.92 (12.73) Female sex 19 794 (73.3) 1030 (70.1) 19 013 (86.9) 3788 (88.1) 41 952 (75.2) 4783 (79.7) Analgesics 13 993 (51.9) 757 (51.5) 1667 (7.6) 304 (7.1) 3081 (5.5) 332 (5.5) Sedatives/hypnotics 10 639 (39.4) 504 (34.3) 2155 (9.9) 362 (8.4) 4555 (8.2) 449 (7.5) Anticoagulants 11 503 (42.6) 615 (41.9) 385 (1.8) 79 (1.8) 1031 (1.8) 155 (2.6) Antidepressants 12 983 (48.1) 627 (42.7) 2199 (10.1) 393 (9.1) 4731 (8.5) 467 (7.8) Antiepileptics 4627 (17.1) 229 (15.6) 651 (3.0) 113 (2.6) 1409 (2.5) 128 (2.1) Antipsychotics 11 650 (43.2) 596 (40.6) 224 (1.0) 43 (1.0) 409 (0.7) 57 (0.9) Antacids 13 970 (51.8) 731 (49.8) 1495 (6.8) 213 (5.0) 3471 (6.2) 269 (4.5) Systemic corticoids 950 (3.5) 77 (5.2) 177 (0.8) 34 (0.8) 451 (0.8) 48 (0.8) Oral antidiabetics 4003 (14.8) 206 (14.0) 552 (2.5) 70 (1.6) 941 (1.7) 58 (1.0) Insulin 2260 (8.4) 146 (9.9) 191 (0.9) 30 (0.7) 397 (0.7) 28 (0.5) Lipid modifying agents 5494 (20.4) 248 (16.9) 1122 (5.1) 117 (2.7) 2582 (4.6) 147 (2.4) α blockers 308 (1.1) 14 (1.0) 16 (0.1) 5 (0.1) 52 (0.1) 3 (<0.1) Other antihypertensives 115 (0.4) 7 (0.5) 5 (<0.1) 0 19 (<0.1) 2 (<0.1) β blockers 4965 (18.4) 292 (19.9) 482 (2.2) 68 (1.6) 1361 (2.4) 93 (1.5) Calcium channel blockers 3940 (14.6) 204 (13.9) 327 (1.5) 54 (1.3) 692 (1.2) 50 (0.8) Combination antihypertensives 2403 (8.9) 99 (6.7) 627 (2.9) 83 (1.9) 1086 (1.9) 77 (1.3) Diuretics 8707 (32.3) 507 (34.5) 410 (1.9) 44 (1.0) 775 (1.4) 67 (1.1) ACE inhibitors/ARBs 7860 (29.1) 405 (27.6) 1119 (5.1) 131 (3.0) 2393 (4.3) 156 (2.6) Chronic obstructive pulmonary disease/asthma inhalers 3465 (12.8) 201 (13.7) 862 (3.9) 150 (3.5) 2262 (4.1) 241 (4.0) Atrial fibrillation 4491 (16.6) 286 (19.5) 56 (0.3) 7 (0.2) 224 (0.40) 14 (0.2) Osteoarthritis 11 636 (43.1) 584 (39.8) 1833 (8.4) 252 (5.9) 2452 (4.4) 228 (3.8) Asthma 1228 (4.6) 65 (4.4) 1185 (5.4) 229 (5.3) 3467 (6.2) 366 (6.1) Ischaemic heart disease 2313 (8.6) 116 (7.9) 89 (0.4) 13 (0.3) 257 (0.5) 19 (0.3) Dementia 11 175 (41.4) 613 (41.7) 4 (<0.1) 0 3 (<0.1) 0 Diabetes mellitus 7053 (26.1) 392 (26.7) 710 (3.2) 93 (2.2) 1204 (2.2) 91 (1.5) Liver disease 1005 (3.7) 54 (3.7) 665 (3.0) 93 (2.2) 1042 (1.9) 96 (1.6) Hypertension 18 454 (68.4) 1002 (68.1) 2405 (11.0) 287 (6.7) 4368 (7.8) 318 (5.3) Heart failure 3137 (11.6) 189 (12.9) 20 (0.1) 2 (<0.1) 34 (0.1) 5 (0.1) Cerebrovascular disease 2911 (10.8) 187 (12.7) 92 (0.4) 16 (0.4) 177 (0.3) 12 (0.2) Chronic obstructive pulmonary disease 2249 (8.3) 125 (8.5) 169 (0.8) 30 (0.7) 269 (0.5) 23 (0.4) Chronic kidney disease 6894 (25.6) 404 (27.5) 124 (0.6) 12 (0.3) 282 (0.5) 18 (0.3) Cancer (all except non-melanoma skin cancer) 4864 (18.0) 278 (18.9) 616 (2.8) 84 (2.0) 2080 (3.7) 188 (3.1) Obesity 5493 (20.4) 253 (17.2) 3894 (17.8) 595 (13.8) 5002 (9.0) 480 (8.0) Valvular disease 1651 (6.1) 100 (6.8) 128 (0.6) 24 (0.6) 383 (0.7) 40 (0.7) Hepatitis B 81 (0.3) 7 (0.5) 54 (0.2) 21 (0.5) 61 (0.1) 5 (0.1) Hepatitis C 305 (1.1) 20 (1.4) 85 (0.4) 12 (0.3) 99 (0.2) 11 (0.2) HIV infection 14 (0.1) 1 (0.1) 72 (0.3) 7 (0.2) 69 (0.1) 7 (0.1) ACE=angiotensin converting enzyme; ARB=angiotensin receptor blocker. Table 2 | Number, incidence rates, and adjusted hazard ratios for covid-19 according to vaccination status in nursing home residents, nursing home staff, and healthcare workers Exposure person Exposure Rate per 10 000 Adjusted hazard ratio Cohort and period Population Cases days days (mean) person days (95% CI) Nursing home residents Unvaccinated 28 456 1335 590 956 20.77 22.59 Reference Days 0-14 after first dose 26 044 620 360 880 13.86 17.18 0.77 (0.69 to 0.86) Vaccinated (from day 14 after first dose) 25 375 527 2 841 387 111.98 1.86 0.21 (0.19 to 0.24) Vaccinated—one dose, from day 14 26 044 882 616 788 23.68 14.3 0.53 (0.49 to 0.58) Vaccinated—two doses, from second dose date 24 484 265 2 585 479 105.6 1.03 0.09 (0.08 to 0.11) Nursing home staff Unvaccinated 26 170 1,144 1 121 942 42.87 10.19 Reference Days 0-14 after first dose 21 269 338 291 473 13.70 11.6 0.89 (0.71 to 1.02) Vaccinated (from day 14 after first dose) 20 429 346 2 151 971 105.34 1.61 0.22 (0.19 to 0.24) Vaccinated—one dose, from day 14 21 269 433 490 062 23.04 8.84 0.62 (0.55 to 0.69) Vaccinated—two doses, from second dose date 19 513 251 1 953 382 100.11 1.28 0.20 (0.17 to 0.24) Healthcare workers Unvaccinated 61 791 1961 2 269 003 36.72 8.64 Reference Days 0-14 after first dose 54 848 649 758 224 13.82 8.56 0.97 (0.87 to 1.08) Vaccinated (from day 14 after first dose) 53 585 358 5 528 745 103.18 0.65 0.13 (0.11 to 0.14) Vaccinated—one dose, from day 14 54 848 785 1 409 807 25.70 5.57 0.60 (0.55 to 0.66) Vaccinated—two doses, from second dose date 51 019 222 4 877 162 95.6 0.46 0.13 (0.11 to 0.16) the bmj | BMJ 2021;374:n1868 | doi: 10.1136/bmj.n1868 5 RESEARCH Residents hazard ratio of 0.49 (0.39 to 0.61), and two doses led 1.00 Survival probability to a hazard ratio of 0.03 (0.02 to 0.04) (table 3). Figure 4 shows Kaplan-Meier plots for mortality in nursing 0.95 home residents by vaccination status. We recorded hospital admissions for 29 (27 0.90 unvaccinated) nursing home staff and 64 (48 unvaccinated) healthcare workers and 0 and <5 deaths respectively. We did not model these owing to limited 0.85 statistical power. P<0.001 0.80 Discussion Staff This is the first report of the clinical effectiveness of 1.00 a covid-19 vaccine in nursing homes globally to our Survival probability knowledge, and the first set of results on the effect of vaccination to prevent covid-19 in Spain. Using 0.95 a comprehensive linked database that combined primary care, screening and diagnostic RT-PCR and 0.90 lateral flow test, hospital, and mortality data, we studied the effects of the BNT162b2 mRNA vaccine 0.85 in the three populations prioritised for vaccination P<0.001 nationally. Partial (single dose) vaccination resulted in 40-50% reductions in SARS-CoV-2 infections, whereas 0.80 full (two dose) vaccination led to 80-90% protection. Healthcare workers The observed effects in nursing home residents were 1.00 Survival probability similar to those seen in healthcare workers and nursing home staff, and in line with previous phase III trials. 0.95 We also found significant, clinically relevant reductions in the risks of severe covid-19 among Strata nursing home residents, amounting to a striking 0.90 Unvaccinated 95% and 97% reduction in hospital admission 1 dose and mortality risks, respectively, after two doses of 0.85 2 doses BNT162b2. A plain English summary of the observed P<0.001 results is reported in supplementary table E. 0.80 0 14 28 42 56 70 84 96 112 126 140 154 Findings in context Time (days) Ambitious covid-19 vaccination campaigns are Fig 3 | Kaplan-Meier estimates of SARS-CoV-2 infection according to vaccination status ongoing around the world, but knowledge is still in nursing home residents (top left), staff (bottom left), and healthcare workers (top scarce on the real world effectiveness of covid-19 right) vaccines, especially in high risk populations such as nursing home residents and healthcare workers. of vaccine, and 33 after the second dose. Incidence Although vaccines have shown high efficacy in clinical rates of covid-19 mortality were 4.3/10 000 person trials, real world evidence on effectiveness is needed days in unvaccinated residents and 0.5/10 000 in all to confirm their effects in routine practice settings vaccinated residents (2.2/10 000 after one dose and and among populations under-represented in pivotal 0.1/10 000 after two doses). Any vaccination led to an trials. A recent study with national population level adjusted hazard ratio for death with covid-19 of 0.31 Scottish data showed that by the fourth week after (0.26 to 0.39). One vaccine dose led to an adjusted the first dose, the BNT162b2 and ChAdOx1 vaccines Table 3 | Number of hospital admissions and deaths, incidence rates, and adjusted hazard ratios according to vaccination status among nursing home residents Exposure person Exposure days Rate per 10 000 Cohort and period Population Cases days (mean) person days Adjusted hazard ratio (95% CI) Hospital admission Unvaccinated 28 456 228 631 699 22.20 3.61 Reference Vaccinated (from day 14 after first dose) 26 780 143 3 009 349 112.37 0.48 0.35 (0.28 to 0.43) Vaccinated—one dose, from day 14 26 887 134 655 958 24.40 2.05 0.43 (0.34 to 0.54) Vaccinated—two doses, from second dose date 25 945 49 2 729 304 105.20 0.18 0.05 (0.04 to 0.07) Death Unvaccinated 28 456 272 639 181 22.46 4.26 Reference Vaccinated (from day 14 after first dose) 26 927 153 3 030 779 112.56 0.51 0.31 (0.26 to 0.39) Vaccinated—one dose, from day 14 27 000 145 662 666 24.54 2.19 0.49 (0.39 to 0.61) Vaccinated—two doses, from second dose date 26 126 33 2 745 713 105.10 0.12 0.03 (0.02 to 0.04) 6 doi: 10.1136/bmj.n1868 | BMJ 2021;374:n1868 | the bmj RESEARCH Hospital admissions Our study differs from the trial in setting, participants, 1.000 Survival probability and outcome ascertainment. Although the trial included only symptomatic covid-19, we included any 0.975 RT-PCR or lateral flow test positive infection, including regular screening testing. The three populations 0.950 Strata included were screened periodically during the study Unvaccinated period to minimise potential outbreaks: whereas 1 dose nursing homes conducted universal RT-PCR among 0.925 staff and residents every time a case was identified, 2 doses P<0.001 healthcare workers were tested (with RT-PCR or lateral 0.900 flow test) every fortnight and after a leave of absence of more than three weeks. The observed reduction Deaths 1.000 in infection rates including screening tests is of high Survival probability relevance and supports recent data suggesting that vaccination could reduce overall transmission of the 0.975 virus.18 0.950 Limitations and strengths of study Our study has several limitations. The observational 0.925 nature of our data may have led to confounding by indication. Comprehensive linkage to primary care P<0.001 and hospital records allowed us to measure differences 0.900 0 14 28 42 56 70 84 96 112 126 140 154 in sociodemographics, comorbidity, and medication Time (days) usage. Despite this, we identified mild (approximately 20%) reductions in SARS-CoV-2 infections among Fig 4 | Kaplan-Meier estimates of hospital admission (left) and mortality with covid-19 healthcare workers during the first 14 days after (right) according to vaccination status in nursing home residents the first dose, suggesting unobserved confounding or nursing home protective effects. Our sensitivity reduced the risk of hospital admission by 85% (95% analysis excluding the “never vaccinated” participants confidence interval 76% to 91%) and 94% (73% to reduced confounding further and showed the expected 99%), respectively. It also showed that two doses of null effect in the 14 days after the first dose of vaccine, the BNT162b2 mRNA vaccine offered more than 85% in line with phase III trials. The findings from these protection against symptomatic infection among analyses are in line with the primary analyses. people aged over 80.17 A prospective cohort study of Changes in community transmission over the study hospital staff in England reported vaccine effectiveness period are depicted in supplementary figure A, which of 72% (58% to 86%) in the 21 days after the first dose shows an improved overall situation in community of the BNT162b2 mRNA vaccine.18 A cohort study of transmission from the second half of February 2021, healthcare workers in Israel reported a 75% reduction with further improvements in nursing homes. Despite in SARS-CoV-2 infection and 85% (71% to 92%) a further increase in the form of a minor wave in reduction in symptomatic covid-19 during the 15-28 community transmission after Easter (April 2021), days after the first dose of BNT162b2.19 A recent study we observed no evidence of this among nursing including more than half a million people in Israel home residents. In addition, cautious behaviour estimated the effectiveness of the BNT162b2 vaccine and increased personal protection during the first against infection at 46% (40% to 51%) 14-20 days two weeks after vaccination could contribute to the after the first dose and 92% (88% to 95%) one week unexpected reduction in infection risk observed during after the second dose.20 Interim results have been that period. Such measures were not recommended in published on the effectiveness of covid-19 vaccines pivotal trials as participants were blinded to vaccine in nursing homes in the United States, with findings exposure and neither participants nor investigators consistent with ours. had knowledge of the lack of effect during the first 12 We found that one dose of the BNT162b2 vaccine days after vaccination. Finally, we observed differences reduced the risk of infection by 47% in nursing home in testing practice and management of visits to nursing residents, 38% in nursing home staff, and 40% in homes over time. Testing practices changed over healthcare workers. These results are equivalent to the time during the study period, particularly in nursing reported vaccine efficacy of 52.4% between the first and homes. Screening of nursing home residents and second doses in the phase III trial of BNT162b2 mRNA.5 staff was reinforced in January 2021 but focused on In line with phase III trial findings, our results highlight unvaccinated staff from mid-February, as shown in that vaccine recipients should be told about the modest supplementary figure C. Similarly, guidelines on visits protection seen between the first and second doses and to nursing homes and on temporary exits from them encouraged to continue shielding, physical distancing, changed over the study period. Visits to nursing homes and other protective measures, especially during the were minimised in January when risk of covid-19 first two weeks after the initial dose. was high, and exits were not allowed unless they the bmj | BMJ 2021;374:n1868 | doi: 10.1136/bmj.n1868 7 RESEARCH were programmed and for a duration of three or more Contributors: CC, DPA, EC, and NCF were responsible for the study design. EC, EH, and NCF did the data analyses. DPA, EC, and XL drafted days. These measures were relaxed and became more the manuscript, and all co-authors reviewed and approved it for flexible when the situation improved from the second submission. DPA, MMP, and CC are the guarantors. The corresponding half of February, and further in mid-May. Incorporating author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted. index month in our random effects model accounted Funding: DPA received partial support from the National Institute for for this. However, higher testing rates in the vaccinated Health Research (NIHR) UK in the form of an NIHR Senior Research population could have resulted in a higher likelihood Fellowship (SRF-2018-11-ST2-004) and the Oxford NIHR Biomedical of diagnosis. The resulting surveillance bias would Research Centre. The funders had no role in considering the study design or in the collection, analysis, or interpretation of data, writing result in an underestimation of vaccine effectiveness. of the report, or decision to submit the article for publication. This study also has strengths. The comprehensive Competing interests: All authors have completed the ICMJE uniform linkage and coverage in our database is unique, disclosure form at www.icmje.org/coi_disclosure.pdf and declare: including primary care, hospital, RT-PCR and lateral no support from any organisation for the submitted work other than that detailed above; DPA’s research group has received grants and flow test results, and mortality data for more than 90% advisory or speaker fees from Amgen, Astellas, Astra-Zeneca, Chiesi- of the regional population. The Catalan health system Taylor, Johnson and Johnson, and UCB; no other relationships or is universal, minimising dropouts and maximising activities that could appear to have influenced the submitted work. the completeness of outcome ascertainment. Access Ethical approval: The study was approved by the Clinical Research to basic sociodemographics and events for people Ethics Committee of the IDIAP Jordi Gol with reference number 21/045-PCV. All result sets represent aggregated, de-identified data excluded from the analysis allowed us to measure that are reported at a minimum cell size of >5 to reduce potential for potential selection bias. Our included study population re-identification. allowed us to study the effects of vaccination against Data sharing: No patient level data can be shared owing to local hospital admission and death in nursing home information governance and data protection regulations. Aggregated data are available and reported in the supplement. residents, a population subgroup extremely vulnerable The lead authors affirm that this manuscript is an honest, accurate, to severe and lethal forms of covid-19 and under- and transparent account of the study being reported; that no represented in previous studies.7 20 The pivotal trial important aspects of the study have been omitted; and that any was underpowered to analyse these outcomes.5 discrepancies from the study as planned (and, if relevant, registered) have been explained. Dissemination to participants and related patient and public Conclusions communities: The study results will be disseminated to the public Our data confirmed that BNT162b2 vaccination through our media channels, including both national (Spain, the UK) strongly reduced the risk of SARS-CoV-2 infection and international media. We have provided a plain English summary table of our findings as supplementary table E. in nursing homes and in healthcare workers, with Provenance and peer review: Not commissioned; externally peer comparable results to those observed in US based reviewed. phase III trials and other international observational This is an Open Access article distributed in accordance with the studies. Hospital admission and death with covid-19 terms of the Creative Commons Attribution (CC BY 4.0) license, which were similarly reduced among nursing home residents, permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: who accounted for a large proportion of deaths with http://creativecommons.org/licenses/by/4.0/. covid-19 in 2020. Although further data and studies 1 World Health Organization. WHO Coronavirus disease (COVID-19) are needed to assess the long term effectiveness and dashboard. 2021. https://covid19.who.int/. safety of this and other covid-19 vaccines, these 2 European Centre for Disease Prevention and Control. COVID-19 findings should reassure the population of the major situation update for the EU/EEA, as of 29 July 2021. https://www. ecdc.europa.eu/en/cases-2019-ncov-eueea. benefits associated with the ongoing vaccination 3 European Medicines Agency. COVID-19 vaccines: authorised. 2021. campaign in Spain and elsewhere. Further research https://www.ema.europa.eu/en/human-regulatory/overview/public- health-threats/coronavirus-disease-covid-19/treatments-vaccines/ is needed to increase our understanding of the effect treatments-vaccines-covid-19-authorised-medicines. of vaccination on the management of nursing homes, 4 Baden LR, El Sahly HM, Essink B, et al, COVE Study Group. Efficacy including visitors, staff, use of protective equipment, and Safety of the mRNA-1273 SARS-CoV-2 Vaccine. N Engl J Med 2021;384:403-16. doi:10.1056/NEJMoa2035389 and residents themselves. 5 Polack FP, Thomas SJ, Kitchin N, et al, C4591001 Clinical Trial Group. AUTHOR AFFILIATIONS Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine. N Engl 1 Public Health Secretariat, Department of Health, Generalitat de J Med 2020;383:2603-15. doi:10.1056/NEJMoa2034577 6 Voysey M, Clemens SAC, Madhi SA, et al, Oxford COVID Vaccine Catalunya, Barcelona, Spain Trial Group. Safety and efficacy of the ChAdOx1 nCoV-19 vaccine 2 Direcció assistencial d’Atenció Primària i a la Comunitat, Institut (AZD1222) against SARS-CoV-2: an interim analysis of four Català de la Salut (ICS), Generalitat de Catalunya, Barcelona, Spain randomised controlled trials in Brazil, South Africa, and the UK. 3 Centre for Statistics in Medicine, NDORMS, University of Oxford, Lancet 2021;397:99-111. doi:10.1016/S0140-6736(20)32661-1 Oxford, UK 7 Prieto-Alhambra D, Balló E, Coma E, et al. Filling the gaps in 4 the characterization of the clinical management of COVID-19: Institut Català de la Salut (ICS), Generalitat de Catalunya, 30-day hospital admission and fatality rates in a cohort of Barcelona, Spain 118 150 cases diagnosed in outpatient settings in Spain. Int J 5 Epidemiol 2021;49:1930-9. doi:10.1093/ije/dyaa190 Idiap Jordi Gol, Universitat Autonoma de Barcelona, Barcelona, Spain 8 Rada AG. Covid-19: the precarious position of Spain’s nursing homes. 6 BMJ 2020;369:m1554. doi:10.1136/bmj.m1554 Chronic Care Program, Integrated Health and Social Care Plan, 9 Morciano M, Stokes J, Kontopantelis E, Hall I, Turner AJ. Excess Department of Health, Generalitat de Catalunya, Barcelona, Spain mortality for care home residents during the first 23 weeks of the 7 Generalitat de Catalunya, Barcelona, Spain COVID-19 pandemic in England: a national cohort study. BMC 8 Department of Medical Informatics, Erasmus University Medical Med 2021;19:71. doi:10.1186/s12916-021-01945-2 Center, Rotterdam, Netherlands 10 Baena-Díez JM, Garcia-Gil M, Comas-Cufí M, et al. Association between chronic immune-mediated inflammatory diseases and We acknowledge English language editing by Jennifer A de Beyer of cardiovascular risk. Heart 2018;104:119-26. doi:10.1136/ the Centre for Statistics in Medicine, University of Oxford. heartjnl-2017-311279 8 doi: 10.1136/bmj.n1868 | BMJ 2021;374:n1868 | the bmj RESEARCH 11 Garcia-Gil M, Elorza J-M, Banque M, et al. Linking of primary 16 Nguyen T-L, Collins GS, Spence J, et al. Double-adjustment care records to census data to study the association between in propensity score matching analysis: choosing a socioeconomic status and cancer incidence in Southern Europe: threshold for considering residual imbalance. BMC Med Res a nation-wide ecological study. PLoS One 2014;9:e109706. Methodol 2017;17:78. doi:10.1186/s12874-017-0338-0 doi:10.1371/journal.pone.0109706 17 Vasileiou E, Simpson CR, Robertson C, et al. Effectiveness of First 12 García-Gil MdelM, Hermosilla E, Prieto-Alhambra D, et al. Construction Dose of COVID-19 Vaccines Against Hospital Admissions in Scotland: and validation of a scoring system for the selection of high-quality National Prospective Cohort Study of 5.4 Million People. 2021. data in a Spanish population primary care database (SIDIAP). Inform https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3789264. Prim Care 2011;19:135-45. 18 Hall VJ, Foulkes S, Saei A, et al. Effectiveness of BNT162b2 mRNA Vaccine 13 Morales DR, Conover MM, You SC, et al. Renin-angiotensin system Against Infection and COVID-19 Vaccine Coverage in Healthcare Workers blockers and susceptibility to COVID-19: an international, open in England, Multicentre Prospective Cohort Study (the SIREN Study). science, cohort analysis. Lancet Digit Health 2021;3:e98-114. 2021. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3790399. doi:10.1016/S2589-7500(20)30289-2 19 Amit S, Regev-Yochay G, Afek A, Kreiss Y, Leshem E. Early rate reductions 14 Duarte-Salles T, Vizcaya D, Pistillo A, et al. Baseline characteristics, of SARS-CoV-2 infection and COVID-19 in BNT162b2 vaccine recipients. management, and outcomes of 55 270 children and adolescents Lancet 2021;397:875-7. doi:10.1016/S0140-6736(21)00448-7 diagnosed with COVID-19 and 1 952 693 with influenza in France, 20 Dagan N, Barda N, Kepten E, et al. BNT162b2 mRNA Covid-19 Germany, Spain, South Korea and the United States: an international Vaccine in a Nationwide Mass Vaccination Setting. N Engl J network cohort study.medRxiv 2020;2020.10.29.20222083. Med 2021;384:1412-23. doi:10.1056/NEJMoa2101765 15 Burn E, Tebé C, Fernandez-Bertolin S, et al. The natural history of symptomatic COVID-19 during the first wave in Catalonia. Nat Commun 2021;12:777. doi:10.1038/s41467-021-21100-y Web appendix: Supplementary materials No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe