Pharmacoepidemiology and Pharmacovigilance Division. Spanish Agency of Medicines and Medical Devices (AEMPS). Madrid, Spain, Spain
Background: Vaccine effectiveness (VE) studies using automatic identification of cases may misclassify the outcome hospitalization ‘with’ Covid-19 as hospitalization ‘for’ Covid-19.
Objectives: Estimate the VE of Covid-19 vaccines against hospitalisations ‘with’ Covid-19 (automatic identification) and ‘for’ Covid-19 (manual review).
Methods: In this cohort study, all individuals with a complete Covid-19 vaccination (time0) in The Primary Care Database For Pharmacoepidemiological Research (BIFAP) in Spain between March-October 2021 were matched 1:1 to unvaccinated controls on time0, by birth year, sex and region. We excluded individuals vaccinated before their age group (prioritised), institutionalised, with prior Covid-19 or visits to primary care in the prior 7 days (to avoid healthy vaccinee effect). We identified hospitalizations with diagnosis of SARS-CoV-2 infection, Covid-19 or pneumonia, with positive Covid-19 test from -30 to +120 days of the admission. A manual review, blinded to vaccination status, of patient clinical notes, was performed to identify any alternative reasons for the hospitalisation. Analysis was restricted to pairs at risk 7 (Pfizer-PF) or 14 days (Moderna-MD, AstraZeneca-AZ, Janssen-JA) after time0. VE (1-adjusted Hazard Ratio-HR; 95% CI) was estimated 1) including all automatic cases identified and 2) excluding admissions for other reasons.
Results: Pairs at risk were 229,859 for PF analysis, 28,994 (MD), 7,727 (AZ) and 33,262 (JA). Admission ‘with’ Covid-19 among vaccinated and controls were 41 and 109 for PF; 4 and 9 (MD); 1 and 10 (AZ) and 33 and 26 (JA), and for other reasons were 6 (15%) and 10 (9%) for PF pairs; 3 (75%) and 1 (11%) for MD; none for AZ pairs; and 14 (42%) and 6 (23%) for JA. VE of PF against admission ‘with’ Covid-19 was 79% (95% CI: 68-87%), being 92% (80-97%) for those aged 18-59 y and 64% (40-79%) for ≥60 y, while VE against admission ‘for’ Covid-19 was 77% (66-85%), being 91% (81-96%) for 18-59 y and 66% (41-81%) for ≥60 y. As for MD, VE against admission ‘with’ Covid-19 was 74% (16-92%) and 93% (47-99%) against admission ‘for’ Covid-19. For AZ, against admission ‘with’ Covid-19, VE was 92% (43-90%). Effectiveness was neither observed for JA vaccine against admission ‘with’ (HR=0.96; 95% CI: 0.56-1.63) nor ‘for’ Covid-19 (HR=0.69; 95% CI: 0.37-1.28).
Conclusions: During alpha and delta variants predominance, effectiveness against severe Covid-19 was observed in 3 of the 4 vaccines although the saimple size was small, and few severe cases occurred. Misclassification of Covid-19 severity was observed to be higher among vaccinated than controls, which underestimated the VE significantly for MD (in a 19%) and JA (in a 27%), overestimated little for Pfizer (in a 2%) and VE for AZ was unaffected.
