Background: As the global roll-out of COVID-19 vaccines has moved from primary series (PS) to booster doses, the unvaccinated population has become increasingly small and select, making it difficult and inappropriate to measure absolute vaccine effectiveness (aVE, the total protection, e.g. booster+PS vs no PS). Relative VE (rVE, the added protection, e.g. booster+PS vs PS alone) is thus an increasingly used and important measure to inform vaccination policy.
Objectives: Use the Brazilian REFORCO study (NCT05697705) as an example to highlight challenges in rVE interpretation.
Methods: In a test-negative case-control study, Brazilian national surveillance data on hospitalisations (Aug21–Sep22) for severe acute respiratory syndrome (SARS) and COVID-19 vaccine records were used to estimate rVE of a 1st booster of any vaccine (≥8d) against COVID-19 hospitalisation, relative to a PS of any vaccine (≥4mths). Individuals (≥18yrs) eligible for a 1st booster when hospitalised for SARS were included. SARS-CoV-2 antigen/RT-PCR test-positive cases were exact-matched to PCR test-negative controls by 10-yr age band, time of PS completion and time of SARS hospitalisation. rVE was estimated using conditional logistic regression, overall, and by age, time from 1st booster to hospitalisation, and dosing interval between PS and 1st booster. Models were adjusted for sex, age, region, race, comorbidities, obesity, and other risk factors.
Results: 13,043 cases were matched to 19,684 controls. Mean age was 71yrs, 49% cases and 53% controls were female. Median days from PS to SARS hospitalisation was 239 (range 120–547). In boosted individuals, median days from 1st booster to SARS hospitalisation was 106 in cases and 89 in controls. CoronaVac (62%) and ChAdOx1-SARS-Cov-2 (33%) were the most common PS vaccines, BNT162b2 (83%) the most common 1st booster. rVE for any 1st booster was 43.8% (95%CI 40.3–47.0) overall, 46.0% (42.6–49.3) in ≥50yrs, and 50.5% (44.7–55.7) in ≥80yrs. Faster waning of aVE in older adults due to immunosenescence may explain the higher rVE in ≥80yrs. rVE by time since 1st booster peaked at 59.6% (55.0–63.8) at 8–35d, declining to 20.3% (1.7–35.3) by 176–203d, then increased to 49.3% (26.3–65.2) by 260–287d. This could reflect changes in demographic, serologic and epidemiologic landscapes over time, posing interpretive challenges. rVE by dosing interval between PS and 1st booster was stable.
Conclusions: The impact of age, time since 1st booster and changing epidemiological context on estimates of rVE pose interpretational challenges. We recommend stratified analyses to support rVE interpretation, particularly in the absence of aVE.
