Background: Although several studies have reported attenuated influenza illness following influenza vaccination, results have been inconsistent and have focused predominantly on adults in the USA. This study aimed to evaluate the severity of influenza illness by vaccination status in a broad range of influenza vaccine target groups across multiple South American countries.

Methods: We analysed data from four South American countries (Argentina, Brazil, Chile, and Paraguay) participating in REVELAC-i, a multicentre, test-negative design, vaccine effectiveness network including 41 sentinel hospitals. Individuals hospitalised at one of these centres with severe acute respiratory infection were tested for influenza by real-time RT-PCR, and were included in the analysis if they had complete information about their vaccination status and outcomes of their hospital stay. We used multivariable logistic regression weighted by inverse probability of vaccination and adjusted for antiviral use, duration of illness before admission, and calendar week, to calculate the adjusted odds ratios (aORs) of intensive care unit (ICU) admission and in-hospital death (and combinations of these outcomes) among influenza-positive patients by vaccination status for three target groups: young children (aged 6-24 months), adults (aged 18-64 years) with pre-existing health conditions, and older adults (aged ≥65 years). Survival curves were used to compare length of hospital stay by vaccination status in each target group.

Findings: 2747 patients hospitalised with PCR-confirmed influenza virus infection between Jan 1, 2013, and Dec 8, 2019, were included in the study: 649 children (70 [10·8%] fully vaccinated, 193 [29·7%] partially vaccinated) of whom 87 (13·4%) were admitted to ICU and 12 (1·8%) died in hospital; 520 adults with pre-existing medical conditions (118 [22·7%] vaccinated), of whom 139 (26·7%) were admitted to ICU and 55 (10·6%) died in hospital; and 1578 older adults (609 [38·6%] vaccinated), of whom 271 (17·2%) were admitted to ICU and 220 (13·9%) died in hospital. We observed earlier discharge among partially vaccinated children (adjusted hazard ratio 1·14 [95% CI 1·01-1·29]), fully vaccinated children (1·24 [1·04-1·47]), and vaccinated adults with pre-existing medical conditions (1·78 [1·18-2·69]) compared with their unvaccinated counterparts, but not among vaccinated older adults (0·82 [0·65-1·04]). Compared with unvaccinated individuals, lower odds of ICU admission were found for partially vaccinated children (aOR 0·64 [95% CI 0·44-0·92]) and fully vaccinated children (0·52 [0·28-0·98]), but not for adults with pre-existing conditions (1·25 [0·93-1·67]) or older adults (0·88 [0·72-1·08]). Lower odds of in-hospital death (0·62 [0·50-0·78]) were found in vaccinated versus unvaccinated older adults, with or without ICU admission, but did not differ significantly in partially vaccinated (1·35 [0·57-3·20]) or fully vaccinated young children (0·88 [0·16-4·82]) or adults with pre-existing medical conditions (1·09 [0·73-1·63]) compared with the respective unvaccinated patient groups.

Interpretation: Influenza vaccination was associated with illness attenuation among those hospitalised with influenza, although results differed by vaccine target group. These findings might suggest that attenuation of disease severity might be specific to certain target groups, seasons, or settings.