Josephine van Dongen

4 Rotavirus vaccine effectiveness 105 Premature infants in our study were generally of lower GA (33.8% 27-30 weeks compared to 20% in the immunogenicity study), therefore, immaturity of the gut and immune system and consequently a poorer rotavirus vaccine response may explain in par t the lower VE in our study. In addition, the gut microbiome, is a known factor of influence on HRV immune response and is different between healthy term and premature infants. Some microbiota species are associated with lower rotavirus vaccine IgA responses, 23 and their presence in premature infants is different than in healthy term infants, depending on gestational maturity. 24 However, we also found low VE in term infants with congenital pathology suggesting that other mechanisms are involved as well. It is impor tant to mention that based on the eligibility criteria, par ticipants in our cohor t represent a par ticularly vulnerable group of infants with prolonged care (between six and 14 weeks of age). For example the median length of post-natal hospital stay was 28 days. In addition, 83.3% of infants in our HRV vaccinated cohor t received the recommended two doses and protection is lower after just one dose of HRV in healthy term infants (range 60-92%). 25,26 A pathogen related factor is genotype diversity, protection is primarily against the outer membrane proteins, defining the antigenetically distinct rotavirus G- and P- genotypes. 27 Although HRV elicits both homotypic and heterotypic immunity, the protective effectiveness may differ by rotavirus genotype. A meta-analysis of strain-specific VE found 94%, 87% and, 71% effectiveness against homotypic, fully – and, par tially heterotypic rotavirus genotypes, respectively with overlapping confidence intervals. 28 In addition, during our study period (2015- 2019) the most dominant circulating strains were all par tially heterotypic (G3P8, G4P8 and, G9P8). The homotypic G1P8 rotavirus genotype did not exceed 14% in any of the study years, 29 as was noticed in more high-income countries. 30,31 VE in our study-population may have been influenced by this genotype distribution. Some limitations need to be addressed. First, fewer than expected rotavirus positive AGE episodes (n=117) were detected, reducing the anticipated statistical power of the study. Cumulative severe rotavirus incidence was assumed at 4% during the 18 months of life for this population, we found only 2.7%.This was however the same for all infants in the Netherlands, the previous expected 3500 annual pediatric rotavirus hospitalizations was adjusted during the study period to 2024. 11,32 This resulted from reduced rotavirus epidemic intensity compared to pre-study years, coinciding with a shift towards a biennial pattern. 33 And fur thermore, from incomplete sampling of AGE episodes during follow-up. In a post-hoc analysis we therefore analyzed all-cause AGE as outcome, which is a non-specific, but more sensitive measure of effect. The results were in line with our analysis of the primary outcome. Fur thermore, the probability that our study had incorrectly estimated lowVE, when trueVE would be more than 60% (the estimate used in the sample size calculation) was small. Second, our study suffered from lower than expected inclusions per hospital and incomplete follow-up. We therefore enrolled five more hospitals, adding to a total of 13 instead of eight, used for sample size calculation. Loss to follow-up occurred in 30% of par ticipants (higher than