Josephine van Dongen

6 Non-specific effects of human rotavirus vaccination 151 enteritis without the rotavirus specific code, misclassification may have occurred and could explain the observation mediated by direct, rather than indirect vaccine effects. Most likely, a reduction in acute hospitalizations resulting from non-specific effects would be mediated by lowering infection incidence. This is however not suppor ted by our findings that show a similar rate of ARI in vaccinated and unvaccinated infants. Fur thermore, when restricting the analysis to hospitalizations for infectious diseases only, an even smaller effect was estimated. Alternatively, one could argue that non-specific effects following rotavirus vaccination reduce severity of ARI, rather than ARI incidence, thereby reducing the risk of hospitalization. However, no reduction was observed in days with ARI symptoms, which could be considered a proxy for severity. Another hypothesis mentions that inactivated childhood vaccinations abrogate the beneficial non-specific effects of earlier administered live vaccines. To investigate this, we conducted a sensitivity analysis in which we restricted the analysis to the time-period up to the next routine childhood vaccination at 11 months, but the effect estimate was unchanged. Jointly, our results do not suppor t the existence of non-specific effects for rotavirus vaccines lowering overall hospitalization rates, infection incidence or severity. For eczema, we did not find a significant effect of rotavirus vaccination either. Evidence on preventing atopy by vaccination so far is based on one underpowered randomized clinical trial and several heterogeneous low-quality observational studies. 29 Unfor tunately, our study was also underpowered to draw any firm conclusion on the effect of rotavirus vaccination on eczema incidence. Our point estimate of 0.89 suggests some benefit may exist, but this warrants confirmation in other controlled studies. In this study the majority of infants was born prematurely.While non-specific effects have also been investigated among low bir thweight infants in multiple randomized trials, 6,30,31 premature infants have a higher risk of respiratory morbidity and hospitalization. 32 Therefore any non- specific effect on ARI incidence should become apparent in this specific study population. However, direct vaccine effectiveness in our study population was lower than previously anticipated ( chapter 4 ). If the direct vaccine response in this population is reduced, this may also apply to non-specific vaccine effects. Overall, there is an additional need for research into non-specific effects of rotavirus vaccination in a healthy term infant population. A few limitations of this study need to be addressed. First, we used complete cases only, missing information is documented and comparing par ticipants with complete versus missing data did not reveal significant differences (data not shown). Loss to follow up occurred in approximately 30% of par ticipants. By using time-to-event analysis and including an offset for observation time this was taken into account in our analyses and we therefore expect this has minimal effect on our results. We were unable to assess severity of ARI, a discrimination in episodes by severity might fur ther explain why the difference in acute hospitalizations was not reflected in occurrence of ARI. Instead, we assessed number of days with ARI complaints and observed no relevant difference