This also increased our ability to allow for variations in diagnoses patterns INK1197 nmr over time. Indeed, the RIRI diagnoses attributable to influenza increased for the latter seasons unlike the specific influenza diagnoses. A weakness of the study is that we included pregnant women with underlying conditions. Therefore our NNV is an underestimate of the NNV among healthy pregnant women.

However, from a policy perspective, we aimed for a minimum NNV estimate in a Swedish context. Even so, in Sweden our NNV estimates were considered high. Other weaknesses relate to underlying assumptions behind our NNV results: that all pregnant women were unvaccinated and at risk of contracting influenza, and that any effect of vaccination of other population groups can be disregarded. All of these assumptions can be debated on different grounds and there is unfortunately limited information to assess their importance. For example, the assumption that all the pregnant MG-132 chemical structure women are unvaccinated is not correct because in Sweden pregnant women belonging to risk-groups were recommended vaccination. Thus, our NNV could be overestimated. However, the vaccine

uptake is unknown but estimated by the profession to be very low (<5%). Finally, the estimates do not take into account that the same individual may be hospitalized repeatedly during one season, nor does the model include other infectious agents that may cause some of the hospitalizations, nor the time-point for vaccination in relation to epidemic influenza activity. This may lead to an underestimate of the NNV. On the other hand, the following may have led to an overestimation of the NNV: hospitalizations with other diagnoses, e.g. exacerbations of pulmonary or cardiac conditions, were not included; neither

were secondary diagnoses, which could have included aminophylline influenza although the main diagnosis did not; nor the effect the vaccine could have on infants, including small-for-gestational-age [26] and symptomatic influenza infection [27]. However, with regard to infant hospitalization, few children <6 months were hospitalized with influenza as main diagnosis. In 2003–2009, 3–15 cases/season were identified, although some cases may be undiagnosed [28]. Our estimate of the absolute risk of hospitalization in an average season with 80% VE resulted in an NNV of 4,138. However, few studies have evaluated the effectiveness of seasonal influenza vaccination during pregnancy, especially there is a paucity of intervention-studies with verified influenza as outcome [29]. If VE instead is 60% then the NNV would exceed 5,500, but in a more severe season NNV could be 3,499.