The safety of influenza vaccination in pregnancy: Examining major congenital malformations as potential adverse outcomes using UK electronic health records

Abstract

The aim of this thesis was to examine the safety of maternal influenza vaccination with respect to major congenital malformations in live-born infants. UK electronic health records from the Clinical Practice Research Datalink were used and work was conducted using linked primary care, hospitalisation and mortality data. The first study systematically reviewed existing methods for identifying congenital malformations in UK electronic health records, and the results of any validation studies. Studies relied on stand-alone primary care or hospitalisation data to identify congenital malformations; none examined linkage between these. Overall, congenital malformations recorded in primary care data had a high positive predictive value (80-100%) but the validity in hospitalisation data was not explored. Methods from these studies informed the development of a comprehensive algorithm to identify major malformations in live-born infants. Using linked primary care, hospitalisation and mortality data, the second study in this thesis demonstrated that just 20% (95% CI, 19-21) of infants with a major malformation had evidence of their condition in both primary care and hospitalisation data. Almost 65% (95% CI, 64-66) only had evidence in hospitalisation data. The third study demonstrated that the overall prevalence of major malformations established in primary care data using this algorithm was slightly higher than published estimates from other studies using UK primary care records (Prevalence ratio, 1.2; 95% CI, 1.2-1.3). Comparisons of linked data with population-based registry data demonstrated a four-fold higher prevalence for major malformations overall in the linked electronic health records (Prevalence ratio, 4.3; 95% CI, 4.1-4.5). This was primarily driven by the high prevalence of some of these conditions in hospitalisation data, which could potentially be explained by nonspecific codes used to record certain malformations that could have related to either major or minor conditions. 5 The fourth study examined the association between the trivalent seasonal inactivated influenza vaccine and major malformations. Among 78,150 live-birth pregnancies, 6,872 (8.8%) were vaccinated in the first trimester whilst 46,669 (59.7%) were unvaccinated throughout pregnancy. There was no evidence to suggest an association between first-trimester vaccination and major malformations recorded in first year of infant life in models adjusted for confounding (HR, 1.06; 99% CI, 0.94-1.19; p=0.23). The fifth study, which examined the safety of the monovalent pandemic inactivated influenza vaccine, showed similar results (HR, 1.02; 99% CI, 0.72-1.46; p=0.86). However, although these vaccine safety studies did not find evidence for an association between vaccination and major malformations, terminations due to foetal anomaly were not included. Therefore, the possibility of an increased risk of the specific subtypes of major malformations typically detected during antenatal scans and subsequently terminated could not be discounted. These results provide additional evidence on the safety of maternal influenza vaccination but highlight the need for further explorations of major malformations among pregnancies that do not result in live-births. The component of this work relating to the methods used to identify major malformations highlights the potential to increase ascertainment through the use of linked data whilst underscoring the need for further studies, particularly in hospitalisation data, to establish the validity of codes used to record these conditions.