H1N1 influenza vaccination during pregnancy

Risks to pregnant women from influenza infection have long been recognized.1 The recent 2009-10 H1N1 pandemic was no exception—pregnant women were at higher risk of severe H1N1 influenza illness compared with the general population,2 and those with H1N1 influenza had higher rates of adverse pregnancy outcomes than did uninfected pregnant women.3 Despite limited safety data for use of the monovalent H1N1 vaccines in pregnancy, pregnant women were widely prioritized for H1N1 vaccination programs.4 Fortunately, enhanced surveillance of pregnant women during the pandemic has enabled retrospective evaluation of the safety of monovalent H1N1 vaccine in obstetric populations around the world.

One such evaluation is published in this issue of The BMJ (doi:10.1136/bmj.g3361). Trotta and colleagues report findings from their large retrospective cohort study assessing the relation between an adjuvanted monovalent pandemic H1N1 vaccine administered during pregnancy and adverse maternal and fetal/neonatal outcomes.5 The study was conducted in the Lombardy region of northern Italy and examined a population of 86 171 women with a singleton pregnancy ending in a live birth or stillbirth between 1 October 2009 and 30 September 2010. About 7% (6246) of women received H1N1 vaccination during pregnancy, predominantly during their second or third trimester (6172 out of 6246 vaccinated women).

As other observational studies have noted, the baseline characteristics of vaccinated and unvaccinated pregnant women differed in numerous important ways—for instance, vaccinated women were more likely to have a university degree, to be employed, and to have taken folic acid supplements during the peri-conceptional period. The authors attempted to mitigate the potential confounding effects of these differences by constructing a propensity score matched cohort in which vaccinated women were matched to unvaccinated women at a one to four ratio. Table 1 of the paper illustrates the improved comparability of vaccinated and unvaccinated women following the application of this approach. Importantly, women were also matched on gestational age on the date corresponding to vaccination for the exposed women. This avoids the survival bias that can be produced when women at lower risk of time dependent adverse outcomes such as fetal death have longer gestations and thus greater opportunity to be vaccinated.

The authors found no association between H1N1 vaccination and neonatal death, stillbirth, or small for gestational age neonates. Other observational studies of these outcomes have generally reported either no association with H1N1 vaccination or, in some cases, a protective effect. In contrast, a modest but significant increased risk of gestational diabetes (adjusted odds ratio 1.26, 95% confidence interval 1.04 to 1.53) and pre-eclampsia/eclampsia (1.19, 1.02 to 1.39) was reported.5

Few studies have examined H1N1 influenza vaccination in relation to gestational diabetes or hypertensive disorders, and no consistent pattern is apparent in those that have. Whether the date of clinical diagnosis of these gestational disorders was available and used to exclude women vaccinated following detection is not clear from the description provided. If not, reverse causation could also be an explanation for the findings, as women may be more likely to be vaccinated because of these conditions. Temporality may also be a factor in the assessment of congenital malformations, in which it seems that exposed women who received their vaccination after the period of first trimester organogenesis have been included.

The vaccines received by pregnant women during the H1N1 pandemic differed from usual seasonal preparations in two ways: they were monovalent (immunizing against a single antigenic strain), and several included adjuvants (substances added to vaccines to amplify the immune response).6 In general, the epidemiologic studies of pandemic monovalent H1N1 vaccines in pregnancy, which now number more than a dozen and reflect diverse obstetric populations, suggest that the H1N1 vaccines administered to pregnant women during the 2009-10 pandemic were not associated with any increased risk of adverse perinatal outcomes.

Reviews of passive surveillance systems also reached this conclusion.6 7 However, high clinical and design heterogeneity across the observational studies, despite assessment of the same antigenic strain during the same influenza season, poses challenges for meta-analyses and vaccine specific risk assessment.8 The number of studies reporting a given fetal/neonatal outcome and using a consistent definition in relation to a specific vaccine formulation is small. For instance, we know of only two other studies that have assessed the same MF-59 adjuvanted H1N1 vaccine (Focetria) studied by Trotta and colleagues, and none of the outcomes is common to all three studies.5 9 10

A need clearly exists for additional research on influenza vaccination during pregnancy. Information pertaining to safety of adjuvants in pregnancy is limited.11 If adjuvants are to be used in obstetric populations in future pandemic or non-pandemic vaccines, additional safety evidence is warranted. Evidence pertaining to outcomes after influenza vaccination in the first trimester also remains limited.12

The certainty that observational studies and “real time” surveillance will continue to play a central role in monitoring the safety of influenza vaccine during pregnancy underscores the importance of ensuring the availability of data sources to support those activities. Ideally, this should include population based surveillance systems that include pregnancy specific information as well as vaccine registries, both with the flexibility to respond to urgent data requirements such as those posed by the 2009-10 H1N1 pandemic. Increased international collaboration to align research methods or share study data would help to reduce heterogeneity across observational studies, benefiting future meta-analyses of study level or individual level data.