Vaccinating Mothers to Protect Their Babies Against Influenza

(See the Brief Report by Walker et al, on pages 16–20.)

Influenza virus infection is responsible for significant morbidity and mortality in children, particularly among those younger than 6 months [1]. Preventing influenza in these young infants is therefore an important health priority. Efforts to prevent influenza in infants aged <6 months through direct vaccination have not been successful, because of inadequate infant immune responses [2]. Providing passive protection to the infant through vaccinating pregnant women may be an attractive alternative to direct immunization.

The efficacy of seasonal influenza vaccination during pregnancy in preventing laboratory-confirmed influenza in infants younger than 6 months has been demonstrated by 4 randomized, controlled trials from Bangladesh, South Africa, Mali, and Nepal [3–6]. It is also critical to evaluate the effectiveness of this strategy when implemented through routine public health programs outside of controlled clinical trials; equally important is documentation of results from different geographical areas and use of different clinical end points.

Influenza vaccine effectiveness may differ from year to year as a result of varying antigenic matches between circulating strains and the recommended annual vaccine [7]. Surveillance data are commonly used to generate annual estimates of vaccine effectiveness, often using the test-negative case-control study design, which is an efficient and feasible approach to generate real-time vaccine effectiveness data. In this issue of The Journal of Infectious Diseases, Walker et al [8] used the screening method, described below, to estimate the effectiveness of maternal vaccination, which is an alternative approach to generating real-time estimates of vaccine effectiveness. They used electronic laboratory surveillance records to identify confirmed influenza cases in infants and a pregnancy register to estimate maternal vaccination uptake. This retrospective study demonstrated that influenza vaccination during pregnancy was associated with an overall 50%–66% reduced risk of laboratory-confirmed influenza virus infection and a similar reduction in influenza-related hospitalization during the first 6 months of life among the infants born to vaccinated mothers during the 2013–2014 and 2014–2015 influenza seasons in England. Higher effectiveness point estimates were detected against the dominant influenza virus strains circulating in both seasons, with an effectiveness of 78% against influenza A(H1N1)virus infection and 60% for influenza A(H3N2) virus infection, although the 95% confidence intervals overlapped.

The screening method is a frequently used method for estimating vaccine effectiveness, which compares the vaccination coverage between the reported cases to that of an external reference group from which the cases emerged [9]. For this particular study, the authors relied on a pregnancy register. This method was designed to provide rapid preliminary estimates of vaccine effectiveness when incidence and attack rate data may not be available. Key features of the screening method include timeliness, inexpensiveness, efficiency, and ease due to reliance on available data and given that individual-level information is only required for cases [9, 10]. Notwithstanding these advantages, reliable individual-level data on case vaccination status and data on population-level coverage, as well as robust information on confounding factors, are required to minimize the danger of biased estimates.

A major limitation of the screening method is that vaccine effectiveness estimates can only be adjusted for covariates measured in the population with vaccine coverage data. Although in the current article, maternal vaccine coverage was matched by week of birth and geographical region of the cases, and analysis was further adjusted for ethnicity, the authors cannot exclude the possibility of residual confounding inherent to ecological studies. Importantly, if unvaccinated mothers visit the doctor with their ill infants less frequently than vaccinated mothers do, this may bias the vaccine effectiveness toward lower estimates. Conversely, an overestimation of the vaccine effectiveness in the current study, which the authors acknowledge, might have resulted from the fact that 17% of cases were born before term. Shorter pregnancies not only have fewer opportunities to receive vaccine, but preterm infants also acquire lower levels of transplacental transferred antibodies [11].

The assessment of the effectiveness of maternal influenza vaccination against severe illness in infants is associated with several specific challenges beyond those of routine observational studies of vaccine effectiveness [12]. These include the fact that laboratory-confirmed hospitalized influenza cases involving infants may be uncommonly identified as clinical presentation can be nonspecific and that, in many settings, influenza viral testing and/or reporting of cases are not routine. Moreover, data on pregnancy status and stage of pregnancy may be lacking in many settings. Complexity is added by the fact that the vaccine is administered to the pregnant women and the outcome is experienced by the infant, introducing an additional need for linkage of data. The timing of pregnancies and subsequent birth of infants are variable in relation to the influenza season. Because of these variables, doses of influenza vaccine may be administered to pregnant women whose infants will not fall into the population at risk for influenza, because they may be born after the influenza season. In addition, infants born before the start of or during the annual influenza vaccination campaign may not have the opportunity to be protected through maternal vaccination. Traditional measures of vaccine effectiveness do not capture these important elements of overall programmatic effectiveness, which are specific to maternal vaccination programs for seasonal diseases.

In 2012, the World Health Organization recommended that countries considering influenza immunization programs place pregnant women at the highest priority for seasonal influenza vaccination [13]. However, data on the real-world impact of these programs remain limited, particularly from low- and middle-income countries. Walker et al describe an approach of using surveillance data combined with electronic data sources, which could potentially be extended to other settings. In recent years, a number of low- and middle-income countries have begun implementing electronic immunization registries to replace paper-based systems, providing the opportunity to enhance decision making around vaccine programs [14] and to design studies using these systems for vaccine evaluation. Although currently these programs are primarily focused on the vaccines included in Expanded Program on Immunization during childhood, because recommendations for vaccination of pregnant women are expanding, it will be important to incorporate these new technologies to capture the immunizations provided to the women.

It is estimated that influenza caused almost 3 million episodes of acute lower respiratory tract infection in infants globally in 2008 [15]. Maternal vaccination is an important strategy to reduce these numbers. Despite this, uptake of influenza vaccination during pregnancy remains low, even in high-income countries such as the United Kingdom and the United States, where coverage remains <50% [8, 16]. Data on real-world effectiveness of maternal influenza vaccination programs, such as those provided by Walker et al [8], are important to sustain and expand existing programs and motivate countries without existing maternal influenza vaccination programs to consider their introduction.

Notes

Potential conflicts of interest. C. C. reports receiving nonfinancial support from Parexel and grants from Sanofi and the Centers for Disease Control and Prevention outside of the submitted work. M. C. N. reports receiving grants from BMGF and MedImmune and personal fees from Sanofi Pasteur and Pfizer outside the submitted work.

Both authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References