UC Berkeley

Pregnant women, infants, and children < 5 years of age are at increased risk of acquiring influenza and its complications. As a result, the World Health Organization recommends that pregnant women and children 6 months to < 5 years of age receive influenza vaccine. However, nearly all of the risk information about influenza in pregnant women, infants, and children comes from temperate regions of the world, where vaccine is plentiful and the disease is seasonal. Very little information exists on influenza in pregnant women and young children in tropical zones, where vaccine coverage is scant and seasonality of influenza transmission is less well understood. Although the highest incidence rates of respiratory illness among children are found in Africa, influenza vaccine use there is rare, and very little research on influenza has been carried out in this region, where co-morbid conditions, such as HIV infection are prevalent, and increase the risk of severe influenza outcomes. Given the limited knowledge of influenza in this region, it is difficult to argue for the use of influenza vaccination.

This dissertation used data collected in Mali from a cohort study of influenza vaccine naïve pregnant and postpartum women and their infants conducted in April 2010 – August 2011 and from a randomized controlled trial of influenza vaccine in pregnant women who were randomized to receive either influenza vaccine or a control vaccine in 2011-14. In both studies, pregnant and postpartum women, their infants, and household contacts < 5 years of age were followed prospectively, and participants were actively assessed for signs of influenza-like-illness (ILI); samples were taken when individuals met the case definition for ILI and tested for influenza by RT-PCR. We also collected a blood smear to exclude cases of ILI due to malaria. In both studies, all households members were followed until the infant turned six months of age. These datasets were used to answer questions regarding influenza and ILI incidence in pregnant women, postpartum women, and infants < 6 months, and household transmission in these groups, as well as predicting the likely impact of modeling influenza vaccine use in the influenza vaccine naïve population.

In Chapter 1, we present results from our analyses that determined the age-adjusted incidence of ILI and laboratory-confirmed influenza (LCI) among pregnant and postpartum women and their newborn infants < 6 months of age in 2010-2011. Although LCI incidence was low, as compared to estimates reported in these same groups in other countries, we found that pregnant and postpartum women < 20 years had the highest rates of LCI, and that rates of LCI decreased with each additional decade of life. We found high rates of ILI in pregnant women, postpartum women and infants. Despite high rates of ILI in postpartum women, rates were higher in pregnant women, and pregnancy was a risk factor for ILI. Women were 2.28 times as likely to develop ILI during pregnancy as during the postpartum period. While pregnant women had a greater risk of LCI than postpartum women, the difference was not statistically significant, possibly due to the low case count we observed in the cohort. We also found that the risk of influenza in adult women was greater in households with a child under the age of five years, compared to households with no children in this age group. Among infants, incidence of LCI and ILI was highest among four and five month olds. The risk of a case of influenza in an infant increased with increasing number of children < 5 years of age residing in the household, although this difference was not statistically significant. Lastly, we found that although we observed few cases of LCI, influenza activity in Mali exhibited a defined bimodal seasonality, with cases peaking in September/October and February.

In Chapter 2, we describe our estimates of the serial interval (SI), the time between primary and secondary cases of influenza in the household, for influenza A (H1N1), influenza A(H3N2), and influenza B viruses. We found that our estimates were, generally, at the shorter end of the range of those previously reported. In this chapter, we also investigated the effect of individual and household factors on the median SI. The presence of cigarette smoking in the household was associated with a shorter SI. SIs were longer in households with moderate and high levels of household crowding. Residing in a household where a pregnant woman had received influenza vaccine during the trial was associated with a shorter SI, although none of the cases of household transmission occurred in pregnant women. Few instances of household transmission involved adults. The age of the index case did not have a statistically significant effect on median SI.

In Chapter 3, we present the predictions from the mathematical modeling analysis in which we used an SEIR compartmental model and ordinary differential equations to model the seasonal peaks of influenza activity in Mali and modeled the likely impact of introducing influenza vaccination in children on the cumulative case count of influenza among children residing with other vulnerable populations at risk for severe influenza-related outcomes (i.e. pregnant, postpartum women and infants). We found that even nominal influenza vaccine coverage (10%), if achieved before transmission began, reduced the total epidemic size considerably, to only one-fourth of the number of cases observed in the same population without vaccination. Our analyses also demonstrated that the timing of vaccination efforts played an important role in the final epidemic size, with implementation of vaccination delayed until after the first case of influenza had appeared resulting in larger projected epidemic sizes than if vaccination had been implemented four weeks before influenza activity began. Vaccinating in late August and the third week of December led to the largest projected reductions in total case counts.

In sum, we found that influenza circulation had a defined bimodal seasonality in Mali, with peak cases occurring in late September/October and February. The risk of influenza in adult women was greater in households with a child < 5 years of age, as compared to households without children in this age group. In our study of household transmission of influenza, we found that presence of cigarette smoking in the household was associated with a shorter SI. The mathematical models suggested that influenza vaccination of children six to fifty-nine months of age administered prior to periods of peak influenza activity (late August and December) may reduce the cumulative number of cases of influenza in children who reside with pregnant and postpartum women and infants under six months of age in Mali.