The burden of influenza among Kenyan pregnant and postpartum women and their infants, 2015–2020

Abstract Background In tropical Africa, data about influenza‐associated illness burden are needed to assess potential benefits of influenza vaccination among pregnant women. We estimated the incidence of influenza among pregnant women and their infants in Siaya County, Kenya. Methods We enrolled women at <31 weeks of gestation and conducted weekly follow‐up until 6‐month postpartum to identify acute respiratory illnesses (ARIs). We defined ARI among mothers as reported cough, rhinorrhoea or sore throat and among infants as maternal‐reported cough, difficulty breathing, rhinorrhoea or clinician diagnosis of respiratory illness. We collected nasal/nasopharyngeal and oropharyngeal swabs from mothers/infants with ARI and tested for influenza A and B using molecular assays. We calculated antenatal incidence of laboratory‐confirmed influenza among mothers and postnatal incidence among mothers and infants. Results During June 2015 to May 2020, we analysed data from 3,026 pregnant women at a median gestational age of 16 weeks (interquartile range [IQR], 13, 18) and followed 2,550 infants. Incidence of laboratory‐confirmed influenza during pregnancy (10.3 episodes per 1,000 person‐months [95% confidence interval {CI} 8.6–11.8]) was twofold higher than in the postpartum period (4.0 [95% CI 2.6–5.5]; p < 0.01). Incidence was significantly higher among human immunodeficiency virus (HIV)‐infected pregnant women (15.6 [95% CI 11.0–20.6] vs. 9.1 [95% CI 7.5–10.8]; p < 0.01). Incidence among young infants was 4.4 (95% CI 3.0–5.9) and similar among HIV‐exposed and HIV‐unexposed infants. Conclusion Our findings suggest a substantial burden of influenza illnesses during pregnancy, with a higher burden among HIV‐infected mothers. Kenyan authorities should consider the value of vaccinating pregnant women, especially if HIV infected.


| INTRODUCTION
Pregnant women have been observed to frequently seek medical care for influenza-associated illnesses and are at high risk of hospitalization. [1][2][3] Studies documented an increased burden of influenza among pregnant women during the 2009 influenza pandemic. 4 However, the few studies reporting seasonal influenza incidence among pregnant women and their infants are typically neither from tropical nor low-and middle-income countries (LMICs). 5 Although studies from temperate regions suggest that pregnant women with comorbidities 6,7 and those in the second and third trimesters have higher risk of influenza-associated hospitalization compared with nonpregnant women, 1,8 pregnant women in LMICs may be disproportionately affected by influenza infections. 9 Many tropical areas of Africa have a high prevalence of comorbidities like human immunodeficiency virus (HIV), tuberculosis (TB) and malaria 10,11 that may influence both incidence and complications of influenza illness. 12,13 Moreover, yearround circulation of influenza viruses 14 may contribute to risk of influenza during pregnancy 7 in ways that might differ from countries with distinct seasonality.
In 2012, the World Health Organization (WHO) recommended prioritization of influenza vaccination among pregnant women in countries initiating or expanding influenza immunization programmes. 15 The recommendation was based on increased risk of severe influenza illness and hospitalization observed among pregnant women during seasonal influenza epidemics and the 2009 pandemic 16,17 and the safety and effectiveness of influenza vaccines during pregnancy. 18,19 WHO also highlighted the potential benefits of maternal vaccination for infants. 20 Children born to mothers vaccinated during pregnancy receive passive protection against influenza during their first months of life, when influenza vaccination is not recommended for infants. 21,22 Despite the WHO recommendation, few LMICs routinely vaccinate pregnant women against influenza, in part because of limited data about the burden of disease in this group. 23 Kenya, like most middle-income countries, has yet to adopt the WHO recommendation for maternal influenza vaccination and currently has no national influenza vaccination programme for any target group. In 2016, Kenya's National Immunization Technical Advisory Group provisionally recommended influenza vaccination among children aged 6-23 months; 24 recommendations for maternal vaccination were deferred until data about influenza burden during pregnancy were available. Here, we estimate the incidence of medically attended influenza illness among Kenyan pregnant women and their infants to inform policy recommendations.

| Study population
In 2015, the Kenya Medical Research Institute (KEMRI) in collaboration with the U.S. Centers for Disease Control and Prevention (CDC) initiated a prospective cohort study to estimate the burden of influenza disease in pregnancy and its impact on birth outcomes in western Kenya. The study was conducted in two public hospitals in Siaya County: Siaya County Referral Hospital and Bondo sub-County Hospital. Siaya County has one of Kenya's highest burdens of maternal and infant mortality; 25 in 2014, maternal mortality was 692 per   100,000 (compared with the national average of 362 per 100,000) and infant mortality was 50 per 1,000 live births (national average, 39 per 1,000). 26 Among women aged 15-49 years, HIV prevalence was 22%, compared with the national average of 5%. 27 Malaria is endemic in the region with an overall prevalence of 37%, 28  Pregnant women were eligible for participation if they (i) were under 31 weeks of gestation (established by ultrasound or fundal height), (ii) aged 15-49 years, (iii) resided within 10 km of study facilities and planned to remain in the study area for the next 12 months, (iv) consented to HIV counselling and testing, (v) agreed to all followup visits, and (vi) were willing to deliver at the study facility. Enrolment was restricted to consenting, women with singleton pregnancies (because of likelihood of delivering low birth weight babies among women with twin pregnancies) and those without leg or spinal deformity or history of fistula repair (because of risks for complicated delivery).

| Participant characteristics
Before enrolment, study staff obtained written informed consent from women who were willing to participate in the study. Nurses and clinical officers then obtained detailed medical and obstetric histories from enrolled participants and provided HIV counselling and rapid diagnostic testing. The staff also collected demographics, participant contact information for follow-up, current and past smoking history, use of medications such as anti-malarial medication (including sulfadoxine-pyrimethamine for intermittent preventive treatment in pregnancy) and other medications, and hospitalizations during the current pregnancy. Health screening included vital signs, physical and obstetric examination (fundal height and presentation, fetal heart sounds and vaginal examination). Mothers were screened for high blood pressure and oedema to evaluate risk of pregnancy-induced hypertension or preeclampsia. Laboratory examinations included blood grouping (ABO and Rhesus factor), haemoglobin count, urinalysis, venereal disease research laboratory tests, blood smear examination and rapid diagnostic tests for malaria. Medical history included history of TB or active TB on treatment, asthma, chronic obstructive pulmonary disease (COPD)/chronic bronchitis, diabetes, hypertension, epilepsy or other chronic illness. For HIV-infected women, we documented information about the use of highly active antiretroviral therapy (HAART) and daily use of cotrimoxazole.

| Surveillance for ARIs
We conducted weekly surveillance among participants and their infants to identify acute respiratory illnesses (ARIs), using participants' preferred contact method (telephone or home visits by CHV or study staff), until delivery and for 24-week postpartum for both mother and infant. Participants not reachable after three consecutive weeks were declared lost to follow-up at the time of last contact. We defined ARIs among pregnant and postpartum mothers as reported cough, runny nose or sore throat and among infants as reported cough, difficulty breathing, runny nose or clinician-diagnosed respiratory illnesses in the past 7 days, during weekly follow up, and within the past 10 days during clinic visits for mothers and infants.
We also inquired about any non-respiratory illnesses among pregnant women, mothers and infants. Patients were tested for malaria if they were symptomatic or if clinicians/nurses suspected malaria infection. Mothers nearing their expected date of delivery were reminded to deliver at study hospitals. At delivery, study staff examined mothers and newborns and documented birth outcomes. For deliveries that occurred outside study facilities, maternal and newborn examinations were conducted within 72 h of birth.
Mothers reporting ARI symptoms or symptoms in their infants during weekly surveillance were asked to come to study clinics for further assessment and management by a study clinician. Respiratory

specimens (nasopharyngeal [NP]/nasal [NS] and oropharyngeal
[OP]/throat swabs) were collected at study clinics from mothers or/and infants who met criteria for ARI. As a benefit of participation, additional specimens including blood smear for malaria tests, and urine for urinalysis was collected for diagnostics and patient management as per Ministry of Health guidelines. Episodes occurring 14 or more days after a previous ARI episode were considered new episodes.

| Laboratory methods
Swabs collected (NS/NP and OP) from a patient presenting with ARI were combined and kept between 2 C and 8 C in cryovials with viral transport medium at field sites, then transported at the end of the day to KEMRI central laboratories located at the main field station in Kisian-Kisumu for testing. The respiratory specimens were tested with real-time reverse transcription polymerase chain reaction (RT-PCR) for influenza A and B viruses, using primers and probes provided by CDC's Influenza Division, USA. 30 Assays were considered positive at cycle thresholds of <40.0.

| Data analysis
We describe demographic and clinical characteristics of pregnant women and pregnancy outcomes, including their infants using counts, percentages, medians, ranges and interquartile ranges (IQRs). We defined the gestational trimesters as first, at ≤13 weeks; second, from 14 to 27 weeks; and third, ≥28 weeks. Periods of increased influenza activity during the year were marked out as months where more specimens were testing positive for influenza than the national average of 7%.
We calculated influenza incidence per 1,000 person-months stratified by maternal HIV status or presence of any chronic condition.
Chronic conditions were defined as asthma, TB, diabetes, COPD/ chronic bronchitis, epilepsy or hypertension (including essential or pregnancy-induced hypertension and preeclampsia/eclampsia). We had no access to HIV testing data for infants; we therefore stratified our analysis by HIV exposure for the infants.
In calculating rates, we divided the number of laboratoryconfirmed influenza episodes by person-months of follow-up and adjusted for the proportion of ARI episodes for which specimens were not tested and for episodes reported during weekly follow-up but not sampled. To estimate the 95% uncertainty interval (confidence interval [CI]), we ran 1,000 iterations allowing for the proportion sampled (tested) and the proportion that tested positive for influenza to vary assuming a binomial distribution defined by the actual observed proportions. We used the 2.5th and 97.5th values to estimate the lower and upper limits of the 95% CI. Antenatal person-time was calculated from enrolment to pregnancy outcome (delivery or miscarriage), maternal death or loss to follow-up; postpartum person-time was calculated from delivery to maternal death, loss to follow-up or study completion at 24 weeks for mothers, and postnatal person-time from birth to infant death, loss to follow-up or study completion at 24 weeks for infants. We performed analyses using Stata version 13.0 (Stata Corp., College Station, TX). Statistical significance was considered at p < 0.05.

| Description of pregnancy and birth outcomes and infant follow-up
Among the 3,026 participating women, 54 (1.8%) had miscarriages prior to 20-week pregnancy, 2,605 (86.1%) gave birth, and 367 (12.1%) were lost to follow-up ( Figure 1A). Of those that gave birth, 55 (2.1%) had stillbirths at a median (IQR) gestational age of 37.5 (32,39) weeks, and the remaining 2,550 (97.9%) bore live term babies at a median (IQR) gestational age of 39 (37,40) weeks. All the 2,550 live infants were included in the study. Five maternal deaths occurred postpartum: Three of these deaths occurred within 14 days after delivery (Table S1). No deaths occurred during pregnancy. None of the deaths were associated with respiratory illness. When we assessed the characteristics of women who were lost to follow-up compared with those who completed the study, younger women <25 years old (63.5% vs. 51.9%, p < 0.01) and those in their first pregnancies (40.1% vs. 24.8%, p < 0.01) were more likely to be lost to follow-up (Table S2).

| ARI and influenza among women
During the study period, our surveillance identified 1,828 ARI episodes among 1,250 women during pregnancy; 400 (32.0%) women women had ≥1 illness episode, and 23 (5%) had ≥3 episodes. Sixteen postpartum women were hospitalized for any illness, and none had a respiratory illness.  (Table S3). Similarly, we did not find any significant differences in rates of influenza infection when we compared women who had malaria and those without malaria infection during pregnancy.

| ARI and influenza among infants
During follow-up, a total of 2,042 ARI episodes occurred among

| Incidence rates of influenza among infants
The incidence of influenza per 1,000 person-months among infants    The maternal burden of influenza in Africa is under-recognized because of lack of epidemiological and laboratory surveillance data. 31 Moreover, relative to the general population, pregnant women are less likely to seek care for a respiratory illness, 32,33 and even when they do, influenza testing is not routinely done. In our study, we found that Kenyan pregnant women experience substantial morbidity from ARI compared with pregnant women in high-income regions 4,34,35 and other tropical upper middle-income countries. 36  Studies conducted in other LMICs have reported similar annual influenza incidence for infants younger than 6 months to those estimated in Kenya. 9 Whereas we observed increased incidence in pregnant women with HIV, we did not identify increased incidence in HIVexposed infants, possibly due to small sample size. Even though we found no difference in risk of influenza among the two groups of infants, prioritizing influenza vaccine for women living with HIV is still a reasonable strategy for both the mother and baby, especially when resources are insufficient to vaccinate all pregnant women and/or their infants aged >6 months.

| Influenza circulation
The incidence of influenza among infants aged 3-6 months in our study was twice that of infants aged <3 months. Our findings suggest that antibody transfer from mothers naturally infected with influenza during pregnancy may offer protection to infants in their early months of life. 42

| Limitations of the study
The following limitations should be considered when interpreting our findings. First, because of the limited size of our cohort, we were unable to assess the impact of influenza on severe outcomes such as hospitalization and death. Second, it is possible that women were more motivated to report ARIs during pregnancy than immediately after birth especially if a report would lead to a clinic visit for a new mother with a neonate and potentially other children to attend to. These would potentially lead to underestimation of the burden among postpartum mothers. Similarly, the median postpartum persontime contributed was significantly less compared with the intended 6-month postpartum follow-up. Finally, it is possible that HIV-infected women were more motivated to report ARIs compared with those who were HIV-uninfected, because they are educated on importance of early healthcare seeking for better treatment outcomes.

| CONCLUSION
Our study shows a substantial burden of influenza-associated illness during pregnancy, especially among HIV-infected mothers and sug-

PEER REVIEW
The peer review history for this article is available at https://publons. com/publon/10.1111/irv.12950.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author after permission from relevant authority to release data upon reasonable request.