Quantifying Factors Associated With Birth Outcomes and Their Implications on Evaluating the Success of a Maternal Respiratory Syncytial Virus (RSV) Vaccine Program in Kenya


 BackgroundMaternal immunisation to prevent respiratory syncytial virus (RSV) associated disease among infants is in focus. However, little is known about adverse birth outcomes and associated factors occurring in a setting with high morbidities of malaria, HIV infection and undernutrition. Quantifying these ahead of introduction of a maternal vaccine would assist in assessing an association between RSV vaccination and adverse birth outcomes. MethodsA cross-sectional survey was conducted to collect data on birth outcomes from women residents of the health and demographic surveillance systems (HDSS) of Siaya and Kilifi, Kenya and from the maternity wards of Siaya County referral hospital and Bondo sub-county hospital. Participants of the HDSS sites had pregnancies registered in the years 2017 to 2020 through census rounds and were traced at home for interview. All women had a birth outcome by the time of data collection. Multiple logistic regression was used to determine independent predictors of adverse birth outcomes. Results A total of 2219 women were interviewed. Median age during pregnancy was 27.7yrs (range: 22.7-32.4), 1857 (83.7%) attended antenatal care clinic (ANC), 1,979 (89.2%) delivered at a health facility and 2204 (99.3%) reported they would take up a new maternal vaccine. Adverse birth outcomes occurred in 781 (35%) of pregnancies; 490 (62.7%) were preterm, 247 (31.6%) low birth weight, 189 (24.2%) macrosomia and 42 (5.4%) still births. Predictors of adverse birth outcomes were, eclampsia (AOR 6.86 (1.40-33.60); p=0.017), gestational diabetes (AOR 3.01 (1.24-7.30; p=0.015), and home delivery (AOR 2.48 (1.20-5.13); p=0.014). Being multiparous (AOR 0.52 (0.33-0.81); p=0.004) was protective. Home delivery was significantly associated with older maternal age 40-49 years (p=0.001), multiparous >5 (p=0.001), level of formal education below primary (p=0.001) and Islamic religion (p=0.001). ConclusionsIn this maternal population, about a third of pregnancies have adverse birth outcomes. Recognizing this baseline prevalence will be important in validating safety of a new maternal vaccine. Monitoring of the actual safety outcomes of the maternal RSV vaccine, will require integrated initiatives to mitigate against factors affecting utilization of maternal healthcare services and individual factors associated with adverse birth outcomes.


Abstract
Background Maternal immunisation to prevent respiratory syncytial virus (RSV) associated disease among infants is in focus. However, little is known about adverse birth outcomes and associated factors occurring in a setting with high morbidities of malaria, HIV infection and undernutrition. Quantifying these ahead of introduction of a maternal vaccine would assist in assessing an association between RSV vaccination and adverse birth outcomes.

Methods
A cross-sectional survey was conducted to collect data on birth outcomes from women residents of the health and demographic surveillance systems (HDSS) of Siaya and Kili , Kenya and from the maternity wards of Siaya County referral hospital and Bondo sub-county hospital. Participants of the HDSS sites had pregnancies registered in the years 2017 to 2020 through census rounds and were traced at home for interview. All women had a birth outcome by the time of data collection. Multiple logistic regression was used to determine independent predictors of adverse birth outcomes.

Conclusions
In this maternal population, about a third of pregnancies have adverse birth outcomes. Recognizing this baseline prevalence will be important in validating safety of a new maternal vaccine. Monitoring of the actual safety outcomes of the maternal RSV vaccine, will require integrated initiatives to mitigate against factors affecting utilization of maternal healthcare services and individual factors associated with adverse birth outcomes.

Background
Maternal immunisation is one of the strategies to help achieve the third sustainable development goal of ensuring healthy lives and promoting well-being for all at all ages by 2030, through reduction of maternal and infant mortality (1). Recent decades have seen an increase in development of maternal vaccines that may reduce infant mortality (2). Maternal vaccines to prevent in uenza and pertussis diseases are licensed (3)(4)(5)(6) and in use in high income countries, while, maternal tetanus toxoid vaccine has successfully reduced the burden of neonatal tetanus in Kenya (7,8). Besides, there are plans to introduce a new maternal vaccination program in low and middle income countries (LMICs) to prevent respiratory syncytial virus (RSV) associated disease in early infancy for which vaccines are at advanced stages of clinical evaluation (2,(9)(10)(11).
It is envisioned that a maternal RSV vaccine will be bene cial if implemented optimally and if the key factors likely to impact its success in a resource-poor setting are clearly understood. Some of the factors that can in uence the successful implementation of a new maternal vaccine are the associated risks or perception of risks of the vaccine to pregnancy outcomes (12)(13)(14). The current maternal RSV vaccine in development has been found to be effective if administered during the third trimester of pregnancy which implies that infants born premature (especially those born less than 33 weeks of gestation) are less likely to bene t from this program (9). Since vaccine safety data among pregnant women is limited, assessment of baseline rates of adverse birth outcomes and associated factors ahead of clinical trials, would bene t validation of risks and safety of the new maternal vaccine (14,15).
Some studies have reported that maternal immunisation may cause adverse outcomes to the mother or infant before or after delivery (16,17). Settings with high rates of morbidities such as HIV, malaria and undernutrition are likely to experience the largest burden of adverse birth outcomes such as preterm births, low birth weight and still births (18), which can obscure outcomes in a maternal vaccine program (19). However, many of these birth outcomes are underreported if they do not occur within a health facility. Furthermore, many pregnant women do not complete all recommended antenatal care (ANC) visits where cost-effective interventions to help prevent adverse birth outcomes can be provided (20). In low resource settings such as Kenya, many births still occur at home (21) and little is known about these pregnancy outcomes, which undermines government efforts of providing free maternal health care services (22,23) and hinders accurate evaluation of the effectiveness of new vaccine programs.
In this study, we quanti ed rates of adverse birth outcomes, where they occur, predictors of adverse birth outcomes, factors that in uence the choice of a place for delivery and the implications of these pregnancy outcomes in validating the safety of a maternal respiratory syncytial virus (RSV) vaccine program in Kenya.

Study site
This study was conducted at Siaya county referral hospital and Bondo sub-county hospital and within the Kili and Siaya HDSS areas, in Kenya.
The Kili HDSS area ( Fig. 1) has previously been described (24). It is situated along the coastal part of Kenya, covering an area of 891km 2 and a population of ~ 300,000 residents as of 2019. Kili HDSS monitors population through census rounds, three times-a-year, and registers about 8000 pregnancies every year. (20,24). The Kili HDSS area is endemic for malaria which has a mortality rate among children aged 6 months to 4 years of 0.57 per 1000 person-years (95% CI 0.2, 1.2) (25) while, neonatal mortality is unclear. However, in the recent years malaria incidence has declined, partly due to interventions that have reduced transmission (26). The KHDSS population is served by over 60 health facilities (both private and public) in which pregnant women attend for care and about 60% of the deliveries at Kili county referral hospital are from this HDSS area (24).
The Siaya HDSS is managed by KEMRI-Centre for Global Health Research (CGHR) with technical and nancial support from US Centers for Disease Control and Prevention (CDC), Kenya (27), and is situated in Siaya County, in the rural western part of Kenya (Fig. 2). This surveillance system covers an area of 700km 2 and monitors a population of about 260000 individuals and records approximately 6000 births per year (28). The Siaya HDSS area has a high burden of malaria, pneumonia and diarrheal diseases (29,30) and is used to conduct longitudinal population based infectious disease surveillance (PBIDS). HIV prevalence in Siaya county is 21% according to 2018 Kenya HIV estimates report, which is among the highest prevalence in Kenya (31). Routine home visits to collect morbidity data are conducted twice a year since 2015 and residents within villages where PBIDS occur are given free care at St. Elizabeth Lwak Mission hospital for all potentially infectious disease syndromes (27).

Study Population
The study population comprised of women residents registered as pregnant during 2017-2020 census rounds in the HDSS areas of Kili and Siaya (Asembo). The study also included, pregnant women presenting at maternity wards of Bondo Sub-county Hospital and Siaya County referral hospital for delivery between February and April 2021. All women had a birth outcome by the time of the interview and data collection.

Study Design
This was a cross sectional survey set up to collect data on birth outcomes and gestational age at attendance for antenatal care among pregnant women as part of a larger study to assess factors that would in uence successful implementation of a maternal RSV vaccine program in Kenya.
The target sample size for women with data on birth outcomes was 1000 per HDSS area. These were randomly selected from census registers, with an equal number of women selected from each of the HDSS administrative locations as previously described (20). However, to replace women that would be missed during home visits, a sample of 2000 women were randomly selected from each of the census registers of Kili and Siaya HDSS areas, and the lists uploaded in the study databases and assigned to trained eld interviewers for tracing. The interviewers visited homesteads of the selected women, consented them for participation into the study and electronically collected data on ANC attendance, ANC services provided, birth outcomes and other obstetric or demographic details using a standardized questionnaire uploaded in computer tablets.
Pregnant women from Bondo and Siaya hospitals maternity wards were also interviewed to provide comparative data on timing for ANC attendance and birth outcomes occurring in referral hospitals within Siaya HDSS. All pregnant women were eligible for enrolment as they presented in the maternity wards of these hospitals. They were all approached after delivery, consented and those who accepted participation were interviewed and their records of birth outcomes collected using similar tools as those used within the HDSS areas (20).

Ethical considerations
Informed consent was obtained in writing from all participants. This study was approved by the KEMRI Scienti c and Ethical Review Unit Committee (SERU #3716).

Statistical Analyses
All data collected from the community in Siaya and Kili HDSS areas were merged for speci c analyses of the differences in participants' characteristics and factors in uencing choice of place for delivery.
Data collected from the hospitals of Bondo and Siaya was merged with datasets form Kili and Siaya HDSS sites for analyses of birth outcomes and predictors of adverse birth outcomes. These analyses focussed on the following birth outcomes: Normal live births, still births, preterm births, macrosomia, and low birthweight. A stillbirth was de ned as the death or loss of a baby before or during delivery after 20 weeks of pregnancy (32). Preterm birth (PTB) was de ned as baby born alive before 37 weeks of pregnancy are completed (33). Preterm births were further categorized into < 32 weeks as very early PTB, 32-<34 weeks as early PTB and 34-36 weeks as late PTB. Low birthweight was de ned as the weight of a neonate below 2500 grams (g) at birth (34).
Descriptive statistics for proportions, mean (standard deviation: SD) and median (Interquartile range: IQR) were used in analysis. The characteristics of pregnant women from Kili and Siaya HDSS sites were compared using a chi-square test. The chi square test was also applied to determine variables which in uence choice of a place for delivery and to assess association between adverse birth outcomes and maternal characteristics. Univariate and multivariable logistic regression models adjusted for each variable category (maternal age, place of delivery, marital status, maternal weight, level of education, occupation, religion, parity, gravida, gestational diabetes, malaria infection, eclampsia, gestational age at delivery, number of ANC visits, timing for ANC initiation and delivery mode) were used to determine predictors for adverse birth outcomes and factors for choice of place for delivery with p < 0.05 taken as statistically signi cant association. All analyses were conducted in Stata version 15 (Stata Corp, College Station, USA).

Characteristics of participants
The median age of these women at the time of delivery was 27.7 years (IQR: 22.7-32.4). 218 (9.8%) women did not have data on gestational age at delivery, while 136 (6.1%) women were missing data on infant's birth weight. Of the 2219 women interviewed, only 4 (0.2%) did not attend antenatal care at all during pregnancy. Of those who reported to have attended ANC, 1,857 (83.7%) had booklets available to con rm attendance. The median birth weight of infants was 3.3 kilograms (kgs) (IQR: 2.9-3.5), while median gestational age at delivery was 37.8 weeks (IQR: [37][38][39][40]. Overall, 1374 (62.0%) of the women had formal education at primary level, 1919 (86.6%) were married, 850 (34.4%) had no formal employment and 1,979 (89.2%) delivered at a health facility. Two hundred and forty (10.8%) delivered at home; 239 (99.6%) were interviewed from the community in HDSS sites, while one was enrolled from the maternity ward of Siaya hospital where she had attended due to retained placenta.
Women from Kili and Siaya HDSS sites were signi cantly different in most demographic characteristics (Table 1). Compared to Siaya HDSS, the Kili site had more women who were older 40-49 years (9.6% vs 4.0%; p = 0.001), fewer women were divorced or separated (0.7% vs 3.4%; p = 0.001), more women with no formal education at all (16.7% vs 0.29%; p = 0.001), more women had more than 6 children (15.2% vs 6.0%; p = 0.001), more babies were born underweight (10.1% vs 4.5%; p = 0.001) and more deliveries occurred at home (29.8% vs 6.0%; p = 0.001). However, women from Kili and Siaya HDSS areas were similar in gestational age at delivery (73.0% vs 73.2%; p = 0.358) and had a similar proportion of adverse birth outcomes (38.2% vs 38.0%; p = 0.931).  Among all 2219 women interviewed in this study, the choice of place for delivery was signi cantly associated with the facility a pregnant woman attended for ANC screening (p = 0.001), being of older maternal age 40-49 years (p = 0.001), having had more than 5 live births (p = 0.001), having level of formal education above primary (p = 0.001), delayed ANC initiation (p = 0.305) being of an Islamic religion (p = 0.001), attending one or no ANC visit (p = 0.001) and having no employment (p = 0.001); but was not associated with marital status (p = 0.134), and gestational age at delivery (p = 0.085) ( Table 3).
Pregnant women who attended ANC in dispensaries (OR 3.3 (95%CI 1.8-6.1); p = 0.000) were more likely to deliver at home than those who attended ANC in a county referral hospital. Pregnant women with formal education to tertiary level were less likely to deliver at home than those who did not have any formal education (OR 0.02 (95%CI: 0.01-0.22; p = 0.001). Muslim women were more likely to have home deliveries than Christians (OR 3.8 (95%CI: 1.9-7.4; p = 0.001), while women who were self-employed were less likely to deliver at home than those who reported not to have any employment (OR: 0.4 (95%CI 0.2-0.5); p = 0.001).

Acceptance of a new maternal vaccine among pregnant women in Kenya
Majority of women 2204 (99.3%) reported they would accept a new maternal vaccine during ANC visits to prevent pneumonia among their infants. However, 803 (36.2%) would rst consult before taking the vaccine. Of these, 499/803 (62.1%) would consult their spouse, 246 (30.6%) would consult their doctor or health care provider or other persons who received the vaccine to con rm it was safe and 58 (7.2%) would consult friends and relatives. The 15 (0.7%) women who reported would not accept the new maternal vaccine,14 were from the community, while one was interviewed from the hospital and the main reason was fear that the vaccine might not be safe.

Discussion
In this sample of women, we found two thirds of pregnancy outcomes are normal live births. The proportion of adverse birth outcomes was signi cantly higher for births occurring at home than in hospital, but similar among all women from the two geographically diverse regions and who were different in demographic characteristics. The difference in characteristics in this study provide a good comparison with the general population as our results are similar with ndings of the 2014 Kenya demographic health survey (21). These results also ascertain possibility of observing equal distribution of safety outcomes across maternal populations from the different geographical regions in Kenya on introduction of a new maternal vaccine.
We found preterm births were the most common adverse birth outcomes in this study accounting to nearly 22% (490/2219) of all infants. Additionally, 2% of the preterm births in this population, occurred within the gestational age period not likely to have an optimum level of protective antibody (10) after maternal vaccination. This implies that, this group of infants will be susceptible to severe RSV disease and might require use of other strategies such as prophylactic monoclonal antibodies for prevention (35,36).
Tertiary level of education, eclampsia, gestational diabetes, delayed ANC initiation, number of ANC visits, being multiparous and home delivery were found to be signi cantly associated with adverse birth outcomes in this study. Eclampsia has been found to cause deterioration of maternal conditions which result into adverse foetal outcomes such as low birthweight and still births (37). In Ethiopia, adverse neonatal outcomes such as macrosomia, preterm births and large for gestational age was found to be signi cantly higher among newborns from mothers with gestational diabetes (38), which is similar to ndings in this study. Women with low level of education might belong to a low socio-economic status group and could not afford the cost of care therefore, missing uptake of preventive services or skilled care during delivery (39). It has also been found that cultural practices during home delivery like massaging of the abdomen to align the baby, which is very common among indigenous populations along the coastal part of Kenya are associated with placenta praevia and abruption, asphyxiating the foetus and increased chances of trauma to the baby and premature delivery (39). Perhaps, a consideration to integrate strategies directed towards mitigating against causes of adverse pregnancy outcomes such as proper management of high-risk pregnancies, educating traditional birth attendants on risks of some of the cultural practices during delivery might be worth an undertaking.
We found 29.8% of births in Kili and 6.0% in Siaya still occurred at home. However, our results show a decline in proportions of home births as observed prior to free care during the 2014 Kenya demographic health survey (47% in Kili vs 27% Siaya) (21); which could be attributed to current government initiatives focused to achieve universal access to maternal and child health services (23). These initiatives include, "Beyond Zero" (22) which was launched in 2014, by the rst lady in Kenya, and aims to prevent maternal and infant deaths by providing mobile clinics to provide care to pregnant women who have no access to hospitals during delivery. In addition, the government of Kenya through the Ministry of Health in 2016, also launched another initiative known as "Linda Mama" (23) which ensures pregnant women and infants have access to free, quality and affordable maternal and child health services by use of a public funded health insurance scheme. The impact of these initiatives seems evident through the observed reduction of home deliveries in this study.
Among the women who delivered at home, 165 (69%) attended ANC and had booklets available, indicating they received ANC services. Multiple ANC visits in this study were also found to be associated with less adverse birth outcomes. It is worth noting that, high rates of ANC attendance among pregnant women enhances uptake of interventions, ensures high vaccine coverage (40) which mitigates against poor outcomes during and after delivery. However, full use of ANC and services alone does not prevent all adverse birth outcomes even in the highest resource settings. This is because, most pregnancy complications occur during delivery and can result in poor pregnancy outcomes (41) which can result into mis-interpretation of safety outcomes of a maternal vaccine. In a survey to nd out birth preparedness and complication readiness among Kenyan women showed only 11.4% (59/519) were well prepared for births and its complications during pregnancy (42). Initiatives such as provision of night transport services to pregnant women during labour, equipping lower levels health facilities with quali ed staff, training, oversight and resources to handle emergencies which will encourage pregnant women utilize hospital services to reduce risks associated with home deliveries, will be reasonable, but should not be a barrier to vaccine rollout. Procedures to detect adverse outcomes should be put into place during trial or within a managed phase iv rollout setting, to provide a reliable system for validating effectiveness and safety of a new maternal vaccine.
The choice of place for delivery was found to be associated with maternal age, facility for ANC attendance, religion, parity and education level. Older women were more likely to deliver at home than in hospital and this is perhaps a result of experience in having previous successful deliveries or uncomplicated pregnancy (39). Similarly, women who have had multiple pregnancies were more likely to deliver at home than in hospital while higher education level was associated with less home delivery as observed in other studies in coastal Kenya (39,43). A study in north eastern Kenya, found male doctors attending to women in labour prevented pregnant women from delivering in hospital (44) because of religious beliefs and in this study, we also nd Muslim women were more likely to deliver at home than in hospital. The role of in uencers such as spouse, healthcare providers and relatives in determining place of delivery is also of much importance and might require empowering of pregnant women in decision making. Introduction of new interventions among these pregnant women may also need consideration of the socio-cultural factors such as religion, individual perceptions on births or cultural beliefs to ensure maximum uptake.
Majority (99%) of pregnant women in this study reported they would accept the maternal RSV vaccine despite a few having concerns that the vaccine might not be safe. Perceptions about risks associated with a vaccine might result to high rates of vaccine refusals which is likely to affect the overall effectiveness. For instance, a study in Quebec found a belief that a H1N1 in uenza vaccine was not adequately tested resulted in its low uptake among pregnant women (45). Most pregnant women appear to have more trust on their health care providers regarding information on interventions available and their uptake within health facilities (46). For a successful implementation of the maternal RSV vaccine program in this setting, we recommend integration of sensitization and education sessions between health care providers and pregnant women perhaps through health talks and information brochures within ANC platforms to create awareness about the safety and e cacy of the new maternal vaccine, resolve doubts and increase con dence before its introduction.
There are some limitations in this study. HIV status was not collected among these women and, proportions of adverse birth outcomes attributed to HIV infection and how HIV would likely have altered the observed associations of other variables in multivariable analyses. The data is drawn from a sample of women from two out of the 47 counties and may not be representative of all Kenyan women. Stillbirths that occur in the community often go unreported and could have also been missed in this study. Hospital enrolment of some of the participants may represent a bias. However, our study provides important baseline data on birth outcomes which has often been missed by studies involving a small sample size of women and gives a detailed description of the baseline proportions of adverse birth outcomes and associated factors in this setting which can guide validation and monitoring of the safety outcome of a new maternal vaccine program.

Conclusions
In this sample of women, about a third of pregnancies had adverse birth outcomes most of which occurred at home. Recognizing this baseline prevalence will be important in validating safety of a new maternal vaccine. Births occurring at home might hinder evaluation of maternal vaccine safety. Acceptance of the maternal RSV vaccine is likely to be high with 99% of women reporting willingness for uptake during pregnancy. However, successful implementation of the maternal RSV vaccine program will require integrated initiatives to mitigate against individual factors associated with adverse birth outcomes and factors affecting utilization of maternal healthcare services such as educating women on the bene ts of ANC services during pregnancy and attending early enough to get bene t and introduction of platforms to create awareness about the safety and e cacy of the new maternal vaccines. All methods were carried out in accordance with relevant guidelines and regulations. Written informed consent was obtained from all women who participated in this study. No consent was sought from parents or guardian of participants under 16 years as these were considered as emancipated minors who were able to consent for themselves and did not require parental consent. This was done following Kenya Ministry of Health guidelines for conducting adolescent HIV sexual and reproductive health research in Kenya (2015) (47). The local ethics committee KEMRI Table3.xlsx