Background: Low birthweight is a major contributor to infant mortality. We evaluated the association between antenatal care (ANC) attendance and low birthweight among newborns in 5 regions of Burkina Faso. Methods: We utilized data from the baseline assessment of a randomized controlled trial evaluating azithromycin distribution during the neonatal period for prevention of infant mortality. Neonates were eligible for the trial if the weighed at least 2500 g at enrollment and were 8–27 days of age. Data on ANC attendance and birthweight was extracted from each child’s carnet de santé, a government-issued health card on which pregnancy and birth-related data are recorded. We used linear and logistic regression models adjusting for potentially confounding variables to evaluate the relationship between ANC attendance (as total number of visits and ≥ 4 antenatal care visits) and birthweight (continuously and categorized into < 2500 g versus ≥2500 g). Results: Data from 21,223 births were included in the analysis. The median number of ANC visits was 4 (interquartile range 3 to 5) and 69% of mothers attended at least 4 visits. Mean birthweight was 2998 g (standard deviation 423) and 8.1% of infants were low birthweight (< 2500 g). Birthweight was 63 g (95% CI 46 to 81 g, P < 0.001) higher in newborns born to mothers who had attended ≥4 ANC visits versus < 4 visits. The odds of low birthweight among infants born to mothers with ≥4 ANC visits was 0.71 (95% CI 0.63 to 0.79, P < 0.001) times the odds of low birthweight among infants born to mothers who attended < 4 ANC visits. Conclusions: We observed a statistically significant association between ANC attendance and birthweight, although absolute differences were small. Improving access to ANC for all women may help improve birth outcomes. Trial registration: The parent trial is registered at clinicaltrials.gov: NCT03682653; first registered 24 September 2018.
We utilized data from the baseline assessment of a randomized controlled trial evaluating whether a single oral dose of azithromycin administered during the neonatal period was effective for reducing infant mortality in Burkina Faso (clinicaltrials.gov {"type":"clinical-trial","attrs":{"text":"NCT03682653","term_id":"NCT03682653"}}NCT03682653) [11]. The trial was conducted in 44 primary healthcare facilities in 5 regions of Burkina Faso, including Centre, Boucle du Mouhoun, Cascade, Centre Ouest, and Hauts-Bassins. Enrollment lasted from April 2019 through December 2020. Facilities were in a mix of urban and rural settings. Because the parent trial was comparing oral azithromycin administered in the neonatal period to placebo, facilities were chosen to be within 4 h of a pediatric hospital that had pediatric surgical facilities in the case of infantile hypertrophic pyloric stenosis, a rare but serious condition that has been linked to azithromycin administration in observational studies [12]. Neonates were eligible for the trial if they were between 8 and 27 days of age, weighed at least 2500 g at enrollment, were able to feed orally, and did not have clinical signs of neonatal jaundice. Low birthweight babies were not excluded from the trial, but the child had to have gained enough weight to meet the trial’s weight-based enrollment criterion (2500 g) by the time they were 27 days of age to be able to participate in the parent trial and be included in this analysis. Only infants with birthweights recorded in the carnet de santé (government-issued health card) with complete data on antenatal care attendance were included in the present analysis. Neonates were recruited via contacting mothers who gave birth in the facilities and outreach during BCG vaccination days. The Institutional Review Boards at the University of California, San Francisco and the Comité National d’Ethique pour la Recherche (National Research Ethics Committee) in Ouagadougou, Burkina Faso reviewed and approved the study. Written informed consent was provided by the caregiver of each neonate enrolled in the trial. All study procedures were carried out in accordance with the Declaration of Helsinki and relevant guidelines and regulations. At enrollment, a baseline questionnaire was completed by the caregiver of each enrolled neonate. The questionnaire included the child’s age at enrollment and sex, maternal age and education (coded as none, primary, secondary, or higher than secondary school), and the number of times the mother had been pregnant. Questions related to the pregnancy included whether the child was a singleton or multiple birth and if the mother had given birth in a healthcare facility. We extracted data from the child’s health and vaccination card (carnet de santé) on the number of antenatal clinic visits attended by the mother and the neonate’s birthweight. We collected data on each healthcare facility included in the study, including whether it was in an urban or rural setting, if the facility has a physician on-site, and the number of non-physician clinicians, including nurses and midwives, employed by the facility. Because infants were not enrolled during pregnancy and gestational age testing is not widely available in the study area, information on gestational age was not collected. We evaluated the relationship between the number of antenatal care visits and 1) birthweight as a continuous variable using linear regression models and 2) low birthweight (defined as < 2500 g) using logistic regression models. All multivariable models were adjusted for maternal age, education, number of previous pregnancies, type of pregnancy (singleton vs multiple), the infant’s sex, region of the facility, if the facility was in an urban or rural setting, and if it had a physician onsite (as a proxy for level of care available at the facility). These variables were chosen because they were hypothesized to be confounders of the relationship between antenatal care attendance and birthweight. We chose this strategy as other strategies that rely on statistical criteria or comparison of adjusted and unadjusted effects may lead to omission of important but non-statistically significant confounders from the analysis and inappropriate adjustment for variables that are on the causal pathway between exposure and outcome [13]. We then evaluated the same outcomes using whether the mother had attended at least four antenatal care visits, adjusting for the same potentially confounding variables. We first ran a series of univariate models for antenatal care use and each confounding variable, and then a multivariable model for each exposure and outcome pair including all covariates. We did not base confounding decisions on statistical significance in the univariate model, due to known biases arising from relying on statistical criteria for confounder selection [14]. For all exposure-outcome pairs, we conducted a subgroup analysis among rural versus urban facilities to evaluate any differences the association between antenatal care use and birthweight by urbanicity. Due to the multilevel nature of the data, all models adjusted for clustering at the facility level using a Huber-White sandwich estimator. All analyses were conducted in Stata version 15.1 (StataCorp, College Station, TX).