Maternal factors of newborn low birthweight in malaria endemic settings of Nanoro, rural Burkina Faso

Background. In sub-Saharan Africa, the intermittent preventive treatment of malaria in pregnancy using sulphadoxine pyrimethamine (IPTp-SP) strategy is recommended to limit malaria adverse effects on birth outcomes. Ten year after IPTp-SP was adopted in Burkina Faso, we assessed the magnitude and maternal factors of low birthweight (LBW) in Nanoro. Methods. A secondary analysis of data from a cross-sectional study was carried out in women who gave birth at Nanoro peripheral health centers using a binary multivariate logistic regression. Maternal sociodemographic factors, gyneco-obstetrical history and relevant medical characteristics were evaluated to identify associated factors. A p-value less than 0.05 was considered statistically signicant. Results. Of 291 delivery records examined, 14 % of women received three or more doses of SP while 80% used bed nets the night before their admission for delivery. Malaria was detected in 36.1% and anemia in 52.9% of women. The average neonate birthweight was 2933 g and 12 % were born with a low birthweight. By multivariate analysis, rst delivery (OR = 8.84, [95% CI: 3.72-21.01]), and being multiparous with history of stillbirth (OR = 5.03, [95% CI: 1.54-16.40]) were signicantly associated with an increased risk of LBW.

implemented during pregnancy. In addition, 25% of all stillbirth occurring in sub Saharan Africa [19] and 40% of all premature births follow a maternal infection with malaria or STI [20,21].
Sulfadoxine-pyrimethamine (SP) is recommended to prevent malaria adverse effects in pregnancy in sub Saharan Africa through the intermittent preventive treatment of pregnant women (IPTp) [22]. The strategy entails the administration of a full treatment course of SP to pregnant women starting early from the second trimester and administered monthly until delivery [23]. Despite the overwhelming evidence supporting the use of SP, the spread of P. falciparum resistance combined to low IPTp-SP uptake however threatens the antimalarial effectiveness of the strategy [24].
In Burkina Faso, IPTp-SP was adopted in 2005 starting at the second trimester of pregnancy with the requirement of 3 doses of SP [25]. Seven years later, the magnitude of LBW did not decrease substantially (from 15.8% in 2005 to 13.4% in 2012) and since then no other evaluation was carried out although it is still the recommended strategy [26,27]. A reevaluation of the impact of the current IPTp strategy is needed to understand its utility and prevent the effects of non-e cacy. The main goal of this analysis was to measure the prevalence of term LBW and to identify associated risk factors in Nanoro ten years after IPTp-SP policy was adopted.  [27]. We hypothesized that the policy change would reduce this proportion to 10% within the ten-year period of implementation. The sample size was estimated to test the hypothesis, with a precision i = 3.5%, a con dent interval of 95 % (α = 0.05, with a critical value Z 2 =1.96), the minimum sample size required was n = 282 participants using the following formula n = Z 2 *p*(1-p)/i 2 . This was a secondary analysis of database collected on 291 participants recruited to estimate malaria prevalence. Therefore, the available sample size was adequate to test the hypothesis.
2.5 Data collection procedures and variables collected. Data were collected from the health center birth registries, the antenatal care cards and by neonate examination. Information not available in these documents were collected by one-on-one person interviews using a standardized questionnaire. Neonate examination was conducted within 24 hours of delivery either in the health facility or by home visit by trained midwives. Variables collected for mothers are presented in table 1 and included age, occupation, education level, parity, history of miscarriage, history of stillbirth, the number of living children at the time of the survey, the use of insecticide treated net (ITN) the night before visiting the health facility for delivery, the number of ANC visits performed, knowledge of the recommended protocol of 3 SP received during the pregnancy. Neonate birthweights were measured using calibrated Seca® 384 electronic scale with 10 g resolution and a precision less than 10 grams (seca gmbh & co. kg, Germany). In addition, blood samples were obtained from nger pricks to measure maternal hemoglobin level using a portable spectrophotometer (HemoCue, Ängelholm, Sweden) and a malaria rapid diagnostic test was performed using SD-Bioline Malaria Antigen Pf ® test strips (sensitivity 99.7% and speci city of 99.5 %) for the identi cation of P. falciparum. 2.6. Data processing and analysis: data were entered into a REDCap database, cleaned and imported to Stata version 15 (StataCorp. 2017, TX, USA) for analysis. The uptake of IPTp-SP was grouped in one, two, and three or more doses. Anemia was de ned as hemoglobin level < 11g/dL, malaria as a positive rapid diagnosis test result. Because ultrasound was not available, pregnancy duration was assessed using the last menstrual period (LMP) or the Ballard score obtained by neonate examination whenever there was uncertainty. A term low birthweight was de ned as any birthweight from a singleton pregnancy of at least 37 weeks of gestational age that was below 2500 g [16,30]. Odds ratios (OR) of LBW and 95% con dent intervals (95% CI) were calculated according to each maternal factor using univariate logistic regression. Adjusted odd ratios (aOR) and 95% CI were derived by a backward elimination regression of variables with a p-values < 0.20 and retention of variables with statistically signi cant p-values. For the multivariable analysis, variable age was not included due to its strong correlation with the parity. Because of a comparable proportion of LBW if more than one delivery, mothers were subsequently grouped in single and multiple deliveries. In addition, variable history of stillbirth was combined to parity ( rst delivery, and multiple delivery with history of stillbirth). A p-value < 0.05 was considered statistically signi cant.

Ethical considerations:
The study protocol was discussed with local health authorities and community leaders to obtain their assent. Permission was obtained from the national ethics committee of Burkina Faso by the principal investigator (CERS 018.6.078). Informed consent was obtained from all women prior to participation. All participants with anemia or positive malaria test were treated according to the national guidelines.

Results
3.1. Study participants' characteristics: Table 2 summarizes study participants' characteristics. In brief, of the 291 mothers included, the average age (± standard deviation) was 26 (± 6) years and most of them were unemployed (85.2%). The median parity stood at three births (interquartile range, 2-5 births) and the median number of living children was three. One third of participant were tested positive for peripheral malaria (with a lower percentage among bed net users 33% versus 41% among non-users although this difference was not statistically signi cant) and half study participants were anemic. More than 96% of women reported three as the minimum SP doses required during the course of a pregnancy contrasting with the low rate (14%) of those that effectively received at least three doses

Low birthweight prevalence and maternal factors:
Of the 291 neonates, the mean birth weight (± standard deviation) was 2933 ± 390 g and 12.0 % (35/291) were born with a birthweight below 2500 g.
This proportion was higher among teenage mothers (28.8%). Table 3 presents the prevalence of LBW in relation to maternal sociodemographic characteristics, obstetrical history as well as the univariate odds ratios. Younger maternal age, and being primigravid or a multigravida with history of stillbirth were positively associated with the risk of low birthweight, while sleeping under bed net was signi cantly associated with a lower prevalence. Uptake of more than 3 doses of SP was not signi cantly association with LBW despite the downward trend of LBW prevalence observed as the uptake of SP increased. The prevalence of low birthweight was higher among anemic mothers although this association was not statistically signi cant (p=0.052). However, the risk of anemia was signi cantly increased mother with malaria compare to those without malaria (66.7% versus 44.2%, p = 0.01)

Discussion
Low birthweight prevalence was still high in Nanoro particularly given that low birthweight from preterm births were not counted. However, the observed prevalence was 3.8 percentage points lower than the 15,8 % reported among term neonates at the start of the new strategy in the same region [27]. Lower prevalence than that reported in Nanoro was observed in other African countries with a similar malaria patterns and comparable intervention packages for antenatal care services. Indeed, in Ethiopia (10%) in 2014 [31], and Nigeria (7%) in 2013 [32] the reported prevalence was lower than that of the current evaluation. However these estimates could be understated as in the Nigerian's study, data were collected retrospectively from the routine surveillance system and it is reported that half of delivery data in Africa is not recorded due to weaknesses in the reporting system [5]. Similarly, the study in Ethiopia was carried out in an urban area where the economic level of women tend to be higher than those from the rural area and the impact of better economic status on the prevalence of low birthweight is already described [33,34]. In contrast, rate in excess were reported in South Africa (38.5%) in 2014 [35], and in a study conducted in Zimbabwe (16.7%) [36]. The reason behind this higher prevalence in the South African and Zimbabwean studies could be related to the higher prevalence of human immuno-de ciency virus (HIV) infection [37]. HIV is reported in many African studies as a possible risk factor of LBW although the mechanism has not yet been fully elucidated [36,38]. This illustrates the importance of testing HIV when evaluating the impact of interventions allocated to pregnant women.
It is reported that the maternal obstetrical history and events occurring during the course of the pregnancy impact the outcome of pregnancies [39]. However, the speci c factors reducing the birthweight are not always identi ed. Our results showed that combined parity and obstetrical history of stillbirth were signi cantly associated with the occurrence of low birthweight in rural Burkina Faso. Similarly to Hoquet et al. report, our study showed that the risk of LBW signi cantly increased at the rst delivery [38]. The impact of primiparity on low birthweight was reported by several authors [40][41][42]. The biological ground for rst delivery children to be at higher risk of low birthweight may be related to the placenta of rst pregnancy women being at higher risk of parasite-infected red blood cells sequestration, disrupting thus nutrients supply to the fetus [43] or because some physiologic changes occurring during pregnancy to facilitate fetal growth with the process incompletely reversing in postpartum may create a more e cient environment for subsequent fetuses [44]. Indeed, uteroplacental blood ow, which is responsible for delivering oxygen and nutrients to the fetus, is greater during subsequent pregnancies compared to the rst pregnancy [45,46]. In addition, there may be structural factors that limit the uterine capacity in the rst pregnancy as described in animal models [47]. Furthermore, infants born from rst pregnancy may have developed in a different maternal immune environment, contributing to relative growth restriction, compared to subsequent pregnancies [48]. Along with primiparity, mother with multiple pregnancy and an obstetrical history of stillbirths also presented a higher risk of delivering low birthweight neonates compared to multiple pregnancy without history of stillbirths. Similar observation was reported by Chen et al. in 2018 andRozi et al. in 2016 [49,50]. Ahrens et al. have noticed that the risk of low birth weight was highest after history of three stillbirths [51]. Particular attention to pregnant women with such background characteristics during their antenatal care visits could be of greater impact in mitigating the risk of low birthweight among them.
Poor maternal nutritional status is also described as a key factor of poor intrauterine fetal growth [52]. Even though that characteristic was not assessed in the current study, the high prevalence of anemia could indicate a poor nutritional status [53]. Indeed, a study conducted in the same site showed that low maternal weights, low maternal body mass indexes and low brachial perimeters were signi cantly associated with anemia [48]. In the current study, anemia was positively associated with an increased risk of low neonate weights at univariate analysis (p = 0.052) however that trend weakened with adjustment to parity by multivariate analysis and also because of reduced sample size secondary to missing values. Further assessment building on the scope of this nding may be necessary to show the adequate impact of anemia and poor nutritional status on the risk of the occurrence of LBW.
Studies reported that using bed net throughout pregnancy reduces the risk of adverse delivery outcomes including low birthweight [54]. A similar trend was observed by univariate analysis in the current study (p=0.018) which showed bed net use to e caciously prevent adverse birth outcomes by reducing up to 11% of malaria infections. However, this could not be con rmed by multivariate analysis because of the reduced sample size due to missing values. Studies building on a larger sample size may be required to assess the actual impact of ITN use on the reduction of LBW A number of limitations in this study are worth noting. The study was conducted on women who have delivered at term excluding low birthweight from preterm deliveries and that could understate the actual extent of the problem. It remains to bear in mind that gestational age record was collected from parental assessment of last menstrual period (LMP), which could be prone to some errors, as majority of the study population did not recall their LMP. In addition, SP is not indicated during the rst trimester of pregnancy and malaria infection may have already occurred at the start of its administration from the second trimester. Although the analysis had limitations, it is worthy identifying the population that need special attention during antenatal care. Larger studies building on the scope of these factors examined in this report will be valuable in guiding efforts to curtail the impact of low birthweight in sub-Saharan Africa.

Conclusions
This report showed that low birthweight prevalence remained high among neonates in rural Nanoro although its magnitude has considerable decreased. It also showed that the intake of IPTp-SP was still low among the women. Improving the coverage of IPTp combined to interventions aiming to reduce poor maternal nutritional status may be key to successfully reduce the magnitude of LBW in such settings.