Pre-term and post-term births: predictors and implications on neonatal mortality in Northern Ethiopia

This report reflects findings from a portion of a large prospective observational study undertaken in one of the nine administrative regions of Ethiopia to investigate pregnancy outcomes and neonatal survival. The current study is a cross-sectional survey and analysis of prospective data on 1162 neonates delivered at selected hospitals between April and July 2014 in Tigray, Ethiopia.

This study was part of a study conducted in seven randomly selected public hospitals in Tigray Region, located in northern Ethiopia. The Ethiopian healthcare system has a decentralized four-tier system of primary, district, zonal, and specialized hospitals. The primary care includes rural health posts nested into health centers serving populations of approximately 25,000. District hospitals form the tiers expected to serve catchments of 250,000. Zonal hospitals and specialized hospital serve 500,000 and 5,000,000 people, respectively. In Tigray Region, one of the nine administrative regions and two administrative cities in Ethiopia, there are 15 public hospitals categorized as zonal (6), one specialized hospitals (1), and the district and primary health care (8). All hospitals provide comprehensive maternal and newborn services including delivery service free of charge. Seven hospitals were selected by simple random sampling from across the region using lottery method [22].

Data were collected using a pre-tested structured questionnaire and checklist partially based on the World Health Organization’s verbal autopsy for neonatal deaths [23]. Data on maternal and obstetric characteristics were derived from the antenatal care record, through a post-partum interview with the mother and/or assessment of the neonate and mother. Data collectors interviewed all mothers, within the first 6 h after delivery, who experienced a live birth at the selected hospitals. Data related to the newborns were collected at birth by midwives and, during follow up, by either midwives or health extension workers. Mothers and neonates were followed for 28 days using three data collection methods/alternatives. On the first contact, the data collector visited the neonate daily, while he/she was in the hospital. At birth, socio-demographic and economic, neonatal, maternal and health service related characteristics were recorded, measured, and assessed. The second data collection method occurred after the mother was discharged, with the data collector communicating with the mother every 7 days either using a phone call or by creating a liaison with the health extension workers, inquiring about the neonatal condition and survival in every phone call. The third data collection alternative occurred for mothers who were not at home during the scheduled visit, when a revisit was arranged. When neonatal or maternal death occurred, the date and time of death was recorded and further contact was discontinued with the participant. To ensure the quality of data collected, one day training for the data collectors and health extension workers was provided on how to collect the data and how to follow the mothers. A pre-test was conducted on 5 % of the sample size to improve clarity and modify the data collection instruments.

Mothers with neonates born at the selected facilities between April and July 2014 were eligible for the study. Mothers referred from other health facilities or coming directly from their homes and admitted to one of the seven participating hospitals within six hour of birth were also included. Mothers who came from other neighbouring regions of Ethiopia, who were not able to speak, who had a psychiatric diagnosis, and/or who had no companion were excluded from the study.

We used previous literature and Mosley and Chen’s conceptual framework [24] with some modifications as a basis for categorization of the independent variables. This analytical framework is used for child health studies and combines sociological and biological explanatory models into a single conceptual framework by sorting independent variables as either distal or proximal determinants, in creating a hierarchy for analysis. The model set up by Mosley and Chen was adapted to fit specific areas, including neonatal health.

This study had two outcome variables that were analyzed independently. The first outcome variable was neonatal mortality, which is death of a live birth baby within 28 days of birth, categorized into binary outcomes for death or survival. The second outcome variable was birth outcome, which is the onset of the labour determined by gestational age of the live birth. This variable was categorized according to World Health Organization’s definitions of pre-term, term births, and post-term births [25]. Pre-term birth is “any birth before 37completed weeks of gestation or fewer than 259 days since the first day of the women’s last menstrual period” [25]. A term birth is a “birth between 37 completed weeks up to 41 completed weeks of gestation” [25]. A post-term is “any birth after 41 completed weeks of gestation or 294 days and above since the first day of the women’s last menstrual period” [25].

Gestational age was estimated based on the last menstrual period. If mothers were not able to recall their last menstrural period, gestational age was estimated using ultrasound methods.

The independent variables were broadly categorized as distal factors, such as socio-economic and demographic factors (i.e., monthly income, marital status, residence); proximal factors (i.e., maternal, neonatal, and health service related); and immediate variables (i.e., maternal complications).

Maternal complications, which were categorizes into ‘Yes’ or ‘No’ responses on assessment, included presence or absence of obstetric haemorrhage, puerperal sepsis, pyrexia, prolonged labour eclampsia and pre-eclampsia, mal-presentation and mal-positioning, premature rupture of membrane, cord prolapse, obstructed labour, cephalo-pelvic disproportion, emergency Caesarean section, and/or retained placenta. In addition, body mass index (BMI) was one of the maternal factors captured and categorized using the standardized categories of underweight (<18.5), normal (18.5–25), and overweight (>25.0). Neonatal factors considered included birth weight (categorized as low, normal, and overweight), sex, and weight for gestational age.

Birth outcomes were analyzed by categorizing into three outcomes variables (i.e., pre-term, term (base) and post-term) considered at the nominal level. Neonatal mortality was fit in a separate model categorized as death (coded “1”) or survival (coded “0”). Potential predictors for pre-term and post-term births were analyzed on the same model based on the assumption of multinomial logistic regression. The predictors were categorized as 1) immediate factors; 2) proximate factors (i.e., maternal, neonatal, and health service factors); and 3) distal factors (i.e., socio-economic and demographics variables). The independent variables considered for neonatal mortality were pre-term and post-term births.

A bivariate logistic regression was performed, and variables with likelihood ratio test P-values of less than 0.05 were retained in the multiple regression models. Variables considered in the final model by using the hierarchical method of multiple regression reflected immediate factors (Model 1) followed by proximal factors (Model 2) and, finally, distal factors being added (Model 3) as predictors of pre-term and post-term births. Multinomial and binary logistic regression was used to estimate the relative risk ratios (RRR) and relative risk (RR) as well as corresponding 95 % confidence interval (CI) for the predictors of birth outcomes and neonatal mortality at P < 0.05. During model building, cofounders were checked by looking for 15 % change in the regression coefficient. Interaction between predictor variables was tested at a significance level P < 0.05 by including an interaction term in the multivariate model. Multicollinearity among independent predictors was tested using Variance Inflation Factor. Data was entered, cleaned, and analyzed using STATA™ Version 11.0 Statistical Package.

Over half (57.2 %) of the mothers were between 25–35 year. Regarding education, 275 (23.9 %) of the study participants were unable to read and write (no formal education). Of all mothers, 327 (32.2 %) were in the low monthly income category (i.e., <500 Ethiopian birr). In this study, we found a prevalence of 93 (8.1 %) and 69 (6 %) pre-term and post-term births, respectively. Our participants showed high rates of pre-term and post-term births of 21 (22.5 %) and 18 (27 %) in Adwa and Adigrat hospitals respectively. Mothers in the medium and high income categories had lower risk of pre-term and higher risk of post-term births with relative risk ratio (RRR: 0.36, CI: 0.19–0.68; RRR: 1.93, CI:1.01–3.71) respectively. Mothers, who were unmarried, widowed, and divorced [81 (7 %)], were at higher risk of post-term births (RRR: 2.58, CI: 1.25–5.29) (Table 1).

Mothers, who had maternal complications during delivery [221 (19.2 %)], showed higher risk of pre-term delivery (RRR: 1.34, CI: 0.67–2.68) over term births. In relation to maternal BMI, 119 (10.6 %) were in the underweight and 96 (8.6 %) overweight categories; however, overweight mothers were found to be at greater risk for a post-term birth (RRR: 3.55, CI: 1.08–11.5) (Table 2).

On multiple multivariable analyses, using hierarchical modeling of variables, we found BMI of the mother, birth type (i.e., single versus multiple) of neonate, length of neonate, monthly income, and marital status were independently associated with pre-term and post-term births relative to term births.

The risk for pre-term births relative to term births in a multiple birth scenario (i.e., twins and triplets) was 2.86 times greater relative to singleton births (RRR: 2.86, CI: 1.45–5.65); however, post-term birth was not associated with multiple births given the other factors in the model are held constant. Body length of neonates was also associated with pre-term infants, but not with post-term. Neonates, who were 46–56 cm and ≥56 cm at birth, were found to be 88 % (RRR: 0.12, CI: 0.07–0.21) and 99.95 % (RRR: 0.05, CI: 0.01–0.4) at lower risk of pre-term birth relative to term births. Post-term birth was not associated with length of the neonate. Relative to term births, overweight status was significantly associated with post-term births at a 3.88 times higher relative risk (RRR: 3.88, CI: 1.01–15.05). Underweight maternal status relative to normal weight status showed a decrease in relative risk by a factor of 0.47 of delivering pre-term births relative to term births when other factors are held constant (RRR: 0.47, CI: 0.22–0.99). Marital status was also associated with post-term births. Neonates from mothers who were either widowed, divorced, separated, or unmarried relative to married mothers were 2.43 times at higher relative risk for post-term birth than term births (RRR: 2.43,CI:1.02–5.80), but pre-term births were not significantly associated on the final models. Regarding income status, mothers in the medium income category were 74 % less relative risk to experience a pre-term birth than mothers in the low income category (RRR: 0.26, CI: 0.12–0.50). Mothers in the high income category were 2.17 times more at relative risk for delivering a post-term birth relative with mothers in the low income category (RRR:2.17, CI:1.1–4.3) (Table 3).

There were 68 neonatal deaths yielding a neonatal mortality rate of 62.5 per 1000 live births. Of these, 42 (62 %) were pre-term and only one was post-term birth. On bivariate analysis, pre-natal birth (RR: 7.06, CI: 4.5–10.9), history of abortion, history of stillbirth, low birth weight, and being a primiparae were associated with high risk of neonatal mortality (Table 4).

On multivariable regression of the primary independent variables and other secondary variables using stepwise backward elimination, we found pre-term birth as a significant predictor of neonatal mortality, but post-term births were not significantly associated with neonatal mortality. Pre-term births were at a 2.45 times higher risk for neonatal mortality compared to non-pre-term births (RR: 2.45, CI: 1.45–4.04). However, post-term births were not found to be associated with neonatal mortality (RR: 0.45, CI: 0.07–2.96) (Table 5).

No confounding, interaction and multicollinearity were detected in this study.