Neonatal mortality refers to the death of a live-born baby within the first 28 days (1) and measures the quality of maternal and child healthcare (2). Neonatal deaths account for 47% of the global under-five mortality (3). Among these deaths, 36% occur during the first 24 h (2, 4), 37% occur between days 1 and 7, and 27% occur between days 7 and 27 (2). In low- and middle-income countries, neonatal deaths account for 53.1% of all under5 deaths (5). From 1990 to 2019, neonatal deaths worldwide decreased from 5 million to 2.4 million (3), resulting in a decline in the mortality rate from 37 deaths per 1,000 live births in 1990 to 18 deaths per 1,000 live births in 2021 (6).

In East Africa, the Neonatal mortality rate (NMR)increased from 33.6 per 1,000 live births in 2017 (7) to 51.32 per 1,000 live births in 2021 (8). NMRs also vary across countries: Kenya reported a rate of 19.6 per 1,000 live births in 2018 (9), Eritrea reported a rate of 16.5 (10), and Ethiopia reported 26.84 per 1,000 live births (11). In Ethiopia, neonatal mortality is a major public health issue (12). The Amhara region has the highest rate of 46 per 1,000 live births, which is higher than that reported in the other areas (12). There is also a geographical clustering of neonatal deaths in Ethiopia, with Amhara being one of the most affected areas (7, 13), emphasizing the need to address NMRs in this region.

Neonatal mortality in Ethiopia improved from 39 deaths per 1,000 live births in 2005–29 deaths in 2016 (14) but slightly increased to 33 deaths in 2019 (12), which is higher than the global average of 18 deaths per 1,000 live births (4). This suggests inadequate progress toward achieving Sustainable Development Goals (SDG) 3.3, which aims to reduce NMRs to less than 12 by 2030. The identified risk factors were home birth, rural residency, gestational age, sepsis, perinatal asphyxia (PNA), longer birth intervals, and respiratory distress (7, 15). Additionally, mothers of advanced age and primipara mothers are at increased risk of neonatal mortality (11).

Numerous efforts are underway to improve neonatal survival. The SDGs aim to reduce neonatal mortality to 12 per 1,000 live births by 2030 (4). Additionally, the Every Newborn Action Plan (ENAP) focuses on preventing stillbirths and newborn deaths (16). Ethiopia has committed to integrating life-saving measures into the national strategy outlined in the Health Sector Transform Plan (HSTP), to achieve an NMR of 10 per 1,000 live births by 2020 (17).

Despite efforts, neonatal mortality remains high in Ethiopia, with a current rate of 33 deaths per 1,000 live births (14). To reduce NMRs, it is crucial to identify and address the underlying causes of neonatal deaths starting from primary hospitals, which are the first point of care for newborns (18), and provide basic neonatal care (19). Therefore, this study aimed to investigate the survival status and predictors of mortality among neonates admitted to Bichena Primary Hospital in Northwest Ethiopia. These findings support the goals of the Ethiopian Hospital Alliance for Quality (EHAQ) program by generating new evidence to support evidence-based care in neonatal care (20).

The study took place at the Bichena Primary Hospital neonatal intensive care unit located in Bichena Administrative Town in the Amhara Region, Northwest Ethiopia. It is approximately 273 km from Addis Ababa, the capital of Ethiopia, and 225 km from Bahir Dar, the capital of the Amhara Regional State. The hospital has been operational since April 2015, providing healthcare services to a population exceeding 450,000. In addition to providing various inpatient and outpatient services, including a neonatal intensive care unit, the hospital provides to the areas surrounding Enemay Woreda and Debay Telatgen.

The neonatal intensive care unit of Bichena Primary Hospital has a maximum of 9 beds, with a monthly average of 35 neonates being admitted, yielding, on average, 400–500 annual admissions of neonates. During the study period from January 1, 2021, to April 2023, the NICU was equipped with four room heaters, one infant radiant warmer, one fan radiant warmer, one incubator, four oxygen concentrators, and oxygen cylinders. However, Bubble CPAP was not available.

A retrospective cohort study was conducted among neonates admitted to the NICU (Neonatal Intensive Care Unit) at Bichena Primary Hospital, Northwest Ethiopia.

The dependent variable for this study was the time to death of the neonate. The independent variables were sociodemographic factors, including maternal age, place of residence, sex of the neonate, and age of the neonate; maternal and obstetric factors, gravidity, parity, Antenatal care (ANC), mode of delivery, place of delivery, duration of labor, time of initiation of breastfeeding, amniotic fluid status, premature rupture of membrane (PROM), antepartum hemorrhage, eclampsia, type of pregnancy, history of abortion, stillbirth, history of fetal death, anemia, hepatitis, Syphilis and HIV/AIDS; and neonatal factors, including date of admission, weight, Apgar (Appearance, Pulse, Grimace, Activity, Respiration) score, and gestational age. Neonatal medical conditions such as sepsis, perinatal asphyxia (PNA), respiratory distress syndrome (RDS), sepsis, hypoglycemia, and temperature at admission.

Neonatal anemia was defined as a hemoglobin level of less than 13 gr/dl or a hematocrit of less than 39% (22).

Birth asphyxia was diagnosed when the newborn had at least one of the following signs: not breathing or gasping, <30 breaths per minute, or an APGAR score <7, neonatal neurologic sequelae (seizures, coma, and hypotonic), or multiple organ involvement (kidney, lungs, liver, heart, and intestines) (23).

Censored: a neonate who was alive at the end of the study or lost to follow-up, including discharged to home, discharged against medical advice, or transferred to other health institutions without knowing the outcome (24).

Event: death of a neonate after NICU admission as evidenced by physician confirmation.

Follow-up time-From the time of admission until either an event or censorship occurs.

Hepatitis: Diagnosed as positive if the hepatitis antigen test is positive during pregnancy or during neonate admission if not tested previously.

Low birth weight: defined as a birth weight admission weight of less than 2,500 grams.

Preterm: refers to births occurring before 37 completed weeks of gestation (25).

Survival status: outcome of the neonate; either death or censored.

Survival time: Measures the follow-up time from admission to the NICU to the occurrence of an event or censoring (26).

Syphilis: Diagnosed as positive if the VDRL test is positive during pregnancy or during neonate admission if not tested previously.

Time origin: date of admission.

Time scale: days from the admission of a neonate to the occurrence of an outcome.

The data extraction tool was adapted from the NICU registration book (27), which has been used by other studies previously (26, 28–30). The questionnaire was prepared in the English language. Four BSC nurses currently working in the NICU of Bichena Primary Hospital were involved as data collectors, and two MSC nurses who are hospital staff were supervisors. Survival time was obtained by considering the time between the date of admission and the date of discharge (the date of death or censoring).

A pretest was performed on 5% (32) of the sample size in the hospital before the actual data were collected. The necessary modifications and corrections were made to ensure its validity. Before starting the data collection, one-day training was given on the objectives of the study, the contents of the checklist, and issues related to the confidentiality of the neonates’ profiles. The principal investigator and supervisors spot-checked and reviewed the completed checklist to ensure completeness and were examined through a random selection of cards by the principal investigator.

The data were cleaned, coded, and edited to ensure completeness and consistency. The cleaned data were entered into Epi-Data version 3.1 and exported to STATA 14.0 for analysis.

Descriptive statistics, such as frequencies with percentages and means with standard deviations, were used to describe the study population. Tables and figures were also used for data presentation. A variance inflation factor (VIF) >10 and tolerance (T) <0.1 were considered suggestive of the existence of multicollinearity. We found that none of the variables in our model had a VIF greater than 10 or a T less than 0.1, indicating that multicollinearity was not a concern.

A Kaplan–Meier curve was used to estimate the survival status of neonates admitted to the NICU at different times. The log-rank test was used to compare the survival experience of categorical predictor variables.

The assumptions of proportional hazards were verified via statistical tests and graphical tests. The log minus log plots were assessed to confirm whether they were approximately parallel to each other over time, indicating that the proportional hazards assumption was met. The p values of all variables and the overall global test results in the final multivariable Cox regression model were assessed, with values greater than 0.05 indicating that the proportional hazard assumption was fulfilled. Schoenfeld residual proportional hazard test was used to confirm the independence between the residuals and time. A slope of zero between the residuals and time indicated a good model fit. The Cox-Snell residuals test was also used to evaluate the model's goodness of fit.

Factors associated with outcome variable with a p-value less than 0.25 in bivariate Cox regression were analyzed via multivariate Cox regression. The adjusted hazard ratios (AHR) with 95% confidence intervals (CIs) from the multivariate Cox regression were used as a measure of association, and variables with a p-value <0.05 were considered significant predictors.

A total of 638 neonates were followed during the study period. Among the participants, the majority of neonates in the study were male (61.44%), while the percentage of deaths was slightly greater in females (22.76%) than in males (20.66%). Furthermore, a notable 72.1% of neonates were admitted within 24 h of birth. Concerning residence, approximately half (55.96%) of the neonates were from urban areas. Additionally, the majority of mothers (42.32%) were in the age range of 25–34 years. Finally, 61.6% of the newborns were born at term, whereas 38.4% were born preterm (Table 1).

Among the 638 neonates in the study, the majority of mothers (94.51%) received ANC, whereas 5.49% did not. The majority of neonates were born spontaneously (66.77%), followed by cesarean section (12.7%), instrumental delivery (17.71%), and assisted breech delivery (2.82%). The place of delivery varied, with 63.48% of neonates born at the same facility, 29.31% from other facilities, and 7.21% born at home. Regarding the history of abortion, 97.18% of the neonates were born to mothers with no history of abortion. The majority of neonates (88.71%) were born without any obstetric complications, although 11.29% experienced complications (Table 2).

In the present study, among mothers of neonates, 2.51% were HIV positive, whereas 97.49% were not exposed. Hepatitis was present in a small fraction (1.41%) of mothers. Eclampsia was diagnosed in 3.13% of the mothers of the neonates (Table 3).

Among the admitted neonates, the majority (86.05%) were hypothermic during admission. Neonates admitted with suspected neonatal sepsis accounted for 67.55% of the neonates. Almost three-fourths (74.29%) of the neonates were asphyxiated, and 18.03% of them received resuscitation. Birth injury or trauma was observed in 18.97% of the neonates (Table 4).

As shown in the table below, prematurity (34.31%) was the leading cause of neonatal death, followed by birth asphyxia, which accounted for 26.28% of neonatal deaths, whereas anemia was the least common cause of neonatal death (Table 5).

The study included 638 neonates, resulting in 2,046 neonate days. The study period was from January 1, 2021, to April 20, 2023, with a follow-up period of 1–21 days. The neonates’ discharge statuses were as follows: 37.1% (237 neonates) recovered, 21.5% (137 neonates) died, 34.0% (217 neonates) were transferred, and 7.4% (47 neonates) left against medical advice. The main reasons for transfer were prematurity and low birth weight (<1,500 g) requiring specialized NICU care, uncontrolled seizures in asphyxiated neonates, severe jaundice needing intensive phototherapy, respiratory distress (especially RDS) requiring CPAP, and severe sepsis. The total incidence rate of neonatal mortality at the NICU of Bichena Primary Hospital was 66.96 (CI 56.64, 79.17) per 1,000 neonate days.

The estimated cumulative survival probability during the follow-up period was 92.88% (95% CI: 90.46%–94.71%) on the first day, 84.33% (95% CI: 80.83%–87.24%) at the end of the second day, 77.35% (95% CI: 73.08%–81.03%) at the end of the third day, 59.73% (95% CI: 53.27%–65.60%) from days four to seven, and 55.61% (95%: 45.41%–64.66%) from days eight to 28 (Table 6).

The Kaplan–Meier survival estimate revealed that the survival probability of neonates was high on the first day after admission. However, this probability decreased over time. According to the graph, nearly all newborn deaths occurred within the first week of hospitalization. The total neonatal survival rate was 61.65%, with a standard error of 0.0416 (95% CI 0.52, 0.69) during the follow-up period (Figure 1).

Among the numerous predictor variables, log-rank tests revealed statistically significant differences in newborn survival probabilities. A lower birth weight (log-rank test, χ2 = 15.07, p < 0.001; Figure 2), respiratory distress (log-rank test, χ2 = 21.77, p < 0.001; Figure 3), preterm gestational age (log-rank test, χ2 = 24.61, p < 0.001; Figure 4), hypothermia (log-rank test, χ2 = 11.70, p < 0.001; Figure 5), and a history of abortion in a previous pregnancy (log-rank test, χ2 = 19.25, p < 0.001; Figure 6) were associated with a significantly lower probability of survival.

In this study, bivariate Cox regression analysis was initially conducted with several variables, including hypothermia, birth injury/trauma, perinatal asphyxia, gestational age, history of abortion, birth weight, gravidity, premature ROM, prolonged ROM, suspected sepsis, amniotic fluid status, hepatitis exposure, residence, antepartum hemorrhage, and hypoglycemia. Variables showing a significance level of p < 0.25 in the bivariate analysis were further considered for the multivariable Cox regression model. The backward elimination method was employed for variable selection, with biological plausibility and insights from previous studies guiding the process. The final adjusted hazard ratios (AHRs) with 95% confidence intervals (CIs) were determined, and variables with a p-value < 0.05 were identified as significant predictors of neonatal mortality.

In this study, neonates with hypothermia had a 2.59-fold greater risk of death than those without hypothermia (AHR: 2.59, 95% CI: 1.17–5.70). Similarly, the presence of birth injury/trauma and perinatal asphyxia increased the risk of death by 1.82 (AHR: 1.82, 95% CI: 1.2, 2.76) and 1.65 times (AHR: 1.65, 95% CI: 1.11–2.45), respectively. Preterm neonates were at a 2.29-fold greater risk of death than term neonates (AHR: 2.29, 95% CI: 1.55, 3.38). Neonates from mothers with a history of abortion had a 2.13 times greater risk of death than those from mothers without a history of abortion (AHR: 2.13, 95% CI: 1.06, 4.25). Low birth weight increased the risk of death by 1.6 times (AHR: 1.6, 95% CI: 1.09, 2.34). Finally, neonates with hypoglycemia had a 1.71-fold greater risk of death than those without hypoglycemia (AHR: 1.71, 95% CI: 1.06 2.74) (Table 7).