Incidence and Predictors of Mortality Among Neonates with Perinatal Asphyxia, Northwest Ethiopia, 2021: An Institution Based Retrospective Cohort Study.


 Background: - Perinatal asphyxia is the third leading cause of neonatal mortality globally, which usually happens within the first week of life. Therefore, evidence-based estimation of neonatal mortality is a cornerstone for tracking progress towards child survival goal and identifying priority areas to improve progress towards eliminating preventable deaths due to perinatal asphyxia. Objective: To assess incidence and predictors of mortality among neonates with perinatal asphyxia admitted to the neonatal intensive care unit of Debre Markos Comprehensive Specialized Hospital, 2021.Methods: An institutional-based retrospective follow-up study was conducted among 402 neonates with perinatal asphyxia (PNA) admitted to the neonatal intensive care unit (NICU) of Debre Markos Comprehensive Specialized Hospital from January 1st, 2018 to the 30th of December, 2020. A simple random sampling technique was used to select the estimated sample. Data were entered using Epi data Version 4.6.0.0 and analyzed using Stata Version 14. The Kaplan–Meier and log-rank test were used to estimate and compare the survival time. Both the bi-variable and multivariable Weibull regression models were fitted to identify predictors of mortality. Finally, the Hazard ratio with a 95%CI was computed, and variables with p-values <0.05 were considered as statistically significant predictors of mortality. Results: A total of 125 (31.09%) neonates died during the follow-up period. The overall incidence rate of mortality was found to be 53.49 per 1000 neonate-days of observations (95%CI: 44.89-63.74). Neonatal sepsis (AHR=2.13;95%CI: 1.38-3.27), preterm birth (AHR= 3.42 95%CI; 2.13- 5.48), Hypoxic Ischemic Encephalopathy stage II (AHR=6.65 95%CI: 2.57-17.26), and III (AHR=16.8 95%CI; 6.28- 44.9), Antepartum hemorrhage (AHR=2;95%CI: 1.13-3.92), the induced onset of labor (AHR=3.90;95%CI; 1.83-8.27), and post-partum hemorrhage (AHR=2.12;95%CI: 1.32-3.38) were significant predictors of mortality among neonates with perinatal asphyxia.Conclusion: The study found that the overall incidence rate of mortality among neonates with PNA remains high. Neonatal sepsis, Hypoxic-ischemic encephalopathy stage II and III, preterm birth, antepartum hemorrhage, postpartum hemorrhage and induced onset of labor were independent predictors of mortality. Therefore, early anticipating high-risk pregnancies and newborns with the respective intervention could reduce neonatal due to perinatal asphyxia.

Globally, perinatal asphyxia is the major cause of neonatal mortality and morbidity. According to 2014, WHO report, between 4 and 9 million newborns develop birth asphyxia each year. Among these, around 1.2 million newborns died during the neonatal period. Moreover, at least a similar number of neonates develop severe consequences, such as developmental delay, epilepsy, cerebral palsy, and 42 million disability-adjusted life years (6-8). Indeed, it is a responsible factor for 24% of all neonatal deaths and 11% of deaths of children under 5 years of age (4,9). Approximately, three-fourths (75%) of such deaths happen on the rst day and less than 2% after 72 hours of birth (10).
The impact of perinatal asphyxia is not restricted at neonatal age but also results in the long-term and short-term neurodevelopmental sequelae, counting cognitive and engine incapacities which are mostly untreatable (11,12). Nearly a quarter of newborns who survived with perinatal asphyxia faced neurological disorders, such as cerebral palsy, certain neurodevelopment learning disabilities (13), and post-traumatic stress disorders (14). Moreover, medical-related costs for the treatment of perinatal asphyxia were high, spent 9.1% of their annual income on acute care (15).
Sustainable development goal (SDG) 3 sets a target of reducing neonatal mortality to 12 deaths or less per 1000 live births and to end all preventable neonatal deaths including PNA (42)(43)(44). Despite this, neonatal mortality increases from time to time in Ethiopia since 2016 (24,45). Moreover, limited studies were conducted in the study area to emphasize mortality and its predictor among newborns with perinatal asphyxia. Therefore, estimating the incidence and identifying predictors of mortality are cornerstone for tracking progress towards child survival goal and identifying priority areas to improve progress towards eliminating preventable deaths due to perinatal asphyxia. Improvement of neonatal survival is a good indicator of quality care. Hence, it is important for the health care providers and health care commissioners to early detection, diagnosis, management and have follow-up care and have also practical vital value for patients, health care providers, researchers and policy-makers in the Ministry of Health.
Hence, this study aimed to determine mortality and its predictors among neonates with perinatal asphyxia admitted in the neonatal intensive care unit of Debre Markos Comprehensive Specialized Hospital.

Study design, period and population
A retrospective follow-up study was carried out from April to May, 2021 at Debre Markos compressive specialized hospital which serves more than 5 million people in its catchment area. Currently, about 100 health centers and 9 district hospitals are available in the catchment area of the referral hospital. The hospital has a total of 489(282 male, 207 female) clinical staffs. Furthermore, it provides neonatal intensive care services for critically ill neonates and those who need neonatal care. The NICU has 27 nurses, one pediatrician, and two general practitioners. In addition, the neonatal intensive unit has 28 neonatal beds and 27 maternal beds(46). The source population was all neonates with perinatal asphyxia admitted to the NICU of Debre Markos comprehensive specialized hospital. The study population included randomly selected eligible neonates with perinatal asphyxia admitted to NICU of Debre Markos comprehensive specialized hospital from January 1, 2018, to December 30, 2020.

Inclusion and exclusion criteria
All neonates with perinatal asphyxia admitted to the NICU of Debre Markos comprehensive specialized hospital from January 1, 2018, to December 30, 2020, who had registration were included. While records of children whose admission date and discharge date were not recorded, missing of outcome variable and neonates with major congenital malformation were excluded in the study.

Sample size and sampling technique
Sample size was computed using single proportion formula with the following assumptions: death rate= 38.7% (p=0.387 and q=0.613) (35), 95% con dence level, 5% margin of error. A total of 789 neonates were admitted to the hospital from January 1/2018 to December 30/ 2020. A sample of 402 neonates with asphyxia were selected from 789 neonates using random number computer generated method, the subsequent unique chart numbers from the registration le were extracted.
Data collection procedure A data extraction format was developed from standardized management protocol for neonates and reviewing applicable articles to assemble the required individual information. The data extraction format consists of socio-demographic characteristics of the mother and the newborn (age, religion, residence, sex of the newborn, and age at admission), antepartum related variables (ANC follow-up, pregnancy induced hypertension, Antepartum hemorrhage, Anemia during pregnancy, gestational diabetic mellitus, Parity, Gestational Hypertension, diabetic mellites, tetanus vaccination, pregnancy screening test (Rh factor, syphilis screening test, HIV, and Hepatitis), Intra-partum related variables (mode of delivery, place of delivery, premature rapture of membrane, Malpresentation, postpartum hemorrhage, prolonged labor, obstructed labor, types of pregnancy, and onset of labor) and neonatal related variables (gestational age, birth weight, Meconium aspiration syndrome, anemia, clinically diagnosis neonatal sepsis, Hypoxic ischemic encephalopathy, birth injury, hypothermia, respiratory distress syndrome, Jaundice, phototherapy, and Random blood sugar). Three experienced nurses were recruited for data collection and supervision, and two days training was given for them to standardize and agree on the way to review medical records. The data extraction tool was checked for completeness and consistency using 5% preliminary reviewed randomly selected charts to maintain data quality assurance.
Data processing and analysis Data were entered into Epi data Version 4.6 and exported to Stata version 14 statistical software for further analysis. Descriptive statistics of numeric variables presented in medians with interquartile range (IQR), categorical variables were presented using frequency tables and percentages. Multicollinearity was checked between independent variables. The incidence rate of mortality was also calculated for the entire follow-up by dividing the total number of new cases of PNA to the total person-days of follow-up. The Kaplan-Meier failure curve was used to estimate survival time. The Log-rank test was used to compare Survival experiences between independent groups. Proportional hazard assumption was checked by graphically using the plot of log [-log (survival probability) versus the log of survival time, by interacting each covariate with time. The Schoenfeld residual global test for the proportionality assumption was checked, and (p-value =0.7554) nding was used as a suggestive of satisfying assumption. The loglikelihood and Akaike Information Criteria (AIC) were applied to select the best tted model, and a model with minimum AIC value was considered as the best tted model. As a result Weibull regression model was selected with AIC=586. 39. The goodness of model tness was also checked using the Cox-Snell residual test. Variables with P < 0.2 in the bi-variable analyses were potential candidate variables for the multivariable analysis, and stepwise forward variable selection was done so as to identify eligible variables using Weibull regression. Variables having p-value ≤0.05 in the multivariable model were considered to be signi cantly associated with the outcome. Finally, the crude and adjusted hazard ratio (HR) with 95% CI was calculated to determine statistical signi cance level.

Operational de nitions
Perinatal asphyxia:-is considered when the 5th minute APGAR score is <7 (21,47). Event(death):-Neonate died in the hospital and death summary written on a chart due to PNA (48).
Censored newborns with PNA who did not develop the outcome of interest (death) until the end of follow-up period or lost to follow-up, recovery from illness, discharged against medical advice or transfer out to other health institutions without knowing the outcome

Survival time
The time in days from admission to the development of outcome variable (Death) within 28 days followup time.
Defaulted-PNA cases that are signed (parents on behalf of their child) against treatment to leave treatment before cure.

Socio-demographic and admission characteristics
A total of 402 newborns were enrolled into the study. Nearly two-thirds, 260(64.68 %) of neonates were males. About 340 (84.58%) of neonates were admitted within the rst 24 hours immediately after birth. The median age of mothers was 25 (IQR: 23, 30) years old, and 37.81% were belonged to the age group of 20-24 years old. Majority, 348 (86.57%) of the respondents were Orthodox Christianity followers, and nearly two-thirds, 252(62.69) of them were rural resident (Table 1). Antepartum characteristics of the mother Nearly two-thirds, 262(65.17%) of the participants were primipara and almost all, 387(96.27%) were attending ANC follow-up. Among these who attend ANC nearly two third 253 (65.37%) had four and above ANC visit ( Table 2). Intrapartum obstetric factors of the mother As shown on the (    The test of equality for survival distribution for different categories of variables were performed with Kaplan-Meier and the log-rank test. In general, the pattern that one group survivorship function lying above another group showed that the upper curve had a better probability of survival compared to the lower curve. On the other hand, the probability of death was high among the lower groups described by Kaplan-Meier survival curve. Furthermore, the log-rank test con rmed the observed difference seen on the KM graph statistical difference or not. In the current study, newborns who faced HIE stage III has lower survival time compared with stage II and I, and newborns who developed HIE stage II were lower survival time compared to stage I. Moreover, neonates without sepsis have favorable survival probability than those neonates with sepsis. This study also revealed that neonates born from mothers who hadn't PPH at admission had better probability of survival compared with those mothers faced PPH.. Furthermore, neonates born from mothers who had antepartum hemorrhage (APH) had better probability of survival compared with newborns born from mothers who did not face APH among neonates admitted with PNA. Thesedifferences were statistically signi cant with p value < 0.001 in log rank test (Figure 3)

Model comparison
After the proportional hazard assumption was checked, both semi-parametric and parametric proportional hazard models were tted to estimate the survival time to death and identify its predictors among neonates with perinatal asphyxia. As shown in gure 3 model Comparession was checked graphically.
Besides, Model Comparession was checked statistically by using information criteria (AIC, BIC) and loglikelihood to select the most parsimonious models for the data set. Based on this, the Weibull regression with the AIC= 586.39 model was best tted model ( Table 5).
This nding is higher than a study conducted in Brazil 0.95-1.38/1000 live birth (29), 0.65-0.81/1000 live birth in 2017 (16), and 9.7/1000 live birth in Southern Nepal (32). This marked difference might be attributed to a number of factors such as source population of the study participants was live birth, while the currents study was PNA, another possible justifying may be due to a study area was population based. On the other hand, the current study was intuitional based.
On the other hand, this nding was lower than the studies conducted at India 40.6(21) and Nigeria 38.7% (35). This may be due to the fact that other studies included only outborn newborns; while in this study participants were both inborn and outborn newborns. Therefore, including outborn overestimated death rate. Moreover, The discrepancy might be variation in different studies was due to different operational de nitions for birth asphyxia and adopted by different researchers (30,33,34,39,50). Moreover, this high death emanated from the poor neonatal resuscitation skills, the lack of resuscitative equipment, lack of skilled manpower in neonatal resuscitation at birth and delay in transporting the asphyxiated newborn to a higher health facilities (51,52).
Neonatal sepsis at the time of admission signi cantly increased the hazard of death among neonates with PNA. This nding was supported by the studies conducted in developed and developing countries (32,38,40,41,50). The possible reason might be associated with neonatal sepsis causes hypothermia/fever, poor feeding, hypotension respiratory distress, shock, multi-organ failure, and meningitis. Besides, poor quality of care and treatment delay also the major gaps of developing counties including Ethiopia. Moreover, the unnecessary overuse of antibiotics can increase the chances of severe candidiasis and multi-drug resistant organisms (53)(54)(55)(56). These collectively resulted to have high mortality due to sepsis among neonates with PNA. This study implied that there is a clear gap in the management and prevention of sepsis. This nding emphasizes the need to improve the quality of care in health facilities, in particular, we strongly believe that achieving a high-quality intrapartum and postnatal care is required to improve neonatal health.
Preterm neonates with PNA increased the hazard of death compared to full-term births. This was in accordance with the studies conducted in Nigeria and South Asian (30,32,35). Higher risk of mortality among premature neonates could be explained by different factors, like newborns with GA less than 37 weeks (preterm) were higher risk to develop complications, such as, infections, hypothermia, and hypoglycemia due to limited number of immunoglobulins at birth. Another possible explanation may be due to lack of feasible, cost-effective care, such as breastfeeding support, basic care for infections and breathing di culties were practiced in developing counties like Ethiopia (57,58). Given these factors are the most common causes of morbidity and mortality in neonates with PNA (59)(60)(61). This implies that anticipating high-risk newborn babies and taking combined approaches to reduce the death of such physiologically and anatomically vulnerable neonates are needed (62, 63) In the current study, hypoxic ischemic encephalopathy (HIE) stage II and III were strong (leading) independent predictors of death, that stage II and III increase the hazard of death by seven times and 17fold respectively compared with stage I by holding another variables constant. This nding is supported by the studies conducted in Nigeria (22,31,34,39), South Africa (33) and India (20,21,23). The possible reason might be once HIE stage II and III occur, they result in poor feeding, seizure, brain damage, and multiorgan dysfunction. In addition, due to lack of feasible, cost-effective care like therapeutic hypothermia, supportive management of seizure and adequate resuscitation in developing countries including Ethiopia (51,64). Besides, the probability that infants with the greatest hypoxic-ischemic brain damage will bene t least by any speci c intervention (65, 66).
The collectively increased death among neonates with PNA who faced HIE Stage II and III during admission. This study implies that more attention should be focused on the early assessment of high-risk mothers and newborns and timely referral to the tertiary care center. Moreover, adequate resuscitation, therapeutic hypothermia, supportive management of seizure and uid balance is essential in ensuring optimal outcomes (67).
In the current study, induced onset of labor increases the hazard of death by nearly four times compared with spontaneous onset by adjusting effects of other variables. This may be associated with induced labor was indicted for high-risk mothers like (Preeclampsia, eclampsia, gestational hypertension…), intrauterine growth restriction, and post -term pregnancy (68-71). As well, induction increases the greater risk for instrumental birth, postpartum hemorrhage, affects the natural process of pregnancy and labour and increases hospitalization for maternal as well as newborn. These collectively increase the risk of death among newborns with PNA (72)(73)(74)(75).
Antepartum hemorrhage (APH) was increasing the hazard of death among neonates with PNA. This may be associated with APH increases premature delivery, low birth weight, and fetal growth retardation (76).
Moreover, APH leads neonatal anemia and decreases placental perfusion as well as hypoxia then which leads to bradycardia, and multiorgan failure (76-79).
Postpartum hemorrhage (PPH) was also independent predictor of neonatal mortality among neonates admitted with PNA, which increases the hazard of death by two times compared with mother who were not faced for PPH during delivery. This may be associated with PPH management delays the time for essential newborn care as well as neonatal resuscitations. Moreover, postpartum hemorrhage is the leading cause of maternal mortality (80, 81), which also resulted neonatal death. since the survival of the mother was vital for their child survival (82, 83). This study implies that regular antenatal care, early detection of high-risk mothers, allowing them to be timely admitted for schedule, availability of blood banks, and early referral to a higher center to reduce APH and PPH related neonatal death (79,84,85).

Limitation
As the study was based on secondary data, the main limitation could be incomplete medical records. In this regard, some important predictors, such as maternal nutritional status, mothers' educational level, birth interval, and monthly household income were not available from medical recorded data. In addition, the exact causes of death were not recorded. Thus, we assumed that all causes of the deaths were related to admission diagnosis. Furthermore, as the study was conducted at hospital, neonatal mortality might be underestimated because neonates delivered and died at home could be underreported. On the other hand, since the study included sick neonates, the mortality could be overestimated

Conclusion
The study found that the overall incidence rate of mortality among neonates with PNA remains high.
Neonatal sepsis, HIE II and III, prematurity, postpartum hemorrhage, antepartum hemorrhage, and induced onset of labor were signi cant predictors of mortality. The study was carried out after getting approval from the university of Gondar institutional review board (IRB). Then, data were collected after getting consent from the hospital manager. This study didn't in ict or exposes children to unnecessary risk as a result of reviewing their medical records. To maintain con dentiality, each and every one collected data were coded and locked in a separate room prior to enter into the computer. Following entered into the computer all data were protected by password. Names and unique medical registered numbers (MRN) were not incorporated in the data collection set-up.

Consent for publication
Not applicable Availability of data and material Additional le: Data abstraction tool Data set: The data sets used and/or analyzed during the current study are available from the corresponding author on reasonable request.