Study area and period
This study was conducted in Mekelle general and Ayder comprehensive specialized hospitals, Tigray regional state; which is located at 783km from Addis Ababa (the capital city of Ethiopia) from January to March, 2018. According to Central Statistics Agency 2015,Tigray region has an estimated total population of 5,056,000 with an area of 84,722 km2(10,54,55).
The Ethiopia health care has a three-tier health-service delivery system. The primary level consists of primary healthcare units (health posts and health centers) and primary hospitals, secondary level services are provided by general hospitals and tertiary services by specialized hospitals. Tertiary level specialized hospital serving for 3.5 – 5 million people, Secondary level general hospital to serve 1–1.5 million catchment populations and Primary level: rural primary hospital60 000 –100 000 people, health Centre15 000 – 25 000 people and health post3 000 – 5 000 people and urban health center serve for 40 000 people(10).
A prospective cohort study was conducted among a cohort of early neonates who were admitted to neonatal intensive care unit (NICU) from January to March, 2018 in Mekelle general and Ayder comprehensive specialized hospitals. The minimum follow up time for this study was 7 days until the event of interest occurs (neonatal death).
All early neonates admitted to neonatal intensive care unit in Mekelle general and Ayder comprehensive specialized hospitals, Tigray, Ethiopia
All early neonates admitted to neonatal intensive care unit in Mekelle general and Ayder comprehensive specialized hospitals, Tigray, Ethiopia from January to March 2018.
Study participants selected by systematic sampling method from early age neonates in neonatal intensive care units of Mekelle general and Ayder comprehensive specialized hospitals, Tigray, Ethiopia from January to March 2018.
Inclusion and exclusion criteria Inclusion criteria
All early neonates in neonatal intensive care units of Mekelle general and Ayder comprehensive specialized hospitals, Tigray, Ethiopia from January to March2018.
All neonates’ greater ≥7 days of life admitted to neonatal intensive care units
Sample size determination
The general formula for the required number of subjects in a survival study expressed as (56).
WhereE () is the number of events required to be observed in a study, and pE () is the probability of observing an event in a study. δ is log (HR)
Sampling technique and procedure
The study hospitals selected purposely due to the service provision is given with well skilled health providers and better technology that use as referral for the entire region. Two hundred fifty three (253) study participants also selected proportional from Ayder comprehensive specialized hospital (202) and Mekelle general hospitals (51).
The conceptual framework proposed by Mosley and Chen(57),was used in this study to identify and classify the factors that potentially influence ENM.
Early neonatal mortality was categorized into ‘Yes’ if the neonate died ‘No’ if the neonate censored. Censored births were the neonates, who were alive at the end of follow up.
The independent variables in this study were demographic and socio-economic, maternal, delivery factors, neonatal factors, and NICU factors.
Data Collection Procedure
Data collection in NICU was carried out prospectively by interviewing mothers using structured questionnaires and from the medical record charts using the data extraction checklist of the neonates by trained nurses. During the process, the principal investigator was supervised these trained data collectors and assisted by intensive care unit head nurse. The questionnaire was derived from WHO standard verbal autopsy questionnaires(8). The questionnaire was initially developed in English and translated into local language, Tigrigna. One nurse that can speak the local language from each hospital was assign for data collection. The data collectors were collected information by interviewing all mothers whose neonate admitted to NICU of Mekelle general and Ayder comprehensive specialized hospitals from 0–6days of life and the clinical information was extracted from medical record charts and by assessment of neonates. The mother and neonate follow for seven days and whenever the neonate discharged to home data collection was completed using phone call. Data collection included demographic and socio-economic characteristics, maternal, delivery, new born and NICU factors. The outcome variable was death in the first 0–6 days after birth.
Data Quality control and management
Various efforts were made to assure the data quality. Personal supervision especially in the data collection process, training was given to data collectors about the research protocols and also oriented on how to appropriately collect the data. Data quality process starts from the check list to the computer entry. The principal investigator inspects all the performances of the data collectors and measure data. Before the data were entering, it was checked for completeness and consistency by the PI.The process of data generation and entry was under close supervision of the investigator. Data were cleaned, sorting and analysis using STAT software version 12.
Death of neonate was the event of interest and the coding were “1” for death and “0” for censored. The survival time was calculated in days using the time interval between the date of birth and the date of death. Information about the number of deaths before seven complete days after birth and right-censoring (survival beyond seven days).Data was entered; cleaned, recoded and analyzed using STATA Version 12 statistical software. The statistical analysis of survival was based on Cox proportional hazards regression model. Cox-proportional hazard model was used to identify the predictors of early neonatal death. The Cox proportional hazards model is such a model: H (t) = h0 (t) exp (β1x1 + · ·· + βkxk)(58,59).
Descriptive statistics used to describe the frequencies, percentages, rates and to calculate the mean and standard deviation after checking the distribution of the data. Kaplan Meier curves were used to show the pattern of death, estimate probability of survival and to compare the survival curves. Log-rank test was used to look statistical differences among or between the categories of variables to be included in the Cox-proportional hazard model. Cox regression model was used to identify potential predictors of early neonatal deaths.
In the Univarete or Bivariate Cox analysis the covariates; residence, mother educational status, number of newborn, distance from health facility, Birth weight, Gestational age, Apgar score in one minute, sex, admission diagnosis and medication were identified as statistically significant at p_ value ≤ 0.25 level of significance. These significant explanatory variables were included in the Cox-proportional hazard model to identify potential predictors of ENM.
After confounding variables were identified, the variables with p-values of 0.05 or less were retained in the final model analysis.Confounding and effect modification was checked by looking at regression coefficient change if greater than or equal to 15 % and multi-co linearity was checked using variance inflation factor and value of <10 was used as a cutoff point, indicating no co linearity.
Proportional hazard assumption was tested by using covariate specific proportional hazard assumption test. Residuals were checked using graphs and goodness-of-fit test by Cox Snell residuals. Upon the finding HR and 95% CI interpretation for statistically significant predictors were perform.
Birth asphyxia/shortness of breathing: Any respiratory problem, fast breathing, and or difficulty of breathing.
Early Neonatal Mortality: Defined as the probability of dying between (0–6) days or before seven completed days of life.
Premature: Any viable neonate before term (<37weeks of gestation)