Incidence of respiratory distress and its predictors among neonates admitted to the neonatal intensive care unit, Black Lion Specialized Hospital, Addis Ababa, Ethiopia

Background Although respiratory distress is one of the major causes of neonatal morbidity and mortality throughout the globe, it is a particularly serious concern for nations like Ethiopia that have significant resource limitations. Additionally, few studies have looked at neonatal respiratory distress and its predictors in developing countries, and thus we sought to investigate this issue in neonates who were admitted to the Neonatal Intensive Care Unit at Black Lion Specialized Hospital, Ethiopia. Methods An institution-based retrospective follow-up study was conducted with 571 neonates from January 2013 to March 2018. Data were collected by reviewing patients’ charts using a systematic sampling technique with a pretested checklist. The data was then entered using Epidata 4.2 and analyzed with STATA 14. Median time, Kaplan-Meier survival estimation curves, and log-rank tests were then computed. Bivariable and multivariable Gompertz parametric hazard models were fitted to detect the determinants of respiratory distress. The hazard ratio with a 95% confidence interval was subsequently calculated. Variables with reported p-values < 0.05 were considered statistically significant. Results The proportion of neonates with respiratory distress among those admitted to the Black Lion Specialized Hospital neonatal intensive care unit was 42.9% (95%CI: 39.3–46.1%) The incidence rate was 8.1/100 (95%CI: 7.3, 8.9). Significant predictors of respiratory distress in neonates included being male [Adjusted hazard ratio (HR): 2.4 (95%CI: 1.1, 3.1)], born via caesarean section [AHR: 1.9 (95%CI: 1.6, 2.3)], home delivery [AHR: 2.9 (95%CI: 1.5, 5,2)], maternal diabetes mellitus (AHR: 2.3 (95%CI: 1.4, 3.6)), preterm birth [AHR: 2.9 (95%CI: 1.6, 5.1)], and having an Apgar score of less than 7 [AHR: 3.1 (95%CI: 1.8, 5.0)]. Conclusions In this study, the proportion of respiratory distress (RD) was high. Preterm birth, delivery by caesarean section, Apgar score < 7, sepsis, maternal diabetes mellitus, and home delivery were all significant predictors of this condition. Based on our findings this would likely include encouraging more hospital births, better control of diabetes in pregnancy, improved neonatal resuscitation and addressing ways to decrease the need for frequent caesarean sections. An institution-based retrospective follow-up study was conducted among a cohort of neonates from the previous consecutive five years (from January 2013 to March 2018). The study was conducted in Addis Ababa, a capital city of Ethiopia at NICU of black lion hospital. Addis Ababa has ten sub-cities in which the City lies at an altitude of 7,546 feet (2,300metres). It has twelve governmental and nine nongovernmental hospitals. The NICU of black lion hospital ward is able to accommodate a maximum of 60 patients with average of 20–40 patients’ daily admission. There are on average 5000–6000 annual admissions. The study was conducted from March to April 1, 2018. The neonatal chart number were taken from the HMIS- data base. The total patients admitted to NICU from January 2013-last of March 2018 were 5000. We have found the number of admissions for each year. The samples were proportionally allocated for each year, and with systematics sampling; the study participants of each year were selected as follows. First, numbering the units of each year on the frame from 1 to N (N = total admission of each year), then we determine the sampling interval (K) by dividing the number of units in the population by the desired sample size of each year (n = sample size of each year) which gives 8. Then number between one and 8 at random was selected (2 were selected). This number is called the random start and the first number included in the sample. Then later Selection was conducted every 8th unit after that first number. Our source population was all neonates admitted to the NICU at the Black Lion Specialized Hospital, Ethiopia. All neonates who were admitted to the NICU in the previous five consecutive years (from January 2013 to March 2018) were considered as the study population. All targeted neonates’ medical cards documented in the previous five years from the study period were recruited and those with incomplete cards were excluded. The sample size was determined via the double population proportion formula using Epi-Info Version 7 by assuming a one-to-one ratio of exposed to non-exposed, 95% level of confidence, and a power of 80%. We considered four significantly associated factors to calculate the sample size; the largest sample size was 522. After adding a 10% non-response rate, the total sample size became 604. The neonates’ cards were accessed using the systematic random sampling technique after determining the sampling fraction (k = 6) and the first card was selected by the lottery method. Incidence of RD Socio-demographic factors. Neonatal-related variables included age at admission, gestational age, sex, the weight of the neonate, date of NICU admission and discharge. Maternal-related variables were age and residency. Gynecologic-obstetric related factors. Antenatal care (ANC) follow up, gravidity, parity, mode of delivery, multiple pregnancies, PROM, preeclampsia, abruption placenta, and breastfeeding initiation. Medical disorders in mother. Hypertension, diabetes mellitus (DM), human immune virus/acquired immune deficiency syndrome (HIV)/(AIDS). Neonatal outcome condition. Apgar score, sepsis, jaundice, hypothermia, prenatal asphyxia (PNA), hypoglycemia, meningitis, esophageal atresia. Data collection tools. A pretested checklist was used to collect the required data from the neonates’ charts. The checklist was translated to the local language of Amharic and back to English. The consistency of this translation was checked to ensure its accuracy. Data were collected by reviewing the complete patients’ cards from the previous five consecutive years from the study period. RD was confirmed by reviewing neonate medical charts. Data quality control. Data quality was assured by designing proper data abstraction tools. The checklist was evaluated by experienced researchers. The data collection instrument was pretested on 5% of the sample size. Rigorous training was given regarding the data abstraction checklist and data collection process for both data collectors and supervisors. During the data collection time, close supervision and monitoring were carried out by the supervisors and investigator. Double data entry was also done using Epi Data 4.2.0 software. Data processing, analysis, and presentation. Before analysis, data was cleaned, edited, and coded. Any errors identified at this time were corrected after review of the original data using the code numbers that we had assigned during the data collection period. Data were entered using Epi-Data version 4.2.0 and analyzed using STATA 14 statistical software. Incidence density rate (IDR) was calculated for the entire study period. Subsequently, the number of cases of RD within the follow-up period was divided by the total person-time at risk on follow-up and reported per 100-person day. Kaplan-Meier survival curves were used to estimate the mean survival time and the log-rank tests were used to compare survival curves. Proportional hazard assumption was tested both graphically and through the Schoenfeld residual test for all predictors, revealing that the proportional hazard assumption was met. After checking this assumption, by comparing models, a more effective hazard model was selected using the log likelihood ratio (LR) test and the Akaike Information Criterion (AIC). In this parametric approach, the baseline hazard and the vector of its parameters were assessed together with the regression coefficients. The best-fit model was chosen using AIC; selecting those having the smallest AIC. Subsequently, parametric models were completed for neonates to ascertain the possible predictors. Variables having a p-value less than or equal to 0.05 in the bivariate analysis were fitted to the multivariable Gompertz hazard distribution regression model with a 95% confidence interval. A p-value less than 0.05 was considered statistically significant. Ethical consideration. Ethical clearance was obtained from Addis Ababa University, College of health science ethical review board. Letters of cooperation were written to Black Lion Specialized Hospital by the ethical review board members and subsequently permission was obtained from the clinical director and relevant department and unit heads of the hospital. Since the study was conducted by taking appropriate information from medical chart, it will not inflict any harm on the patients. The name or any other identifying information was not be recorded on the checklist and all information that was taken from the chart was kept strictly confidential and in a safe place. Following these approvals, access to the medical charts was provided and we did our utmost to maintain participant confidentiality by storing in a file cabinet and kept in a key and locked system with computer pass ward. The presence of two or more of the following signs: an abnormal respiratory rate (tachypnea > 60 breaths/min, bradypnea < 30 breaths/minute, respiratory pauses, or apnea) or signs of labored breathing (expiratory grunting, nasal flaring, intercostal recessions, xyphoid recessions), with or without cyanosis. presence of two or more of the following signs: an abnormal respiratory rate, expiratory grunting, nasal flaring, chest wall recessions, and cyanosis as per patient chart information.