Time to recovery and predictors of neonatal sepsis among neonates admitted to intensive care units in Southern ethiopia: multicenter study Authors: Elias Geneti, Tolossa Dibissa, Tesfu Zewdu, Teklu Gemechu, Ayele Mamo, Mohamed Tessema, Bethelhem Birhanu Published: 2025 ABSTRACT Neonatal sepsis is a serious problem in the first 28 days of life which is a leading cause of morbidity and mortality in developing counties. Neonatal sepsis remains a leading cause of neonatal morbidity and mortality, particularly in Sub-Saharan Africa. There is paucity of multicenter studies in this region. Therefore, this study aimed to assess the burden and predictors of recovery time among neonates with neonatal sepsis admitted to neonatal intensive care units in South Ethiopia. A retrospective follow up study was conducted among 643 charts of neonates with neonatal sepsis who were admitted to NICU in hospitals of Southern Ethiopia using systematic random sampling. Data were inputted into Epi data version 3.1 and exported to STATA 14 for analysis. Cox Proportional Hazard regression model was used. P-value < 0.05 in multivariable analysis was used to declare the result as statistically significant predictors of recovery from neonatal sepsis. The overall median recovery time was 8 days. The incidence rate of recovery from neonatal sepsis was 9.86 per 100 neonate day’s observation. Factors significantly associated with recovery time of neonatal sepsis includes; respiratory distress (AHR: 0.71, 95% CI (0.59, 0.85)), resuscitation at birth (AHR: 0.65, 95% CI (0.49, 0.87)), having fever (AHR: 0.82, 95% CI (0.67, 0.99)) and frequency of antibiotics change (AHR: 2.0, 95% CI (1.57, 2.63)). Time to recovery of neonatal sepsis was moderately acceptable as compared to the previous studies. It was significantly affected by resuscitation at birth, frequent antibiotic change, respiratory distress and fever. Therefore, health care providers must pay more attention for prevention and early intervention of complication.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-29473-6. Introduction Background Neonatal sepsis is a clinical syndrome of bacteraemia characterized by systemic signs and symptoms of infection in the first 28 days of life1,2. Two major categories based on the time of onset: early-onset neonatal sepsis (EONS), which appears within the first seven days of life and most cases appear within 24 h of birth and late onset neonatal sepsis (LONS) occurs after 8 days of infants life3,4. Common clinical manifestations of sepsis on infants are; feeding difficulties, temperature instability, hypoperfusion characterized by pallor and mottled skin, tachycardia or bradycardia, apnea, respiratory distress and grunting2,5. Therefore, it is diagnosed if at least two of the above clinical features and at least two of the following laboratory values is positive such as, complete blood count with differential, blood culture, urine culture, chest radiograph, gram stain, lumbar puncture examination especially for late onset sepsis6. Globally, 15% of neonatal deaths are due to neonatal sepsis and it is a concern for low-and middle-income countries7,8. In developing countries, the rate of neonatal mortality due to sepsis is very high and it ranges from 14.6% to 36%9. Neonatal sepsis is a primary cause of neonatal mortality and an urgent global health concern, especially within low-income and middle-income countries (LMICs), where 99% of global neonatal mortality occurs10. In Africa the case fatality rates of neonatal sepsis was 52%11. A recent estimated global incidence of neonatal sepsis is approximately 2824 per 100,000 live births12, representing a significant increase compared to the 2018 estimate of 2,202 cases per 100,000 live births13. According to the World Health Organization (WHO) report, the global burden of neonatal sepsis was up to 3 million sepsis cases occur per year among neonates, with a global mortality of 400,000–700,000 every year14. In Africa, neonatal sepsis accounts for 28% of neonatal mortality and believed that early detection and treatment of neonatal sepsis (NS) can prevent around 84% of neonatal sepsis-related mortality14. The burden varies from country to country; Cameroon 37.9%15, Ghana 20.2%16, Tanzania 31.4%17, Kenya 23.9%18 and Egypt 8.6%19. Sepsis is the most common cause of neonatal mortality and probably responsible for 30 to 50% of the total neonatal deaths each year in developing countries14. Despite improvements over the past period, Ethiopia’s current neonatal mortality rate is one of the highest in the world and a significant proportion of the under-five mortality in the country20. Ethiopian Demographic and Health Survey (EDHS) 2016 report showed that, the neonatal, infant, and under-5 mortality rates were 29, 48, and 67 deaths per 1,000 live births, respectively. The EDHS 2019 report showed that the neonatal mortality rate was 30 deaths/1000 live births21. In Ethiopia, 40% of under-five mortality occurs during the neonatal period; among this, neonatal sepsis accounts for 30–35% of neonatal deaths7. The magnitude of neonatal sepsis in Ethiopia is still high, which is 48% and varies from region to region; in Gondar 64.4%22, in Mekele 76.8%23, in Shashamane 77.9%24, in Woldia and Dessie Comprehensive Specialized Hospitals 79.4%25, and in Arbaminch 78.3%26, and in Wolaita Sodo 33.8%27. The burden and severity of neonatal sepsis in Ethiopia is high28 and approximately 50% of the neonates in Neonatal Intensive Care Units (NICUs) developed neonatal infection due to long duration of hospital stay(LDOHS)29. Moreover, limited studies have been conducted, especially in the study area and there is a paucity of multicenter studies to generate evidence for action. Therefore, this study aimed to assess the burden and predictors of recovery time among neonates with neonatal sepsis admitted to neonatal intensive care units in public comprehensive specialized hospitals of the South Nation, Nationalities and peoples Regional State. Significance of the study Though Ethiopia has made considerable achievement in the reduction of under-five mortality rate, the neonatal mortality burden has not experienced the same reduction and unacceptably high and the trend in reduction is slower30. To prevent further this neonatal problem, data from different geographical areas of Ethiopia are required. This study will provide valuable information on the recovery time from sepsis and its predictors factors among new born admitted with neonatal sepsis. Understanding indicators of time to recovery in neonate with sepsis can help healthcare professionals and parents to understand how long the new born will be in the hospital. This allows them to mobilize resources for the newborn’s hospital care and prepare for any disruption the admission may cause. It helps health care provider for early initiation of further investigation and treatment. Finally, the results of this study will serve as an input for further studies. Objective of the study General objective To assess time to recovery of neonatal sepsis and its predictors among neonates admitted to the NICUs in public comprehensive specialized hospitals of SNNPRS, Ethiopia, 2023. Specific objective To determine the recovery time of neonatal sepsis among neonates admitted to the NICUs in public comprehensive specialized hospitals of SNNPRS, Ethiopia, 2023. To identify the predictors of the recovery time of neonatal sepsis among neonates admitted to the NICUs in public comprehensive specialized hospitals, SNNR, Ethiopia, 2023. Methods and materials Study area and period This study was carried out from May 1st to May 30th, 2023 among neonates with neonatal sepsis admitted to public comprehensive specialized hospitals in SNNPRS from January 1st, 2021 to April 30, 2023. South Nation Nationality and People`s regional state is one of the administrative regional states in southwestern Ethiopia. It is the most diverse region in the country in terms of culture, language and ethnicity. There are four comprehensive specialized hospitals in the region namely; Wolaita Sodo University comprehensive specialized hospital, located in Sodo town of Wolaita zone, which is 380 km away from Addis Ababa. The NICU ward of this hospital has 34 patient beds for newborn infants. Wachemo University Nigist Eleni Mohammad memorial comprehensive specialized hospital, located in Hadiya zone in SNNPR which is 232 km far from Addis Ababa, Ethiopia. It is one of the frontline hospitals in Ethiopia. The NICU ward of this hospital has 40 patient beds. It also has a very high patient flow (200 neonates per month on average). Werabe comprehensive specialized hospital, found in Silte zone. The town, werabe is located 172 km south of Addis Ababa. The NICU ward of this hospital has 36 patient beds for newborn infants. Another area of this study was, Walkitie University Comprehensive Specialized Hospital, located in Gurage zone, Walkitie town. It is 132 km far away from Addis Ababa. Study design Retrospective follow up study was used. Population Source population The source population comprised of all neonates with sepsis who were admitted for neonatal sepsis management and treated at public comprehensive specialized hospitals in SNNPRS. Study population All neonates admitted for neonatal sepsis management and treated at public comprehensive specialized hospitals of SNNPRS from January 1st, 2021 to April 30, 2023. Study unit Each selected neonate’s chart that fulfills the inclusion criteria. Eligibility criteria Inclusion criteria Neonates admitted to NICUs in Public Comprehensive Specialized Hospitals of SNNPRS with the diagnosis of neonatal sepsis. Exclusion criteria Neonates who died before taking the treatment and incomplete records were excluded from the study. Sample size determination The sample size was determined based on common predictors that have been found to be significantly associated with time-to-recovery from neonatal sepsis among newborns admitted to Neonatal intensive care units as reported in previous studies. From all predictors considered for sample size calculation, the largest sample size (n = 643) was selected. This predictor was taken from the retrospective cohort study conducted in Gondar (Early recognition of illness as a major predictor)31. The Freedman method of sample size determination was used. The total sample size was calculated as follows:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:E=\frac{\left(\text{Z}\frac{{\upalpha\:}}{2}+\text{Z}{\upbeta\:}\right)2}{\:\:\:\:\left(\text{l}\text{n}\:\text{H}\text{R}\right)\:2\:\text{p}(1-\text{p})\:}\:\text{a}\text{n}\text{d}\:P\left(E\right)=\frac{1-\:\left(\:\text{e}\text{x}\text{p}\text{S}1\right(\text{t})\:+\:\text{e}\text{x}\text{p}\:\text{S}2(\text{t}\left)\text{*}\text{H}\text{R}\right)}{2}$$\end{document} The total sample size was calculated by; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:n=\frac{\text{E}}{\text{p}\text{r}\left(\text{E}\right)}$$\end{document} Where: n = total sample size needed. E = the number of events. Pr(E) = the probability that the event of interest (recovery in this context) occurs. Zα/2: value at 95% confidence interval = 1.96. Zβ: the power of the study = 0.84. AHR: Adjusted Hazard Ratio. Sampling procedure All comprehensive specialized hospitals were included in our study. For record reviews, two consecutive years (January 2021 to January 2023) was chosen deliberately because they provide the latest data on the issue under investigation. During the two-year period, the total numbers of neonatal sepsis admissions were nine thousand four hundred seventy one (9,471). By calculating the interval from the sampling frame, the total sample size for each year was proportionally allocated for (WUNEMMCSH, N = 3200), (WSUCSH, N = 2156), (WCSH, N = 2170) and (WUCSH, N = 1945) and the interval was calculated as (k = N/n) while N = 9471 and sample size n = 643. The interval (k = 9471/643 = 14) so the data were collected every 14th charts which was similar for each hospital. The first chart to start the data collection was selected randomly. Then, every 14th medical chart within the two consecutive years period was reviewed using systematic random sampling (Fig. 1). Fig. 1Schematic representation of sampling procedure for determining time to recovery and its predictors among neonates admitted with neonatal sepsis in Southern, Ethiopia, 2023. Schematic representation of sampling procedure for determining time to recovery and its predictors among neonates admitted with neonatal sepsis in Southern, Ethiopia, 2023. Variables Dependent variable Time to recovery of neonatal sepsis. Independent variable Socio demographic factors Maternal age in year, neonatal age in days and neonatal sex. Neonatal related factors Gestational age, comorbidity, Apgar score, History of resuscitation, Birth weight, History of previous hospitalization, Exclusive breast feeding(EBF) initiation within one hour, sepsis type and BCG Vaccine. Maternal related factors Mode of delivery, Place of delivery, number of antenatal care(ANC) visits, twin pregnancy, urinary tract infection(UTI/STD) during pregnancy, Pregnancy-Induced Hypertension (PIH), antepartum hemorrhage, diagnosed chorioamniotis, Duration after the rupture of membranes(ROM), maternal infection history and presence of chronic illness. Clinical presentation/features-related variables Fever, respiratory distress, tachycardia, poor feeding, dehydration, vomiting, hypothermia, meconium aspiration syndrome, chest indrawing, lethargy, jaundice and cyanosis. Medical treatment or care Medications type, Duration of antibiotic treatment, Number of antibiotic switches/change and assisted with bag and mask. Operational definitions Neonate A new born from birth until 28 days old. Neonatal sepsis Sepsis diagnosed and confirmed by physician among neonates. Event Recovery of neonate admitted with neonatal sepsis. Comorbidities In addition to sepsis the neonate had; necrotizing enterocolitis, congenital heart diseases and neurodevelopmental anomalies32. Early-onset neonatal sepsis Sepsis diagnosed and confirmed by physician among neonates age less than seven days. Late-onset neonatal sepsis Sepsis diagnosed and confirmed by physician among neonates age seven days or more. Recovery If a neonate is recovered from the infection after completing the treatment according to physician diagnosis. Defaulter Refers to neonate left (or stops treatment) the treatment unit against medical advice or the treatment. Death A neonate died by NS during the treatment or in the treatment unit. Censored It refers to a neonate defaulted from the treatment, referred, died, or transferred. Time to recovery A time from the admission date by NS to the discharge date while the neonate is recovered. It was measured by subtracting the date of admission from the discharge date (time in days until recovery/discharge). Baseline period Starting from the first day of admission to NICU with a diagnosis of Sepsis. Data collection tool and procedure Data collection tool A data collection tool was adapted from another peer reviewed article conducted in Ethiopia3,31 (Supplementary file). Data abstraction was designed based on study objectives, and contains five parts; checklist related to socio-demographic information, maternal related information, neonatal related and clinical presentation and medical care related information which was collect from medical records. Data collection procedure All available information on patient records was checked in the selected public comprehensive specialized hospitals from January 1st, 2021 to April 30, 2023 was retrieved from the inpatient Health Management Information System (HMIS) registry book. The time from the diagnose of sepsis was the starting point for retrospective follow-up and the end point was the date of recovery, the date of lost to follow-up or censored (improved, referred or against medical treatment) or the date of death. All neonatal sepsis charts admitted to the selected four hospitals from January 1, 2021 to April 30, 2023 was checked from the (HMIS) registration book. The records of all study participants were selected according to the eligibility criteria. Recovery was confirmed by discharge note complemented and registration book. Recovery time was calculated from the date of admission to the date of discharge. Two Masters of Science for supervision and six BSc Nurse for data collection were assigned. One day training was given to data collectors and supervisors regarding significance of the study and ways of data collection process. The supervisors monitored the data collection process. Data quality control To ensure the quality of the data, supervisors and data collectors were trained on how and what information they should collect from the targeted data sources. Prior to data collection, data extraction form was tested on 5% (32 charts) of neonatal charts with neonatal sepsis at Butajira hospital before the data collection to ensure that the data abstraction format was consistent with study requirements. The completeness of the collected data were checked on site daily basis during data collection and gave prompt feedback by the supervisors and the principal investigator. During data management, storage, cleaning and review, all completed data collection forms were checked for completeness and accuracy. The principal investigator examines consistency by selecting cards at random and comparing their similarity. Data processing and analysis Data were entered into Epi data version 3.1. The consistency of data were also checked before analysis to avoid errors during data entry. After that, the data were exported to SPSS version 25 for analysis. Descriptive statistics (frequency and percentages) were used to summarize categorical variables. Before performing Cox-proportional hazard regression, model and goodness-of-fit was checked by Cox Snell residuals and assumptions were also checked by using Schoenfeld residual test. For each independent predictor bivariate cox proportional Hazard regression was performed. Multicollinearity was checked for estimating the correlation between all pairs of variables by variance inflation factor value ≥ 533. Variables with p-value < 0.25 in bivariate analysis was included in multivariable cox proportional hazard regression. And p-value < 0.05 was used to declare the strength of association. Outcome was dichotomized into censored “0” or recover “1”. Kaplan Meier survival curve and log-rank test was used to describe the survival function for each predictor. Results were presented in text, tables, and graphs based on the type of variable. Median recovery time and Incidence density rate (IDR) was calculated. Ethical consideration Ethical clearance and approval for this study was obtained from the research ethical review committee of College of Medicine and Health Sciences, Debre Brehan University (protocol number IRB-148). Also, a permission letter was obtained from each hospital manager before the full scale data collection. An informed consent was waived by the ethics committee of College of Medicine and Health Sciences, Debre Brehan University because the study was conducted through review of medical charts. Patients’ names were not recorded. All procedures were performed in accordance with the relevant guidelines and regulations. Dissemination of the result The findings of this study disseminated to Debre Brehan University Asrat Woldeyes health science campus school of Nursing and Midwifery, department of pediatrics and child health nursing, to all public comprehensive specialized hospitals in SNNPRS health bureau and other organizations working on related area to be used as a baseline for intervention. Results Socio-demographic characteristics of the study participants In this study, 643 neonatal charts were reviewed. More than half 399(62.1%) were males which accounted for 52.7% of recover neonates as compared with female neonates. The mean (± SD) age of the neonates was 179.33 ± 192.67 h. About 64.7% of the neonates had age less than seven days. Concerning to mothers’ socio demographic characteristics, the mean (± SD) age of mothers was 27.07 ± 0.192 years range from 16 to 39 years (Table 1). Table 1Socio-demographic characteristics of neonates admitted with neonatal sepsis at public comprehensive specialized hospitals of SNNPRS, Ethiopia, 2023 (n = 643).Variables CategoryStatusRecovered censoredResidenceUrban339(52.7%)56(8.7%)Rural196(30.5%)52(8.1%)Sex of the neonateMale336(52.3%)63(9.8%)Female199(30.9%)45(7.0%)Place of delivery/birthHome18(2.8%)5(0.8%)Health institution517(80.4%)103(16.0%)Age of Mother< 2057(8.9%)17(2.6%)20–34440(68.4%)76(11.8%)> 3438(5.9%)15(2.3%) Table 1Socio-demographic characteristics of neonates admitted with neonatal sepsis at public comprehensive specialized hospitals of SNNPRS, Ethiopia, 2023 (n = 643).Variables CategoryStatusRecovered censoredResidenceUrban339(52.7%)56(8.7%)Rural196(30.5%)52(8.1%)Sex of the neonateMale336(52.3%)63(9.8%)Female199(30.9%)45(7.0%)Place of delivery/birthHome18(2.8%)5(0.8%)Health institution517(80.4%)103(16.0%)Age of Mother< 2057(8.9%)17(2.6%)20–34440(68.4%)76(11.8%)> 3438(5.9%)15(2.3%) Socio-demographic characteristics of neonates admitted with neonatal sepsis at public comprehensive specialized hospitals of SNNPRS, Ethiopia, 2023 (n = 643). Maternal related characteristics In considering the number of ANC visits, about 10 (1.6%) of the mothers had no antenatal visits, while, 66(10.3%), 38 (5.8%), 288(44.8%) and 241(37.5%) of them had one, two, three and four and above visits respectively. With regard to mode of delivery, more than half 432 (67.2%) of the mother were give birth by spontaneous vaginal delivery and the remaining cesarean section, forceps delivery and vacuum assisted were accounted for 112(17.4%), 69 (10.7%) and 30(4.7%) respectively. Regarding rupture of membrane, 240 (37.3%) had history of PROM and out of them 115 (47.9%) were for more than 18 h duration (Table 2). Table 2Maternal-related characteristics of mothers having neonates admitted with neonatal sepsis at public comprehensive specialized hospitals of SNNPRS, Ethiopia, 2023. (n = 643)Variables CategoryStatusRecoveredCensoredTwin pregnancyYes40(6.2%)7(1.1%)No495(77.0%)101(15.7%)UTI/STD during pregnancyYes87(13.5%)41(6.4%)No448(69.7%)67(10.4%)Pregnancy-induced hypertensionYes51(7.9%)15(2.3%)No484(75.3%)93(14.5%)Antepartum hemorrhageYes34(5.3%)4(0.6%)No501(77.9%)104(16.2%)Intra-partum feverYes53(8.2%)11(1.7%)No482(75.0%)97(15.1%)Diagnosed ChorioamnionitisYes105(16.3%)19(3.0%)No430(66.9%)89(13.8%)GravidityPrimi290(45.1%)56(8.7%)Multi224(34.8%)45(7%)Grandmulti gravida21(3.3%)7(1.1%)Parity Primi para259(57.6%)53(11.8%)Multi para 108(24%)23(5.3%)Grandmulti para6(1.3%)1(0.2%) Table 2Maternal-related characteristics of mothers having neonates admitted with neonatal sepsis at public comprehensive specialized hospitals of SNNPRS, Ethiopia, 2023. (n = 643)Variables CategoryStatusRecoveredCensoredTwin pregnancyYes40(6.2%)7(1.1%)No495(77.0%)101(15.7%)UTI/STD during pregnancyYes87(13.5%)41(6.4%)No448(69.7%)67(10.4%)Pregnancy-induced hypertensionYes51(7.9%)15(2.3%)No484(75.3%)93(14.5%)Antepartum hemorrhageYes34(5.3%)4(0.6%)No501(77.9%)104(16.2%)Intra-partum feverYes53(8.2%)11(1.7%)No482(75.0%)97(15.1%)Diagnosed ChorioamnionitisYes105(16.3%)19(3.0%)No430(66.9%)89(13.8%)GravidityPrimi290(45.1%)56(8.7%)Multi224(34.8%)45(7%)Grandmulti gravida21(3.3%)7(1.1%)Parity Primi para259(57.6%)53(11.8%)Multi para 108(24%)23(5.3%)Grandmulti para6(1.3%)1(0.2%) Maternal-related characteristics of mothers having neonates admitted with neonatal sepsis at public comprehensive specialized hospitals of SNNPRS, Ethiopia, 2023. (n = 643) Neonatal related characteristics From 643 neonates admitted with neonatal sepsis, the mean (± SD) weight of neonates at admission was 2906.43 ± 616.611 g and more than half, 385 (63.79%) of recovery was seen among neonates with birth weight greater than 2,500 g. Regarding gestational age, the mean (± SD) gestational age was 37.26 ± 2.138 weeks ranging from 31 to 43 weeks. Of all neonate who were Resuscitated at birth 59(9.2%) of them were recovered while 18(2.8%) were censored (Table 3). Table 3Neonatal-related characteristics admitted with neonatal sepsis at public comprehensive specialized hospitals of SNNPRS, Ethiopia, 2023. (n = 643)Variables CategoryCategory of statusRecovered CensoredEBF initiated within one hourYes167(26.0%)38(5.9%)No368(57.2%70(10.9%)BCG vaccination administrationYes229(35.6%55(8.6%)No306(47.6%)53(8.2%)Sepsis typeClinical sepsis526(81.8%)106(16.5%)Culture positive9(1.4%)2(0.3%)Time of the infection onsetEONS354(55.1%)62(9.6%)LONS181(28.1%)46(7.2%)Birth weight< 2,500150(23.3%)35(5.4%)≥ 2,500385(59.9%)73(11.4%)Gestational age< 37292(45.4%)71(11.0%)37–42242(37.6%)37(5.8%)> 421(0.2%)0(0.0%)Temperature< 36.5308(47.9%)56(8.7%)36.5–37.599(15.4%)15(2.3%)> 37.5128(19.9%)37(5.8%)Frist min Apgar score<7511(79.5%)104(16.2%)≥724(3.7%)4(0.6%)Fifth min Apgar score<7439(68.3%)92(14.3%)≥796(14.9%)16(2.5%) Table 3Neonatal-related characteristics admitted with neonatal sepsis at public comprehensive specialized hospitals of SNNPRS, Ethiopia, 2023. (n = 643)Variables CategoryCategory of statusRecovered CensoredEBF initiated within one hourYes167(26.0%)38(5.9%)No368(57.2%70(10.9%)BCG vaccination administrationYes229(35.6%55(8.6%)No306(47.6%)53(8.2%)Sepsis typeClinical sepsis526(81.8%)106(16.5%)Culture positive9(1.4%)2(0.3%)Time of the infection onsetEONS354(55.1%)62(9.6%)LONS181(28.1%)46(7.2%)Birth weight< 2,500150(23.3%)35(5.4%)≥ 2,500385(59.9%)73(11.4%)Gestational age< 37292(45.4%)71(11.0%)37–42242(37.6%)37(5.8%)> 421(0.2%)0(0.0%)Temperature< 36.5308(47.9%)56(8.7%)36.5–37.599(15.4%)15(2.3%)> 37.5128(19.9%)37(5.8%)Frist min Apgar score<7511(79.5%)104(16.2%)≥724(3.7%)4(0.6%)Fifth min Apgar score<7439(68.3%)92(14.3%)≥796(14.9%)16(2.5%) Neonatal-related characteristics admitted with neonatal sepsis at public comprehensive specialized hospitals of SNNPRS, Ethiopia, 2023. (n = 643) Medical treatment/care related factors Majority of the neonate 543(84.4%) of newborns were treated with Ampicillin + Gentamicin and the remaining 63(9.8%) were treated with Vancomycin + Ceftazidime and 36(5.6%) of them were treated with Ampicillin + Ceftriaxone. Of all neonate who were treated with those antibiotics, for newborns 107(16.6%) the drug was changed once and above. At the time of follow up, one third 187(29.1%) of newborn were assisted with bag and mask/requirement of invasive Ventilation (Table 4). Table 4Treatment/care-related factors of neonate admitted with neonatal sepsis at public comprehensive specialized hospitals of SNNPRS, Ethiopia, 2023. (n = 643)Variables CategoryStatusRecovered CensoredAssisted with bag and mask/requirement of invasive VentilationYes157(24.4%)30(4.7%)No378(58.8%)78(12.1%)Antibiotics givenAmpicilline + Gentamycin455(70.8%)88(13.7%)Vancomycine + Ceftazidime24(3.7%)12(1.9%)Ampicilline + ceftriaxone55(8.6%)8(1.2%)Vancomucine + Cefotaxime1(0.2%)--Number of antibiotic switches/changeNo406(63.2%)60(9.3%)Once & above129(20%)48(7.5%) Table 4Treatment/care-related factors of neonate admitted with neonatal sepsis at public comprehensive specialized hospitals of SNNPRS, Ethiopia, 2023. (n = 643)Variables CategoryStatusRecovered CensoredAssisted with bag and mask/requirement of invasive VentilationYes157(24.4%)30(4.7%)No378(58.8%)78(12.1%)Antibiotics givenAmpicilline + Gentamycin455(70.8%)88(13.7%)Vancomycine + Ceftazidime24(3.7%)12(1.9%)Ampicilline + ceftriaxone55(8.6%)8(1.2%)Vancomucine + Cefotaxime1(0.2%)--Number of antibiotic switches/changeNo406(63.2%)60(9.3%)Once & above129(20%)48(7.5%) Treatment/care-related factors of neonate admitted with neonatal sepsis at public comprehensive specialized hospitals of SNNPRS, Ethiopia, 2023. (n = 643) Clinical features related factors In this study about one-fourth 152(23.6%) of newborns were having hypothermia among those 121(18.8%) of them were recovered and the remaining 31(4.8%) of them were censored and among neonates with comorbidity 62(9.6%) were recovered and 21(3.3%) were censored (Table 5). Table 5Clinical features/presentation of neonate-related factors among neonate admitted with neonatal sepsis at public comprehensive specialized hospitals of SNNPRS, Ethiopia, 2023.Variables CategoryStatusRecovered CensoredFeverYes163(25.3%)44(6.8%)No372(57.9%)64(10.0%)TachycardiaYes117(18.2%)27(4.2%)No418(65.0%)81(12.6%)Poor feedingYes344(53.5%)71(11.0%)No191(29.7%)37(5.8%0DehydrationYes29(4.5%)11(1.7%)No506(78.7%)97(15.1%)VomitingYes104(16.2%)23(3.6%)No431(67.0%)85(13.2%)LethargyYes72(11.2%)21(3.3%)No463(72.0%)87(13.5%)Meconium Aspiration SyndromeYes130(20.2%)24(3.7%)No405(63.0%)84(13.1%)Respiratory distressYes281(43.7%)73(11.4%)No254(39.5%)35(5.4%)CyanosisYes95(14.8%)18(2.8%)No440(68.4%)90(14.0%)Severe jaundiceYes30(4.7%)8(1.2%)No505(78.5%)100(15.6%)Chest indrawingYes159(24.7%)38(5.9%)No376(58.5%)70(10.9%) Table 5Clinical features/presentation of neonate-related factors among neonate admitted with neonatal sepsis at public comprehensive specialized hospitals of SNNPRS, Ethiopia, 2023.Variables CategoryStatusRecovered CensoredFeverYes163(25.3%)44(6.8%)No372(57.9%)64(10.0%)TachycardiaYes117(18.2%)27(4.2%)No418(65.0%)81(12.6%)Poor feedingYes344(53.5%)71(11.0%)No191(29.7%)37(5.8%0DehydrationYes29(4.5%)11(1.7%)No506(78.7%)97(15.1%)VomitingYes104(16.2%)23(3.6%)No431(67.0%)85(13.2%)LethargyYes72(11.2%)21(3.3%)No463(72.0%)87(13.5%)Meconium Aspiration SyndromeYes130(20.2%)24(3.7%)No405(63.0%)84(13.1%)Respiratory distressYes281(43.7%)73(11.4%)No254(39.5%)35(5.4%)CyanosisYes95(14.8%)18(2.8%)No440(68.4%)90(14.0%)Severe jaundiceYes30(4.7%)8(1.2%)No505(78.5%)100(15.6%)Chest indrawingYes159(24.7%)38(5.9%)No376(58.5%)70(10.9%) Clinical features/presentation of neonate-related factors among neonate admitted with neonatal sepsis at public comprehensive specialized hospitals of SNNPRS, Ethiopia, 2023. Survival status of neonate admitted neonatal sepsis The overall median recovery time or hospital stay was 8 (IQR: 7, 10) with a minimum and maximum follow-up time of 2 and 21 days respectively. During follow up time, 535(83.2%) neonates with neonatal sepsis were recovered while 108(16.8%) of neonates with neonatal sepsis were censored. From those censored, 33(5.1%) of the neonates were died, 47(7.3%) of the neonates were defaulted/lost to follow up, 26(4%) transferred and 2(0.3%) of the neonates were referred to other hospitals). The total follow-up time is 5,424 with the incidence rate of 9.86 per 100 person/day of observation. Kaplan-Meier survival recovery estimates of neonatal sepsis The overall Kaplan Meier estimates shows the probability of recovery of patients admitted with NS is high on the first day of admission, which relatively decreases as hospital stay increase. The overall median recovery time of admitted neonates with NS in this study was 8(IQR: 7, 10) days (Fig. 2). Fig. 2Overall Kaplan-Meier estimates of study participant that was admitted to Neonatal Intensive Care Unit in Southern, Ethiopia, 2023. Overall Kaplan-Meier estimates of study participant that was admitted to Neonatal Intensive Care Unit in Southern, Ethiopia, 2023. The probability of recovery of neonatal sepsis from day one up to day 5 was about 0.94 CI (0.94–0.97) (Table 6). Table 6A life table showing survival probability to recovery of neonate who was admitted to public comprehensive specialized hospitals in SNNPRS, Ethiopia, 2023.IntervalTotalRecoverCensoredProbability to recovery(95% Conf. Int)0-564321270.96660.94930.97815-10595315490.43290.39270.472410-15231182190.07720.05630.102215-203012120.03860.02250.061320-256510.00350.00010.0298 Table 6A life table showing survival probability to recovery of neonate who was admitted to public comprehensive specialized hospitals in SNNPRS, Ethiopia, 2023.IntervalTotalRecoverCensoredProbability to recovery(95% Conf. Int)0-564321270.96660.94930.97815-10595315490.43290.39270.472410-15231182190.07720.05630.102215-203012120.03860.02250.061320-256510.00350.00010.0298 A life table showing survival probability to recovery of neonate who was admitted to public comprehensive specialized hospitals in SNNPRS, Ethiopia, 2023. Survival function and comparison for different categorical variables The Kaplan-Meier estimator survival curve gives the estimate of survival function among different categories to make a comparison. Survival function line lying below another means, groups defined by the lower line curve has better recovery time than other group line curve within the category (Fig. 3). Fig. 3The Kaplan-Meier survival curve compare the survival time of the category of hypothermia for participant admitted with NS at public Comprehensive specialized hospitals in SNNPRS, Ethiopia, 2023. The Kaplan-Meier survival curve compare the survival time of the category of hypothermia for participant admitted with NS at public Comprehensive specialized hospitals in SNNPRS, Ethiopia, 2023. In this study, participants who have no a history of antibiotic change have better recovery time as compared to those who have history of antibiotic change once and above (Fig. 4). Fig. 4The Kaplan-Meier survival curve compare the survival time of the category of number of antibiotics change for participant admitted with NS at public Comprehensive specialized hospitals in SNNPRS, Ethiopia, 2023. The Kaplan-Meier survival curve compare the survival time of the category of number of antibiotics change for participant admitted with NS at public Comprehensive specialized hospitals in SNNPRS, Ethiopia, 2023. In this study, neonates with respiratory distress have lower recovery time as compared to those neonates who had no respiratory distress (Fig. 5). Fig. 5The Kaplan-Meier survival curve compare the survival time of the category of having respiratory distress for participant admitted with NS at public Comprehensive specialized hospitals in SNNPRS, Ethiopia, 2023. The Kaplan-Meier survival curve compare the survival time of the category of having respiratory distress for participant admitted with NS at public Comprehensive specialized hospitals in SNNPRS, Ethiopia, 2023. In this study, neonates with history of resuscitation at birth have lower recovery time as compared to those neonates without history of resuscitation at birth (Fig. 6). Fig. 6The Kaplan-Meier survival curve compare the survival time of the category of neonates having resuscitation at admission weight at public Comprehensive specialized hospitals in SNNPRS, Ethiopia, 2023. The Kaplan-Meier survival curve compare the survival time of the category of neonates having resuscitation at admission weight at public Comprehensive specialized hospitals in SNNPRS, Ethiopia, 2023. Log-rank estimate of recovery time among neonates with neonatal sepsis across predictors As shown in the table below the log-rank test estimate between categories of different predictor or variables; those variables like; residence, place of delivery, intrapartum fever, resuscitation at birth, BCG vaccine, assisted with bag and mask, co-morbidity and hypothermia were significantly different with p-value < 0.05 as there is a significant difference among survival curves. The Kaplan Meier together with the log-rank test shows the effect of each predictor on the neonatal recovery with NS (Table 7). Table 7Log-rank test for equality of different categorical determinants of neonates admitted with neonatal sepsis at public comprehensive specialized hospitals, SNNPRS, Ethiopia, 2023.VariablesCategoryMedian recovery timeX2 P-valueResidenceUrban8(8,9)4.41 0.0357Rural9(8,10)Place of deliveryHome10(9,14)7.46 0.0063Health institution8(8,9)Intrapartum feverYes8(7,9)6.07 0.0137No9(8,9)Twin pregnancyYes10(8,10)0.77 0.3806No8(8,9)EBFIYes8(8,9)3.17 0.0749No9(8,9)Resuscitated at birthYes8(8,9)35 <0.001No10(10,12)BCG vaccineYes8(8,9)4.17 0.0411No8(8,10)Time of sepsis onsetEONS9(9,10)2.18 0.1398LONS8(8,9)Assisted with bag & maskYes10(9,10)13.33 <0.001No8(8,9)FeverYes9(8,10)0.25 0.6152No8(8,9)HypothermiaYes10(9,10)21.92 0.0000No8(8,9)ComorbidityYes10(10,12)12.95 0.0003No8(8,9)Age of neonate in days<7 Days9(8,10)0.8 0.3691>7Days8(8,9) Table 7Log-rank test for equality of different categorical determinants of neonates admitted with neonatal sepsis at public comprehensive specialized hospitals, SNNPRS, Ethiopia, 2023.VariablesCategoryMedian recovery timeX2 P-valueResidenceUrban8(8,9)4.41 0.0357Rural9(8,10)Place of deliveryHome10(9,14)7.46 0.0063Health institution8(8,9)Intrapartum feverYes8(7,9)6.07 0.0137No9(8,9)Twin pregnancyYes10(8,10)0.77 0.3806No8(8,9)EBFIYes8(8,9)3.17 0.0749No9(8,9)Resuscitated at birthYes8(8,9)35 <0.001No10(10,12)BCG vaccineYes8(8,9)4.17 0.0411No8(8,10)Time of sepsis onsetEONS9(9,10)2.18 0.1398LONS8(8,9)Assisted with bag & maskYes10(9,10)13.33 <0.001No8(8,9)FeverYes9(8,10)0.25 0.6152No8(8,9)HypothermiaYes10(9,10)21.92 0.0000No8(8,9)ComorbidityYes10(10,12)12.95 0.0003No8(8,9)Age of neonate in days<7 Days9(8,10)0.8 0.3691>7Days8(8,9) Log-rank test for equality of different categorical determinants of neonates admitted with neonatal sepsis at public comprehensive specialized hospitals, SNNPRS, Ethiopia, 2023. Factor associated to neonatal sepsis recovery time The Cox proportional hazard regression model was used to identify determinants of recovery of neonatal sepsis. Bivariate analysis was performed first using cox proportional hazard regression to identify which variable had an association with time to recovery from neonatal sepsis. Those variables like; residence, number of ANC visit, Twin pregnancy, intrapartum fever, EBF initiation within one hour, having history of resuscitated at birth, BCG vaccination, time of infection onset, assisted with bag and mask/requirement of invasive ventilation, having history of fever, respiratory distress, hypothermia, comorbidity, age of neonate, birth weight, 5th minute APGAR score, number of anti-biotic change and place of delivery were statistically significant with a p-value ≤ 0.25 and candidate for multivariate cox regression model. In a multivariable proportional hazard model; those variables like; resuscitated at birth, having history of fever, frequent antibiotics change and respiratory distress were remains independent determinants of recovery time among neonates with neonatal sepsis with (p value < 0.05). The multivariable analysis result shown that; the hazard of prolonged time to recovery of NS among neonates with fever was 18% delayed than its counterparts. Neonates with history of resuscitation at birth were delayed by 35% in time to recovery of NS as compared to neonates without history resuscitation at birth. Similarly, the time to recovery of NS among neonates with respiratory distress was prolonged by 29% as compared to its counterparts. The neonates who had no history of antibiotic change were 2 times shorter time to recovery from NS (Table 8). Table 8Bivariate and multivariable Cox proportional hazard regression of neonates admitted with neonatal sepsis at public comprehensive specialized hospitals, SNNPRS, Ethiopia, 2023.VariablesCategoryCHRP-valueAHR P-valuePlace of deliveryHome0.65(0.40,1.04)0.070.71(0.44, 1.15)1.171Health institution11Twin pregnancyYes0.74(0.54,1.03)0.070.84(0.60, 1.17)0.310No11Intrapartum feverYes1.31(0.98,1.74)0.061.23(0.93, 1.65)0.145No11Resuscitation at birthYes0.60(0.45,0.79) 0.00 0.65(0.49, 0.87) 0.004* No11BCG vaccineYes1.17(0.99,1.40)0.0631.00(0.81, 1.24)0.985No11Sepsis typeEONS0,89(0.074,1.06)0.210.96(0.77, 1.22)0.792LONS11Oxygen supplyYes0.79(0.66,0.96)0.020.87(0.72, 1.06)0.183No11Fever during admissionYes0.98(0.74,1.07) 0.22 0.82(0.67, 0.99) 0.046* No11Respiratory distressYes0.67(0.56,0.79) 0.00 0.71(0.59, 0.85) 0.000* No11HypothermiaYes0.75(0.62,0.93)0.010.83(0.65 1.05)0.136No11Birth weight< 25000.78(0.64,0.94)0.010.89(0.71, 1.10)0.299> 2500115th minute APGAR score< 70.77(0.62,0.97)0.030.88(0.70, 1.10)0.272> 711Frequency of antibiotic changeNo2.57(2.0, 3.30) 0.00 2.0(1.57, 2.63) 0.000* Once & above11 Table 8Bivariate and multivariable Cox proportional hazard regression of neonates admitted with neonatal sepsis at public comprehensive specialized hospitals, SNNPRS, Ethiopia, 2023.VariablesCategoryCHRP-valueAHR P-valuePlace of deliveryHome0.65(0.40,1.04)0.070.71(0.44, 1.15)1.171Health institution11Twin pregnancyYes0.74(0.54,1.03)0.070.84(0.60, 1.17)0.310No11Intrapartum feverYes1.31(0.98,1.74)0.061.23(0.93, 1.65)0.145No11Resuscitation at birthYes0.60(0.45,0.79) 0.00 0.65(0.49, 0.87) 0.004* No11BCG vaccineYes1.17(0.99,1.40)0.0631.00(0.81, 1.24)0.985No11Sepsis typeEONS0,89(0.074,1.06)0.210.96(0.77, 1.22)0.792LONS11Oxygen supplyYes0.79(0.66,0.96)0.020.87(0.72, 1.06)0.183No11Fever during admissionYes0.98(0.74,1.07) 0.22 0.82(0.67, 0.99) 0.046* No11Respiratory distressYes0.67(0.56,0.79) 0.00 0.71(0.59, 0.85) 0.000* No11HypothermiaYes0.75(0.62,0.93)0.010.83(0.65 1.05)0.136No11Birth weight< 25000.78(0.64,0.94)0.010.89(0.71, 1.10)0.299> 2500115th minute APGAR score< 70.77(0.62,0.97)0.030.88(0.70, 1.10)0.272> 711Frequency of antibiotic changeNo2.57(2.0, 3.30) 0.00 2.0(1.57, 2.63) 0.000* Once & above11 Bivariate and multivariable Cox proportional hazard regression of neonates admitted with neonatal sepsis at public comprehensive specialized hospitals, SNNPRS, Ethiopia, 2023. Test of proportional-hazards assumption To use of fitted proportional hazard models, the proportional hazard assumption should be checked. As a result, in this study the Schoenfeld residuals proportional hazard assumption test was used for individual covariates as well as global tests. The proportional hazard assumption is not met if the P-Value is less than 0.05. In this study each covariate has P-Value > 0.05 and all covariates (Global test for Cox proportional hazard P-Value = 0.2754 which is also > 0.05 which met the proportional hazard assumption as shown in (Table 9). Table 9Test of proportional hazard by Schoenfeld residuals for each predictor and global test.Variablesrhochi2Prob > chi2Birth weight0.014870.130.7207Age of neonate0.088733.600.0579Age of mother0.027920.430.5107Comorbidity0.037170.780.3757Chest indrowing0.065772.460.1171Jaundice0.023700.310.5803Cyanosis0.013670.110.7443Residence0.037700.810.3689Place of delivery0.021600.240.6210Number of ANC visit0.040720.870.3497BCG vaccine0.040380.930.3341Sepsis type0.011300.060.8044Assisted with mask & bag0.065542.390.1220Global test15.53130.2754Notice: Rho is the correlation coefficient between the residuals and time. Table 9Test of proportional hazard by Schoenfeld residuals for each predictor and global test.Variablesrhochi2Prob > chi2Birth weight0.014870.130.7207Age of neonate0.088733.600.0579Age of mother0.027920.430.5107Comorbidity0.037170.780.3757Chest indrowing0.065772.460.1171Jaundice0.023700.310.5803Cyanosis0.013670.110.7443Residence0.037700.810.3689Place of delivery0.021600.240.6210Number of ANC visit0.040720.870.3497BCG vaccine0.040380.930.3341Sepsis type0.011300.060.8044Assisted with mask & bag0.065542.390.1220Global test15.53130.2754Notice: Rho is the correlation coefficient between the residuals and time. Test of proportional hazard by Schoenfeld residuals for each predictor and global test. Notice: Rho is the correlation coefficient between the residuals and time. Testing overall fitness of the model In this study goodness-of-fit (gof) was checked graphically using the Cox Snell residuals plot to check the assumption of the Cox proportional hazard model. For the residuals test, the hazard function follows 45º close to the baseline hazard, which indicates that the model was well-fitted. The jagged line with the reference line (Cox Snell residual line) follows the 45˚ line closely. Hence, the overall cox- regression model fits to the data (Fig. 7). Fig. 7Cox Snell residual test for proportional assumption of each covariant and overall Cox proportional hazard model. Cox Snell residual test for proportional assumption of each covariant and overall Cox proportional hazard model. Discussion This study assessed the time to recovery of neonatal sepsis and determinant factors among neonates admitted in Public Comprehensive Specialized Hospitals of SNNPRS, Ethiopia. In this study, the neonates with sepsis were followed for a total of 5424-neonate day observations. The results of this study showed that the overall recovery time of neonatal sepsis was 8 days (IQR; 7, 10). This result in line with the previous study done in Addis Ababa which was reported as 8.92 days34. And lower than the previous study conducted in Arbaminch 12.74 days7; the possible reason for discrepancy might be due to sample size difference and level of hospitals that the study was carried out (general and primary hospitals) while the current study was conducted in comprehensive specialized hospital which have trained and specialized care giver on newborn care. And this is also lower than the study done in Brazil that report 19 days which was two times more than the current study35. The possible difference may be due to sample size difference and different source population that it assesses the LDOHS of all neonates admitted to NICU. This results of this study is higher than the previous studies conducted in Uganda and India which reported 5.5 and 5.4 days respectively36,37; the reason for this discrepancy might be due to difference in study design, sample size and age range of the participants they used(less than 4 month (birth to 120 days)), this may help them for strength of their immunization and early recovery. The result of this study is also higher than the previous studies done in Bahir Dar34, Gondar31 and Dire Dawa20, 6 days, 7 days and 7 days respectively. During the follow up period; recovery probability at the beginning, half and end of study was 99.84%, 18.26% and 1.98% respectively. This finding was in line with the previous study conducted in Gondar31. The possible similarity may be due to almost similar sample size (631) and model of analysis. In this study the incidence rate of recovery is 9.86 per 100 neonate days. This result is consistent with the study conducted in Gondar which was 10.78 per100 neonate days. The variation might be due to gestational age < 37 weeks in the previous study was small in number than the current one, this may cause them to have increased rate of recovery31. In this study neonates with fever during admission were delayed by 18% in time to recover from NS as compared to their counter parts. This finding is supported by previous studies conducted in Nigeria38 and Iraq39. The reason may be due to the fact that; as the neonates have an infection, the body’s temperature usually rises as it tries to fight off the host causing the infection as a result there is the exhaustion of immunity which leads to late recovery. In this study those neonates who had no history of antibiotic change were 2 times shorter time to recover from NS than those neonates with history of one and above antibiotic change during their stays. This finding is inconsistence with the study done in Philadelphia40. The observed difference may be due to small sample size (n = 194) and age cut of point (< 1year) on the pervious study. In this study the hazard of prolonged time to recovery of neonatal sepsis among neonates with respiratory distress were 29% less likely to recover fast as compared to those neonates without respiratory distress. This finding is in line with the study conducted in Nigeria38. The possible reason for consistence may be due to similar age limit of the study participant. However this result was inconsistence with the study done in Bahir Dar Felege Hiwot Referral Hospital41. The observed difference may be due to the variation in study design and model of analysis that the previous used logistic regression for outcome of neonate. The possible explanation is that; respiratory distress may leads to low gas exchange and increased physiological dead space, which causes multiple organ hypoxia and dysfunction; as a result neonates with such complication have prolonged hospital stay. The time to recovery among neonates with resuscitation at birth is delayed by 35% from NS compared to their counterparts. This finding was supported by previous studies done in Woldia and Dessie comprehensive specialized hospital25, Tanzania17 and Kaffa zone42. The possible fact may be due to the lumen of the peripheral airway of the newborn is thin and respiratory secretions are overflowing which could predispose to atelectasis. For the collapsed lungs, performing different procedures may cause for an entry point for microbial agents. And also resuscitation might be done with unsterile procedure, which could introduce microbes into the respiratory system of the neonate whose immune system is not well developed and unable to fight against other nosocomial infection and become risk for prolonged hospital admission. In this study those neonates who had no history of antibiotic change were 2 times shorter time to recovery from NS than those neonates with history of one and above antibiotic change during their stays. This finding is inconsistence with the study done in Philadelphia40. The observed difference may be due to small sample size (n = 194) and age cut of point (< 1year) on the pervious study. This finding is also inconsistent with the previous study conducted in Nigeria38 which report it as it has no effect on the patient outcome and recommended for further validation, this inconsistence may be due to difference in model of analysis that it was used. However this finding is supported by WHO guideline that recommends parenteral antibiotics treatment for a neonate having known risk of sepsis for no less than 2 days43. This might be due to fear of antibiotic resistance to the host causing the infection. Conclusion and recommendation Duration of recovery from neonatal sepsis is acceptable according to previous literatures but still it needs hastening the recovery time. The time it took to recovery from neonatal sepsis had affected by, having history of fever, respiratory distress and frequent/number of antibiotic change. Therefore, health care providers, especially those working in NICU must pay more attention for prevention and early intervention of complication. The above-mentioned factors could be used for the early identification of neonates with sepsis at risk for delayed recovery time. Strength and limitation The study was conducted at the multicenter scope with different variable categories, which was indicated as a limitation by most studies. The study was retrospective and centered on secondary data, it did not take into account all possible predictors of the outcome variable, such as blood chemistry tests, other health professions skill was a limitation. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1