**2. Determinants of neonatal survival at a tertiary hospital in Johannesburg, South Africa**

Although VLBW mortality at CMJAH has been studied, the overall neonatal survival has not been audited. The aim of this study is to review neonatal survival at CMJAH and to determine important modifiable factors to inform protocols and budgeting for neonatal care. The objectives of this study were to:

**•** Describe the patient population with regard to demographic information, clinical characteristics, and outcome at discharge.


#### **2.1. Subjects and methods**

4.8 million of the almost 6 million children under the age of five years who died in 2015, died within the neonatal period [1]. Concentrating resources on newborns is therefore essential to

88 Epidemiology of Communicable and Non-Communicable Diseases - Attributes of Lifestyle and Nature on Humankind

The causes of neonatal mortality vary considerably among different units and different countries. The United Nations MDG 2015 report [1] states that "better data are needed for the post-2015 development agenda" and "real-time data are needed" to guide policy makers. Most data have a time lag of between 2 and 3 years before the policies are implemented. The MDGs formed the foundation of the so-called Sustainable Development Goals (SDG) [2]. The SDGs are less specific than the MDGs, but include health targets, one of which is to reduce both

Regular audits of neonatal mortality are required to identify the causes of death so that proper interventions can be implemented to reduce neonatal deaths. It is essential to have local data to address local health issues; transposing mortality data from another country will not necessarily solve local problems. This is particularly true when using data from a high-income country to address problems experienced in low- to middle-income countries (LMICS). A recent review of the mortality rates in neonatal intensive care units showed that the rate varied considerably between different countries [3]; the mortality rate was generally high, but greater in developing than developed countries. Issues such as the lack of antenatal care and inadequate health facilities are the causes of neonatal mortality in LMICS. A recent review from The Gambia [4] showed a high neonatal mortality rate – 35% of admitted neonates died. The important causes of neonatal death included lack of antenatal care, birth weight below 1500 g,

Previous studies done in very low birth weight (VLBW) neonates – birth weight below 1500 g – at Charlotte Maxeke Johannesburg Academic Hospital (CMJAH) have shown that birth weight was the most significant predictor of survival [5, 6]. Resuscitation at birth, the use of nasal continuous positive airways pressure (NCPAP) and the mode of delivery were also important factors affecting survival. Survival of extremely low birth weight (ELBW) neonates was particularly low at CMJAH [7]. The provision of NCPAP to this category of neonates more

**2. Determinants of neonatal survival at a tertiary hospital in Johannesburg,**

Although VLBW mortality at CMJAH has been studied, the overall neonatal survival has not been audited. The aim of this study is to review neonatal survival at CMJAH and to determine important modifiable factors to inform protocols and budgeting for neonatal care. The

**•** Describe the patient population with regard to demographic information, clinical charac-

neonatal mortality and mortality of children under the age of 5 years.

hypothermia at birth, and delivery outside a teaching hospital.

than doubled their survival to discharge [6].

**South Africa**

objectives of this study were to:

teristics, and outcome at discharge.

further reduce childhood mortality.

The study was conducted in the neonatal unit of a tertiary academic hospital (CMJAH) in Johannesburg, South Africa. All neonates admitted within 48 h of birth, between 1 January 2013 and 31 December 2015, were included in the study. Neonates with a birth weight below 400 g and those with important missing data, particularly birth weight, gestational age, and outcome at discharge were excluded.

#### **2.2. Study design**

This was a secondary analysis of an existing neonatal database. Data were collected upon discharge for each neonate admitted to the CMJAH neonatal unit and entered on to a database. The database was managed using Research Electronic Data Capture (REDCAP) [8] hosted by the University of the Witwatersrand. The information collected included demographic details, maternal information, delivery room data, clinical information, and outcome at discharge. Data from VLBW neonates was contributed to the Vermont Oxford Network (VON) (www.vtoxford.org), a multinational neonatal collaboration. A paper computer summary form was completed for each patient, using the patient file. Data were checked against the patient file and then entered on to the database. The information on the database was then checked against the paper form. Any discrepancies noted were verified against the patient files. Definitions and codes for congenital defects or surgical procedures were obtained from the VON.

Neonates were classified by weight using standard definitions—term large for gestational age (TLGA) neonates weighed above 4000 g at birth, term appropriate for gestational age (TAGA) infants weighed between 2500 and 3999 g at birth, low birth weight (LBW) neonates had a birth weight less than 2500 g, very low birth weight (VLBW) included those weighing less than 1500 g at birth and extremely low birth weight (ELBW) less than 1000 g at birth. Term was considered to be a gestation age between 37 and 42 weeks, preterm below 37 weeks, and post-term to be above 42 weeks.

The unit participated in a national perinatal mortality audit – the perinatal problem identification programme (PPIP)(www.ppip.co.za). The broad causes of neonatal death were categorized using standard PPIP definitions.

#### *2.2.1. Neonatal unit*

The neonatal unit was situated in large tertiary academic hospital in a metropolitan setting. Neonatal facilities included a transitional nursery in labour ward, a shared paediatric/neonatal intensive care unit (PNICU) with 15 ventilator beds, a neonatal high care unit with 40 beds, low-care facility with 25 beds, and nine kangaroo mother-care (KMC) beds. Nasal continuous positive airways pressure (NCPAP) and therapeutic hypothermia for perinatal asphyxia were provided in high care. The neonatal unit was staffed by neonatologists, registrars, and house staff. There were various paediatric sub-specialities in the hospital including nephrology, neurology, cardiology, endocrinology, and infectious diseases. There was a large paediatric surgery service and paediatric surgical neonates were admitted to the neonatal unit and jointly managed with the neonatal staff.

Neonates who were observed in the transitional unit and then discharged to their mothers were not included in the study. Neonates who died in the delivery room and transitional nursery were considered to be admissions and were included in the study. Owing to resource constraints, there were insufficient ventilator beds for the number of neonates requiring ventilation. The PNICU functioned essentially as a ventilator unit; high-care observation was not possible due to limited facilities. The neonatal unit had a policy of rationing care based on birth weight—babies weighing below 750 g at birth would not be offered surfactant or NCPAP, but only given supplemental oxygen, intravenous fluids, and antibiotics; babies weighing between 750 and 900 g would be given surfactant and NCPAP, but would not be provided with mechanical ventilation if required. All neonates with respiratory distress syndrome were initially managed with NCPAP and early rescue surfactant; those who failed would be transferred to the PNICU for mechanical ventilation. The use of NCPAP at CMJAH has recently been reviewed [9].

#### *2.2.2. Statistical analysis*

Data were exported to IBM SPSS version 22 for the purpose of analysis. The standard statistical methods were used to describe the data—continuous variables were described using measures of central tendency and dispersion, mean and standard deviation (SD), or median and interquartile range (IQR) as appropriate. Categorical variables were described using frequency and percentages.

The primary endpoint was whether a neonate survived to hospital discharge. Univariate analysis was done considering different maternal, demographic, and clinical variables as independent factors of survival. Differences in outcome for continuous variables were compared using unpaired *t*-tests or Mann Whitney *U*-test as appropriate. Associations of outcome with categorical variables were investigated using Chi-squared test. A factor with a *p*-value of 0.05 was considered statistically significant. Variables with a *p*-value <0.1 on the univariate analysis were entered into a multiple logistic regression model considering whether a child survived to discharge as the outcome variable. Factors associated with neonatal mortality were determined separately for VLBW and bigger babies.

The possible sources of bias were identified and excluded from the analysis. Conditions which were only present in neonates who were survivors and approaching discharge were identified and excluded from the analysis of deaths. These conditions included supplementary oxygen at 28 days, home oxygen and steroids for chronic lung disease. Maternal and delivery room conditions were compared between those neonates who died in the delivery room and those who died in the neonatal ward.

#### **2.3. Ethics**

provided in high care. The neonatal unit was staffed by neonatologists, registrars, and house staff. There were various paediatric sub-specialities in the hospital including nephrology, neurology, cardiology, endocrinology, and infectious diseases. There was a large paediatric surgery service and paediatric surgical neonates were admitted to the neonatal unit and jointly

90 Epidemiology of Communicable and Non-Communicable Diseases - Attributes of Lifestyle and Nature on Humankind

Neonates who were observed in the transitional unit and then discharged to their mothers were not included in the study. Neonates who died in the delivery room and transitional nursery were considered to be admissions and were included in the study. Owing to resource constraints, there were insufficient ventilator beds for the number of neonates requiring ventilation. The PNICU functioned essentially as a ventilator unit; high-care observation was not possible due to limited facilities. The neonatal unit had a policy of rationing care based on birth weight—babies weighing below 750 g at birth would not be offered surfactant or NCPAP, but only given supplemental oxygen, intravenous fluids, and antibiotics; babies weighing between 750 and 900 g would be given surfactant and NCPAP, but would not be provided with mechanical ventilation if required. All neonates with respiratory distress syndrome were initially managed with NCPAP and early rescue surfactant; those who failed would be transferred to the PNICU for mechanical ventilation. The use of NCPAP at CMJAH has recently

Data were exported to IBM SPSS version 22 for the purpose of analysis. The standard statistical methods were used to describe the data—continuous variables were described using measures of central tendency and dispersion, mean and standard deviation (SD), or median and interquartile range (IQR) as appropriate. Categorical variables were described using frequency

The primary endpoint was whether a neonate survived to hospital discharge. Univariate analysis was done considering different maternal, demographic, and clinical variables as independent factors of survival. Differences in outcome for continuous variables were compared using unpaired *t*-tests or Mann Whitney *U*-test as appropriate. Associations of outcome with categorical variables were investigated using Chi-squared test. A factor with a *p*-value of 0.05 was considered statistically significant. Variables with a *p*-value <0.1 on the univariate analysis were entered into a multiple logistic regression model considering whether a child survived to discharge as the outcome variable. Factors associated with neonatal

The possible sources of bias were identified and excluded from the analysis. Conditions which were only present in neonates who were survivors and approaching discharge were identified and excluded from the analysis of deaths. These conditions included supplementary oxygen at 28 days, home oxygen and steroids for chronic lung disease. Maternal and delivery room conditions were compared between those neonates who died in the delivery room and those

mortality were determined separately for VLBW and bigger babies.

managed with the neonatal staff.

been reviewed [9].

and percentages.

who died in the neonatal ward.

*2.2.2. Statistical analysis*

Data were de‐identified and the key to patient details was kept separately and only known to the principal investigator. Ethical clearance for the study was obtained from the Human Research Ethics Committee of the University of the Witwatersrand. Permission to conduct the study was obtained from the Chief Executive Officer of CMJAH. One of the authors was the gatekeeper of the neonatal database; additional permission to access the database was not required.
