Retrospective review of neonatal morbidity and mortality at public referral hospitals in Greater Accra Region of Ghana:2013–2014

The regional and district hospitals recorded only the birth weight in the register while the tertiary hospital reweighed 98.9% of all newborns at admission and also documented the birth weight. The overall Pearson correlation coefficient (R) between the birth weight (weight measured in the labour room of referring hospital) and weight at the time of admission at the tertiary hospital for infants hospitalised within 24 hours after birth was = 0.979 but varied between 0.441 for the smallest infants (BW <750grams); 0.478 for ELBW, 0.565 for VLBW; 0.842 for LBW and 0.924 for infants weighing 2500grams and higher (Fig. 2).

OPEN IN VIEWER

Fig.2

Correlation between weight at birth and at admission on the day of birth.

The smallest surviving infant during the period weighed 600grams. Only 4.8% (3/63) of newborns who weighed between 500grams to 699grams survived, 50/63 of these were hospitalised at the tertiary hospital and 2 out of the 3 that survived were at the tertiary hospital (Table 2). Of the 44 infants with birth weight between 700grams to 749grams, 28 were hospitalised at the tertiary hospital and reweighed within four hours after birth, 2/28 had no outcome data, 8/28 weighed over 750grams on admission and two survived, 4/28 weighed less than 700grams and none survived, and 14/28 were within the weight category and one survived. The 16/44 infants hospitalised at the regional and district hospitals had only birth weight data, 12 were documented as discharged (survived) and four had no outcome data.

The odds of survival for infants with birth weight <750grams whose weight were cross-checked at admission was 1:25 for those less than 700grams and 1:16 for infants weighing 700grams– 749grams. The odds of survival increased with increasing birth weight and was essentially similar for infants with birthweight of 1800grams and higher; this group comprised 69.8% (5731) of all admissions and had an overall survival of 88.8% (Table 2).

Outcome data for 8205 neonates (excluding 23 infants weighing <500grams) showed that LBW infants accounted for 46.7% of admissions and 67% of deaths. Normal birth weight infants accounted for 44.1% of total admissions and 27% of deaths, infants weighing 4000grams and higher accounted for 5.8% of admissions and 2.7% of deaths and those without documented birth weight data accounted for 3.4% of admissions and 3.3% of deaths (Table 2).

There was no significant difference in mortality between males and females (676/4253 vs. 515/3495; p = 0.16). Normal birth weight infants accounted for 45.5% (3619/7953) of all admissions with documented birth weight (Table 2). Of these 74% (2680) had documented provisional diagnosis or reason for referral and the major causes of morbidity were perinatal asphyxia 49% (1312), respiratory distress 14.7% (533), suspected sepsis 11.9% (430), and suspected meconium aspiration syndrome 7.5% (273). Other causes of morbidity included congenital anomalies, gastrointestinal disorders, infant of diabetic mother, and birth trauma. The following provisional diagnosis or reason for referral entries in the register were grouped as perinatal asphyxia: birth asphyxia, blue asphyxia, white asphyxia, low Apgar score, weak cry at birth, or unable to cry at birth. There was no entry described as neonatal encephalopathy, hypoxic-ischaemic encephalopathy or intrapartum-related brain injury.

Among infants with birth weight data, date of birth and date of discharge or death was documented in 79.6% (6328/7953) of cases; 78.8% (4983) were discharged (or died) by the tenth day, 16.6% (1050) were hospitalised up to 27 days, 2.6% (168) up to 42 days, and 2% (127) beyond 42 days. All birth weight categories were represented among those with length of stay beyond 42 days including cases with congenital abnormalities and infants who were abandoned. Among infants with less than 24-hours length of stay (407), 65.9% had outcome documented as death. The length of stay for the weight categories is summarized in Table 3.

Of the 1203 documented deaths, 528 (55.1%) had documented provisional diagnosis or reason for referral; 555 (46.1%) including the 23 infants <500grams were VLBW; 958 (79.6%) weighed 1000grams and higher. Overall, the most common provisional diagnosis in infants who died were perinatal asphyxia and respiratory distress (Table 3). Among infants weighing 2500 grams and higher, 398 died, provisional diagnosis data was documented in 309 and 82.8% (256) had perinatal asphyxia, 10% (31) respiratory disorders, 3.2% (10) congenital anomalies and 2.3% (7) suspected sepsis. (Table 3). Among LBW infants, 58.4% (146/250) of those who died had provisional diagnosis and perinatal asphyxia (54.1%) and respiratory distress (26.7%) were the major probable underlying causes of death (Table 3). Perinatal asphyxia was the major probable underlying cause of death for infants with birth weight of 1000grams and higher.

Data was available for one year at the district hospital and for 2 years at the regional and tertiary levels of care (Table 4). Overall less than 1% of infants were transferred out to other departments or hospitals and most transfers were for surgical care. There were many gaps in the registers at all three levels of care. None of the hospitals documented head circumference. Clinical complications such as sepsis, and clinical procedures such as oxygen therapy, blood transfusion, kangaroo care or phototherapy were rarely documented in the register. Human resource was severely limited at all levels of care. The poorest outcome was among VLBW infants; survival rate increased with increasing birth weight at all levels of care (Table 4). Of the two hospitals with two years’ data, the tertiary hospital had a higher burden of ELBW infants and the regional hospital had no outcome data for 18.7% of VLBW infants compared to 5.1% at the tertiary hospital. There was no statistically significant difference (p = 0.9) in the outcome of infants weighing more than 1499grams at the regional and tertiary levels of care.

4.1 Underlying causes of morbidity and mortality

The high burden of neonatal morbidity and mortality in sub-Saharan Africa including Ghana is well known [7]. Although Ghana is one of the countries with relatively low neonatal death burden [7] and high rates of antenatal care attendance in the sub-region [8, 9], there is urban-rural disparity [10–12] and the higher rate of mortality in urban settings has been attributed to limited resources and substandard care in hospitals [13]. In Ghana, the high rates of antenatal care attendance and facility-based births do not seem to reflect in health indicators on care around the time of birth. Maternal morbidity and mortality remain relatively high [9], and as shown in this study, perinatal asphyxia, a surrogate indicator of antenatal and intrapartum care, was a major cause of morbidity and mortality for all birth weight categories at all levels of neonatal care. The limited specialised neonatal services and human resource for the volume of hospitalised newborns in this study attest to overstretched resources [13]. It also provides insight to underlying factors for poor neonatal outcome and highlights health system blocks for targeted investments to reduce newborn deaths in GAR [14–17]. For example, there is need to urgently address the upstream factors associated with the high burden of perinatal asphyxia. Respiratory distress, a common nonspecific sign of neonatal illness [18], was a significant cause of morbidity and mortality in this cohort. It is common in preterm/LBW infants because of immaturity of their respiratory system including surfactant deficiency and may occur as a result of infections, birth complications, congenital and metabolic abnormalities and other factors irrespective of birth weight or maturity [19, 20]. Reducing the burden of respiratory distress will require preventive measures for reducing the risk of preterm birth, effective use of antenatal steroids, reducing birth injuries, and minimizing maternal-associated and hospital-acquired neonatal infections. Severe cases of respiratory distress require respiratory support with simple technologies like continuous positive airway pressure (CPAP) ventilation, a recognized service that improves outcome [20, 21], or more advanced support with mechanical ventilation. In 2013/2014, none of the 3 hospitals had the capacity to provide CPAP respiratory support to affected infants. The national action plan includes equipping referral centres to provide respiratory support [5]; if implemented, the impact of the service on the outcome of cases with respiratory distress at the neonatal units in this study will need evaluation.

4.2 Survival rates

The overall survival for each birth weight category in this study was relatively low especially among normal birth weight infants who were mostly afflicted with perinatal asphyxia, a largely preventable condition [17]. In 2013/2014, the rate of survival to discharge for infants with birth weight of 500–1499grams (VLBW) in Ghana was significantly lower than rates in United States in 1988 [22], New Zealand in 1986 [23], a public hospital in Republic of South Africa in 2000–2002 [24] and Bangladesh in 2013/2014 [25]; but higher than some sub-Saharan African countries [26–28]. In this cohort, VLBW infants accounted for 17.4% of admissions and 45.1% of deaths. National investments in maternal-newborn health should prioritize cost-effective services that reduce morbidity and mortality in VLBW infants [14, 28–30]. The survival rate for VLBW infants in this cohort could be lower than described in this paper. The low correlation between the weight measured at place of birth and at admission in the tertiary hospital for VLBW cases indicates uncertainty about the accuracy of birth weight documented in the registers of all 3 neonatal units. Differences in the precision of weighing scales in labour rooms and neonatal units and inadequate fluid therapy before and during transfer of the smallest neonates could have contributed to the incongruent weight measurements on the day of birth among VLBW at the tertiary hospital. Birth weight is the most useful predictor of neonatal outcome and the most useful tool for classifying newborns to determine the level of hospital care required [28]. It is also critical for calculating the dosage of medicines for sick newborns, hence inaccurate body weight measurement is potentially unsafe for newborns [31]. The World Health Organization has recommended standards for measuring weight in newborns [32]. Implementing these standards with clinical infant weighing scales in every health facility where birth occurs is vital for improving the quality of care and assessing the impact of neonatal interventions. Reweighing all newborns with standardized high precision digital infant weighing scales at admission in neonatal referral centres in GAR is recommended.

4.3 Study limitations

The study has several limitations. First, the inherent disadvantages of retrospective design [33] and considerable data gaps in the hospital registers is a major limitation to making definitive causal inferences. Nevertheless, the large cohort of over 8000 newborns, reasonably reliable birth weight data in infants weighing 1500grams and higher, and outcome data in over 90% of cases provides a good estimate of burden of disease irrespective of possible errors in documentation, diagnosis and birth weight measurements. These data gaps are important findings of this study as it highlights areas for quality improvement in the health system. Comprehensive medical records, an essential tool for planning and evaluating healthcare services, have not been a priority in many developing countries although it is vital for evaluating regional and national programs and interventions to reduce newborn deaths [34, 35]. Secondly, no inference could be made on the burden of prematurity in this cohort as described elsewhere [36], or the prevalence of neonatal infections because of the gaps in documentation. This further highlights the need to review the tools and processes for data capture at the study sites including the content of the hospital register or database, and diagnostic criteria for clinical conditions recorded in the register. Thirdly, the criteria for the diagnosis of perinatal asphyxia in this cohort could not be verified retrospectively. However, most facility-based maternal-newborn services in Ghana are provided by skilled health workers [37] and doctors are the only clinicians who statutorily give a diagnosis to a clinical condition at each of the 3 hospitals and it is likely that majority of the provisional diagnosis were rational.