Discussion
We sought to describe the characteristics, disease patterns and outcomes of inborn neonates admitted in NBUs in a set of Kenyan counties’ referral hospitals. This study has demonstrated that even in low-resource settings, it is possible to have clinical networks enabling multisite data collection that yields quality data that could inform policy and help in planning newborn services that are context sensitive. Neonates contribute to almost half (46%) of all admissions among medical patients aged 0–13 years, although this proportion considerably varies across hospitals. The reasons for this variation are likely to be complex and depend on admission policies in NBUs and paediatric wards, different disease patterns in paediatrics wards (eg, prevalence of malaria) and local referral patterns especially for outborn neonates.10 For example, we have previously demonstrated considerable variation in classification of paediatric illnesses’ severity (eg, malaria and pneumonia) in Kenyan hospitals. This may lead to higher admission rates in paediatric wards in some settings that result in a relatively lower proportion of newborn admissions.17–19 Importantly, neonates contribute to two-thirds of the mortality among all patients aged 0–13 years admitted in CIN hospitals. This high proportion of deaths may reflect changing patterns of mortality at a population level in Kenya especially among children aged 1–4 years.1 This is important as hospitals in general may not be well equipped, staffed or organised to cater for the needs of the large proportion of small and sick neonates admitted in either paediatric wards or NBUs.20–22
Five conditions, all to some degree preventable, accounted for 80% of the disease episodes at admission among inborn neonates in NBUs, with intrapartum-related complications being the most common (30%) cause of admission. Reducing intrapartum-related complications means having a functional health system to provide quality care during the antenatal period, labour, childbirth and immediate postnatal period.23–25 Other common causes of admission are LBW/prematurity-related complications such as RDS, a condition that hospitals were poorly prepared to manage. Majority (10/16) of the NBUs had none or just one continuous positive airway pressure (CPAP) machine. Managing RDS with conventional oxygen therapy, as high-income countries (HICs) practised in 1960s, is estimated to be associated with a survival rate of less than half that associated with high-quality care supported by CPAP.26
However, while better technologies may help in the management of these common neonatal conditions, they need to be carefully introduced. Where human resources for health (HRH) are limited, adding new tasks can exacerbate existing challenges of delivering high-quality care.20 While half (8/16) of the CIN NBUs (often in large county hospitals) had a paediatrician dedicated to the NBU, the other half had a paediatrician who was also responsible for providing daily services to the paediatric wards. Only 7/16 NBUs had at least one neonatal nurse. Upgrading these facilities means that HRH challenges must be addressed in tandem to efforts to upgrade the equipment.14 27
There is high in-patient neonatal fatality in the NBUs studied across all birth weight bands. However, ELBW (<1000 g) with median mortality of 80% is almost five times higher that of HICs, while median mortality of VLBW (1000–1499 g) of 40% is about 10 times higher that of HICs.28 Given the limited resources in Kenya and many LMICs, it seems the considerable room for improvement in survival of VLBW infants should be given greater priority than efforts to tackle mortality in those with birth weights <1000 g. However, it is also worth noting that in our study, two-thirds of the babies who died had a birth weight >1500 g, with intrapartum-related complication being the leading cause of death. This is a reversal of what is observed in HICs where babies with GA <33 weeks account for about four-fifths of neonatal deaths.28
Majority (90%) of inborn neonates in the NBUs studied are admitted on their first day of life, and three out of every five deaths in the NBUs occur on the first day of admission. Poor quality of intrapartum and immediate postnatal care makes the day of birth the riskiest period for a neonate. WHO has developed guidance for continued care along the life course by integrating maternal and newborn care and promoting maternal perinatal death surveillance and response. However, the ‘P’, other than newborn resuscitation, is a weak component of these strategies.29–31 As in other LMICs, but in contrast to HICs, congenital conditions are not among the top 5 causes of admissions/deaths in Kenyan NBUs.28 32 This could be explained by the high numbers of preventable causes of deaths (>95%) in this study.
There are several limitations to this study. First, these 16 study hospitals are not a representative sample of Kenyan public hospitals. So, we cannot ascertain generalisability of estimates, although there have been other reports on high NBU mortality in Kenya.33 Second, we have used routine data from records reviewed. These can prove to be inaccurate when coupled with missing data and imprecise diagnoses due to limited diagnostic capacity in these low-resource settings. We tried to address these problems by using a stringent data quality assurance system and building leadership in the facilities to improve documentation practices.6 10 11 However, we used syndromic diagnoses. For example, diagnosis of neonatal sepsis was based was on clinical signs as described by WHO and national clinical guidelines as all hospitals had limited capacity and half had no capacity to do blood cultures.34 35 We used birth weight in our analysis instead of GA. LBW, though accurate, is a composite measure of SGA and prematurity and the proportion that are SGA/preterm will influence disease pattern and outcome. The different survival rates among the VLBWs could, in part, be due to variation in occurrence of SGA with lower mortality in places with a high prevalence of SGA. We did not collect data on stillbirths, but recent work by Hagel et al36 indicates that there may be misclassification of very early deaths as stillbirth in the study setting, which may result in our data underestimating NBU mortality rates. Lastly, we excluded the ‘outborn neonates’ in morbidity and mortality analysis due the variable admission policies for these neonates across hospitals; thus, our data cannot be generalised to the whole population of the NBUs. However, what we have done is to highlight that hospitals must be prepared to provide high-quality care for large numbers of small and sick neonates, but often they are not.20 22 29 36
In conclusion, neonatal fatality is unacceptably high, but the varied KMC and NBU admission policies challenge comparison of fatality rates across facilities. The fact that almost all the deaths are preventable offers opportunities to improve newborn survival. Care of the small and sick newborn must be included in efforts to strengthen peripartum care, and this goes well beyond neonatal resuscitation, which is the current ‘signal newborn care function’ for comprehensive emergency obstetric and newborn care training. Nevertheless, in view of the high burden of intrapartum-related complications, more work needs to be done on neonatal resuscitation training of the care providers in these health facilities. Neonates comprise almost half of all admissions and two-thirds of the deaths in the paediatric age group (0–13 years) in the hospitals studied. Thus, there is need for governments to prioritise newborn care services and appropriate resources and staffing as a key element of evidence-based hospital care.15 Intentional efforts must therefore be made to support generation and capture of high-quality patient data that are credible, complete, analysable and provide opportunities for learning. CIN has enabled generation and use of local clinical information and aims to promote adoption of better practices and wider health system improvements and performance monitoring as part of efforts to reduce neonatal mortality.11