Table 1

Summary of key hospital characteristics and important illness specific research findings

District/county hospitals—the context
Network hospitals were purposefully selected to include counties in high and low malaria endemic settings with moderate to high workloads of between 1300 and 3500 paediatric ward admissions per year.6 Pneumonia, diarrhoea with dehydration, suspected meningitis, severe anaemia and severe malnutrition are major causes of admission with comorbidity common, including with malaria and in some settings clinically diagnosed rickets.6 57 Laboratories offer limited diagnostic capacity, malaria smears and HIV rapid testing are typically available but simple bedside glucose or urine testing or access to pulse oximetry are often unavailable with almost no access to reliable culture of blood or cerebrospinal fluid microbiology.5 6 More specific work demonstrates good access to printed national paediatric and neonatal guidelines but overcrowding and considerable weaknesses in organisation of and material for infection prevention and control in a number of sites.58 59 Although larger county hospitals were selected for the network several had no paediatrician in charge of services for long periods and most had high patient to nurse ratios (10–20 patients per nurse being common).
Inpatient paediatric care—condition specific insights (total admissions in the studies outlined ranged from 40 000 to 87 000 from which condition specific populations were 16 162 and 1832 for pneumonia in ages 2–59 m and 5–13 years respectively, 8562 and 7657 in two studies of diarrhoea and dehydration, 5766 and 13 104 in two studies of malaria, and of 622 for a study on shock and 5306 on severe acute malnutrition)
Pneumonia18 57Although based on admission data only, analyses of data from 17 000 children with pneumonia suggested lower-chest wall indrawing is likely to be a risk factors for poor outcome. Findings prompted policy discussions in Kenya with local adaptation of WHO guidance and ongoing discussion around WHO policy.26Those aged 5–13 years represent 20% of all medical admissions with pneumonia responsible for 12% admissions of these and with mortality 8%, there are no evidence-based guidelines for this population.
Malaria21 22In 5 hospitals with large numbers of clinically diagnosed malaria admissions 62% had a confirmatory test result but 69% with a negative result were prescribed antimalarials, most of whom met criteria for severe illness, and only 3.5% cases with an initial negative test had a repeat test as guidelines recommend. In 4 hospitals malaria admissions were further characterised in preparation for evaluating introduction of the malaria vaccine, 40.6% children had severe malaria with a case fatality rate 7.0% and median age 33 months.
Diarrhoea and dehydration19 20Diarrhoea and dehydration remain common causes of admission. Severe illness has a mortality rate of 9%. However, while signs used to diagnose dehydration help guide fluid therapy mortality is more strongly linked to age <12 m and additional signs of severe illness including abnormal circulatory, respiratory or neurological signs. Analyses suggest that mortality is lower in cases where fluids are correctly prescribed and that feedback is associated with documentary evidence of better case management.
Severe acute malnutrition23Severe Acute Malnutrition (SAM) is important in all settings with prevalence and outcomes amongst admissions aged 1–59 m varying from 4.6% to 18.2% and 6% to 28.6%, respectively. Feedback seems to promote use of key anthropometric assessments in diagnosis and while indicators suggest adherence to 5 of the 10 recommended steps in case management is often quite poor there was some evidence it was associated with in small to modest improvements in case management.
Shock60Trials indicated unexpected harm from iv fluid bolus in children with severe febrile illness but no dehydration. Amongst admissions with an overall mortality of 5% fluid bolus was rarely used (0.85%) and 89% boluses given were for severe dehydration; data suggest harm from overuse of boluses is likely to be uncommon where guidelines are clear.