The Etiology of Pneumonia in HIV-uninfected Children in Kilifi, Kenya: Findings From the Pneumonia Etiology Research for Child Health (PERCH) Study.

BACKGROUND: In the 1980s, Streptococcus pneumoniae and Haemophilus influenzae were identified as the principal causes of severe pneumonia in children. We investigated the etiology of severe childhood pneumonia in Kenya after introduction of conjugate vaccines against H. influenzae type b, in 2001, and S. pneumoniae, in 2011. METHODS: We conducted a case-control study between August 2011 and November 2013 among residents of the Kilifi Health and Demographic Surveillance System 28 days to 59 months of age. Cases were hospitalized at Kilifi County Hospital with severe or very severe pneumonia according to the 2005 World Health Organization definition. Controls were randomly selected from the community and frequency matched to cases on age and season. We tested nasal and oropharyngeal samples, sputum, pleural fluid, and blood specimens and used the Pneumonia Etiology Research for Child Health Integrated Analysis, combining latent class analysis and Bayesian methods, to attribute etiology. RESULTS: We enrolled 630 and 863 HIV-uninfected cases and controls, respectively. Among the cases, 282 (44%) had abnormal chest radiographs (CXR positive), 33 (5%) died in hospital, and 177 (28%) had diagnoses other than pneumonia at discharge. Among CXR-positive pneumonia cases, viruses and bacteria accounted for 77% (95% CrI: 67%-85%) and 16% (95% CrI: 10%-26%) of pneumonia attribution, respectively. Respiratory syncytial virus, S. pneumoniae and H. influenza, accounted for 37% (95% CrI: 31%-44%), 5% (95% CrI: 3%-9%), and 6% (95% CrI: 2%-11%), respectively. CONCLUSIONS: Respiratory syncytial virus was the main cause of CXR-positive pneumonia. The small contribution of H. influenzae type b and pneumococcus to pneumonia may reflect the impact of vaccine introductions in this population.

Hypothetical performance of syndrome-based management of acute paediatric admissions of children aged more than 60 days in a Kenyan district hospital.

OBJECTIVE: To investigate whether the outpatient, syndrome-based approach of the Integrated Management of Childhood Illness (IMCI) protocol could be extended to the inpatient arena to give clear and simple minimum standards of care for poorly resourced facilities. METHODS: A prospective, one-year admission cohort retrospectively compared hypothetical performance of syndrome-based management with paediatrician-defined final diagnosis. Admission syndrome definitions were based on local adaptations to the IMCI protocol that encompassed 20 clinical features, measurement of oxygen saturation, and malaria microscopy. FINDINGS: After 315 children with clinically obvious diagnoses (e.g. sickle cell disease and burns) were excluded, 3705 admission episodes were studied. Of these, 2334 (63%) met criteria for at least one severe syndrome (mortality 8% vs <1% for "non-severe" cases), and half of these had features of two or more severe syndromes. No cases of measles were seen. Syndrome-based treatment would have been appropriate (sensitivity >95%) for severe pneumonia, severe malaria, and diarrhoea with severe dehydration, and probably for severe malnutrition (sensitivity 71%). Syndrome-directed treatment suggested the use of broad-spectrum antibiotics in 75/133 (56% sensitivity) children with bacteraemic and 63/71 (89% sensitivity) children with meningitis. CONCLUSIONS: Twenty clinical features, oxygen saturation measurements, and results of malaria blood slides could be used for inpatient, syndrome-based management of acute paediatric admissions. The addition of microscopy of the cerebrospinal fluid and haemoglobin measurements would improve syndrome-directed treatment considerably. This approach might rationalize admission policy and standardize inpatient paediatric care in resource-poor countries, although the clinical detection of bacteraemia remains a problem.