Mortality of sick children after outpatient treatment at first-level health facilities in rural western Kenya

Objectives: (1) To determine whether mortality rates were raised in sick children in the 30 days after visiting first-level health facilities in an area under demographic surveillance in western Kenya, (2) to identify the types of illnesses associated with increased mortality and (3) to estimate the effectiveness of appropriate treatment. Methods: All sick children (2-59 months of age) who attended one of the seven participating first-level health facilities from May to August 2003 were identified. A standardized mortality ratio was computed to compare their mortality rate in the 30 days after a sick visit with that of the community under active demographic and health surveillance. A multivariate Cox Proportional Hazards model was used to identify illnesses associated with death and to estimate the protective effectiveness of appropriate treatment for potentially life-threatening diseases. Results: A total of 1383 eligible children made 1697 sick visits; 33 (2.4%) died within 30 days. Compared with children 2-59 months in the general population, sick children had a 5.3 times greater mortality rate [95% confidence interval (CI) 3.8-7.5]. In a multivariate survival analysis, significant risk factors for mortality included age <24 months [Hazard Ratio (HR) 4.4, 95% CI 1.5-12.6], malnutrition (HR 15.5, 95% CI 6.1-39.8), severe pneumonia (HR 12.9, 95% CI 3.0-56.4) and anaemia (HR 3.3, 95% CI 1.5-7.2). Appropriate treatment for a child's most severe illness reduced mortality by 78% (95% CI 57-89%). Conclusion: We estimate that improvements in diagnosis and appropriate treatment at first-level health facilities for children 2-59 months could reduce overall under-5 mortality in the area by 12-14%.