Incidence of childhood pneumonia: facility-based surveillance estimate compared to measured incidence in a South African birth cohort study

BACKGROUND Pneumonia is the leading cause of childhood mortality and a major contributor to childhood morbidity, but accurate measurement of pneumonia incidence is challenging. We compared pneumonia incidence using a facility-based surveillance system to estimates from a cohort study conducted contemporaneously in the same community in Cape Town, South Africa. METHODS A surveillance system was developed in six public sector primary care clinics and in a regional referral hospital, to detect childhood pneumonia cases. Nurses recorded all children presenting to facilities who met WHO case definitions of pneumonia, and hospital records were reviewed. Estimates of pneumonia incidence and severity were compared with incidence rates based on active surveillance in the Drakenstein Child Health Study. RESULTS From June 2012 until September 2013, the surveillance system detected 306 pneumonia episodes in children under 1 year of age, an incidence of 0.20 episodes/child-year (e/cy) (95% CI 0.17 to 0.22 e/cy). The incidence in the cohort study from the same period was 0.27 e/cy (95% CI 0.23 to 0.32 e/cy). Pneumonia incidence in the surveillance system was almost 30% lower than in the birth cohort; incidence rate ratio 0.72 (95% CI 0.58 to 0.89). In the surveillance system, 18% were severe pneumonia cases, compared to 23% in the birth cohort, rate ratio 0.81 (95% CI 0.55 to 1.18). CONCLUSIONS In this setting, facility-based pneumonia surveillance detected fewer cases of pneumonia, and fewer severe cases, compared to the corresponding cohort study. Facility pneumonia surveillance using data collected by local healthcare workers provides a useful estimate of the epidemiology of childhood pneumonia but may underestimate incidence and severity.