Evaluation of the ARI program: a health facility survey in Simbu, Papua, New Guinea.

The purpose of this health facility survey was to evaluate how the ARI (Acute Respiratory Infection) Program actually works in an everyday, non-research setting. We surveyed 33 clinics and aid posts, including 223 children with ARI and 104 health workers. In this primary health care setting, health workers diagnosed 37% of ARI cases as pneumonia, compared to 69% in the same children assessed independently by trained ARI surveyors using Papua New Guinea case management, which defines fast breathing as > or = 40 per minute for children 1 month to 5 years of age. Agreement between health workers and surveyors was reasonably good (kappa > or = 0.6) for the history of symptoms, but was poor (kappa < 0.3) for diagnoses, treatments and signs such as respiratory rate (RR) and chest indrawing. Health workers counted the RR in only 14% of cases in the survey. In essence, we found that health workers in Simbu are not practising ARI case management. We conclude that case management guidelines which define fast breathing as a rate of > or = 40 per minute classify too many obviously well children as pneumonia. Furthermore, we encountered difficulties in measuring RR accurately, and documented marked inter-observer variation in this setting. Consequently, we are concerned about the ARI Program's excessive reliance on RR and rigid protocols at the expense of clinical sense. Case management guidelines developed for aid post orderlies or village health workers may need to be modified, or used differently by experienced nurses at health centres, who recognize sick children better than by following a standardized protocol. Despite better overall ARI knowledge and practice by nurses, we could demonstrate an impact of the ARI Training Program only on community health workers (CHWs). In order to improve ARI clinical practice, we recommend that the ARI Program in PNG initiates regular on-site clinical supervision of nurses and CHWs at health centres.