Utility of health facility-based malaria data for malaria surveillance.

Currently, intensive malaria control programs are being implemented in Africa to reduce the malaria burden. Clinical malaria data from hospitals are valuable for monitoring trends in malaria morbidity and for evaluating the impacts of these interventions. However, the reliability of hospital-based data for true malaria incidence is often questioned because of diagnosis accuracy issues and variation in access to healthcare facilities among sub-groups of the population. This study investigated how diagnosis and treatment practices of malaria cases in hospitals affect reliability of hospital malaria data.The study was undertaken in health facilities in western Kenya. A total of 3,569 blood smears were analyzed after being collected from patients who were requested by clinicians to go to the hospital's laboratory for malaria testing. We applied several quality control measures for clinical malaria diagnosis. We compared our slide reading results with those from the hospital technicians. Among the 3,390 patients whose diagnoses were analyzed, only 36% had clinical malaria defined as presence of any level of parasitaemia and fever. Sensitivity and specificity of clinicians' diagnoses were 60.1% (95% CI: 61.1-67.5) and 75.0% (95% CI: 30.8-35.7), respectively. Among the 980 patients presumptively treated with an anti-malarial by the clinicians without laboratory diagnosis, only 47% had clinical malaria.These findings revealed substantial over-prescription of anti-malarials and misdiagnosis of clinical malaria. More than half of the febrile cases were not truly clinical malaria, but were wrongly diagnosed and treated as such. Deficiency in malaria diagnosis makes health facility data unreliable for monitoring trends in malaria morbidity and for evaluating impacts of malaria interventions. Improving malaria diagnosis should be a top priority in rural African health centers