P O'Campo1, B Guyer. 1. Department of Population and Family Health Sciences, Johns Hopkins School of Hygiene and Public Health, Baltimore, Md., USA.
Abstract
OBJECTIVES: Monitoring health in small localities such as cities or local communities is important, because rates of adverse outcomes often vary widely by geographic area. This article explores the utility of CUSUM (cumulative summation), a method developed and refined in industry, for monitoring health outcomes in cities and smaller geographic areas. METHODS: CUSUM monitoring methods were applied to rates of late or no prenatal care initiation and very low birthweight for the city of Baltimore as a whole and for a cluster of high-risk areas within the city. The performance of supplementary runs criteria was also assessed. The ability of both methods to flag significant increases or decreases in prenatal care initiation and very low birthweight rates was assessed. RESULTS: CUSUM and runs criteria detected most significant rate changes. The 2 methods performed better in regard to outcomes with higher prevalence and in larger geographic areas. CONCLUSIONS: CUSUM methods are convenient and reliable for use in the monitoring of moderately low prevalence outcomes in small geographic areas. Future research should examine their applicability to other health outcomes and further refine these methods, especially for rarer outcomes.
OBJECTIVES: Monitoring health in small localities such as cities or local communities is important, because rates of adverse outcomes often vary widely by geographic area. This article explores the utility of CUSUM (cumulative summation), a method developed and refined in industry, for monitoring health outcomes in cities and smaller geographic areas. METHODS: CUSUM monitoring methods were applied to rates of late or no prenatal care initiation and very low birthweight for the city of Baltimore as a whole and for a cluster of high-risk areas within the city. The performance of supplementary runs criteria was also assessed. The ability of both methods to flag significant increases or decreases in prenatal care initiation and very low birthweight rates was assessed. RESULTS: CUSUM and runs criteria detected most significant rate changes. The 2 methods performed better in regard to outcomes with higher prevalence and in larger geographic areas. CONCLUSIONS: CUSUM methods are convenient and reliable for use in the monitoring of moderately low prevalence outcomes in small geographic areas. Future research should examine their applicability to other health outcomes and further refine these methods, especially for rarer outcomes.