The application of multidimensional scaling methods to epidemiological data

Identifying at-risk communities for action on cancer prevention: a case study in new brunswick (Canada) communities

Health statistics reported by large geographic area such as province, state, county or health region offer little insight into disease conditions at the community level where people live and work, where occupational and environmental exposures occur, and where industrial emissions are often concentrated. This study investigated overall patterns of cancer incidence and socioeconomic status (SES) among 14 communities in the province of New Brunswick (Canada). A multivariate ordination technique, hierarchical clustering, and permutation procedures were used to identify and test significance of community clusters and whether the overall pattern of SES was correlated with patterns of cancer among communities. Communities with significantly high or significantly low overall rates of cancers were identified, patterns that were not related to SES. The potential influence of age, small populations, diagnostic screening, smoking and environmental risk factors contributing to locally elevated cancer rates are discussed. Cancer incidence reported at smaller spatial scales provides health officials and researchers with a basis for identifying communities potentially at-risk and aids in the development of appropriate community-based risk reduction actions and cancer prevention.

A monitoring system for detecting aberrations in public health surveillance reports

Routine analysis of public health surveillance data to detect departures from historical patterns of disease frequency is required to enable timely public health responses to decrease unnecessary morbidity and mortality. We describe a monitoring system incorporating statistical 'flags' identifying unusually large increases (or decreases) in disease reports compared to the number of cases expected. The two-stage monitoring system consists of univariate Box-Jenkins models and subsequent tracking signals from several statistical process control charts. The analyses are illustrated on 1980-1995 national notifiable disease data reported weekly to the Centers for Disease Control and Prevention (CDC) by state health departments and published in CDC's Morbidity and Mortality Weekly Report. Published in 1999 by John Wiley & Sons, Ltd. This article is a U.S. Government work and is in the public domain in the United States.