Related Experiment Videos
Comparison of the entropy technique with two other techniques for detecting disease clustering using data from
J B Swartz1, S J Rothenberg, S Teklehaimanot
1Charles R. Drew University of Medicine and Science, Research Centers in Minority Institutions/Epidemiological and Statistical Research Group, Los Angeles, CA 90059, USA.
Abstract:
The entropy technique was compared with two other case-control techniques for detecting disease clustering using data on blood lead levels of children who were patients at the King/Drew Medical Center in South-Central Los Angeles in 1991 to 1994. The other two methods are the nearest neighbor technique (NNT) and Moran's IPOP technique, a variation of Moran's I test, in which rates are adjusted for population size. Four different blood lead levels (15 microg/dl, 20 microg/dl, 30 microg/dl, 35 microg/dl) were used as cutoff levels to designate cases. Persons with blood lead levels greater than or equal to the cutoff level were designated as cases. The authors found significant clustering for all four cutoff levels using the entropy method, and for the first three cutoff levels using the NNT. They found significant clustering with Moran's IPOP for some scales for two of the cutoff levels. While performance of the entropy technique and the NNT were independent of scale, that of Moran's IPOP was highly scale-dependent.