Local clustering in breast, lung and colorectal cancer in Long Island, New York

Methods Comparisons

Martin Kulldorff   (2003-02-28 04:43)  University of Connecticut

In geographical disease surveillance it is sometimes of interest to study disease incidence and mortality in different but overlapping geographical areas. For example, if a geographical cluster is found in a particular location, it may be of interest to study the geographical distribution within that specific location. Jacquez’ and Greiling’s study of breast cancer incidence on Long Island is therefore an interesting follow-up to the state-wide analyses done by the New York State Health Department, which had found a cluster on Long Island.

The comparison that the authors make between the spatial scan statistic and the local Moran method is not informative though. The former was applied to New York State as a whole and the latter only to Long Island. In order to compare methods, the methods need to be applied to the same data sets. While different methods will always produce different results, no conclusions about the nature of these differences can be made from the current study.

Competing interests

Dr Kulldorff is the developer of the spatial scan statistic used for the state-wide analysis.

top