Approaches like the one used in this paper have utility for
policy, practice and future research. For example, data
mining can be used to support a benchmarking approach
in order to compare home healthcare agencies. Benchmarking
initiatives have been mentioned in the literature;
for example in New Jersey where a benchmarking project
examined the Outcome and Assessment Information Set
(OASIS) and Outcome Based Quality Improvement
(OBQI) reports published by home healthcare agencies
[24]. In some ways, the publicly reported home healthcare
outcomes are benchmarks in that national results are
presented for consumers and others to use to compare a
specific agency's performance with national results,
although benchmark infers that a "correct" or "ideal" rate
has been established, which is not the case. Thus, the use
of a data mining approach could identify the dependencies
and interactions that influence outcomes so that risk
adjustment methodologies can be improved in accuracy
and actual benchmarks could be established.