Techniques for defining geographic boundaries for health regions

Many federal and state programs require the geographic partitioning of states into regions for health services planning, monitoring, and/or administration. A common consideration for such programs is that region boundaries should be drawn so as to maximize the proportion of the state's population that receives health care services in its region of residence. Defining region boundaries thus may be viewed as a problem of partitioning a set of N small areal units (e.g. counties) into M subsets (regions) so as to minimize interactions (patient flow) among subsets. This paper describes three algorithms for region design and compares them in terms of computer-processing efficiency and solution value based on results from a number of test cases. Application of two of the algorithms, one based on the greedy heuristic and the other incorporating a max-flow/min-cut procedure, to a problem of dividing a metropolitan region into separate service areas for clusters of hospitals is also described.