Decision Support Analysis

A Part of the Q-DART Toolset

The Q-DART project is developing several tools that go beyond identifying areas or "hotspots" of low quality of care; these tools combine information on health care performance, neighborhood characteristics, and scale of intervention and economic efficiency to identify "actionable hotspots." The intent is to better characterize areas of low performance using rich demographic and socioeconomic data from the US Census Bureau and other sources such that more-effective interventions can be designed, while also providing means to prioritize areas for intervention given limited resources and often competing goals. Geostatistical constructs are employed to further focus attention on subgroups of interest, for example, the area accounting for the majority of Medicare diabetics within a county.

Also part of its decision support analysis efforts, Q-DART is developing tools that employ cluster methods to explore member data at fine spatial scale. One tool under development employs nearest-neighbor search (NNS) methods, among the simplest and most common methods used to reveal spatial clusters. By varying search distance, "hotspots" at different scales emerge among the dataset. Their scale has direct relevance for the scale and type of intervention that an organization may choose to undertake. Decision makers, for example, might be interested in efficient allocation of resources in program design, preferring to target first those neighborhoods with the greatest concentration of members not being screened for diabetes tests. Results of the NNS-based tool can inform such decisions.