For this project, we compiled data for 14 project sites. It is important to recognize there are differences in the populations living in each site, as well as in the sites’ health service systems and data, when examining project findings. In this section of the report, we describe such differences and our approach for addressing them.
The populations living in each site differ with respect to age, income, disease burden, and other characteristics. Features such as location (i.e., urban, rural), geographic size, patient transportation times for accessing services, and proximity to I/T health facilities and non-IHS facilities also differ across the 14 sites.
The health service systems within each project site also vary by organizational type (i.e., IHS or Tribal) and the type and level of inpatient and outpatient services provided. For example, smaller hospitals provide more limited inpatient services and may not provide obstetrical services. At many project sites, the IPC program was implemented. With IPC, patients are empanelled to integrated primary care provider teams that may include physicians, mid-level providers, nurses, case managers, nutritionists, and pharmacists. In addition, types and modalities (e.g., teleradiology, telenephrology) for the provision of specialty services also vary. These differences influence the cost of providing services within each project site. Tribal health providers may differ from IHS providers in a number of ways, including their target populations. For example, a Tribal health program may seek to provide health services for AI/ANs living in the project site (i.e., Service Unit) as well as for AI/ANs from the Tribe who live outside of the IHS Service Unit. In contrast, a Tribal health program may focus the provision of specific services in geographic areas that are smaller than the IHS Service Unit.
Finally, the types and quality of data vary across sites. For example, not all data stored in computer systems at the project sites are extracted and included in the NDW. Project sites may vary with respect to how certain types of services are recorded in their data systems. Differences may also exist in the quality of the data. For example, the number of diagnostic codes recorded in a health utilization record may vary, influencing health status measures derived from the diagnostic codes.
It is important to be mindful of these differences and how they may bias project findings. In some situations, data for specific sites (e.g., laboratory or pharmacy data) may have been excluded from some analyses. In regression analyses, statistical approaches may be used to control for differences across project sites.
