Defining and Validating Criteria to Identify Populations Who May Benefit From Home-Based Primary Care.

Background: Home-based primary care (HBPC) is an important care delivery model for high-need older adults. Currently, target patient populations vary across HBPC programs, hindering expansion and large-scale evaluation.

Objectives: Develop and validate criteria that identify appropriate HBPC target populations.

Research Design: A modified Delphi process was used to achieve expert consensus on criteria for identifying HBPC target populations. All criteria were defined and validated using linked data from Medicare claims and the National Health and Aging Trends Study (NHATS) (cohort n=21,727). Construct validation involved assessing demographics and health outcomes/expenditures for selected criteria.

Subjects: Delphi panelists (n=29) represented diverse professional perspectives. Criteria were validated on community-dwelling Medicare beneficiaries (age ≥70) enrolled in NHATS.

Measures: Criteria were selected via Delphi questionnaires. For construct validation, sociodemographic characteristics of Medicare beneficiaries were self-reported in NHATS, and annual health care expenditures and mortality were obtained via linked Medicare claims.

Results: Panelists proposed an algorithm of criteria for HBPC target populations that included indicators for serious illness, functional impairment, and social isolation. The algorithm's Delphi-selected criteria applied to 16.8% of Medicare beneficiaries. These HBPC target populations had higher annual health care costs [Med (IQR): $10,851 (3316, 31,556) vs. $2830 (913, 9574)] and higher 12-month mortality [15% (95% CI: 14, 17) vs. 5% (95% CI: 4, 5)] compared with the total validation cohort.

Conclusions: We developed and validated an algorithm to define target populations for HBPC, which suggests a need for increased HBPC availability. By enabling objective identification of unmet demands for HBPC access or resources, this algorithm can foster robust evaluation and equitable expansion of HBPC.