How should safety-net ambulatory practices be defined—and how do different definitions compare?

Two ways to define safety-net ambulatory practices—by share (institution based) vs volume (population based) of underserved patients dually eligible for Medicare and Medicaid—identify almost completely different groups of ambulatory practices.

In a cross-sectional analysis of 2022 Medicare data, we examined 106,545 ambulatory practices serving 967,820 dually eligible beneficiaries to compare institution-based and population-based definitions of safety-net practices.

Using an institution-based definition (>80% of services delivered to dually eligible patients), 7,338 practices (6.9%) were identified as safety-net. In contrast, using a population-based definition (practices accounting for a cumulative 80% of total services to dually eligible patients nationwide), only identified 302 practices (0.3%) as safety-net. Strikingly, just 5 practices (0.005%) overlapped between the two definitions.

The characteristics of practices identified by each definition also differed substantially. Institution-based safety-net practices were smaller, with fewer clinicians and Medicare beneficiaries, but served a much higher proportion of dually eligible patients (89.6% vs 17.7%). In contrast, population-based safety-net practices were larger and delivered a high absolute volume of care to dually eligible individuals despite a lower proportional share. Patients served by institution-based practices were younger (64 vs 73 years old) and more likely to be Black (15.4% vs 7.4%) or Hispanic (22.2% vs 4.8%).

Institution-based definitions using patient share may better identify practices at financial risk and in need of support, while population-based definitions using an absolute volume may better capture where most underserved patients actually receive care.

Future policy efforts should align safety-net definitions with specific goals—whether supporting financially vulnerable practices or reaching the largest number of underserved patients.

Article link: https://jamanetwork.com/journals/jama-health-forum/fullarticle/2847568