TABLE 2.

Variation in AWV use at provider and clinic levels.

Source of variation in regular vs. low/moderate AWV use during 2018–2022	Proportion of variation in low vs. regular AWV use, % [95% confidence interval]	Proportion of variation in moderate vs. regular AWV use, % [95% confidence interval]
Unadjusted multilevel model
Proportion of variation at clinic level	44.5% [33.7%, 55.8%]	21.1% [14.1%, 30.3%]
Proportion of variation at provider and clinic levels	57.1% [48.2%, 65.5%]	30.1% [24.2%, 38.6%]
Adjusted multilevel model
Proportion of variation at clinic level	43.9% [31.4%, 57.2%]	19.6% [13.2%, 28.0%]
Proportion of variation at provider and clinic levels	56.4% [45.3%, 66.9%]	29.6% [23.7%, 36.3%]

Note: (1) Analysis was performed using two subcohorts. The first subcohort included those with low AWV and regular AWV use, and was used to examine the dichotomous indicator of low AWV use (1) treating regular AWV use as the reference cohort (0). The second subcohort included those with moderate AWV and regular AWV use, and was used to examine the dichotomous indicator of moderate AWV use (1) treating regular AWV use as the reference cohort (0). (2) For the unadjusted models, we used multilevel random effects logit regression models of the dichotomous indicator of low/moderate AWV use, in comparison with regular AWV use, with clustering of individuals at provider and clinic levels. We obtained the intraclass correlation at provider and provider + clinic levels. (3) For the adjusted models, we used multilevel random effects logit regression models noted above, including individual-level covariates: age, sex, race/ ethnicity, plan type (Fee for Service/Medicare Advantage), no. Elixhauser comorbidity conditions, baseline year number of primary E&M visits and specialty E&M visits, and zip code-level indicators of poverty rate, SNAP use rate, adults without high school education, proportion non-White, and nearest distance to healthcare clinic over 6 miles.

Abbreviation: AWV, Annual Wellness Visit.