Practice-Level Clustering of Industry Payments to Clinicians

Abstract

Objective: To test whether industry payments to clinicians are clustered at the level of the medical practice.

Study setting and design: We performed a cross-sectional study of clinicians who billed Medicare Part B in 2021 to test whether the receipt of an industry payment, log total value of industry payments, or log total number of industry payments to clinicians were clustered at the level of the medical practice. We used mixed effects linear regression to analyze practice-level clustering, controlling for clinician sex, age, urbanicity, state, and specialty, as well as practice size and specialty.

Data source and analytic sample: We used the 2021 Medicare Data on Provider Practice and Specialty file to assign clinicians to medical practices, and the 2021 General Payment Data from the Open Payments Program to calculate the total value and number of industry payments to each clinician.

Principal findings: We identified 996,982 clinicians who billed Medicare Part B in 2021, of whom 679,577 (68.2%) were physicians and 317,305 (31.8%) were advanced practice clinicians. These clinicians worked across 109,952 medical practices. In total, 474,312 (47.6%) clinicians received an industry payment in 2021. The average total value of industry payments was $1497 (SD $54,823), and the average total number of industry payments was 9.4 (SD 27.5). Regression analysis of each outcome identified significant clustering at the level of the medical practice, including 24.8% of the variation in the receipt of an industry payment, 36.8% in the log total value of industry payments, and 60.5% in the log total number of industry payments.

Conclusions: Industry payments to clinicians are strongly clustered by medical practice. Future research should examine the role of the medical practice in facilitating financial conflicts of interest between industry and clinicians.

Keywords: advanced practice clinicians; clustering; conflicts of interest; industry payments; medical practice; physicians.

FIGURE 1

Decreasing clustering in industry payments at the level of the medical practice as practice size increases (N = 996,882). TINs are units of financial organizations [14]. They are not necessarily single‐site practices and may instead be multisite practices (e.g., medical group) or multiple practices at a single site (e.g., academic medical center). As TINs with fewer physicians should more often be single‐site practices, and as the clustering of industry payments should be greatest in single‐site practices due to the nature of pharmaceutical representative visits (e.g., catered lunch) [9], we repeated the mixed effects regression analysis for each outcome in four strata defined by practice size.

FIGURE 2

Clustering for food and beverage and not food and beverage industry payments at the level of the medical practice (N = 996,882). We did not anticipate a meaningful difference in the ICCs for log total value and number of industry payments (0.368 [SE 0.002] and 0.605 [SE 0.002], respectively; Table 3). We hypothesized that the difference in the ICCs was the consequence of low frequency, high value payments whose nature was not food and beverage. To test this hypothesis, we stratified the outcomes for each clinician by the nature of the payment—food and beverage versus not food and beverage—and we repeated the mixed effects regression analysis.