Physicians payment in the United States between 2014 and 2018: An analysis of the CMS Open Payments database

Abstract

The Open Payments database reports payments made to physicians by industry. Given the potential for financial conflicts of interest relating to patient outcomes, further scrutiny of these data is valuable. Therefore, the objective of this study was to analyze physician-industry relationships by specialty type, payment type, geospatial trend, and longitudinal trend between 2014-2018. We conducted an observational, retrospective data analysis of payments from the Open Payments database for licensed United States physicians listed in the National Plan & Provider Enumeration System (NPPES). Datasets from 2013-2018 were joined using the Python programming language. Aggregation and sub-setting by characteristics of interest was done in R to calculate means and frequencies of reported general physician payments from industry across different specialties, locations, timeframes, and payment types. Normalization was applied for numbers of physicians or payments. Geospatial statistical hot spot analysis was conducted in ArcGIS. 51.73 million payment records were analyzed. In total, 50,047,930 payments were issued to 771,113 allopathic or osteopathic physicians, representing $8,702,631,264 transferred from industry to physicians over the five-year period between 2014 and 2018. The mean payment amount was $179, with a standard deviation of $12,685. Variability in physicians' financial relationships with industry were apparent across specialties, regions, time, and payment type. A limited match rate between records in the NPPES and Open Payments databases may have resulted in selection bias of trends related to physician characteristics. Further research is necessary, particularly in the context of changing industry payment trends and public perceptions of the appropriateness of these relationships.

Fig 1. In blue-red gradient, results of a hot spot analysis using zip code averages for payment amount per physician in the contiguous United States.

Hatched overlay represents zip codes where hot spot reached statistical significance, and gray zip codes are those with no data. Cartographic boundaries obtained from the U.S. Census Bureau and reprinted with permission in accordance with OMB Memorandum M-10-06 and Executive Order 13642.

Fig 2. In blue-red gradient, annual variation in geospatial clustering of average individual payments per zip code, with hatched overlay representing statistical significance.

Cartographic boundaries obtained from the U.S. Census Bureau and reprinted with permission in accordance with OMB Memorandum M-10-06 and Executive Order 13642.