Screening for Rate of Ghost Physicians in Provider Directories

ABSTRACT Objective To provide a conceptual framework for understanding ghost networks and propose a new methodology for estimating ghost physician prevalence in health plans' provider directories. Study Setting and Design We focused on providers listed as primary care physicians in Medicare Advantage (MA) plans' provider directories. Our framework categorizes ghost PCPs into general ghosts—those listed as PCPs but unavailable to any Medicare beneficiaries for primary care—and network‐specific ghosts—those available to Medicare beneficiaries but inaccessible to enrollees with a certain MA network. We identified general ghosts with multiple data sources. In estimating network‐specific ghost prevalence, to separate those who were truly unavailable from those who were accessible but saw no patients simply due to chance, we estimated a logistic model predicting being low‐volume among the zero‐volume and low‐volume PCP‐networks. Data Sources and Analytic Sample We used the 2019 Ideon MA provider directory data. For physician information, we used the National Plan and Provider Enumeration System National Provider Identifier registry and OneKey Healthcare Industry Database. To estimate the patient volume of listed PCPs, we extracted from the 2019 MA encounter data carrier file beneficiaries' primary care visits to physicians in the office, hospital outpatient, or clinic setting. Principal Findings We found that 17.5% of the listed PCPs in an average MA network were general ghosts and 11.5% were network‐specific ghosts. Health maintenance organization networks listed more ghost PCPs than preferred provider organization (30.5% vs. 26.9%). Networks associated with high star rating contracts had substantially fewer ghost PCPs than those associated with low star rating contracts (26.5% vs. 37.2%). Our methodology for screening for ghost prevalence reduces the penalty on networks offering more choice, such as those serving urban markets. Conclusions Policymakers should ensure that provider directories reflect the physicians available to provide care. Our methodology may facilitate targeted network audits.