Care Experiences and Visit Use of Physician and Non-Physician Medicare Patients

A central concept in health economics is that physicians act as agents on behalf of patients, imparting their knowledge about medicine and the health care system to guide patients to optimal care.¹ Comparisons of care received by physicians when they are patients with care received by non-physician patients can help characterize the advantages of being highly informed, and thus the potential gains from better informing or guiding patients. Building on research examining physicians as uniquely informed patients,^2-4 we compared care experiences and outpatient visit use between physician patients and comparable non-physician patients.

Methods

We analyzed survey and linked enrollment data for respondents to the 2010-2018 fee-for-service Medicare Consumer Assessment of Healthcare Providers and Systems (CAHPS) survey.⁵ We identified respondents who had billed Medicare as a solo practitioner physician (eMethods). To limit differences between physician and non-physician patients, we restricted the sample to respondents reporting high education (more than college) and without end-stage renal disease, disability, dual/Medicaid eligibility, or help completing the survey. All but 3 physician patients met these criteria.

To facilitate interpretation, we dichotomized patient experience item responses into indicators of the highest possible rating (eTable 1). Primary outcomes were 1) a composite of item-specific indicators across five domains (eMethods), and 2) self-reported visits with primary physicians and specialists. We used linear regression to estimate differences in outcomes between physician and non-physician patients, controlling for county of residence and sex (88.3% of physician patients were male). We adjusted for additional sociodemographic and clinical characteristics (Table 1) to gauge confounding (eMethods). The study followed STROBE guidelines and was approved by the Harvard Faculty of Medicine IRB.

Table 1.

Characteristics of Physician Patients and Non-Physician Patients with More Than a College Education in the Study Sample, Adjusted for County and Sex^a

Characteristic	Physician Patients N=239	Other High- Education Patients N=45,855	Standardized Difference (SDs)
Age, years	72.5 ± 5.8	73.8 ± 6.7	−0.19
Self-Reported race and ethnicity, % b
White, non-Hispanic	78.5	87.1	−0.26
Other race or ethnicity c	21.5	12.9	0.26
Chronic conditions d
No. of conditions	3.4±3.4	4.3 ± 37	−0.26
Functional limitations, %
Difficulty with ≥ 1 ADL	19.7	20.7	−0.03
Difficulty with ≥ 2 ADLs	15.5	13.5	0.06

Abbreviations: SD, standard deviation; ADL, activities of daily living.

^aPlus-minus values represent mean ± standard deviation. Means and percentages were adjusted for county of residence and patient sex to reflect comparisons within geographic areas and balanced on sex and weighted using CAHPS survey weights. Adjustment for sex was included in baseline comparisons as 88.3% of physician patients were male. The full set of covariates adjusted for in the main analysis were county of residence, sex, age, race, chronic condition indicators, smoking status, and functional limitations. However, because cell sizes for smokers and some race categories among physician patients were smaller than required by CMS privacy and reporting requirements, we omitted smoking status from reporting and collapsed self-reported race into White, non-Hispanic vs. other race or ethnicity to ensure cell sizes remained in accordance with CMS requirements.

^bParticipant race and ethnicity was identified using designations self-reported by CAHPS survey respondents. We included race and ethnicity in the analysis to assess balance in patient characteristics between physician and non-physician patients and adjust for any differences.

^cThe full set of racial and ethnic categories included in the main analysis were American Indian or Alaska Native; Asian or Pacific Islander; Black or African-American; Hispanic or Latino; White, non-Hispanic; and Other or Unknown.

^dSee eMethods for the full list of 27 chronic conditions from the Chronic Conditions Data Warehouse.

Results

The sample included 239 physician and 45,855 non-physician patients (Table 1). Survey response did not differ significantly between groups (−3.1%; P=0.20). In adjusted comparisons, physician patients rated their care substantially higher than non-physician patients (overall composite difference: 9.9 percentage points [pp]; 95% CI: 5.2, 14.6; P<0.001). Differences ranged from 6.0-14.2pp across domains (7.6-39.4% of sample means) (Figure 1). These adjusted differences were slightly larger than those adjusted only for county and sex. Physician patients reported substantially fewer visits to their primary physician over 6 months (−0.62 visits/beneficiary; 95% CI; −0.79,−0.44; P<0.001) and modestly fewer visits to specialists (−0.23 visits/beneficiary; 95% CI; −0.39,−0.09; P=0.002).

Figure 1. Differences in Care Experiences and Visit Use Between Physician Patients and Non-Physician Patients with More Than a College Education.

Differences in care experiences and visit use between physician patients and non-physician patients with more than a college education are plotted with 95% confidence intervals (error bars) after adjustment for county of residence and sex (Model 1) and the additional covariates in Table 1 (Model 2). Analyses were weighted using CAHPS survey weights and used a robust variance estimator to account for clustering within counties. Unadjusted means are provided for each measure for reference. Estimates for care experiences (Panel A) describe the difference in the proportion of patients, in percentage points, who reported the highest rating on the scale for a survey item. The overall composite rating averages non-missing ratings across items or subcomposites after subtracting the mean for each. Model 2 adjusted differences for patient experience outcomes were 4.4%-21.4% larger than differences adjusted only for county and sex (Model 1). Estimates for outpatient visit use (Panel B) describe the differences in number of reported visits that occurred during the six months preceding the survey. Model 2 adjusted differences for outpatient visit use were 14.3%- 33.7% smaller than differences adjusted only for county and sex (Model 1).

Results were substantively similar when relaxing inclusion criteria to include all non-physician patients, thereby making the comparison group’s health status worse (overall composite difference: 10.3 pp; 95% CI: 5.7,15.0; P<0.001) and when analyzing CAHPS scores on a 0-10 scale (overall composite difference: 0.46; 95% CI: 0.27,0.65; P<0.001). In a falsification analysis of non-physician patients to assess whether differences were specific to health care knowledge, high education was associated with worse care experiences (−3.1pp; P<0.001) and more visits with primary physicians (0.03; P<0.001) and specialists (0.33; P<0.001).

Discussion

These findings suggest that the more extensive health care knowledge possessed by physician patients may facilitate substantially better care experiences, including higher overall ratings, more effective physician-patient communication, and better understanding of how and when to access care. Better care experiences were observed uniquely for physician patients and not for other highly educated patients relative to those with less education. Physician patients’ visit use – substantially lower for primary physicians but modestly lower for specialists – is also consistent with information differences (more consistently present for general than sub-specialized medical knowledge) influencing care differences.

Other mechanisms may have contributed, though robustness checks suggested health differences were not explanatory. Physician patients may rate their physicians highly out of sympathy or professional courtesy, but such affinity biases would not explain their lower visit use. Alternatively, physician patients may be better equipped to navigate the system and identify higher-quality or better-matched providers. Notably, this explanation still invokes deeper health care knowledge.

Our findings suggest ample room for improving care experiences and efficiency. Although we cannot expect patients to be as knowledgeable as physicians, strategies to help physicians and organizations better inform and guide patients may improve care efficiency and experiences.