Abstract
This cohort study uses Medicare data to assess trends and characteristics among hospitalists who shift practice to settings outside of the hospital.
Introduction
Recent decades have observed a rapid growth in the number of hospitalists.1,2 Receiving care from hospitalists has been associated with shorter length of stay and similar outcomes3 and lower readmission and mortality rates compared with care from nonhospitalist physicians who were not the patient’s primary care physician.4 Furthermore, hospitalist experience (years in practice) has been inversely correlated with patient mortality.5 Surveys indicate high turnover among hospitalists compared with other specialties,6 but those surveys measure turnover within hospitalist groups rather than the specialty overall. Our objective was to characterize hospitalists who shift practice to settings outside of the hospital.
Methods
For this cohort study, we used Medicare’s Provider Utilization and Payment Data records from 2012 to 2018, which aggregate Medicare Part B billings by service. This database was linked to physician demographic details and practice characteristics from the Medicare Data on Provider Practice and Specialty files and to county-level characteristics from the Area Health Resource Files. Generalist physicians (internal medicine, geriatrics, general practice, pediatrics, or family medicine) who billed at least 90% of visits from the hospital were considered hospitalists.4,5 Physicians with 100 visits or fewer per year were excluded.
Hospitalists were identified in 2012, and those who continued to practice as generalists were followed up through 2018 to examine the settings where they provided care. We measured the proportion of visits in the following settings (service codes in eTable in the Supplement): (1) acute care hospital; (2) office (ie, a location where patients are treated on an ambulatory basis); (3) nursing home or skilled nursing facility; and (4) other (all other settings).
We examined physician characteristics (gender, age, time from medical school graduation, medical school rank), practice characteristics (size of group, Medicare beneficiary volume), and county characteristics (urban location, region, household income, and the proportion of the population who were White). The data sets we used do not contain information on physician race and ethnicity. The characteristics were selected a priori based on availability and relevance in physician workforce literature. We compared the characteristics between physicians who continuously practiced as hospitalists vs those who changed practice in any year by using the χ2 test for categorical and the Wilcoxon test for continuous variables. We used multivariable linear probability regression to measure the association between the characteristics and continuous hospitalist practice. We also performed a sensitivity analysis, including in our sample the hospitalists who did not bill Medicare or subspecialized in 2013 to 2018.
Data analyses were conducted from March 15 to June 28, 2021. Two-sided P values <.05 were considered significant. Analyses were conducted using SAS, version 9.4 (SAS Institute Inc). The University of Pennsylvania Institutional Review Board waived review of this study and the need for informed consent because the data used were publicly accessible. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.
Results
Of the 16 985 hospitalists identified in 2012 (mean [SD] age, 41.1 [8.1] years; 6146 women [36.2%] and 10 839 men [63.8%]), 4052 (23.9%) shifted practice to another setting in at least 1 year during the follow-up period. Those physicians tended to be in smaller practices (eg, compared with solo practice, the regression coefficient for practices with ≥300 physicians was 22.06; 95% CI, 18.60-25.52; P < .001) and/or rural practices (regression coefficient for urban vs rural, 3.40; 95% CI, 1.28-5.53; P = .002) (Table 1). Physicians who changed practice were more likely to split time between a hospital and other settings than to practice predominantly outside of the hospital; eg, after hospitalists, the second largest category (601 of 16 985 [3.5%]) was physicians who provided 50% or more but less than 90% of visits in the hospital, with the other visits provided in other settings (Table 2).
Table 1. Baseline Demographic Details and Practice Characteristics of Physicians Who Were Hospitalists in 2012.
| Characteristic | No. (%) | Regression coefficient (95% CI)c | P value | ||
|---|---|---|---|---|---|
| Cohort overall | Always hospitalista | Shifted practice to another settingb | |||
| No. (%) | 16 985 (100) | 12 933 (76.1) | 4052 (23.9) | NA | NA |
| Age, y | |||||
| <30 | 220 (1.3) | 170 (1.3) | 50 (1.2) | [Reference] | NA |
| 30-39 | 8224 (48.4) | 6371 (49.3) | 1853 (45.7) | 0.44 (−5.22 to 6.10) | .88 |
| 40-49 | 5904 (34.8) | 4487 (34.7) | 1417 (35.0) | 0.08 (−5.63 to 5.79) | .98 |
| 50-59 | 2168 (12.8) | 1589 (12.3) | 579 (14.3) | −1.29 (−7.18 to 4.60) | .67 |
| ≥60 | 469 (2.8) | 316 (2.4) | 153 (3.8) | −4.66 (−11.5 to 2.14) | .18 |
| Gender | |||||
| Female | 6146 (36.2) | 4756 (36.8) | 1390 (34.3) | [Reference] | NA |
| Male | 10 839 (63.8) | 8177 (63.2) | 2662 (65.7) | −0.77 (−2.12 to 0.59) | .27 |
| Top 20 medical school | 727 (4.3) | 589 (4.6) | 138 (3.4) | 2.63 (−0.56 to 5.81) | .11 |
| Practice size, No. of physiciansd | |||||
| Solo practice | 746 (4.4) | 445 (3.4) | 301 (7.4) | [Reference] | NA |
| 2-9 | 1288 (7.6) | 804 (6.2) | 484 (11.9) | 2.46 (−1.43 to 6.34) | .22 |
| 10-49 | 4550 (26.8) | 3366 (26.0) | 1184 (29.2) | 13.24 (9.87 to 16.60) | <.001 |
| 50-99 | 2382 (14.0) | 1842 (14.2) | 540 (13.3) | 16.50 (12.92 to 20.08) | <.001 |
| 100-299 | 4399 (25.9) | 3466 (26.8) | 933 (23.0) | 18.01 (14.62 to 21.40) | <.001 |
| ≥300 | 3620 (21.3) | 3010 (23.3) | 610 (15.1) | 22.06 (18.60 to 25.52) | <.001 |
| Annual Medicare beneficiary volume, mean (SD)e | 379.0 (212.5) | 379.4 (208.4) | 377.5 (225.3) | 0.06 (0.03 to 0.09) | <.001 |
| Regionf | |||||
| Northeast | 3684 (21.7) | 2870 (22.2) | 814 (20.1) | [Reference] | NA |
| Midwest | 3777 (22.2) | 2898 (22.4) | 879 (21.7) | −0.59 (−2.56 to 1.38) | .56 |
| South | 6291 (37.0) | 4720 (36.5) | 1571 (38.8) | −2.01 (−3.87 to 0.14) | .03 |
| West | 3193 (18.8) | 2428 (18.8) | 765 (18.9) | −0.78 (−2.82 to 1.27) | .46 |
| Urban (vs rural) | 15 112 (89.0) | 11 626 (89.9) | 3486 (86.0) | 3.40 (1.28 to 5.53) | .002 |
| Median household income, mean (SD), $ | 53 216 (13 210) | 53 326 (13 161) | 52 865 (13 364) | −0.002 (−0.05 to 0.05) | .94 |
| Proportion of county population who are White, mean (SD) | 71.1 (15.8) | 70.8 (15.9) | 72.0 (15.6) | −0.10 (−0.14 to −0.05) | <.001 |
Abbreviations: MD-PPAS, Medicare Data on Provider Practice and Specialty; NA, not applicable.
An “always hospitalist” is defined as a physician in the cohort who continuously practiced as a hospitalist (billing ≥90% of patient visits from an acute care hospital).
Represents physicians in the cohort who were hospitalists in 2012 and shifted practice to another setting or settings in any of the study years (billing <90% of patient visits from an acute care hospital).
Multivariable linear probability models were used to measure the probability of a physician remaining in hospitalist practice throughout the study period.
Physician group size was measured by summing the number of National Provider Identifiers associated with each Tax Identification Number using the MD-PPAS.
Patient volume was measured using the number of unique Medicare beneficiaries.
Physicians practicing in US territories were not categorized into regions (n = 40).
Table 2. Practice Settings of Physicians Who Were Hospitalists in 2012 by Year of Follow-up.
| Setting | Description | No. (%) of physicians (N = 16 985) | |||||
|---|---|---|---|---|---|---|---|
| 2013 | 2014 | 2015 | 2016 | 2017 | 2018 | ||
| Hospitalist | ≥90% of visits to hospitalized patients | 15 997 (94.2) | 15 378 (90.5) | 14 987 (88.3) | 14 603 (86.0) | 14 396 (84.8) | 14 164 (83.4) |
| Hospital-SNF practice | ≥50% to <90% of visits to hospitalized patients, followed by visits to SNF as the second largest visit type | 255 (1.5) | 360 (2.1) | 424 (2.5) | 504 (3.0) | 473 (2.8) | 454 (2.7) |
| Hospital-office practice | ≥50% to <90% of visits to hospitalized patients, followed by office visits as the second largest visit type | 165 (1.0) | 213 (1.3) | 223 (1.3) | 272 (1.6) | 263 (1.6) | 271 (1.6) |
| Hospital-other practice | ≥50% to <90% of visits to hospitalized patients, followed by “other”a visits as the second largest visit type | 335 (2.0) | 438 (2.6) | 445 (2.6) | 492 (2.9) | 519 (3.1) | 601 (3.5) |
| SNF practice | ≥50% of visits to SNF patients | 34 (0.2) | 77 (0.5) | 129 (0.8) | 153 (0.9) | 203 (1.2) | 254 (1.5) |
| Office practice | ≥50% of visits provided in outpatient office or clinic | 53 (0.3) | 155 (0.9) | 224 (1.3) | 335 (2.0) | 388 (2.3) | 415 (2.4) |
| Other practice | ≥50%of visits provided in “other” settings | 73 (0.4) | 188 (1.1) | 275 (1.6) | 333 (2.0) | 423 (2.5) | 462 (2.7) |
| Mixed | Physician provided services in ≥3 settings with <50% of services provided in each setting | 73 (0.4) | 176 (1.0) | 278 (1.6) | 293 (1.7) | 320 (1.9) | 364 (2.1) |
Abbreviation: SNF, skilled nursing facility.
Includes all other settings not otherwise specified.
Our sensitivity analyses that included physicians who did not bill Medicare (n = 2994) or subspecialized (n = 2644) in any year during study follow-up were generally consistent with the primary analyses. In addition to smaller practice size and rural location, physicians who shifted practice were more likely to be aged 60 years or older (7.1% vs 2.4%, P < .001) and less likely to come from a top 20 medical school (3.6% vs 4.6%, P < .001) compared with physicians who continuously practiced as hospitalists, which were not statistically significant coefficients in our primary analysis.
Discussion
This study found that between 2012 and 2018, nearly 1 in 4 hospitalists shifted practice to other settings, although most hospitalists continued to practice in the hospital at least part time. Our findings have important implications for hospitals, especially during infectious disease outbreaks, such as the COVID-19 pandemic, that place extra demands on hospital-based clinicians. Understanding the drivers behind these trends may help mitigate the high costs of physician turnover in health systems.
Limitations of this study include the use of Medicare fee-for-service claims but not claims for other payers. However, Medicare is the most common payer for inpatient care in the US.
eTable. HCPCS Codes Used to Identify Practice Settings
References
- 1.Kuo YF, Sharma G, Freeman JL, Goodwin JS. Growth in the care of older patients by hospitalists in the United States. N Engl J Med. 2009;360(11):1102-1112. doi: 10.1056/NEJMsa0802381 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Wachter RM, Goldman L. Zero to 50,000—the 20th anniversary of the hospitalist. N Engl J Med. 2016;375(11):1009-1011. doi: 10.1056/NEJMp1607958 [DOI] [PubMed] [Google Scholar]
- 3.Lindenauer PK, Rothberg MB, Pekow PS, Kenwood C, Benjamin EM, Auerbach AD. Outcomes of care by hospitalists, general internists, and family physicians. N Engl J Med. 2007;357(25):2589-2600. doi: 10.1056/NEJMsa067735 [DOI] [PubMed] [Google Scholar]
- 4.Stevens JP, Nyweide DJ, Maresh S, Hatfield LA, Howell MD, Landon BE. Comparison of hospital resource use and outcomes among hospitalists, primary care physicians, and other generalists. JAMA Intern Med. 2017;177(12):1781-1787. doi: 10.1001/jamainternmed.2017.5824 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Goodwin JS, Salameh H, Zhou J, Singh S, Kuo YF, Nattinger AB. Association of hospitalist years of experience with mortality in the hospitalized Medicare population. JAMA Intern Med. 2018;178(2):196-203. doi: 10.1001/jamainternmed.2017.7049 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Vuong K. Turnover rate for hospitalist groups trending downward. The Hospitalist. February 1, 2017. Accessed May 21, 2021. https://www.the-hospitalist.org/hospitalist/article/130462/turnover-rate-hospitalist-groups-trending-downward
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable. HCPCS Codes Used to Identify Practice Settings