Abstract
Importance
Substantial numbers of hospitalists are fresh graduates of residency training programs. Current data about the effect of hospitalist years of experience on patient outcomes are lacking.
Objective
To describe the association of hospitalist years of experience with 30-day mortality and hospital mortality of their patients.
Design, Setting, and Participants
We used a 5% sample of national Medicare data of patient and hospital characteristics to build a multilevel logistic regression model to predict mortality as a function of years of experience of the hospitalists. We created 2 cohorts. The first was a cross-sectional cohort of 21 612 hospitalists working between July 1, 2013, and June 30, 2014, with a 5-year look-back period to assess their years of prior experience as a hospitalist, and the second was a longitudinal cohort of 3860 hospitalists in their first year of practice between July 1, 2008, and June 30, 2011, who continued practicing hospital medicine for at least 4 years.
Main Outcomes and Measures
Thirty-day postadmission mortality adjusted for patient and hospital characteristics in a 3-level logistic regression model. Hospital mortality was a secondary outcome.
Results
Among 21 612 hospitalists caring for Medicare inpatients from July 1, 2013, to June 30, 2014, 5445 (25%) had 1 year of experience or less, while 11 596 (54%) had 4 years of experience or more. We then identified 3860 physicians in their first year as hospitalists who continued to practice as hospitalists for 4 years. There was a significant association between hospitalist experience and mortality. Observed 30-day mortality was 10.50% for patients of first-year hospitalists vs 9.97% for patients of hospitalists in their second year. The mortality odds for patients of second-year hospitalists were 0.90 (95% CI, 0.84-0.96) compared with patients of first-year hospitalists. Observed hospital mortality was 3.33% for patients cared for by first-year hospitalists vs 2.96% for second-year hospitalists. (odds ratio, 0.84; 95% CI, 0.75-0.95). For both 30-day and hospital mortality, there was little change in odds of mortality between the second year and subsequent years of experience.
Conclusions and Relevance
Patients cared for by hospitalists in their first year of practice experience higher mortality. Early-career hospitalists may require additional support to ensure optimal outcomes for their patients.
This cohort study of Medicare data describes the association of hospitalist years of experience with 30-day mortality and hospital mortality of their patients.
Key Points
Question
Does the mortality of patients vary by the amount of prior experience of the hospitalists providing their care?
Findings
In this multicohort analysis of Medicare data, patients cared for by hospitalists in their first year of practice experience higher mortality than patients cared for by more experienced hospitalists.
Meaning
Hospitalists very early in their career may need additional support to ensure optimal outcomes for their patients.
Introduction
Starting in the 1990s. the shift to managed care and pressures to reduce hospital costs in the United States coincided with the growth of hospitalists. Since then, the total number and percentage of Medicare beneficiaries cared for by hospitalists have been increasing. The potential advantages of hospitalist care stem from greater access to and expertise of physicians concentrating on inpatient care. A physician specializing in hospital medicine should be more effective in initiating quick and appropriate treatment, as well as in recognizing and preventing complications. Hospitalists are also more likely to participate in multidisciplinary teams to address early discharge planning, quality of care, and patient safety.
Hospitalists tend to have high turnover. Many change their career by either switching to primary care medicine or starting their fellowship training. Some international medical graduates may spend 2 years as hospitalists in medically underserved areas to meet their visa requirement before pursuing other training. The turnover leads to substantial numbers of hospitalists who are early in their careers, frequently right out of residency. This in turn raises concerns that the outcomes of care provided by inexperienced hospitalists might be different from care provided by more experienced hospitalists. This question has not been adequately studied.
In the current study, we characterized the years of experience of hospitalists practicing from January 2013 to December 2014. We then examined the association of patient mortality with the years of experience of the hospitalist providing their care, using a 5% sample of national Medicare data from January 2008 to December 2014. We hypothesized that adjusted 30-day postadmission mortality and hospital mortality would decline with increasing hospitalist experience.
Methods
Overview
We first used Medicare data from January 2007 to December 2013 to estimate the years of prior experience of hospitalists practicing from July 1, 2013, to June 30, 2014. July 1 defined the beginning of each year because it coincides with the start of the new academic year. Next, we constructed 3 cohorts of hospitalists who had their initial year of experience starting July 1, 2008, through June 30, 2009, July 1, 2009, through June 30, 2010, or July 1, 2010, through June 30, 2011, and who continued to work as hospitalists for 3 additional and consecutive years. We then assessed the mortality of the patients of those physicians over their first 4 years of experience as hospitalists. We restricted the analyses to physicians who practiced hospital medicine for at least 4 years to compare outcomes of the same hospitalists longitudinally as they gained additional experience. Approval for this study was granted by the University of Texas Medical Branch institutional review board.
Data Source
We used a 5% national sample of Medicare data from July 1, 2003, through June 30, 2014. These include the Medicare Denominator File for demographic and enrollment information, the Carrier File for physician claims, the Outpatient Statistical Analysis File for outpatient claims, the Medicare Provider Analysis and Review (MEDPAR) File for inpatient claims and the Provider of Service (POS) file for hospital characteristics.
Study Patients
We first identified all acute hospitalizations with at least 3 days’ length of stay that met the following criteria: (1) 66 years or older; (2) Medicare Part A and B coverage with no health maintenance organization (HMO) enrollment in the 12 months before the acute hospitalization; and (3) evaluation and management (E&M) charges during the hospitalization by a generalist physician (general internist, family physician, general practitioner, or geriatrician). We further limited the admission to those in which 1 hospitalist was responsible for at least 2 E&M charges, and also responsible for at least half of all inpatient E&M charges generated by generalist physicians, resulting in 304 579 hospitalizations. We also conducted sensitivity analyses with a cohort limited to hospitalizations with a medical Diagnosis-Related Group (DRG), using the selection criteria above (n = 238 723).
Identification of Hospitalists
As in our prior work, hospitalists were defined as generalist physicians (general internist, family physician, general practitioner or geriatrician) who had at least 20 E&M billings in each year studied and at least 90% of the E&M billings in that year were from services to hospitalized patients. The percentage was calculated by using inpatient E&M billing codes (Current Procedural Terminology [CPT] codes: 99221-99223, 99231-99233, 99238-99239, 99251-99255) and outpatient E&M billings (CPT codes: 99201-99205, 99211-99215, 99241-99245). This method has a sensitivity of 84% and positive predictive value of 89%, compared with self-identification.
Years of Experience of Hospitalists
For each year in which we assessed care by hospitalists, we conducted a 5-year look-back to determine how many years a given hospitalist had provided hospitalist services in the 5 years prior to the study year. In preliminary analyses, we explored different cut points of the minimum number of Medicare E&M billings and the percentage of those billings devoted to hospitalized patients. Our intent was to use looser criteria to identify years of prior experience than used in initially identifying hospitalists to avoid underestimating years of experience. The different cut points had minimal effect on the results. We chose a minimum of 10 E&M charges in a given year (from the 5% Medicare data) with at least 80% devoted to hospitalized patients. We also used the cut point of 50% or more of E&M visits generated on hospitalized patients in a sensitivity analysis.
Hospitalist Cohorts
For each of three years (July 1, 2008 to June 30, 2009; July 1, 2009 to June 30, 2010; and July 1, 2010 to June 30, 2011), we selected physicians who were in their first year as hospitalists. We also required that those physicians work as hospitalists for at least 3 additional and consecutive years. This produced a longitudinal cohort of 3860 physicians for whom we had data on their first 4 years of working as hospitalists.
Measures
Patient Characteristics
Beneficiary age, sex, race/ethnicity, and Medicaid eligibility were extracted from the Medicare Denominator File. Education information at the beneficiary’s zip code of residence was obtained from the 2013 American Community Survey estimates of the US Census Bureau. Elixhauser comorbidities were identified by reviewing all diagnoses associated with hospitalizations, physician services, and outpatient claims over the prior 12 months. The MEDPAR file provided the DRG Major Diagnostic Category (DRG-MDC), length of stay, and the number of hospitalizations in the prior 12 months. Residence prior to hospitalization was categorized as community vs nursing facility or other institution based on the MEDPAR inpatient source code.
Hospital Characteristics
Information on location (urban or rural), type (for profit, nonprofit, government), bed size (≤200, 201-350, 351-500, or >500 beds), and medical school affiliation (major, limited, graduate, no affiliation) was extracted from the POS file.
Study Outcomes
The primary outcome was 30-day posthospital admission mortality. We extracted death information from the Medicare Denominator File. We also assessed hospital mortality, as reported in MEDPAR as discharged dead.
Statistical Analysis
Descriptive analyses were used to summarize patient and hospital characteristics and mortality. We used multilevel logistic regression modeling, in which hospitalizations were nested in hospitalists, and hospitalists nested in hospitals, to evaluate the association between patient, hospitalist, and hospital characteristics and mortality. Patient characteristics included age, sex, race, Medicaid eligibility, education, the 31 Elixhauser comorbidities, length of stay, number of acute hospitalizations in the 12 months before the index hospitalization, residence prior to hospitalization (community vs other), medical or surgical DRG type and DRG-MDC. The years of prior experience (0, 1, 2, or 3) of the hospitalist caring for the patient was a hospitalist level characteristic. Hospital characteristics included hospital bed size, location, type and medical school affiliation. The 1833 hospitalists who worked in multiple hospitals in the 4 years were assigned to the first hospital. In a sensitivity analysis we studied the 2027 hospitals who worked in only one hospital.
We estimated the excess deaths attributed to first-year hospitalists in the longitudinal cohort as the product of the number of admissions cared for by first-year hospitalists multiplied by 20, to inflate from a 5% to a 100% sample, multiplied by the mean 30-day mortality rate, multiplied by 1 minus the adjusted odds ratio (OR) for 30-day mortality of second- vs first-year hospitalists.
Results
Table 1 summarizes the years of experience of the hospitalists identified as practicing in the most recent time period, July 1, 2013, to June 30, 2014. Of the 21 612 hospitalists, 2796 (12.9%) had no prior experience as hospitalists and 2649 (12.2%) had 1 year of prior experience. In contrast, 11 596 hospitalists (53.7%) had 4 or more years of experience.
Table 1. Total Prior Years of Experience for 21 647 Hospitalists Treating Medicare Patientsa.
| Years of Experience | Hospitalists, No. (%) | Hospitalizations, No. (%) |
|---|---|---|
| 0 | 2796 (12.94) | 10 444 (10.28) |
| 1 | 2649 (12.26) | 11 286 (11.11) |
| 2 | 2434 (11.26) | 11 008 (10.84) |
| 3 | 2137 (9.89) | 10 076 (9.92) |
| ≥4 | 11 596 (53.66) | 58 761 (57.85) |
| Total | 21 612 | 101 575 |
Data are from a 5% sample of US Medicare enrollees, so the 101 575 hospitalizations represent 2 031 500 total hospitalizations of Medicare enrollees cared for by these hospitalists from July 1, 2013, to June 30, 2014.
In sensitivity analyses we liberalized the criteria defining years of experience to count any year when at least half of physician E&M charges were generated on hospitalized patients. Using this definition, 2490 hospitalists (11.5%) had no prior years of experience and 2468 (11.4%) had 1 year.
In this sample of 21 612 hospitalists working July 1, 2013, to June 30, 2014, there were significant differences in the types of patients seen and in the hospitals served, depending on the experience of the hospitalists (Table 2). This was particularly true for hospital characteristics, with more experienced hospitalists working in smaller, nonacademic hospitals. For example, 55.6% of the hospitalizations cared for by hospitalists with 4 or more years’ experience were in hospitals with no academic affiliation, compared with 49.7% of hospitalizations cared for by hospitalists in their first year. In contrast, only 19.3% of hospitalizations by the most experienced hospitalists were in hospitals with a major academic affiliation vs 25.3% for first year hospitalists, which is consistent with the higher turnover of hospitalists in academic hospitals.
Table 2. Cross-sectional Analysis of the Association of Patient and Hospital Characteristics With the Year of Experience of the 21 647 Hospitalists Providing Carea.
| Characteristic | Mean (SD) or No. (%) | Hospitalist Experience, Years | P Value | |||
|---|---|---|---|---|---|---|
| 0 (N = 10 206) |
1 (N = 11 571) |
2-3 (N = 21 421) |
≥4 (N = 59 630) |
|||
| Age, mean (SD) | 80.55 (8.16) | 80.64 (8.10) | 80.54 (8.14) | 80.65 (8.19) | 80.50 (8.15) | .01 |
| Sex, No. (%) | .31 | |||||
| Male | 36 717 (36.15) | 3820 (36.58) | 4132 (36.61) | 7661 (36.34) | 21 104 (35.91) | |
| Female | 64 858 (63.85) | 6624 (63.42) | 7154 (63.39) | 13 423 (63.66) | 37 657 (64.09) | |
| Medicaid eligible, No. (%) | .89 | |||||
| No | 79 856 (78.62) | 8218 (78.69) | 8843 (78.35) | 16 597 (78.72) | 46 198 (78.62) | |
| Yes | 21 719 (21.38) | 2226 (21.31) | 2443 (21.65) | 4487 (21.28) | 12 563 (21.38) | |
| Length of stay in hospitalization, mean (SD) | 5.84 (4.30) | 5.75 (4.10) | 5.75 (4.10) | 5.77 (4.01) | 5.90 (4.48) | <.001 |
| No. of hospitalizations in 12 mo before admission, mean (SD) | 1.17 (1.66) | 1.18 (1.68) | 1.18 (1.66) | 1.16 (1.65) | 1.18 (1.67) | .49 |
| DRG weight, mean (SD) | 1.63 (1.25) | 1.60 (1.23) | 1.61 (1.23) | 1.62 (1.24) | 1.64 (1.26) | .04 |
| No. of comorbidities, mean (SD) | 5.14 (3.65) | 5.16 (3.68) | 5.12 (3.62) | 5.08 (3.63) | 5.16 (3.65) | .04 |
| Race, No. (%) | <.001 | |||||
| White | 85 234 (83.91) | 8846 (84.70) | 9560 (84.71) | 17 829 (84.56) | 48 999 (83.39) | |
| Black | 8935 (8.80) | 933 (8.93) | 956 (8.47) | 1718 (8.15) | 5328 (9.07) | |
| Hispanic | 4700 (4.63) | 407 (3.90) | 482 (4.27) | 1043 (4.95) | 2768 (4.71) | |
| Other | 2706 (2.66) | 258 (2.47) | 288 (2.55) | 494 (2.34) | 1666 (2.84) | |
| Education (persons in zip code >25 years with high school education or higher, %), mean (SD) | 86.62 (8.61) | 86.52 (8.37) | 86.66 (8.46) | 86.68 (8.54) | 86.61 (8.71) | .46 |
| DRG type, No. (%) | <.001 | |||||
| Medical | 79 019 (77.79) | 8259 (79.08) | 8810 (78.06) | 16 466 (78.10) | 45 484 (77.41) | |
| Surgical | 22 556 (22.21) | 2185 (20.92) | 2476 (21.94) | 4618 (21.90) | 13 277 (22.59) | |
| Residence prior to hospitalization, No. (%) | .01 | |||||
| Community | 91 252 (89.84) | 9310 (89.06) | 10 102 (89.51) | 19 037 (90.29) | 52 812 (89.88) | |
| Nursing facility and other institutions | 10 323 (10.16) | 1143 (10.94) | 1184 (10.49) | 2047 (9.71) | 5949 (10.12) | |
| Bed size, No. (%) | <.001 | |||||
| >500 | 27 465 (27.04) | 3115 (29.83) | 3118 (27.63) | 5328 (25.27) | 15 904 (27.07) | |
| 351-500 | 17 617 (17.34) | 1891 (18.11) | 1990 (17.63) | 3529 (16.74) | 10 207 (17.37) | |
| 201-350 | 28 883 (28.39) | 2724 (26.08) | 3023 (26.79) | 6401 (30.36) | 16 685 (28.39) | |
| ≤200 | 27 660 (27.23) | 2714 (25.99) | 3155 (27.95) | 5826 (27.63) | 15 965 (27.17) | |
| Urban vs rural indicator, No. (%) | <.001 | |||||
| Rural | 14 909 (14.68) | 1723 (16.50) | 1930 (17.10) | 3427 (16.25) | 7829 (13.32) | |
| Urban | 86 666 (85.32) | 8721 (83.50) | 9356 (82.90) | 17 657 (83.75) | 50 932 (86.68) | |
| Type of insurance provider, No. (%) | <.001 | |||||
| For profit | 15 072 (14.84) | 1281 (12.27) | 1508 (13.36) | 3418 (16.21) | 8865 (15.09) | |
| Government | 11 853 (11.67) | 1222 (11.70) | 1260 (11.16) | 2449 (11.62) | 6922 (11.78) | |
| Nonprofit | 74 650 (73.49) | 7941 (76.03) | 8518 (75.47) | 15 217 (72.17) | 42 974 (73.13) | |
| Medical school affiliation, No. (%) | <.001 | |||||
| Major | 21 167 (20.84) | 2644 (25.32) | 2733 (24.22) | 4447 (21.09) | 11 343 (19.30) | |
| Limited | 20 022 (19.71) | 2018 (19.32) | 2248 (19.92) | 3646 (17.29) | 12 110 (20.61) | |
| Graduate | 5042 (4.96) | 589 (5.64) | 542 (4.80) | 1272 (6.03) | 2639 (4.49) | |
| No affiliation | 55 344 (54.49) | 5193 (49.72) | 5763 (51.06) | 11 719 (55.58) | 32 669 (55.60) | |
Abbreviation: DRG, Diagnosis-Related Group.
Data were generated from a 5% sample of national Medicare data; thus, the 101 575 admissions represent 2 031 500 admissions between July 1, 2013, and June 30, 2014.
Because of these differences, we restricted subsequent analyses to hospitalists whom we followed through the first 4 years of their experience as hospitalists. Our assumption was that there would be smaller differences in the types of patients and hospitals by year of experience in a cohort of hospitalists followed longitudinally. We constructed a cohort of 3860 physicians in their first year working as hospitalists from either July 1, 2008, to June 30, 2009, July 1, 2009, to June 30, 2010, or July 1, 2010, to June 30, 2011, and who worked as hospitalists for at least the following 3 years. This allowed us to examine the outcomes of the same cohort of physicians as they gained experience over their first 4 years of practice.
eTable 1 in the Supplement presents the patient and hospital characteristics as a function of whether hospitalist were in their first, second, third, or fourth year of experience. There were still small differences in characteristics by year of experience, and, given the large number of patients, many of these were statistically significant. The subsequent analyses controlled for all measured patient and hospital characteristics.
Table 3 presents the 30-day and hospital mortality rates for the 84 947 admissions cared for by the 3860 hospitalists during their first 4 years of experience, stratified by whether they were in their first, second, third, or fourth year of hospitalist experience. Also presented are the mortality rates by patient and hospital characteristics and the odds of mortality generated from a multilevel multivariable model, with patients clustered within hospitalists, and hospitalists within hospitals, including all the variables in the table, plus the MDC-DRG and 31 comorbidities, each entered individually. There was a significant association between hospitalist experience and 30-day and hospital mortality. The major change occurred between the first and second year experience as hospitalists. Observed 30-day mortality was 10.50% for patients of first year hospitalists vs 9.97% for patients of hospitalists in their second year. In the multivariable analysis, the 30-day mortality odds for patients of second year hospitalists were 0.90 (95% CI, 0.84-0.96) compared with patients of first year hospitalists. The observed hospital mortality was 3.33% for patients cared for by first year hospitalists vs 2.96% for second year hospitalists. In the multivariable analysis, the odds of mortality was 0.84 (95% CI, 0.75-0.95) for patients cared for by hospitalists in their second vs their first year in practice. For both 30-day and hospital mortality, there was little change in odds of mortality between the second year and subsequent years of experience.
Table 3. Mortality in the 30 Days After Admission and Hospital Mortality as a Function of the Number of Years of Experience of the 3860 Hospitalists Followed Over Their First 4 Years of Practicea.
| Characteristics | No. (%) | 30-Day Mortality Rate After Hospital Admission | Hospital Mortality Rate | ||
|---|---|---|---|---|---|
| Observed Rate, % | Odds Ratiob (95% CI) | Observed Rate, % | Odds Ratio (95% CI) | ||
| All | 83 831 | 10.21 | ND | 3.07 | ND |
| Age, y (mean SD) | 80.5 (7.96) | 10.21 | 1.05 (1.05-1.06) | 3.07 | 1.03 (1.02-1.04) |
| Sex | |||||
| Female | 54 061 (64.49) | 9.52 | 1 [Reference] | 2.81 | 1 [Reference] |
| Male | 29 770 (35.51) | 11.46 | 1.21 (1.15-1.27) | 3.53 | 1.14 (1.05-1.25) |
| Medicaid eligible | |||||
| No | 65 096 (77.65) | 10.17 | 1 [Reference] | 3.06 | 1 [Reference] |
| Yes | 18 735 (22.35) | 10.33 | 0.99 (0.94-1.06) | 3.08 | 0.92 (0.83-1.03) |
| Education (% of persons in zip code >25 years with high school education or higher [per percent]), mean (SD) | 86.4 (8.55) | 10.21 | 0.99 (0.99-1.01) | 3.07 | 0.99 (0.98-0.99) |
| Hospitalizations in 12 mo before admission (per hospitalization), mean (SD) | 1.3 (1.73) | 10.21 | 0.97 (0.96-0.99) | 3.07 | 0.95 (0.92-0.99) |
| Race | |||||
| White | 71 803 (85.65) | 10.34 | 1 [Reference] | 3.07 | 1 [Reference] |
| Black | 7050 (8.41) | 9.32 | 0.85 (0.77-0.93) | 3.06 | 0.94 (0.81-1.10) |
| Hispanic | 3173 (3.78) | 9.05 | 0.89 (0.78-1.03) | 2.80 | 0.90 (0.71-1.15) |
| Other | 1805 (2.15) | 10.47 | 1.06 (0.90-1.25) | 3.43 | 1.13 (0.87-1.48) |
| DRG type | |||||
| Medical | 66 203 (78.97) | 11.44 | 1 [Reference] | 3.28 | 1 [Reference] |
| Surgical | 17 628 (21.03) | 5.58 | 0.78 (0.71-0.84) | 2.26 | 1.26 (1.10-1.43) |
| Residence prior to hospitalization | |||||
| Community | 75 151 (89.65) | 9.55 | 1 [Reference] | 2.81 | 1 [Reference] |
| Nursing facility and other institutions | 8680 (10.35) | 15.89 | 1.62 (1.51-1.73) | 5.26 | 1.66 (1.48-1.86) |
| Bed size | |||||
| >500 | 21 189 (25.28) | 9.98 | 1 [Reference] | 3.21 | 1 [Reference] |
| 351-500 | 15 343 (18.30) | 10.13 | 0.99 (0.91-1.07) | 3.07 | 0.94 (0.83-1.09) |
| 201-350 | 24 536 (29.27) | 9.92 | 0.94 (0.87-1.02) | 3.01 | 0.91 (0.79-1.05) |
| ≤200 | 22 763 (27.15) | 10.79 | 1.00 (0.92-1.09) | 2.99 | 0.87 (0.75-1.03) |
| Urban/rural indicator | |||||
| Rural | 12 739 (15.20) | 11.34 | 1 [Reference] | 3.52 | 1 [Reference] |
| Urban | 71 092 (84.80) | 10.01 | 0.90 (0.83-0.97) | 2.98 | 0.82 (0.72-0.94) |
| Type of provider | |||||
| For profit | 11 849 (14.13) | 10.26 | 1 [Reference] | 3.03 | 1 [Reference] |
| Government | 9714 (11.59) | 10.76 | 1.02 (0.92-1.13) | 3.21 | 1.01 (0.85-1.21) |
| Nonprofit | 62 268 (74.28) | 10.11 | 0.93 (0.86-1.01) | 3.05 | 0.94 (0.83-1.08) |
| Medical school affiliation | |||||
| Major | 16 213 (19.34) | 9.55 | 1 [Reference] | 3.05 | 1 [Reference] |
| Limited | 15 211 (18.14) | 10.35 | 1.16 (1.06-1.27) | 3.06 | 1.06 (0.91-1.23) |
| Graduate | 4922 (5.87) | 11.19 | 1.24 (1.10-1.40) | 3.74 | 1.27 (1.04-1.57) |
| No affiliation | 47 485 (56.64) | 10.28 | 1.15 (1.06-1.24) | 3.00 | 1.06 (0.92-1.22) |
| Hospitalist experience | |||||
| First year | 18 370 (21.91) | 10.51 | 1 [Reference] | 3.33 | 1 [Reference] |
| Second year | 22 263 (26.56) | 9.97 | 0.90 (0.84-0.96) | 2.96 | 0.85 (0.76-0.95) |
| Third year | 22 395 (26.71) | 10.19 | 0.89 (0.84-0.96) | 3.11 | 0.86 (0.77-0.97) |
| Fourth year | 20 809 (24.82) | 10.23 | 0.89 (0.83-0.95) | 2.90 | 0.79 (0.69-0.88) |
Abbreviations: DRG, Diagnosis-Related Group; ND, no data.
The 83 831 admissions in this 5% sample represent a total of 1 676 620 admissions cared for by the 3860 hospitalists over their first 4 years working as hospitalists.
Odds ratio are from a multilevel model adjusted for all characteristics presented in the Table, as well as patient comorbidities and DRG-Major Diagnostic Category.
We performed 2 sensitivity analyses. The first restricted the sample to the 2027 hospitalist who worked in only 1 hospital over the 4 years (eTable 2 in the Supplement). The results are similar to those found in the main cohort. In the second we further restricted the sample used in eTable 2 in the Supplement to hospital discharges with a medical diagnosis. Once again the results were quite similar to those found in the main cohort (eTable 3 in the Supplement).
The 83 831 admissions in Table 3 are a 5% sample of the 1 676 620 Medicare admissions for the 3860 hospitalists over their first 4 years of practice. Using the 0.90 adjusted odds of 30-day mortality for patients treated by second vs first year hospitalist, we estimate an excess 3856 deaths among the 367 280 admissions cared for by first year hospitalists, with 1957 of the excess deaths occurring during hospitalization.
Discussion
Most prior studies that have examined the association of physician experience with Medical processes or outcomes have compared longer categories of experience. For example, the “least experience” category in many studies is the first 5 years or longer. The most common finding of such studies is that older physicians, defined by age or years since graduation, often do less well than younger and more recently trained physicians. For example, Southern et al in a study of 59 physicians at one hospital found that patients of physicians with more than 20 years in practice had higher mortality compared with patients of physicians in practice for 0 to 5 years. Similarly, Tsugawa et al analyzed national Medicare data and reported that patients cared for by hospitalists who were younger than 40 experienced lower 30-day mortality than those cared for by older hospitalists.
Few studies have looked at shorter lengths of experience. An exception is studies of postgraduate medical training, where investigators have compared hospital mortality, complications, and other outcomes in the first month(s) of the academic year vs those later in the year. While several studies found no differences, others showed worse outcomes for the new medical trainees and/or house staff.
An association of experience of hospitalists with mortality was found in the early prospective trial of hospitalist care conducted by Meltzer et al. The trial was 2 years long and the investigators found significantly lower 30-day mortality in patients on the hospitalist service in year 2 of the trial but not year 1, compared with patients treated by nonhospitalists. In contrast, a 2-year observational study in one hospital by Auerbach et al found no difference in mortality of patients cared for by hospitalists between years 1 and 2 but demonstrated improvements in efficiency.
The average yearly turnover rate for hospitalists ranges from 13% to 22% in a report on hospitalist workload, with many returning to subspecialty postdoctoral fellowships. Individual hospitals vary in the experience of their hospitalists. Hospitalists at academic hospitals appear to have briefer careers in that specialty.
In our cross-sectional analysis of all hospitalists practicing during the July 1, 2013, to June 30, 2014, academic year (Table 2), less experienced hospitalists were more likely to practice at large academic hospitals. Patients at such hospitals experience lower mortality than patients in community hospitals. This would artifactually produce lower mortality rates for less experienced hospitalists. We addressed this bias in 2 ways. First, we restricted our analyses to a longitudinal cohort of physicians for whom we had 4 consecutive years’ experience as hospitalists, starting with their initial year. We did this to minimize any selection bias involving differences in characteristics and outcomes between hospitalists who practiced for only a year or 2 vs those who practiced longer. Second, we used a multilevel analysis, with patients clustered within hospitalists and hospitalists within hospitals. This implicitly controls for differences in mortality across hospitals.
Limitations
The study has several limitations. It is an observational study, subject to selection bias. As noted, our method of following the same hospitalists over 4 years should minimize those biases. Also, the patient sample was limited to older patients with fee-for-service Parts A and B Medicare. Another limitation is that the method we used may underestimate the years of experience of the hospitalist, as we required at least 10 E&M charges in the 5% sample of Medicare data, with at least 80% of them for services to hospitalized patients. This means that a physician with few Medicare inpatients in a prior year or who devoted, for example, 60% time to hospital care and 40% to outpatient care, would not be counted as being a hospitalist for that year. However, a sensitivity analysis using 50% hospital care as the cut off had similar results.
There are potential explanations of our findings other than lack of experience producing increased deaths. Less experienced hospitalists might be assigned less stable patients. Working in a new hospital with an unfamiliar environment might contribute to the excess mortality associated with first year hospitalists. Hospitalists might become better at coding diagnoses with experience, resulting in artificially lower adjusted mortality for their patients. To reduce this possibility, we did not include discharge diagnoses from the incident admission in assessing comorbidity.
The 21 612 hospitalists that we identified as practicing in 2013 is about half the number estimated by American Hospital Association surveys from that time. This presumably reflects the data we used, Medicare fee-for-service patients, and the criteria employed, such as restricting the analyses to hospitalists who were general internal medicine, family medicine, or geriatric physicians. Hospitalists working predominantly in managed care or with younger patients would not be counted.
Our study has implications for future research, quality improvement and medical education. The worse outcomes for patients cared for by hospitalists in their first year of experience need further investigation into the factors that contribute to this finding. Many new hospitalists may experience dramatic change in their work environment after transitioning from residency. The faster pace and novel systems of hospital care may be difficult for physicians making a transition from primary care to hospital medicine. Our findings should prompt interventions to bolster support systems for new hospitalists such as mentoring or coaching programs. Additionally, medical education changes may have a role to play in mitigating the worse outcomes of first year hospitalists. For example, involving more hospitalist faculty in residency training and creation of hospitalist tracks in residency or hospitalist fellowship programs may prepare individuals better for the challenges early in their hospitalist career.
Conclusions
Patients cared for by physicians in their first year as hospitalists have worse 30-day and hospital mortality when compared with more experienced hospitalists. Hospitalists very early in their careers may benefit from additional support and reduced caseloads.
eTable 1. Association of patient and hospital characteristics with the year of experience of hospitalists for all hospitalizations cared for by 3,860 physicians during their first four years working as hospitalists.
eTable 2. Mortality in the 30 days after admission, and hospital mortality, as a function of the number of years of experience of 2,027 hospitalists who practiced in the same hospital for four years, adjusted for patient and hospital characteristics, in a three level model.
eTable 3. Mortality in the 30 days after admission, and hospital mortality, as a function of the number of years of experience of the 2,026 hospitalists who practiced in the same hospital over their first four years, adjusted for patient and hospital characteristics, and limited to patients with a medial DRG.
References
- 1.Coffman J, Rundall TG. The impact of hospitalists on the cost and quality of inpatient care in the United States: a research synthesis. Med Care Res Rev. 2005;62(4):379-406. [DOI] [PubMed] [Google Scholar]
- 2.Freed DH. Hospitalists: Evolution, evidence, and eventualities. Health Care Manag (Frederick). 2004;23(3):238-256. [DOI] [PubMed] [Google Scholar]
- 3.Wachter RM, Goldman L. The emerging role of “hospitalists” in the American health care system. N Engl J Med. 1996;335(7):514-517. [DOI] [PubMed] [Google Scholar]
- 4.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] [PMC free article] [PubMed] [Google Scholar]
- 5.Pete Welch W, Stearns SC, Cuellar AE, Bindman AB. Use of hospitalists by Medicare beneficiaries: a national picture. Medicare Medicaid Res Rev. 2014;4(2):mmrr2014.004.02.b01. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Wachter RM. The state of hospital medicine in 2008. Med Clin North Am. 2008;92(2):265-273, vii. [DOI] [PubMed] [Google Scholar]
- 7.Rohde JM, Jacobsen D, Rosenberg DJ. Role of the hospitalist in antimicrobial stewardship: a review of work completed and description of a multisite collaborative. Clin Ther. 2013;35(6):751-757. [DOI] [PubMed] [Google Scholar]
- 8.Whelan CT. The role of the hospitalist in quality improvement: systems for improving the care of patients with acute coronary syndrome. J Hosp Med. 2010;5(suppl 4):S1-S7. [DOI] [PubMed] [Google Scholar]
- 9.Thrall TH. Encouraging hospitalists to stick with you. Turnover rate cause for concern; workload big reason for leaving a job. Hosp Health Netw. 2009;83(3):16. [PubMed] [Google Scholar]
- 10.Centers for Medicare & Medicaid Services 2013. Provider of Services Current Files (POS). https://data.medicare.gov/data/archives/hospital-compare. Accessed June 27, 2017.
- 11.Quan H, Sundararajan V, Halfon P, et al. . Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43(11):1130-1139. [DOI] [PubMed] [Google Scholar]
- 12.Research Data Assistance Center. MEDPAR Source inpatient admission code. https://www.resdac.org/cms-data/variables/MEDPAR-Source-Inpatient-Admission-Code. Accessed September 15, 2017.
- 13.Tom S, Roel B, eds. Multilevel Analysis: An Introduction to Basic and Advanced Multilevel Modeling. 2nd ed London, England: Taylor & Francis; 2012. [Google Scholar]
- 14.Flanders SA, Centor B, Weber V, McGinn T, DeSalvo K, Auerbach A. Challenges and opportunities in academic hospital medicine: report from the Academic Hospital Medicine Summit. J Hosp Med. 2009;4(4):240-246. [DOI] [PubMed] [Google Scholar]
- 15.McAlister FA, Youngson E, Bakal JA, Holroyd-Leduc J, Kassam N. Physician experience and outcomes among patients admitted to general internal medicine teaching wards. CMAJ. 2015;187(14):1041-1048. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Southern WN, Bellin EY, Arnsten JH. Longer lengths of stay and higher risk of mortality among inpatients of physicians with more years in practice. Am J Med. 2011;124(9):868-874. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Burns LR, Wholey DR. The effects of patient, hospital, and physician characteristics on length of stay and mortality. Med Care. 1991;29(3):251-271. [DOI] [PubMed] [Google Scholar]
- 18.Blanc PD, Trupin L, Earnest G, et al. . Effects of physician-related factors on adult asthma care, health status, and quality of life. Am J Med. 2003;114(7):581-587. [DOI] [PubMed] [Google Scholar]
- 19.Tsugawa Y, Newhouse JP, Zaslavsky AM, Blumenthal DM, Jena AB. Physician age and outcomes in elderly patients in hospital in the US: observational study. BMJ. 2017;357:j1797. doi: 10.1136/bmj.j.1979 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Choudhry NK, Fletcher RH, Soumerai SB. Systematic review: the relationship between clinical experience and quality of health care. Ann Intern Med. 2005;142(4):260-273. [DOI] [PubMed] [Google Scholar]
- 21.Barry WA, Rosenthal GE. Is there a July phenomenon? The effect of July admission on intensive care mortality and length of stay in teaching hospitals. J Gen Intern Med. 2003;18(8):639-645. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Finkielman JD, Morales J, Peters SG, et al. . Mortality rate and length of stay of patients admitted to the intensive care unit in July. Crit Care Med. 2004;32(5):1161-1165. [DOI] [PubMed] [Google Scholar]
- 23.Ford AA, Bateman BT, Simpson LL, Ratan RB. Nationwide data confirms absence of ‘July phenomenon’ in obstetrics: it’s safe to deliver in July. J Perinatol. 2007;27(2):73-76. [DOI] [PubMed] [Google Scholar]
- 24.Jen MH, Bottle A, Majeed A, Bell D, Aylin P. Early in-hospital mortality following trainee doctors’ first day at work. PLoS One. 2009;4(9):e7103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Phillips DP, Barker GE. A July spike in fatal medication errors: a possible effect of new medical residents. J Gen Intern Med. 2010;25(8):774-779. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Meltzer D, Manning WG, Morrison J, et al. . Effects of physician experience on costs and outcomes on an academic general medicine service: results of a trial of hospitalists. Ann Intern Med. 2002;137(11):866-874. [DOI] [PubMed] [Google Scholar]
- 27.Auerbach AD, Wachter RM, Katz P, Showstack J, Baron RB, Goldman L. Implementation of a voluntary hospitalist service at a community teaching hospital: improved clinical efficiency and patient outcomes. Ann Intern Med. 2002;137(11):859-865. [DOI] [PubMed] [Google Scholar]
- 28.Soong C, Welker JA, Wright SM. Hospitalist organizational structures in the Baltimore-Washington area and outcomes: a descriptive study. J Clin Outcomes Manag. 2011;18(3):107-111. [Google Scholar]
- 29.Burke LG, Frakt AB, Khullar D, Orav EJ, Jha AK. Association between teaching status and mortality in US hospitals. JAMA. 2017;317(20):2105-2113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.American Society of Anesthesiologists, Health Policy Research Prevalence of Hospitalists in U.S. Community Hospitals: Data from the American Hospital Association, 2012-2013. http://www.asahq.org/resources/health-policy-research/hpr-resources. Accessed June 27, 2017.
- 31.Singh S, Fletcher KE, Schapira MM, et al. . A comparison of outcomes of general medical inpatient care provided by a hospitalist-physician assistant model vs a traditional resident-based model. J Hosp Med. 2011;6(3):122-130. doi: 10.1002/jhm.826 [DOI] [PubMed] [Google Scholar]
- 32.Roy CL, Liang CL, Lund M, et al. . Implementation of a physician assistant/hospitalist service in an academic medical center: impact on efficiency and patient outcomes. J Hosp Med. 2008;3(5):361-368. doi: 10.1002/jhm.352 [DOI] [PubMed] [Google Scholar]
- 33.Walker C, Catignani K. Academic Hospitalist On-Boarding Mentor Program evaluation. J Hosp Med. 2013;8(Suppl 2). http://www.shmabstracts.com/abstract/academic-hospitalists-onboarding-mentor-program-evaluation/. Accessed March 19, 2017. [Google Scholar]
- 34.Iyasere CA, Baggett M, Romano J, Jena A, Mills G, Hunt DP. Beyond continuing medical education: clinical coaching as a tool for ongoing professional development. Acad Med. 2016;91(12):1647-1650. [DOI] [PubMed] [Google Scholar]
- 35.Ranji SR, Rosenman DJ, Amin AN, Kripalani S. Hospital medicine fellowships: works in progress. Am J Med. 2006;119(1):72.e1-72.e7. [DOI] [PubMed] [Google Scholar]
- 36.Beasley BW, McBride J, McDonald FS. Hospitalist involvement in internal medicine residencies. J Hosp Med. 2009;4(8):471-475. doi: 10.1002/jhm.522 [DOI] [PubMed] [Google Scholar]
- 37.Glasheen JJ, Siegal EM, Epstein K, Kutner J, Prochazka AV. Fulfilling the promise of hospital medicine: tailoring internal medicine training to address hospitalists’ needs. J Gen Intern Med. 2008;23(7):1110-1115. doi: 10.1007/s11606-008-0646-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Kumar A, Smeraglio A, Witteles R, et al. . A resident-created hospitalist curriculum for internal medicine housestaff. J Hosp Med. 2016;11(9):646-649. doi: 10.1002/jgm.2590 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1. Association of patient and hospital characteristics with the year of experience of hospitalists for all hospitalizations cared for by 3,860 physicians during their first four years working as hospitalists.
eTable 2. Mortality in the 30 days after admission, and hospital mortality, as a function of the number of years of experience of 2,027 hospitalists who practiced in the same hospital for four years, adjusted for patient and hospital characteristics, in a three level model.
eTable 3. Mortality in the 30 days after admission, and hospital mortality, as a function of the number of years of experience of the 2,026 hospitalists who practiced in the same hospital over their first four years, adjusted for patient and hospital characteristics, and limited to patients with a medial DRG.