The Status of Hospital Medicine Groups in the United States

Home , Community hospital, Teaching hospital

ORIGINAL RESEARCH

Peter D. Kralovec1 BACKGROUND: Hospitalists, deﬁned as hospital-based physicians who take respon- 2 Joseph A. Miller, MS sibility for managing the medical needs of inpatients, represent a signiﬁcant trend 2 Laurence Wellikson, MD, FACP in physician specialization. However, only limited anecdotal data quantifying the Jeanne M. Huddleston, MD, FACP3 status of hospital medicine groups around the country is available. OBJECTIVE: To better understand the extent and nature of the hospitalist move- 1 Health Care Data Center, Health Forum, Ameri- ment, utilizing data from the 2003 Annual Survey of the American Hospital Asso- can Hospital Association, Chicago, Illinois ciation (AHA). 2 Society of Hospital Medicine, Philadelphia, Penn- STUDY POPULATION: 4895 acute care hospitals in the United States. sylvania MEASUREMENTS: Number and percentage of hospitals with hospital medicine 3 Mayo Clinic, Rochester, Minnesota groups; mean number of hospitalists per group; hospitalists per average daily census (ADC) of 100 patients; distribution of groups by employment model. DESCRIPTIVE VARIABLES: Census region; rural/urban status; number of beds; or- ganizational control; teaching status. RESULTS: There are approximately 1415 hospital medicine groups and 11 159 hospitalists in the United States. The overall penetration of hospital medicine groups at hospitals is 29% (55% at hospitals with 200 or more beds), and the in-hospital impact at hospitals with hospital medicine groups is 3.93 hospitalists per 100 ADC. The average hospital medicine group has 7.9 hospitalists. There is a fairly equal distribution among the 3 major employment models for hospital medicine groups: hospital employees, independent provider groups, and physician groups. All these measures can vary substantially, depending on the characteristics of individual hospitals. CONCLUSIONS: Hospital medicine appears to have become part of the mainstream delivery of health care in the United States. No employment model of hospital medicine group appears to dominate this specialty. We expect there will continue to be growth and diversity in the implementation of hospital medicine groups. Journal of Hospital Medicine 2006;1:75–80. © 2006 Society of Hospital Medicine.

KEYWORDS: hospitalist, hospital medicine, hospital medicine group, hospitalist movement, AHA survey, hospital medicine status, hospital medicine growth, hospital medicine characteristics, hospitalist movement status, hospitalist movement growth, hospitalist movement characteristics

he term hospitalist was coined in 1996 in an article1 that Tappeared in the New England Journal of Medicine. Robert M. Wachter, MD, and Lee Goldman, MD, of the University of Califor- nia, San Francisco, deﬁned hospitalists as hospital-based physicians who take responsibility for managing medical inpatients. Hospitalists were described as having responsibility for seeing This article contains supplementary material avail- unassigned hospital patients and being available for in-hospital able via the Internet at http://www.interscience.wi- consultations. Several years later, the Society of Hospital Medicine ley.com/jpages/1553–5592/suppmat. posted the deﬁnition of a hospitalist as someone whose primary

© 2006 Society of Hospital Medicine 75 DOI 10.1002/jhm.82 Published online in Wiley InterScience (www.interscience.wiley.com). professional focus is the medical care of hospital- DATA AND METHODS ized patients—in patient care, education, research, The data for our analysis came from the 2003 AHA and administrative activities. Annual Survey. Conducted since 1946, this survey is In January 2002, Wachter and Goldman pub- the principal data collection mechanism of the lished a follow-up article,2 “The Hospitalist Move- American Hospital Association and is a basic source ment 5 Years Later,” in the Journal of the American of data on hospitals in the United States about the Medical Association. This formal review of 19 pub- availability of services, utilization, personnel, fi- lished studies analyzed the impact of hospital med- nances, and governance. Its main purpose is to icine groups on financial and clinical outcomes. provide a cross-sectional view of hospitals and hos- Wachter and Goldman concluded, “Empirical re- pital performance over time. In the 2003 survey, a search supports the premise that hospitalists im- series of items were added about hospitalists in- prove inpatient efficiency without harmful effects cluding whether hospitals had hospital medicine on quality or patient satisfaction.” These studies groups, the number of hospitalists operating in indicate an average reduction of cost per stay of such groups, and the employment model used. 13.4% and an average reduction in length of stay of The study population for this analysis was lim- 16.6%. ited to US community hospitals (n ϭ 4895). Com- The evolution of the hospitalist movement has munity hospitals are defined as all adult and pedi- been fast paced and extensive. Given the recent atric nonfederal, short-term general, and specialty pace of growth, a scholarly analysis estimated that hospitals whose facilities and services are available the mature hospitalist workforce in the United to the public. Excluded from the analysis were all States will eventually total 20,000, making it the federal hospitals, long-term care hospitals, and psy- equivalent of the cardiology specialty.3 Beyond chiatric hospitals. sheer growth, medical literature has demonstrated positive effects of the hospitalist model on patient quality outcomes, including readmission rates, Imputation of Missing Data postoperative complications, and mortality.4–7 In the 2003 survey, 77% of the 4895 US community In addition to peer-reviewed medical literature, hospitals answered the question on specific use of there is anecdotal evidence about the growth and hospitalists. To get a complete picture of the num- effects of the hospitalist movement: ber of groups and hospitalists, we imputed data for the nonresponding hospitals. ● The Society of Hospital Medicine (SHM), the hospi- We performed logistic regression analysis of talist professional society, estimated that in 2003 data from the responding hospitals to estimate the there were 8000 physicians practicing as hospitalists number of nonresponding hospitals that had a in the United States.8 group and the number of hospitalists in these ● Twelve of the country’s “top” 15 hospitals have hos- groups. The dependent variable in the regression pital medicine groups.8 was whether a hospital had a group, and the independent variables included hospital characteristics As hospital medicine groups have proliferated, for which data were available for all US hospitals, 4 major employment models have evolved. Hospi- both survey respondents and nonrespondents. The talists can be employees of: 1) a hospital or a hos- results of the regression analysis were then applied pital subsidiary; 2) a multispecialty or primary care to the data for each nonresponding hospital to es- physician group; 3) a medical group (local or na- timate its probability of having a group. These tional) of independent hospitalists; or 4) a univer- probabilities were summed over the various nonre- sity or medical school. However, there is little pub- sponding hospitals to estimate the total number of lished data on the prevalence of each of these nonresponding hospitals that had groups. hospitalist employment models, nationally or by To impute the number of hospitalists in the type of hospital. nonresponding set of hospitals, the additional To better understand the extent and nature of number of groups was stratified into the 9 US Cen- the hospitalist movement, the American Hospital sus Divisions. On the basis of reported data, the Association (AHA) utilized its 2003 Annual Survey to average number of hospitalists per group was cal- gather data on hospital medicine groups in the culated at the Census Division level. The per-group United States value was then applied to the number of additional

76 Journal of Hospital Medicine Vol1/No2/Mar/Apr 2006 groups, and the result was added to the total num- TABLE 1A ber of reported hospitalists. The Census Division Diffusion of Hospital Medicine Groups by Geographic Categories values were then summed to produce the national Hospitals with total. To produce results for all other control group- Hospital hospital ings, the national total was then apportioned across medicine medicine the categories according to percentage of hospital- Category Hospitals groups groups (%) ists by category on the basis of the reported data. Region 1: Northeast 203 94 46% Analytical Plan 2: Mid-Atlantic 486 172 35% In analyzing the hospitalist movement across the 3: South-Atlantic 731 272 37% country, we realized there are 2 dimensions of dif- 4: East North Central 732 209 29% fusion, which can be characterized as “breadth” 5: East South Central 427 92 22% and “depth.” In the present study: 6: West North Central 675 106 16% 7: West South Central 737 164 22% ● The measure of breadth is the percentage of hospital 8: Mountain 348 83 24% medicine groups in a given group of hospitals. In the 9: Pacific 556 223 40% Rural/urban Results section, this measure is sometimes referred Rural 2166 235 11% to as “penetration.” Small urban 1285 488 38% ● The measure of depth is the number of hospitalists for Large urban 1444 692 48% each average daily census (ADC) of 100 patients. For Total 4895 1415 29% instance, for a hospital with an average daily census of Source: AHA Annual Survey file for 2003 100 that has 4 hospitalists, that measure is 4. To com- Region Definitions: 1 (ME, VT, NH, MA, RI, VT); 2 (NY, NJ, PA); 3 (DE, DC, MD, VA, NC, SC, WV, GA, FL); pute this metric for a given category of hospitals (eg, 4 (OH, IN, IL, MI, WI); 5 (KY, TN, AL, MS); 6 (IA, KS, MO, NE, MN, SD, ND); 7 (LA, AR, OK, TX); 8 (CO, major teaching hospitals), the numerator is the num- WY, UT, AZ, NM, NV, MT, ID); 9 (CA, OR, WA, AK, HW) ber of hospitalists and the denominator is the ADC at hospitals that have hospital medicine groups. The metric reflects the in-hospital impact of hospital medicine groups at their hospitals. TABLE 1B Using these 2 measures, it is possible to differ- Diffusion of Hospital Medicine Groups by Size, Control, and Teaching entiate between a group of hospitals that has many Status hospital medicine groups but each group has a minimal impact at the hospital versus a group of Hospital Hospitals with medicine hospital medicine hospitals that has few hospital medicine groups but Category Hospitals groups groups (%) each group has a major impact at the hospital. The analysis also characterizes the employment Size status of hospitalists by comparing the proportion 6-24 beds 327 18 6% of hospitals in each of the employment models by 25-49 beds 965 88 9% 50-99 beds 1031 168 16% category of hospital. 100-199 beds 1168 372 32% 200-299 beds 624 287 46% RESULTS 300-399 beds 349 183 52% Diffusion and Impact 400-499 beds 172 116 67% ϩ Overall, the penetration of hospital medicine 500 beds 259 183 71% Control groups across the 4895 hospitals in the United Government 1121 161 14% States is 29% and the in-hospital impact at hospitals Not for profit 2984 1032 35% with hospital medicine groups is 3.93 hospitalists For profit 790 222 28% per 100 ADC. The average hospital medicine group Teaching status has 7.9 hospitalists at a hospital with an ADC of Nonteaching 3800 823 22% Other teaching 779 382 49% 200.6. Major teaching 316 210 66% Total 4895 1415 29% Geographic Categories (Tables 1A and 2A) The Northeast (46%) and the Pacific (40%) divisions Source: AHA Annual Survey file for 2003. have the greatest penetration of hospital medicine Major teaching defined as a member of the Council of Teaching Hospitals and Health Systems (COTH).

Status of US Hospital Medicine Groups / Kralovec et al. 77 TABLE 2A TABLE 2B Impact of Hospitalists on Their Hospitals by Geographic Categories Impact of Hospitalists on Their Hospitals by Size, Control, and Teaching Status Groups Hospitalists Hospitalists per Category (hospitals) Hospitalists per group 100 census Groups Hospitalists per Hospitalists per Category (hospitals) Hospitalists group 100 census Region 1: Northeast 94 669 7.1 3.62 Size 2: Mid-Atlantic 172 1133 6.6 2.42 6-24 beds 18 38 2.1 46.34 3: South Atlantic 272 1933 7.1 3.21 25-49 beds 88 260 3.0 17.94 4: East North Central 209 2087 10.0 4.65 50-99 beds 168 885 5.3 12.75 5: East South Central 92 433 4.7 2.83 100-199 beds 372 1757 4.7 5.29 6: West North Central 106 887 8.4 4.37 200-299 beds 287 2308 8.0 4.72 7: West South Central 164 1828 11.1 6.24 300-399 beds 183 1,553 8.5 3.29 8: Mountain 83 644 7.8 4.43 400-499 beds 116 1751 15.1 4.35 9: Pacific 223 1546 6.9 4.56 500ϩ beds 183 2,607 14.2 2.47 Rural/urban Control Rural 235 893 3.8 4.85 Government 161 1,674 10.4 5.85 Small urban 488 3236 6.6 3.03 Not for profit 1032 8,481 8.2 3.64 Large urban 692 7030 10.2 4.43 For profit 222 1,004 4.5 4.47 Total 1415 11 159 7.9 3.93 Teaching Status Nonteaching 823 4,910 6.0 4.85 Source: AHA Annual Survey file for 2003. Other teaching 382 2,678 7.0 3.25 Region definitions: 1 (ME, VT, NH, MA, RI, VT); 2 (NY, NJ, PA); 3 (DE, DC, MD, VA, NC, SC, WV, GA, FL); Major teaching 210 3,571 17.0 3.57 4 (OH, IN, IL, MI, WI); 5 (KY, TN, AL, MS); 6 (IA, KS, MO, NE, MN, SD, ND); 7 (LA, AR, OK, TX); 8 (CO, Total 1415 11 159 7.9 3.93 WY, UT, AZ, NM, NV, MT, ID); 9 (CA, OR, WA, AK, HW). Source: AHA Annual Survey file for 2003 Major teaching defined as a member of the Council of Teaching Hospitals and Health Systems (COTH). groups. The West North Central Division (16%) has the lowest penetration of hospital medicine groups. 19% of hospitals with fewer than 200 beds. As would Hospital medicine groups in the West South Cen- be expected, larger hospitals have larger hospital tral Division average 11.1 hospitalists, which par- medicine groups: hospitals with 6-24 beds average tially explains why this region has the greatest in- 2.1 hospitalists, whereas hospitals with 500ϩ beds hospital impact (6.24 hospitalists per 100 ADC). At average 14.2 hospitalists. However, hospitalists the other end of the spectrum are the Middle At- have a proportionately greater impact at smaller lantic and East South Central divisions with (2.42 hospitals. Their greatest impact is at hospitals with and 2.83 hospitalists per 100 ADC, respectively. 6-24 beds (46.34 hospitalists per 100 ADC); their There are more hospital medicine groups in smallest impact is at hospitals with 500ϩ beds (2.47 urban locations. The penetration of hospital medi- hospitalists per 100 ADC). cine groups is 48% at hospitals in large metropoli- Of the 3 categories of control, government tan locations (ie, with a population of more than 1 groups have the lowest penetration of hospital million), 38% at hospitals in small metropolitan medicine groups (14%). However, the hospital locations, and 11% at hospitals in rural areas. How- medicine groups at these government-controlled ever, rural hospitals have a relatively high in-hos- hospitals are large (10.4 hospitalists), and they have pital impact (4.85 hospitalists per 100 ADC), ex- a significant in-hospital impact on care at these plained by an average group size of 3.8 and an hospitals (5.85 hospitalists per 100 ADC). Not-for- average ADC of 78.4. profit hospitals have the highest penetration of hospital medicine groups (35%), whereas hospital Hospital Size, Control/Ownership, and Teaching Status medicine groups at for-profit hospitals have the (Tables 1B and 2B) lowest average size (4.5 hospitalists). The penetration of hospital medicine groups in- There appears to be a relationship between creases as the size of the hospital increases. Six teaching status and the likelihood that a hospital percent of hospitals with 6-24 beds have groups, has a hospital medicine group. The penetration of whereas 71% of hospitals with 500ϩ beds have hospital medicine groups is 66% at major teaching groups. Among hospitals with 200 or more beds, hospitals, 49% at other teaching hospitals, and 22% 55% have hospital medicine groups compared to at nonteaching hospitals. However, nonteaching

78 Journal of Hospital Medicine Vol1/No2/Mar/Apr 2006 hospitals have a relatively high in-hospital impact landscape. We expect that there will continue to be (4.85 hospitalists per 100 ADC). This is explained by diversity among the organizations that choose to their having an average group size of 6.0, but an establish hospital medicine groups. average ADC of only 123.0 (compared to 477.0 for In light of this growth and diversity, hospital major teaching hospitals and 215.7 for other teach- medicine groups appear to be valued by a wide ing hospitals). range of stakeholders in the health care industry. The potential benefits provided by hospitalists in- Employment Models clude financial savings, improved throughput effi- The results of the analysis of hospitalist employ- ciency, improved quality and safety, improved ment models (data not shown) can be summarized medical education, and better provider satisfaction. as follows: Despite this success story, the hospitalist movement has maintained a relatively low profile among ● Employees of hospitals: This employment model av- consumers and some segments of the health care eraged 33% of all groups, with an average size of 9.8 industry. This is likely to change. As the hospital hospitalists. The employees of hospital model was medicine specialty gains recognition, hospitalists more prevalent in the Mid-Atlantic (56%), New En- will receive increased scrutiny and attention. This gland (49%), and West North Central (45%) regions emerging specialty will need to be able to clearly and in rural hospitals (45%). The East South Central define its role and document its performance in the (16%) and West South Central (12%) regions and constantly changing health care industry. for-profit hospitals (20%) had fewer hospital employee groups. ADDENDUM ● Employees of medical groups: This employment Subsequent to the acceptance of this manuscript, model averaged 29% of all groups, with an average the authors received results of the 2004 Annual of 7.4 hospitalists. More hospitals in the East South Survey of the American Hospital Association. Some Central (35%) and New England (34%) regions had highlights of the new data and comparisons to the this employment model. Fewer hospitals in the Mid- 2003 results are as follows: Atlantic (18%) and West North Central (18%) regions ● and rural (18%) hospitals had medical group-based The penetration of hospitals with hospital medicine groups. groups grew from 29% to 34% (for hospitals with ϩ ● Employees of independent hospitalist groups: This 200 beds, the penetration grew from 55% to 63%) ● employment group averaged 25% of all groups and An estimated 1,661 hospitals have hospital medicine had the smallest mean number of hospitalists (6.9). groups (an increase of 17% from 2003) ● This employment model was more prevalent in for- The average size of a hospital medicine group de- profit hospitals (43%) and was less prevalent in the creased from 7.9 physicians to 7.5 physicians (a de- New England (9%) and Mid-Atlantic (11%) regions crease of 5%) ● and in major teaching hospitals (11%) and govern- It is estimated that there are 12,504 hospitalists in ment hospitals (19%). the U.S. (an increase of 12% from 2003) ● Hospital medicine groups remain equally distrib- CONCLUSIONS uted among the three employment models: employees of hospitals 30%, employees of medical groups Hospital medicine groups appear to have become 29%, employees of independent hospitalist groups part of the mainstream delivery of health care. With 29% more than 11 000 hospitalists, the specialty is equivalent in size to the gastroenterology medical These updated results indicate strong hospitalist specialty.9 Fifty-five percent of hospitals with more growth over the one year period and continued than 200 beds have hospital medicine groups. Fur- diversity among hospital medicine programs, rein- thermore, it appears that the growth of the hospi- forcing the conclusions of the manuscript. talist movement has not peaked. It is likely that the number of hospitals with hospital medicine groups APPENDIX will increase and that existing hospital medicine AHA Annual Survey Overview groups will continue to add hospitalists. Conducted since 1946, the AHA Annual Survey is No one employment model of hospital medi- the principal data collection mechanism of the cine group appears to dominate the health care American Hospital Association and is a basic source

Status of US Hospital Medicine Groups / Kralovec et al. 79 of data on hospitals in the United States. Its main ᮀ Independent provider group purpose is to provide a cross-sectional view of the ᮀ Employed by your hospital hospital field each year and to make it possible to ᮀ Employed by a physician group monitor hospital performance over time. The infor- ᮀ Employed by a university or school program mation that it gathers from a universe of approxi- ᮀ Other mately 5700 hospitals concerns primarily the avail- It is the results from these questions that are ability of services, utilization, personnel, finances, the subject of this analysis and the manuscript. and governance. Newly added to the 2003 survey were the following questions regarding hospitalists: Address for correspondence and reprint requests: Joseph A. Miller, MS, Society of Hospital Medicine, 190 N Independence Mall West, Suite 340, Philadelphia, Do hospitalists provide care for patients in your PA 19106; Fax: (215) 351-2536; E-mail: [email protected] hospital? YES ᮀ NO ᮀ Received 3 November 2005; revision received 27 December 2005; accepted 22 ● Hospitalist is defined as a physician whose primary January 2006. professional focus is the care of hospitalized medical REFERENCES patients (through clinical, education, administrative 1. Wachter RM, Goldman, L. The emerging role of “hospital- and research activity). ists” in the American health care system. N Eng J Med. 1996;335:514-517. If yes, please report the number of full time and 2. Wachter RM, Goldman, L. The hospitalist movement 5 years part time hospitalists? later. JAMA. 2002;287:487-494. 3. Lurie JD, Miller DP, Lindenauer PK, Wachter RM, Sox HC. Full-time ______The potential size of the hospitalist workforce in the United Part-time ______States Am J Med. 1999;106:441-445. 4. Auerbach, A. Implementation of a hospitalist service at a ● Full-time equivalent (FTE) is the total number of community hospital: evolution of service utilization, costs, hours worked by all employees over the full (12 and patient outcomes [abstract]. National Association of month) reporting period divided by the normal Inpatient Physicians, 3rd Annual Meeting. Philadelphia, number of hours worked by a full-time employee for Penn, April 11-12, 2000. 5. Meltzer D, Manning W, Shah M, Morrison J, Jin L, Levinson that same period. For example, if your hospital con- W. Decreased length of stay, costs, and mortality in a ran- siders a normal workweek for a full-time employee domized trial of academic hospitalists [abstract]. National to be 40 hours, a total of 2080 hours would be Association of Inpatient Physicians, 4th Annual Meeting, worked over a full year (52 weeks). If the total num- Atlanta, GA, March 27-28, 2001. ber of hours worked by all employees on the payroll 6. Diamond HS Goldberg E, Janosky JE. The effect of full-time faculty hospitalists on the efficiency of care at a community is 208 000, then the number of FTEs is 100 (employ- teaching hospital. Ann Intern Med. 1998;129:197-203. ees). The FTE calculation for a specific occupational 7. Aberhaim HA, Kahn SR, Raffoul J, Becker MR. Program category such as registered nurses is exactly the description: a hospitalist run, medical short-stay unit in a same. The calculation for each occupational cate- teaching hospital. CMAJ. 2000;163:1477-1480. gory should be based on the number of hours 8. Society of Hospital Medicine. Growth of hospital medicine nationwide. July 2003. Available at: http://www.hospital- worked by staff employed in that specific category. medicine.org/presentation/apps/indlist/intro.asp?flagϭ18. Accessed February 2005. If yes, please select the category below that best 9. American Medical Association. Physician characteristics describes the employment model for your hospital- and distribution in the US, 2004. Chicago, Ill: American ists: Medical Association, 2004.

80 Journal of Hospital Medicine Vol1/No2/Mar/Apr 2006