The Hospitalist Movement 5 Years Later FREE

Five years ago, we described a burgeoning model of care in the United States, in which a new breed of physicians, whom we dubbed "hospitalists," provide inpatient care in place of primary care physicians or academic 1-month-per-year attendings.¹ We cited numerous forces fomenting this change, including cost pressures on hospitals, physician groups, and managed care organizations; the increased acuity of hospitalized patients and the accelerated pace of their hospitalizations; the time pressures on primary care physicians in the office; the decreasing inpatient volumes of most primary physicians; and the evidence that practice makes perfect in other medical fields. We predicted that the hospitalist movement would grow and ultimately become a, if not the, dominant model for inpatient care in the United States. Finally, we argued that judgments regarding the hospitalist model should be informed by data on cost, quality, education, and patient satisfaction.

Our original article generated several negative responses,^2- 4 largely focused on the purposeful discontinuity in care introduced by the hospitalist model. Later, concerns were raised when a few managed care organizations attempted to mandate the use of hospitalists.^5- 7 Several medical societies mounted vigorous opposition to mandatory hospitalists, including the National Association of Inpatient Physicians (NAIP), the hospitalists' newly established professional society. Since then, attempts to create mandatory programs by managed care organizations have become unusual,⁸ and physicians and their societies have increasingly accepted voluntary hospitalist systems.^9- 11

This acceptance is reflected in the remarkable growth in the hospitalist model over the past several years. By 1999, 65% of internists had hospitalists in their community and 28% reported using them for inpatient care.¹⁰ The referral rate was even higher (62%) among primary care physicians in California.¹¹ There are hospitalist programs at many prestigious US hospitals (BOX 1). In its first 3 years, NAIP has grown to more than 2300 members. A recent analysis projected an ultimate US hospitalist workforce of about 19 000 (up from 5000 presently), which would make it comparable in size to cardiology.¹⁹

Barnes-Jewish Hospital
Brigham and Women's Hospital¹³
Cleveland Clinic¹⁴
Duke University Medical Center
Hospital of the University of Pennsylvania¹⁵
Massachusetts General Hospital
Mayo Clinic
Stanford University Hospital
UCLA Medical Center¹⁶
University of California, San Francisco Medical Center¹⁷
University of Chicago Hospitals¹⁸
University of Michigan Medical Center

*As defined by US News & World Report¹² as of December 2001 and presented in alphabetical order.

Most of the early growth of the hospitalist movement occurred in the absence of data. Published data now illuminate some key successes and challenges for the movement as it enters its next 5 years.

We searched MEDLINE, BIOSIS, EMBASE, and the Cochrane Library from 1996 to September 2001 to identify studies, published in either manuscript or abstract form. When abstracts led to manuscripts reporting the same population of patients, only the manuscript was included. When preliminary results were cited, we contacted authors for updated results. We searched on both title and key words for "hospitalist" or "hospitalists," and included studies that reported data comparing hospitalists with an appropriate nonhospitalist control group on resource use and/or clinical outcomes. We extracted information regarding study design, nature of hospitalist and control groups, use of risk or other adjustment, and key outcomes. Because of the newness of the literature and the relative dearth of methodologically rigorous studies, a formal statistical analysis (eg, meta-analysis) was not attempted.²⁰

We also aimed to describe and analyze several additional hospitalist-related thematic issues, including training, expanding roles for hospitalists, patient and provider satisfaction, and financing hospitalist programs. We performed focused literature searches in these areas in addition to the systematic review of efficiency and quality.

We found 19 published studies (12 articles, 7 abstracts) regarding the impact of hospitalist programs on financial and clinical outcomes (Table 1a).^18,21- 44 Viewed together, the studies cover all health care delivery settings: teaching and nonteaching hospitals, urban and rural locations, adult and pediatric services, and managed care and nonmanaged care environments. Fifteen of the 19 studies found significant decreases in both hospital costs (average decrease, 13.4%) and lengths of stay (average decrease, 16.6%).^{18,21- 22,25,28- 38,40- 42,44} Two studies demonstrated decreases in average length of stay but no decreases in hospital costs.^23,27 Only 2 of 19 studies found neither cost nor length of stay reductions.^39,43 In 1 study, the nonhospitalists had attended an average of 2.5 months in the year before the intervention,³⁹ a very different group than the more typical comparison groups of primary care physicians or academic 1-month-per-year attendings. In the other study,⁴³ the control group was a family practice teaching service in which a rotating group of 4 physicians served as dedicated inpatient attendings 2 weeks out of 8 weeks, thus meeting one of the published definitions of hospitalist.⁴⁵

Proponents of the hospitalist model postulate that hospitalists improve inpatient care quality, both through practice makes perfect and through their onsite availability throughout the day. Critics have raised concerns that quality might be sacrificed on the altar of efficiency. Thus far, there is little to suggest that hospitalist-generated savings come at the expense of quality (Table 1). Most studies have found no change in quality measures. One study demonstrated a significant decrease in readmission rates,²² while another showed an increase in potentially preventable readmissions.³³ One very small observational study (n = 99 hospitalist patients) of heart failure patients showed an increased hospital mortality rate in the hospitalist group (2.0% vs 1.5%).²³ However, 2 much larger (averaging 1600 hospitalist patients per study) and more methodologically rigorous studies (1 quasi-randomized, the other adjusted for case-mix) found significant decreases in inpatient and short-term mortality, which grew to approximately a 33% relative reduction in the hospitalist programs' second year (4.2% vs 6.0% in 1 study, 4.8% vs 7.2% in the other).^18,37 These results, while provocative, are insufficient to support an unqualified statement that hospitalists improve quality. In addition to more large and rigorous outcome studies, future studies should use more refined measures of quality, including functional status outcomes and process measures associated with improved outcomes, such as the use of appropriate medications for heart failure, rapid administration of antibiotics in pneumonia, and smoking cessation counseling for patients admitted with asthma.^46- 47

The published outcomes literature has generally upheld the promise of the hospitalist model, as demonstrated by impressive and consistent cost savings associated with no decrease in quality. Assuming the average US hospitalist cares for 600 inpatients yearly⁴⁸ and generates a 10% savings over the average medical inpatient cost of $8000, the nation's 5000 hospitalists would reduce inpatient costs by approximately $2.4 billion per year.

A major early concern was that patients accustomed to having their primary physician as their inpatient attending would not accept hospitalists.⁴⁵ In general, however, surveys of patients who were cared for by hospitalists show high levels of satisfaction, no lower than that of similar patients cared for by their own primary physicians^28,31- 32 or by traditional academic ward attendings.^18,21 We have postulated that patients may be willing to trade off the familiarity of their regular physician for the availability of the hospitalist.⁴⁵

Initial resistance to the hospitalist movement among physicians often focused on the unavoidable discontinuity in care created by the model and the potential loss of information across the hospital threshold.^45,49- 52 Effective hospitalist programs have created mechanisms to mitigate the impact of this discontinuity, including calling primary care physicians on admission and discharge, faxing daily progress notes, and encouraging primary care physicians to visit or call their hospitalized patients. Though some concerns about information transfer linger, 2 recent surveys suggest that most physicians now accept the hospitalist model. In a national telephone survey of 400 internists, 51% (204) thought hospitalists might provide better care and 46% (184) thought patients might get more cost-effective care. Although 73% were concerned about the impact of hospitalists on continuity, physicians with hospitalists in their community were more approving.¹⁰ In a survey of 524 California primary care physicians, physicians perceived hospitalists as increasing (41%) or not changing (44%) the overall quality of care and most (55%) thought that hospitalists increase inpatient efficiency.¹¹ In both surveys, primary care physicians stated their belief that patients generally preferred to be cared for in the hospital by their regular physician. Surveys of both generalists and specialists at Park Nicollet showed high levels of physician satisfaction several years after the implementation of a hospitalist program.²⁸

Although concerns remain about potential burnout, surveys of NAIP members conducted in 1997⁵³ and 2000 (A. Auerbach, MD, MPH, written communication, December 2001) found that more than 80% expected to be hospitalists in 2 to 3 years. In a 1999 NAIP survey, 64% still expected to be hospitalists in 7 years.⁵⁴ Using a standard scale, the 1999 survey found that 62% were at no current risk of burnout, 25% were at risk for burnout, and 13% were in fact burned out. These burnout rates are significantly lower than those documented in similar surveys of intensivists and emergency medicine physicians,^55- 57 although rates may climb as the hospitalist field matures.⁵⁴ Most hospitalists chose the field for its "intellectual stimulation and enjoyment," "chance to be involved in an emerging specialty," and "chance to have a job with predictable hours and lifestyle."⁵⁴

Until recently, there were no specific training programs for hospitalists. Thus, hospitalists have developed from diverse training backgrounds. The 1997 NAIP survey indicated that 90% of hospitalists were internists: 50% general internists and 40% medical subspecialists. Of this latter group, half were pulmonologists and/or critical care specialists.⁵³ The 1999 survey found that the generalist fraction had grown to 75%, while subspecialists fell to 15%.⁵⁴ Our impression is that early programs often took advantage of the onsite presence of pulmonary physicians, intensivists, and other subspecialists. The recent data indicate that the maturing field is attracting more general internists who view it as a long-term career option. In both surveys, about 3% to 5% of hospitalists were family physicians and another 5% to 7% were pediatricians.⁵⁸

There are now several early training programs for hospitalists, including a residency track and fellowship program at the University of California, San Francisco.⁵⁹ In developing training modules for hospitalists, researchers have been guided by a survey in which practicing hospitalists rated areas of importance to their practice and sufficiency of their training (Table 2).⁶⁰ Interestingly, hospitalists thought that clinical skills (ie, managing heart failure, inserting central lines, interpreting electrocardiograms) were very important and had been well taught during residency. Conversely, they cited major educational deficits in their training regarding communication skills, end-of-life care, quality improvement and patient safety, medical economics, care of surgical patients, and postacute care. These topics are likely to form the core of future hospitalist curricula for both trainees and practicing physicians. Although graduating residents raised concerns about underpreparedeness in some of the same areas, such as in nursing home care and quality improvement, in a recent national survey,⁶¹ other educational deficits cited by hospitalists (eg, palliative care) were not raised to the same extent by these graduating residents. Moreover, the illustrative cases and settings used to train future hospitalists would undoubtedly be different than those used to educate outpatient generalists, even within the same general content area, such as pain management or patient safety.

Initially, hospitalists engaged primarily in traditional hospital internal medicine or pediatric practice, caring for medical inpatients and performing medical consultations on nonmedical patients. The data showing improvements in efficiency with at worst stasis (and possibly improvements) in outcomes and satisfaction relate to these traditional functions. Like a drug that is approved for specific indications but finds broader use in practice, hospitalists are being employed in a number of off-label functions, including emergency department triage and managing hospital-to-hospital transfers.⁵³

An area of controversy is the role of hospitalists in the intensive care unit. Approximately 80% of hospitalists care for their own patients in intensive care units,⁵³ frequently or always calling on intensivists as consultants. Although the literature demonstrates improved outcomes and efficiency when intensivists participate in the care of critically ill patients,^62- 66 the comparison groups in these studies were either nonhospitalist ward attendings or community-based primary care physicians. There are no data comparing hospitalist-based intensive care (with intensivist consultation) to intensivist-based care. Further research is needed to determine the optimal models for intensive care when both hospitalists and intensivists are available. Collaborative hospitalist-intensivist models will become increasingly important if the projected national shortage of critical care physicians materializes.⁶⁷

Finally, in many systems hospitalists are moving beyond the traditional role of medical consultant to a new role as physician-of-record for nonmedical inpatients with medical comorbidities. Just as with the primary care physician whose office obligations make inpatient management and discharge planning difficult during daylight hours, the surgeon is similarly obligated in the operating room. In the 1997 NAIP survey, nearly half the hospitalists had added this comanagement role to their repertoire.⁵³ To date, there are no data regarding the impact of hospitalists on the care of surgical patients, making this an important area for future research.

Most early hospitalist groups formed in community hospitals. Over the past 5 years, many academic medical centers have adopted hospitalist models for inpatient care and teaching. Beyond the legitimacy that this expansion into academia affords the movement,⁶⁸ the 2 greatest effects are on education and research.⁶⁹

Academic hospitalists are emerging as core teachers of inpatient medicine. For example, at the University of California, San Francisco, 15 faculty hospitalists now staff about two thirds of ward-attending months and all medical consult months. Preliminary evidence indicates that resident teaching evaluations of hospitalists are significantly higher than those of our highly selected nonhospitalist 1-month-per-year ward attendings (K. Hauer, MD, written communication, December 2001), an effect seen elsewhere as well.³² The impact on medical student education has not been empirically studied; we recently described several potential advantages and disadvantages of hospitalists in student education.⁷⁰

The development of hospitalist groups in academic medical centers may have its most far-reaching effects in defining a new research agenda.⁶⁹ No longer limiting their research to the impact of the hospitalist model itself, academic hospitalists are now applying the tools of health services and outcomes research, ethics, and clinical epidemiology to critical inpatient issues, such as preventing nosocomial infections,⁷¹ end-of-life care,⁷² and hospital quality measurement.⁴⁶ Over time, hospitalists are likely to become increasingly engaged in patient-centered research, clinical trials, and genetic epidemiology (BOX 2), taking advantage of their onsite availability and large numbers of potential subjects.

1. Studies of the efficiency, cost, and quality of hospitalist services.
2. Studies of new approaches to systems issues that arise in and around the hospital (eg, medical errors, pain management, palliative care, inpatient-outpatient communication).
3. Clinical trials (single site or multicenter) involving diseases that hospitalists commonly encounter (eg, pneumonia, sepsis, gastrointestinal bleeding, heart failure, thromboembolism, asthma).
4. Multicenter trials involving interventions with smaller expected benefits or less common diseases.
5. Multicenter studies of the etiology of disease and disease susceptibility (eg, genetic analyses of why some patients become severely ill with pneumococcal pneumonia while exposed family members do not).

We originally defined a hospitalist as a physician who spends more than 25% of his/her time caring for inpatients, referring them back to their primary care physician at discharge.⁴⁵ The NAIP later defined hospitalists as physicians whose primary professional focus (clinical, teaching, research, or administration) was general inpatient care. This distinction is usually unimportant in community settings, since most hospitalists serve in their clinical role for the vast majority of their workday. However, the advent of research-oriented academic hospitalists, who may spend substantial amounts of time engaged in nonclinical activities related to hospital medicine (and perhaps only 2-3 months on the wards) does highlight this definitional question. Currently, the data (Table 1) indicate that hospitalists defined either way achieve efficiency and perhaps other advantages over 1-month-per-year ward attendings or primarily outpatient-based primary care physicians. Preliminary evidence from our institution and elsewhere seems to indicate that faculty who serve as ward attendings for several months each year but whose professional focus is hospital medicine can achieve clinical results comparable with those of more clinically active hospitalists, and therefore such faculty are included in the University of California, San Francisco hospitalist group. It is possible that above a certain level (ie, 2-3 inpatient months per year), a physician's overall focus on the inpatient setting may be as important as his/her aggregate hospital clinical work in the pathogenesis of a hospitalist effect.

Primary care physicians' willingness to forgo management of their hospitalized patients has been facilitated by the relatively low professional fee compensation for nonprocedural inpatient care. One analysis showed that the average primary care physician would realize a yearly net gain of about $40 000 by forgoing hospital care, simply by replacing wasted commute time with increased ambulatory productivity.⁷³

Unfortunately, these low reimbursements, coupled with the performance of few procedures and the poor payer mix of most general hospital practices, also mean that few hospitalist programs can support themselves on professional fee revenues alone.^74- 75 Moreover, the effective hospitalist, who may see an inpatient twice a day and talk to the family, primary care physician, radiologist, case manager, and 2 consultants in order to expedite a hospitalization, is reimbursed only for a single inpatient encounter. As a result, the average hospitalist has a shortfall (salary minus collections) of 10% to 50%, depending on the clinical workload, teaching obligations, payer mix, and reimbursement rates.^74- 76

Medical groups, managed care organizations, or, most commonly, hospitals often find it attractive to support hospitalist programs. If hospitalists improve quality, shorten lengths of stay, and decrease costs while satisfying patients and other providers, the return on these organizations' investments in hospitalist programs is highly favorable. Over time, it will be critical that professional fee reimbursement rates be adjusted so that a sustainable hospitalist workload creates sufficient income to support a full salary.

The hospitalist movement mirrors the health care trend toward ever-increasing specialization. However, hospitalists are fundamentally generalist physicians who provide and coordinate inpatient care, often aided by myriad subspecialists. How can a generalist be a specialist?

Specialties in medicine are traditionally defined by organ (eg, cardiology), disease (oncology), population (pediatrics), or procedure/technology (surgery or radiology). The hospitalist, on the other hand, is a "site-defined generalist specialist" (similar to emergency medicine physicians or critical care specialists), caring for patients with a wide array of organ derangements, illnesses, and ages within a specific location.⁴⁵ Accordingly, the hospitalist should not be seen as a retreat from generalism and its emphasis on coordination and integration^9,77 but rather as an affirmation of these values and as a surrogate for the primary care physician in the hospital. The competing pressures resulting from the distance between office and hospital as well as the requirement of around-the-clock availability make the hospital-based generalist a logical evolution.

Hospital medicine has already satisfied many of the requirements of a specialty. A large and enthusiastic group of practitioners identify themselves not according to their training background but as hospitalists. The NAIP is almost certainly the fastest growing physician society in the United States. The field hosts several successful meetings each year and has its own clinical textbook.⁷⁸

To establish themselves as members of a recognized medical specialty, hospitalists must identify a core skill set or body of knowledge and obtain the approval of credentialing organizations. Advocates of specialty status for hospitalists should be encouraged by the history of 2 other site-defined inpatient specialties: emergency medicine and critical care medicine. Like these relatively young fields, it seems probable that hospitalists will ultimately define a unique set of skills and competencies that will distinguish their field. The identification of practice-training mismatches (Table 2) represents an important first step.

Credentialing organizations deliver the final stamp of approval on new specialties by creating a board certification or added qualification. Most new fields quickly agitate for such status, their motivation both practical and visceral. However, for unique reasons, few hospitalists are pressing this point. Many physicians—hospitalists and nonhospitalists—worry that if a credentialing body (such as the American Boards of Internal Medicine or Pediatrics) created a hospital medicine credential, health maintenance organizations might require that physicians possess this credential to care for inpatients. This would be unacceptable to many primary care physicians, who would be excluded from the hospital despite their desire and competence to continue practicing there. For this reason, we expect neither NAIP nor the relevant boards to promote separate credentials in the near future. Nevertheless, as evolutionary forces lead to specialized training, some formal specialty designation may emerge.⁷⁹

In summary, the hospitalist model is growing rapidly and appears to have achieved its minimum goal of improving efficiency without adverse effects on quality, teaching, or patient satisfaction. Ongoing larger studies should help determine whether preliminary data suggesting an improvement in some of these outcomes, such as quality and teaching, can be confirmed. The dominant questions no longer relate to whether the hospitalist model is here to stay—even skeptics concede that it is—but rather, the myriad organizational, financial, ethical, educational, and clinical issues that arise with a major change in the organization of US hospital care. We hope that these issues will continue to be settled on the basis of rigorous analysis of the evidence.