Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
9 Resources to Implement Improvements for Patient Safety and Resident Training Improved residency training conditions for patient safety, enhanced resi- dent well-being, and excellent educational outcomes are the committeeâs goals. Recommendations for duty hour adjustments, enhanced supervi- sion, and workload reduction will best achieve the targeted goals when implemented in concert, and implementation of all the committeeâs recom- mendations will require a significant investment in personnel to substitute for the hours that residents are no longer available. To cover some of the excess resident hours with substitute personnel would cost an estimated $1.7 billion dollars, the equivalent of about 9 percent of direct and indi- rect graduate medical education payments made to teaching facilities from public and private sources. To help institutions implement the changes, the committee recommends that additional funding be provided by all sup- porters of graduate medical education. Systematic collection of pertinent data would help monitor and evaluate the effects of implementing the recommendations, and research would provide an evidence base for future changes to duty hours or educational strategies. As the preceding chapters demonstrate, this Institute of Medicine (IOM) committee found the issues of patient safety, resident safety, duty hours, and schedules closely related to broader issues of how graduate medical education is structured, including the work environment, the super- vision of residents, and safety practices throughout the training institutions. While recommendations on a particular topic are embedded in the scientific evidence of its corresponding chapter and presented serially in this report, the committee intends the report and all of its recommendations to be considered as a whole. This chapter discusses the possible consequences of 295
296 RESIDENT DUTY HOURS implementing a single recommendation (e.g., duty hour limits) in isolation and an approach to phase-in the recommendations in a practical fashion. Thus, the committee presents first a preamble to the recommendations: Preamble to Recommendations To promote conditions for safe medical care, improve the education of doctors in training, and increase the safety of residents and the general public, the committee offers the (previous and) following recommenda- tions, which should be implemented with all deliberate speed. While some recommendations should be implemented immediately, changes to duty hours, adjustments in workload, and the funding needed for these changes might require an integrated phase-in. The recommendations will require additional resourcesâboth financial and human. Without the necessary restructuring in resource allocation, attempts to implement the recommen- dations will fail to have the desired benefits and could even reduce patient safety. The committee believes that the Accreditation Council for Gradu- ate Medical Education and the other organizations charged to implement aspects of the recommendations should begin their work with urgency, and that action on all recommendations should be taken within 24 months. The committee took a broader perspective than just the duty hours and schedules, looking at related aspects of the work and learning environment of residents. It found little detailed information available on the educational outcomes of training programs since the 2003 limits because many of the first cohort of doctors fully trained under the 2003 limits are just complet- ing their training. Supervision and workload were not addressed when duty hours were set in 2003, and lack of supervision at critical junctures, excess workload, and fatigue can all contribute to error and to reduced learning. Thus, the committee believes that to minimize unintended consequences, recommendations to prevent and mitigate fatigue through adjusting resi- dent duty hours, enhancing supervision, and reducing workload should be implemented in concert. Implementing the duty hour adjustments without the others could prevent achievement of important goals identified by the committee. Recommendations by the committee reflect the best ways to achieve performance and learning goals based on sleep science, learning theory, and the close observations of medical educators. In designing its recommendations to achieve the targeted goals outlined in the preamble, the committee considered strong evidence from the litera- ture concerning the impact of sleep and fatigue on human performance and the occurrence of error, and based several recommendations on this evidence. Although fatigue creates an unsafe condition in the work environ- ment, there is insufficient evidence to determine the degree to which resident
resources to implement improvements 297 hours of work translates into patient harm. For this reason, the committee did not change the current weekly duty hour limit of 80 hours or the limit of 30 hours for extended duty periods. Rather, it chose to create better opportunities for fatigue prevention and mitigation within the basic duty hour structure and to focus on supervision, handovers, and other systemic changes to enhance learning and safety. The intent in adjusting the 2003 Accreditation Council for Graduate Medical Education (ACGME) duty hour limits is to: â¢ Prevent fatigue whenever possible; â¢ Recognize that some fatigue is inevitable and provide measures to relieve both acute and chronic sleep deprivation and reduce its negative effects; â¢ Be practical to schedule; â¢ Be feasible to monitor; â¢ Enhance the learning experience; â¢ Preserve the ability of residency programs of various sizes and dif- ferent specialties to adapt the changes to their circumstances by not mandating a single schedule for all and by allowing limited opportunities for exceptions based on patient need and unusual learning opportunities; and â¢ Maintain the spirit of residency and the excitement of being a doctor. The committee was also asked to consider the potential cost impact of its recommendations, and it is the estimated cost and the limited availability of the healthcare workforce that the committee believes are the greatest bar- riers to further changing resident duty hours. Having an adequate workforce of physician extenders, residents, and physicians alike takes planning to develop incentives and remove disincentives to grow the labor force that the country and individual labor markets require to support reductions in resi- dent hours and serve other healthcare needs. Based on a commissioned cost model, an estimate of the personnel substitution costs associated with several duty hour and workload limits shows that the costs would be substantial, in the ballpark of $1.7 billion in 2008, with variations in that amount depend- ing on who substitutes for residents and how programs choose to schedule residents. Other recommendations of the committee could require additional funds. The committee, while recognizing that funds for health care are in great demand, recommends adequate support by all funders of graduate medical education (GME) and related research so that the recommendations can be fully implemented and have the desired impact. The committee acknowledges that there are objections from some members of the graduate medical training community to any changes to the
298 RESIDENT DUTY HOURS 2003 duty hour limits and related aspects of GME, just as there were objec- tions to those limits initially (see Chapters 1 and 2). A crucial objective of graduate medical training is to ensure that the country will produce increas- ingly competent physicians for independent practiceâa long-term patient safety goal. With this in mind, the committee found that there are good reasons to take a deliberate approach to changing resident duty hours. This chapter reviews current funding for GME and projections for the costs and workforce needs associated with the committeeâs proposed adjustments to hours and workload. The chapter concludes with a phased implementation strategy and addresses the necessity of further research, data collection, and evaluation that would allow consideration of the ap- propriateness of resident duty hours in the future. COST IMPLICATIONS OF CHANGES TO DUTY HOURS Implementing workload reductions and adjustments to the 2003 duty hour limits will require replacing residentsâ time with that of other workers and entails substantial costs for society but also potential benefits to patient and resident safety. This section first looks at what the United States invests in graduate medical training and then examines projected estimates of per- sonnel costs to implement the committeeâs recommendations on hours and workload adjustments. Funding for Graduate Medical Education GME is paid for largely through insurance premiums and payroll taxes. A number of parties specifically contribute to GME: the Centers for Medicare and Medicaid Services (CMS), the Department of Veterans Af- fairs (VA), the Department of Defense (DOD), the Health Resources and Services Administration (HRSA) of the U.S. Public Health Service, states, and private funders. Available data on the level of funding for GME over the past few years is limited. Medicare is the largest single payer of GME ($8.5 billion in 2007), and its funds come in two forms (see Table 9-1). Some of the other funds, particularly from private and state sources, are harder to identify and estimate at the national level. Wynn and colleagues estimate that for all sources of support, direct and indirect expenditures as- sociated with training residents were approximately $18.7 billion in 2003 (Wynn et al., 2006). Some of this funding comes indirectly through payments for patient care in teaching institutions (e.g., from Medicare indirect medical educa- tion [IME] payments, state Medicaid, private payers) to cover the increased costs of care associated with resident training in teaching hospitals. The Medicare IME funds ($5.7 billion in 2007) are provided through higher
resources to implement improvements 299 TABLE 9-1â Sources of GME Funding FY 2007 Funding Source (billion dollars) CMS â Medicare 8.50 ââ IME (5.70) ââ DGME (2.80) VA 1.0 â Direct (0.50) â Indirect (0.50) DOD NA HRSA â Childrenâs Hospital 0.28 â GME â Training in primary care, Medicare, and dentistry 0.05 StatesâMedicaid 3.20a Private payer (not direct payments, but imputed from higher Unknown reimbursement to teaching hospital) NOTE: CMS = Centers for Medicare and Medicaid Services; DGME = Direct graduate medi- cal education payment; GME = Graduate Medical Education; HRSA = Health Resources and Services Administration; IME = Indirect medical education payment; NA = Not available; VA = U.S. Department of Veterans Affairs. aEstimated spending in 2005 (Henderson, 2006). SOURCES: CBO, 2008; Chang, 2007; HRSA, 2008a,b. hospital payments. The size of the operating adjustment to a hospitalâs payment rate is based on teaching intensity and the number of residents per bed, with limits on the rate of increase. It is included in inpatient operat- ing and capital payments under the Medicare inpatient hospital payment system and includes subsidies to hospitals treating patients of the Medical Advantage Program (MedPAC, 2008). Other funding, such as direct GME (DGME) payments are made to the training institution for support of training such as the residentsâ stipends, teaching physiciansâ salaries and benefits, and administrative overhead of GME offices. The DGME payment is based on historic, hospital-specific costs per trainee, with maximum limits on the number of trainees. Some teaching hospitals receive very little, if any, support from the Medicare GME funding stream, even though they train many residents, because they do not serve a large Medicare population (Opas, 2008). Other federal financial sources of support for GME include the VA and DOD (which both also provide sites for residency training), and HRSA. The VA has approximately 9,500 residency slots in its healthcare facili- ties (9 percent of U.S. medical resident slots) and, with multiple residents rotating through each slot, participates in the instruction of approximately one-third of U.S. doctors in training (about 34,000 medical residents) per
300 RESIDENT DUTY HOURS year. The DOD supports approximately 3,000 residency positions in mili- tary facilities accredited by the ACGME. HRSA manages the Childrenâs Hospital Graduate Medical Education program, authorized to support the training of stand-alone childrenâs hospitals that do not receive Medicare reimbursements (and therefore do not receive the DGME and IME pay- ments described earlier). States may voluntarily provide GME funding through their Medicaid programs. Up until at least 2005, all but three states (Illinois, North Da- kota, and Texas) did so (Henderson, 2006). However, there has been a debate in recent years about whether the use of federally matched dollars for GME reimbursements is an appropriate use of Medicaid funds. In June 2008, Congress placed a moratorium on a proposed rule until April 1, 2009, to block CMSâs prohibition of GME payments from these matched funds (P.L. 110-252, June 30, 2008). Private payersâ contributions are difficult to determine. These pay- ments compensate GME providers for the care received by their benefi- ciaries rather than paying for GME directly. This is similar in concept to Medicare IME. Private sector sources and the federal government provide the preponderance of funds for GME (Knapp, 2002; Wynn et al., 2006). Finally, economists consider that residents support some of the cost of their education through their own efforts. By providing service in hospitals at a relatively low hourly wage rate, residents, in effect, subsidize the institu- tion for some of the costs of their education. Given the cost of replacing residents with other personnel, they are not a trivial source of support for their own graduate medical training although they pay no tuition. The committee considered the impact of its recommendations on the fi- nancial status of hospitals. Figure 9-1 compares the median margins of total income and patient revenues of hospitals by teaching or non-Âteaching status as well as whether they are safety net hospitals (Andrews et al., 2007). While there is not a consensus on what constitutes a safety net hospital ( Â Siegel et al., 2004), the study by Andrews and colleagues bases its definition on the proportion of uncompensated care that a hospital provides. For most hospitals, except the category of safety net teaching hospitals, the negative patient revenue margin is smaller than the positive total income margin. Safety net teaching hospitals, however, have a substantial negative patient revenue margin (â9.6 percent), in part because of the uncompensated care they deliver, and a positive total income margin of only 1.2 percent. It is a positive total income margin only because safety net teaching hospitals, â Personal communication, J. P. Bagian, National Center for Patient Safety, July 31, 2008. â âThe total income margin is the total income for a hospital (i.e., net patient revenue plus contributions, government appropriations, and other income), divided by the total expenses (i.e., operating costs and other expenses)â (Andrews et al., 2007, p. 13).
6 .0 4.0 3. 2 % 2.9 % 3.1% 2.5% 1.9 % 2 .0 1. 2% 0.0 â 0.7% â 0.4% â 2 .0 â1.6 % â1.3 % â 2 .0 % â 4.0 Patient Revenue Total Income â 6 .0 M e d i a n M ar g i n s ( p e r c e n t ) landscape â 8 .0 Figure 9-1.eps â10.0 â 9.6 % â12.0 N o n - S a fe t y - N e t Secondar y S a fe t y- N e t N o n - S a fe t y - N e t Secondar y S a fe t y- N e t S a fe t y - N e t S a f e t y- N e t N o n -t e a c h i n g H o s p i t a l s Teaching Hospitals Type of Hospital FIGURE 9-1 Median margins of hospitals by teaching status. SOURCE: Andrews et al., 2007. 0
302 RESIDENT DUTY HOURS on average, successfully obtain sufficient other revenue such as government allocations and other subsidies. Approximately one-third of safety net hos- pitals, whether teaching or not, have a negative total income margin, and 20 percent of all safety net hospitals are teaching hospitals (Andrews et al., 2007). Teaching hospitals offer a substantial amount of charity careâcare provided without expectation of payment. Major teaching hospitals make up only 6 percent of the acute hospitals in the country, yet they deliver 41 percent of all hospital-based charity care, and all teaching facilities provide 71 percent of that charity care. In 2006, the 274 members of the Council of Teaching Hospitals and Health Systems (COTH) provided an estimated $6.3 billion in uncompensated charity care, non-COTH teaching hospitals an estimated $4.5 billion, and non-teaching hospitals $4.5 billion, totaling about $15.3 billion in hospital charity care (COTH, 2008). The committee is concerned that some safety net teaching hospitals may not have the resources under current funding mechanisms to provide the additional supports necessary to allow residents recommended oppor- tunities to rest, transfer noneducational tasks to others, or offer residents sufficient supervision because their resources already are spread too thinly (Werner et al., 2008). Cost Model As previously mentioned, to implement the recommendations of this report, some of the work presently performed by residents will have to be done by others. The committee commissioned a health services researcher, Teryl Nuckols, M.D., MSHS, and a health economist, JosÃ© Escarce, M.D., Ph.D., both at the David Geffen School of Medicine at the University of C Â aliforniaâLos Angeles as well as the RAND Corporation, to construct a model that would provide an estimate of the costs and workforce that would be needed to replace resident work under various duty hour scenarios. This section of the chapter is based on their methods and results. The complete pa- per is published on the following websites: www.iom.edu/residenthours and www.iom.edu/hcs. A committee member, economist Jayanta Bhattacharya, M.D., Ph.D., performed sensitivity analyses on some of the main assumptions of this model. His discussion and figures are posted with the main paper. The model, based on existing literature and explicit assumptions, de- rives estimates from four scenarios, called âcomponents,â related to resi- dent hours and workload, which were specified before the committee had formulated its final recommendations. The model provides an indication of the level of expected substitution costs based on 2006 dataâapproximately $1.6 billion dollars ($1.7 billion when inflated to 2008 dollars)âif all four components are adopted. The costs are reasonably similar whether excess hours of residentsâ time are replaced by hiring other healthcare providers
resources to implement improvements 303 or additional residents. The simplified components of the model related to the committeeâs recommendations are: 1. Bringing all residency programs into compliance with the existing 2003 ACGME 80-hour duty limit, since not all programs and resi- dents now comply. 2. Having any extended duty period beyond 21 hours incorporate a 5-hour undisturbed sleep period. 3. Reducing the workload of postgraduate year 1 (PGY-1) residents by 10 percent. 4. Limiting shifts to a maximum of 16 hours for residents beyond PGY-1. The model assumed that after hospitals achieved compli- ance with the previous three components, they might choose to reduce all shifts to 16 hours. Since they would have to include a 5-hour rest period for shifts lasting longer than 16 hours, eliminat- ing shifts of 16 to 21 hours would require no more substitute hours than would limiting the shifts to 16 hours. Available literature reports that most PGY-1 extended duty shifts exceed 21 hours, so the model assumed hospitals would not limit the shifts of PGY-1s to 16 hours. Implementing these reforms with the substitution of personnel tailored to the tasks that residents currently perform would require the following increases nationally in full-time equivalents (FTEs): nursing aides, 229; lab- oratory technicians, 45; licensed vocational nurses, 320; midlevel providers, such as physicians assistants and nurse practitioners, 5,984; and attending physicians, 5,001. Given that there are 1,206 teaching hospitals across the country, less than one FTE would be needed on average per hospital for several of these provider types. If instead more residents were to be added, this would require approximately 8,247 new residents (specialty and sub- specialty) in addition to the existing pool of more than 105,000 residents. Please note that the cost model calculates the economic costs of only certain aspects of the committeeâs recommendations and does not attempt to predict which elements of the recommendations will be adopted (e.g., keep extended duty periods with protected sleep periods or use only shorter shift schedules). Precise cost estimates of every recommendation were not feasible given time and data constraints and were beyond the committeeâs statement of task, but the four components do reflect the likely magnitude of costs for a number of the reportâs major recommendations. Lack of comprehensive nationwide data on the actual hours residents now work, the frequency of their overnight work, and other factors required some as- sumptions in the modelâs design based on the existing literature. Sensitivity analyses of selected assumptions allow for a range of estimates to address
304 RESIDENT DUTY HOURS these assumptions (e.g., different substitution ratios, call frequency dur- ing residentsâ inpatient months, current compliance levels). In addition, the model could not anticipate fully programsâ choices of implementation strategies (e.g., if the programs shifted workload from residents with longer hours to those with shorter ones), which could possibly produce lower cost estimates. Estimating Baseline of Resident Duty Hours, Rate of Violation, and Workload To estimate the direct annual costs of the proposed reforms, the differ- ence between what residents work at baseline and what they would work under the proposed changes is calculated and called âexcess resident workâ hours. As discussed in Chapter 2, there has been no recent rigorous or reli- able collection of data on the total duty hours of residents in all specialties across the country. The economic model uses the best available data on mean weekly duty hours (66.6 hours) and the hours worked by PGY-1s in excess of 80 hours from the first year of implementation: 29.0 percent of the workweeks were longer than 80 hours, 12.1 percent were 90 or more hours, and 3.9 percent were 100 or more hours (Landrigan et al., 2006). The study did not give details on workweeks that fell below the 80-hour limit. Other more recent studies have found relatively similar rates of duty hours and degree of violations (AMA Division of Market Research and Analysis, 2005; Jagsi et al., 2008). The baseline calculation makes accommodation for the portion of the year that residents spend on inpatient months (e.g., 50.6 percent of PGY-1 residentsâ months in the Landrigan paper), the num- ber of nights that a resident might be on extended duty, how long residents are staying over the 30-hour limit, and differences between PGY-1s and other years (Nuckols and Escarce, 2008). The values for these assumptions are detailed in the paper describing the model. Hierarchical Nature of the Model The costs of the four reform components are estimated in a sequential, hierarchical fashion to prevent counting excess hours twice. Several basic assumptions are built into this hierarchy: â¢ That achieving compliance with the 80-hour workweek would make it possible to achieve the 30-hour duty period limit at no ad- ditional cost because violations of the 30-hour limit are generally what push residents over 80 hours; â¢ That residency programs would choose to implement a nap only
resources to implement improvements 305 for residents already working more than 21-hour duty periods at baseline; and â¢ That reducing average workload by 10 percent would be equivalent to reducing weekly duty hours by the same amount; there are no widely accepted measures or estimates of the workload of residents across specialties (e.g., average daily census, number of admissions or procedures per call day) available. PGY-1 residents are more likely to violate duty hour limits than more se- nior residents or fellows because PGY-1s tend to have more inpatient months, more frequent extended duty periods, and more direct patient care respon- sibilities. Reducing the workload of only first-year residents is factored into the cost model, but the committee recognizes that it may become necessary in some or all of the specialties for workload to be reduced in other years of residency as well. The final component of the hierarchy for cost estimating is a 16-hour shift maximum for residents beyond their intern year. Substitution Scenarios As discussed in Chapter 4, many training programs hired replacements to assume âexcess resident workâ in response to the 2003 limits. This model builds on substitution ratios available in the published literature. Each step, or component, in the hierarchy of hours reduction has its own combination of resident substitutes (Table 9-2). Alternatively, the model estimates having each resident hour replaced by additional residents with no work transferred to other types of personnel. Here the term âresident substitutionâ means transferring residentsâ clinical care-related work to other providers or sharing it among a larger population of residents. Mid- level providers (nurse practitioners and physician assistants) have often been considered the prototypical resident substitutes (Stoddard et al., 1994; Whang et al., 2003), but publications following the 2003 ACGME reform and recent testimony from hospital administrators suggest that there are a few basic strategies for reducing resident duty hours or workload: â¢ Task-tailored substitutes: Transferring noneducational patient care tasks to the lowest-level personnel qualified to perform them. â¢ Midlevel substitutes: Transferring work to midlevel providers. â¢ Midlevel and attending substitutes: Transferring work to a mixture of midlevels and attending physicians. â¢ Attending substitutes: Transferring work to attending physicians. â¢ Resident substitutes: Hiring new residents to share the work of existing ones.
306 RESIDENT DUTY HOURS TABLE 9-2â Methods: Application of Substitution Strategies to Base Case Scenario Base Case Scenario Using Substitutes Other Than Reform Component Additional Residents 1.â Achieve compliance with Specialty residents: Task-tailored substitutes 80-hour workweek Subspecialty residents: Attending physician substitutes 2.â When shifts last 21 to 30 All residents: Attending substitutes hours, include a 5-hour nap 3.â Reduce workload of PGY-1 PGY-1 specialty residents: Midlevel substitutes residents by 10% 4.â Optional: reduce maximum Specialty residents: 50% midlevel and 50% attendings shift length to 16 hours Subspecialty residents: attendings The committee has concluded that transferring noneducational routine âscutâ work to the lowest-level personnel possible (âtask-tailored substitu- tionâ) would enable residents to preserve their educational patient care ex- periences and reduce their duty hours. This transfer could be accomplished at relatively modest cost compared to hiring other physicians to do resi- dentsâ work. Substantial reductions in resident duty hours and workload, such as those represented by the cumulative total of the four proposed reform components, would probably not be achievable using midlevel providers alone. Patient care responsibilities would likely require attending supervision for complex or unusual tasks. Further, the work of subspecialty residents is complex, suggesting that only substitution by attending physi- cians would be appropriate. The nap requirement (Component 2) would likely require the substitution of attending-level physicians so that residents would feel comfortable signing out their patients and would take advantage of naps during extended duty periods. Unwillingness of residents to sign out to other night-float residents has been a barrier to incorporating pro- tected sleep periods into long duty periods (Arora et al., 2006). Similarly, Component 4 would be expected to lead to redesign of overnight call and require greater attending physician presence. The model assumes that substitution does not affect length of stay, test ordering, or other hospitalization cost determinants. Substitution or resi- dent work by others would occur at a hospital levelâmeaning that excess work from multiple residents would naturally distribute among substitutes within each hospital. Work would be transferred to substitutes in a 1 hour to 1 hour ratio (i.e., this assumes residents and substitutes would perform tasks at the same speed).
resources to implement improvements 307 Hourly wage substitutions for the model were based on data from the Bureau of Labor Statistics (BLS). According to BLS, in 2006, mean hourly wages for potential substitutes were as follows: nursing aides, $11.21; medical and clinical laboratory technicians, $16.55; licensed vocational nurses, $18.12; midlevel providers, $37.84; and physicians, $58.76; ben- efits were 30.1 percent of total compensation (43.1 percent of wages) (U.S. Department of Labor, 2007a,b). Having residents incurs training expenses as well as salary and benefits expenses. The average salary for PGY-1s in the 2007-2008 academic year was $44,747 plus benefits (AAMC, 2007a). The hourly rate of an intern receiving this average stipend and working 66 hours per week throughout the entire year (the mean found in one study after implementation of duty hours) (Landrigan et al., 2006) would be $12.92. However, dividing the estimated total payments for GME by the total number of residents nationally suggests that salary, benefits, and train- ing expenses together cost $187,000 per resident year (this figure includes all public and private sources of direct and indirect payments) (Wynn et al., 2006). Thus, the cost model assumed that the $187,000 costs would apply if additional residents were used to replace reduced hours of current residents. The model includes replacement costs for both âspecialtyâ resi- dents and âsubspecialtyâ fellows. The $187,000 figure was used because no uniform data were available from teaching programs on the incremental cost of training residents. Costs of Replacing âExcess Resident Dutyâ Hours Table 9-3 shows the results of calculations for the four reform scenarios using other personnel as substitutes for the âexcess resident workâ hours. The cost projections are presented for two groups of teaching hospitals: all hospitals with ACGME-accredited programs (1,206 hospitals) and COTH members (367 hospitals). Three-quarters of residents in academic year 2006-2007 were trained at COTH hospitals (AAMC, 2008c; Nuckols and Escarce, 2008). The total U.S. cost of all four reform components in 2006 dollars is estimated to be $1.6 billion for the 1,206 hospitals with ACGME programs including $1.2 billion for COTH hospitals. In Table 9-3 the total dollars are then broken down into an average cost per hospital and per admission. As an alternative approach, the model assumes that excess resident hours and work would be distributed among a larger pool of residents than are being trained today. Table 9-4 shows the number and cost of additional residents that would be needed nationally to substitute for the excess resi- dent work hours. At least an additional 8,247 residents would be necessary. Salary and benefits for the 7,639 specialty residents and 608 subspecialty fellows would total more than $500 million when excluding IME payments,
308 RESIDENT DUTY HOURS TABLE 9-3â Results: Costs of Hiring Other Providers to Assume Excess Resident Work, Base Case Scenario (2006) Hospitals with ACGME- Accredited Programsa Baseline Work Total U.S. Cost Component 1. Achieve compliance with 80-h workweek â 24,772 PGY-1 specialty residents $209,742,405 â 22% of 64,497 specialty residents above PGY-1 $120,139,928 â 22% of 15,610 subspecialty residents $45,915,338 ââ Subtotal $375,797,671 Component 2. When shifts last 21 to 30 h, include 5-h nap â 24,772 PGY-1 specialty residents $319,707,737 â 23.1% of 64,497 specialty residents above PGY-1 $192,284,187 â 23.1% of 15,610 subspecialty residents $46,537,919 ââ Subtotal $558,529,843 Component 3. Reduce workload of PGY-1 residents by 10% â 24,772 PGY-1 specialty residents $391,736,621 ââ Subtotal $391,736,621 Component 4. Optional: Reduce maximum shift to 16 h â 30.6% of 64,497 specialty residents above PGY-1 $192,950,559 â 30.6% of 15,610 subspecialty residents $56,812,524 ââ Subtotal $249,763,084 Total, components 1-3 $1,326,064,134 Total, components 1-4 $1,575,827,218 NOTE: Totals may reflect rounding. aIncluding COTH hospitals. although the total including costs from public and private payers could be $1.5 billion, if the associated patient care costs are included. The cost of training residents is challenging to estimate accurately, as has been explored by other sources (Knapp, 2002; Wynn et al., 2006). Residentsâ salaries and benefits are significantly lower than the costs involved in training them. The $187,000 estimate for hiring each additional resident, based on the $18.7 billion in payments related to GME (Wynn et al., 2006), may overstate the amount necessary to support training per resident; thus, the $1.5 billion cost for adding residents would be an overestimate. Overall, in 2006-2007, resident stipends and benefits represented 9.0 percent of institutionsâ oper- ating expenses (AAMC, 2008c). Expanding the population of residents as an approach to achieving reform has some short-term appeal because residents provide highly skilled labor at a low hourly cost relative to other substitutes, but more precise estimates would be needed to determine the incremental costs of training
resources to implement improvements 309 COTH Hospitals Cost per Cost per Cost per Cost per Hospital Admission Total U.S. Cost Hospital Admission $311,607 $21.37 $281,848,253 $767,979 $33.68 $463,126 $31.76 $418,897,382 $1,141,410 $50.05 $324,823 $22.28 $293,802,466 $800,552 $35.11 $207,100 $14.20 $187,322,313 $510,415 $22.38 $1,099,556 $75.41 $994,548,101 $2,709,940 $118.84 $1,306,656 $89.61 $1,181,870,414 $3,220,355 $141.22 residents on top of their salary and benefits. Such an option would also have long-term implications for the national supply of physicians. In addi- tion, increasing the total number of residents would not necessarily ensure the number and distribution of residents by specialty or geographic area where they are deemed to be needed most. The cost estimates of the model do not include any assumptions of sav- ings from work and education redesigns. Many institutions may find ways to streamline the work and training of residents to eliminate excess hours without having to hire substitutes for each and every resident hour reduced and without burdening existing residents with increased workloads. Cer- tainly some institutions may experience labor shortages for some of the potential substitutes and many may have serious cost constraints, providing an added incentive to reduce costs through efficiencies and systems rede- sign. The committee took a conservative approach and did not project or assume any such cost savings.
TABLE 9-4â Results: Reducing Resident Duty Hours by Increasing Number of Residents Nationally 310 Total Using Total Using Annual Annual Number Total Using Expenditures by Expenditures by of New Increase Annual Salary Public Payers All Public and Residents from 2006 and Benefits Excluding IMEa Private Payers Baseline Work Needed (%) per Resident per Resident per Resident Component 1. Achieve compliance with 80-h workweek â Specialty residents 2,121 2.6 $126,108,713 $140,005,876 $396,683,314 â Subspecialty residents 187 1.2 $13,144,147 $12,340,648 $34,965,169 â Subtotal $139,252,860 $152,346,524 $431,648,483 Component 2. When shifts last 21 to 30 h, include 5-h nap â Specialty residents 2,085 2.3 $130,228,151 $137,607,874 $389,888,977 â Subspecialty residents 190 1.2 $13,322,373 $12,507,979 $35,439,273 â Subtotal $143,550,524 $150,115,853 $425,328,249 Component 3. Reduce workload of PGY-1 residents by 10% â Specialty residents 2,477 3.1 $142,992,976 $163,495,200 $463,236,400 â Subtotal $142,992,976 $163,495,200 $463,236,400 Component 4. Reduce maximum shift to 16 h â Specialty residents 956 1.1 $59,706,617 $63,090,050 $178,755,142 â Subspecialty residents 231 1.5 $16,263,676 $15,269,481 $43,263,528 â Subtotal $75,970,293 $78,359,531 $222,018,670 Total for Components 1-4 â Specialty Residents Needed 7,639 8.6 $459,036,458 $504,199,000 $1,428,563,833 â Subspecialty Residents Needed 608 3.9 $42,730,197 $40,118,107 $113,667,970 Cost $501,766,655 $544,317,107 $1,542,241,803 NOTE: Totals may reflect rounding. aMedicare indirect graduate medical education payment.
resources to implement improvements 311 Sensitivity Analyses The model of the cost implications of resident duty hour and work- load reform requires assumptions about a wide range of parameters. To the extent that such data are available, Nuckols and Escarce derive these parameters from the published literature. However, the committee recog- nizes that there is uncertainty about these assumptions and that in many cases the published literature contains little relevant information. In other cases, the literature cited by Nuckols and Escarce provides estimates on key parameters such as the frequency of night shifts or extended duty periods or the proportion of a year spent in an inpatient service for only a subset of residents, such as PGY-1s, or from a very limited number of institutions. Hence the committee decided to test the sensitivity of the cost impact estimates by testing some of the assumptions in the model. Jayanta Bhattacharya, an economist and a member of the committee, conducted the sensitivity analyses. The committee provided guidance on the range of assumptions to use for each parameter, with some higher and some lower than the baseline assumption. Dr. Bhattacharya designed the sensitivity analyses to test one param- eter at a time through 11 âthought experiments.â The assumptions tested include the following: â¢ The rate of violations by PGY-1s and more senior residents of the 80-hour week of the 2003 ACGME reform affects the costs of complying with current rules, or Component 1 of the model. â¢ The frequency of extended duty periods during inpatient rotations for PGY-1s and all other residents, which affects the number of hours that would have to be replaced to accommodate the required sleep period for duty periods lasting from 21 to 30 hours, impacts the costs of Component 2. â¢ The use of various healthcare professionals as task-tailored substi- tutes, such as laboratory technicians and licensed vocational nurses; midlevel professionals, such as registered nurses and physician as- sistants; and attending physicians or other residents affects the cost of covering the excess hours of residents and influences the costs of all four components, based on their assumed substitution patterns. â¢ In addition to the different hourly costs of various substitutes to cover excess resident hours, the substitution costs could vary de- pending on the efficiency of the substitutesâwhether they produce the same amount of work per hour as the residents or not. â¢ The rate of compliance with the new reforms of the model also af- fect the costs of Components 2-4; total costs increase as compli- ance increases.
312 RESIDENT DUTY HOURS Based on the 11 thought experiments used to test these assumptions of the model, it is clear that the cost of implementing each of the four components discussed in the model could vary substantially, depending on the details of the assumptions. The total cost estimates of the potential reforms are surprisingly robust to a wide range of assumptions about cur- rent resident work schedules, which providers would perform the work if the potential reforms were adopted, and the relative efficiency of those substitute workers. The $1.6 billion estimated by the model for the cost of TABLE 9-5â Sensitivity Analyses No. Experiment Sensitivity Range â1 Vary assumption about current 0-100% as many hours above the 80- compliance with ACGME 80-hour hour limit as reported by Landrigan workweek (2006) â2 Vary assumption about noncompliance 0-100% of PGY-2+ specialty and of residents above PGY-1 with 80-hour subspecialty residents worked the hours workweek reported for PGY-1s in Landrigan (2006) â3 Vary assumption about call frequency Every third night to every seventh night during inpatient months for PGY-1 â4 Vary assumption about call frequency Every third night to every seventh night during inpatient months for all residents above PGY-1 â5 Vary assumption about frequency of PGY-2+ have 50 to 100% of rotations as inpatient rotations among residents PGY-1 in Landrigan (2006) above PGY-1 â6 For specialty residents, vary percentage 0-100% transferred to midlevels of time transferred to midlevels in Component 1 â7 For all residents, vary percent of time 0-100% transferred to attendings transferred to attendings in Component 4 â8 Vary substitution ratio for task-tailored Substitution ratio of 0.5 to 3.0 hours for substitutes each hour transferred from a resident â9 Vary substitution ratio for midlevel Substitution ratio of 0.8 to 2.4 hours for substitutes each hour transferred from a resident 10 Vary substitution ratio for attending Substitution ratio of 0.5 to 1.0 hour for substitutes each hour transferred from a resident 11 Vary assumption about compliance 0-100% compliance with the changes contemplated in Components 2, 3, and 4
resources to implement improvements 313 implementing all four components is within the range calculated for each parameter in the sensitivity analyses. The cost range for each âexperimentâ is included in Table 9-5. Assuming at least some compliance with the pro- posed reforms (as illustrated with components 2, 3, and 4 of the model), the lowest-cost estimate for implementing the potential reforms is $1.14 billion, while the highest is $2.52 billion. The full discussion of the sensitivity analyses, a description of the 11 experiments that Dr. Bhattacharya conducted, graphs of the results, and the Outcome Range (million dollars) Component 1 Component 2 Component 3 Component 4 Total Cost $376-$0 $559 $392 $250 $1,576-$1, 200 $210-$376 $559 $392 $250 $1,410-$1,576 $376 $878-$559 $392 $250 $1,896-$1,576 $376 $797-$559 $392 $500-$250 $2,064-$1,576 $376 $439-$559 $392 $125-$250 $1,332-$1,576 $376-$381 $559 $392 $250 $1,576-$1,581 $376 $559 $392 $250-$292 $1,576-$1,618 $362-$430 $559 $392 $250 $1,562-$1,630 $334-$665 $559 $313-$940 $235-$356 $1,441-$2,519 $305-$376 $279-$559 $392 $163-$250 $1,138-$1,576 $376 $0-$559 $0-$392 $0-250 $376-$1,576
314 RESIDENT DUTY HOURS summary table of cost ranges is on the projectâs website along with the full paper on the model and can be accessed at www.iom.edu/residenthours or at www.iom.edu/hcs. Cost of Preventable Adverse Events and Possible Net Costs One expectation is that reduced hours of work and workload would reduce errors and improve patient safety by reducing preventable adverse events (PAEs). In addition to harming patients, PAEs increase hospitaliza- tion costs, outpatient medical care costs, and costs associated with patient disability (Thomas et al., 1999; Zhan and Miller, 2003). With this in mind, the economic analysis was designed to estimate the potential net annual costs of the proposed reforms from both direct costs and any changes in costs related to PAEs. The framework for this analysis can be described by the equation below: Net costs = [Cost of resident substitutes] â [(Baseline costs of PAEs)(Î in PAE rate)]. If the proposed reforms succeed at reducing PAEs in teaching hospitals, this would likely yield cost offsets both during and after hospitalization. These cost offsets would reduce the gross costs associated with reduced resi- dent hours. However, the possibility exists that the reforms could increase PAEs rather than reduce them. Reducing duty hours has been associated with increased numbers of handovers, which in turn has been associated in one study with a significant increase in the rate of PAEs, for example, al- though this risk could be mitigated (Petersen et al., 1994, 1998). The com- mittee does not attempt to predict the changes, if any, in PAEs following full implementation of its recommendations. Therefore, this analysis uses a range of possible changes in PAE rates to estimate net costs of implementa- tion of the four components of the model. The costs of PAEs for teaching hospitals and for society as a whole are likely to differ in their magnitude, according to an analysis of the 2003 duty hours reforms (Nuckols and Escarce, 2005). Teaching hospitals would incur costs resulting from PAEs for additional intensive care unit (ICU), inpatient non-ICU, and physician care. From the societal perspective, considering all costs regardless of who bears them, there would be, in addition to the inpatient event costs, costs for outpatient medical care and the non-medical costs of lost wages and lost household production (Thomas et al., 1999). Thus, the main costs of a PAE occur after the patient leaves the hospital and are borne by society. Since most PAE costs occur after hospital discharge, teaching hospitals are not very likely to experience direct and sizable cost offsets if the proposed duty hour reforms succeed in reducing PAEs. Con-
resources to implement improvements 315 sequently, the current analysis considers net costs from both the hospital and the societal perspective. Table 9-6 presents the net costs of reform. Given an absence of litera- ture to suggest the potential effect of the currently proposed reforms, a range of possible changes in PAE rates of â30 percent, â15 percent, 0 per- cent, +15 percent, and +30 percent is examined. If there is no (0 percent) change in PAEs (column 1), the net cost to the hospital is the same as the gross costs estimated, $1.6 billion (rounded from column 2). Table 9-6 (column 1) shows that a reduction of between 15 and 30 percent in PAEs would be necessary to balance the costs of implementing the four reform components from the hospital perspective. Specifically, the savings from a decline in PAEs of 21.4 percent would totally offset the costs of the four components of the model. From a societal perspective, a reduction of PAEs of less than 15 percent, actually 7.2 percent, would be sufficient to offset the $1.6 billion costs of the four components. Please note that the com- mittee is not predicting that all programs will reduce shifts and total hours to those of the modelâs assumptions. Neither is it predicting that harmful errors will drop by 7 percent. However, should this occur, if the other cost assumptions are close to reality then the costs to society of the reforms and the savings from reduced harmful errors, from a societal perspective, would be roughly in balance or cost neutral. Ideally, the current study would be based on the actual rates and costs of PAEs across U.S. teaching hospitals after the 2003 ACGME reform was implemented, but such data are not available at this time. FUNDING THE COMMITTEEâS RECOMMENDATIONS The cost model considers hiring additional residents and hiring other types of providers as mutually exclusive options. In reality, a combination of these two substitution strategies is likely to be used as the committeeâs recommendations are implemented in various ways by programs across the country. The main implication of the economic model is that the proposed reform is costlyâ$1.7 billion in 2008 dollars. The costs of achieving these reforms relative to the total costs of GME would be approximately 9 per- cent of current GME payments now borne by all payers ($1.7 billion of the $18.7 billion estimated for 2003 by Wynn ) based on a substitution strategy of either additional residents or other providers. In comparison to Medicareâs total outlays of $440.6 billion in 2007 (CBO, 2008), the $1.7 billion of substitution costs are 0.4 percent. The costs of adapting to resident duty hours in 2003 were borne by teaching institutions under existing funding. Some institutions may have to make relatively few changes to comply with the committeeâs recommenda- tions; they may have residency programs that do not schedule extended
316 TABLE 9-6â Results: Net Costs of Proposed Changes, Considering Costs of Resident Substitution and Possible Changes in PAEs (2006) Hospitals with ACGME-Accredited Programs COTH Hospitals Possible Change in PAE Net Costs to Cost per Cost per Net Costs to Cost per Cost per Rate After Reform U.S. Hospitals Hospital Admission U.S. Hospitals Hospital Admission Hospital perspective â30% â$633,547,948 â$525,330 â$36.03 $130,424,709 $355,381 $15.58 â15% $471,139,635 $390,663 $26.79 $656,147,561 $1,787,868 $78.40 0% $1,575,827,218 $1,306,656 $89.61 $1,181,870,414 $3,220,355 $141.22 +15% $2,680,514,801 $2,222,649 $152.43 $1,707,593,266 $4,652,843 $204.04 +30% $3,785,202,384 $3,138,642 $215.25 $2,233,316,118 $6,085,330 $266.86 Societal perspective â30% â$4,995,355,121 â$4,142,086 â$284.06 â$1,945,367,346 â$5,300,728 â$232.45 â15% â$1,709,763,951 â$1,417,715 â$97.23 â$381,748,466 â$1,040,187 â$45.62 0% $1,575,827,218 $1,306,656 $89.61 $1,181,870,414 $3,220,355 $141.22 +15% $4,861,418,388 $4,031,027 $276.45 $2,745,489,293 $7,480,897 $328.06 +30% $8,147,009,557 $6,755,398 $463.29 $4,309,108,173 $11,741,439 $514.90
resources to implement improvements 317 duty periods, that already allow sufficient time off, and that provide excel- lent supervision, or they may have sufficient resources on hand to fund the necessary changes. Fortunate hospitals might have sufficient operating funds, the ability to create efficiencies and savings, a growing patient base, or the opportunity to raise private funds to support the recommended changes. Not all hospitals, however, would be able to bear the costs of these new proposals. Determining the financial capacity of teaching institutions to absorb some portion of these costs was beyond the scope of this study. If some hospitals are unable to absorb these costs fully, it could impact other parts of service delivery (e.g., the amount of uncompensated care) or other qual- ity improvements (e.g., adoption of electronic health records). If funds are unavailable to hire substitutes to pick up residentsâ excess hours, hospitals might increase residentsâ workload or overburden other staff. This could lead to a decrease in opportunities for learning and indirectly affect patient safety. Without sufficient staff, patients also might have decreased access to hospital services. The cost projections of the model do not include the additional costs of implementing the committeeâs recommendations that might stem from duty hour adjustments such as the extra day off per month, safe transportation options, more detailed compliance auditing, and faculty supervision. In New York State, an extensive, detailed duty hour compliance audit costs on average $24,000 per hospital. The cost of oversight would depend on the frequency and nature of any inspections, and there may be local costs of compliance monitoring as well. Also, the model does not estimate the re- search costs of monitoring the implementation of all the recommendations and evaluating their impact. On the other side of the ledger, the model does not calculate potential savings. Implementation of system redesign strategies could produce savings stemming from efficiencies and streamlining of work and education systems as well as the prevention of PAEs, but these savings are not projected or offset against the costs of the recommendations. The committee believes that additional funding and personnel should be made available to support workload reduction and compliance with the recommended duty hour limits so that they do not have undesirable effects on patient or resident safety. Residents experienced work compressed into fewer hours after the 2003 duty hour rules and now would have increased pressures. If the recommended duty hour parameters are implemented without additional funds, this would be another unfunded mandate that some training institutions could not afford and workload might be shifted to other staff who are frequently overloaded themselves. The committee recognizes that this is not an opportune time to be ask- ing for additional funds for the health system. It understands that there are strong political pressures on the Medicare budget and all domestic spending
318 RESIDENT DUTY HOURS and that MedPAC has recommended reductions in IME funding. It also recognizes that there are many payers that support GME, various places other than health care where the federal budget could be cut, and many pa- tients who want to receive safer hospital care and have better-trained phy- sicians. The committee has responsibly interpreted the available evidence and reached clear conclusions that implementing its recommendations will require additional funds. It urges Congress and other supporters of GME to carefully consider this report and to seek funds to help hospitals implement the duty hour changes. Recommendation 9-1: All financial stakeholders in graduate medical education, such as the Centers for Medicare and Medicaid Services, De- partment of Veterans Affairs, Department of Defense, Health Resources and Services Administration, states and local governments, private insurers, and sponsoring institutions, should financially support the changes necessitated by the committeeâs recommendations to promote patient safety and resident safety and education, with special attention to safety net hospitals. â¢ An independent convening body should bring together all the major funders of graduate medical education to examine current financing methodologies and develop a coordinated approach to generate needed resources. WORKFORCE IMPLICATIONS Resident duty hour and workload adjustments, as well as the greater degree of supervision by attending physicians recommended by this com- mittee, will create demand for more residents, midlevel providers, and trained physicians to provide 24-hour coverage in training facilities. This demand for staff will come at a time when other national trends are driv- ing demand for hospital services and personnel to staff these facilities. Technology changes, increasing numbers of patients being hospitalized, and an aging population in the United States are primary contributors to the demand for inpatient services (IOM, 2008; Kozak et al., 2006). Calls for patient safety improvements will also create more demand for hospital- based staff (Shulkin, 2008). For example, some hospitals are recognizing the need to increase staff in hospitals on nights and weekends when there is greater mortality, but currently less comprehensive staffing. In addition, calls for residencies to incorporate more diverse settings than hospitals into â Medicare Payment Advisory Commission, 2008 Report to Congress: Medicare Payment Policy, Washington, DC, March 2008.
resources to implement improvements 319 training programs (e.g., ambulatory care settings) would leave less resident time for inpatient coverage and require more personnel to cover existing inpatient facilities (COGME, 2007). Shifting workload to existing team personnel does not appear to be a solution in many settings because of the workload pressures already experienced in hospitals by all staff (Weissman et al., 2007). The issue of the proper size of the physician workforce and the ad- equacy of the supply of particular specialists is a controversial one, lacking consensus (Iglehart, 2008). Many professional organizations, panels of ex- perts, and researchers have identified current and projected workforce short- ages for nurses, nurse practitioners, and physicians (AAMC, 2006, 2007b, 2008a,d; American Association of Colleges of Nursing, 2004; ANSR, 2008; Colwill et al., 2008; Larson and Hart, 2007; National League of Nursing, 2005; Salsberg, 2008). There are projections of a physician shortage across the United States by 2020, especially in certain geographic areas and cer- tain specialties, and calls for increasing the size of medical school classes and residency positions to replace an aging physician workforce and serve the greater care needs of an aging population (AAMC, 2007b, 2008a; COGME, 2007; Colwill et al., 2008; IOM, 2008). On the other hand, a body of evidence indicates that there may be a surplus or at least not a shortage of physicians. Researchers who have analyzed geographic and hospital-specific resource data find that there are vast variations in the size of the physician workforce among geographic regions that are not associated with improved health outcomes and better- quality care (Fisher, 2004; Wennberg et al., 2004). These studies indicate that health care over a period of time beyond just an inpatient stay could be made more efficient, and that effective systems of care provide higher- quality care over the course of a patientâs chronic illness. This longitudinal efficiency relates particularly to the lower use of supply-sensitive services such as inpatient hospital days, imaging and diagnostic tests, and physician visits. Studies also show that solely increasing the total supply of physicians is an inefficient way to benefit the specialties and geographic areas that may need more doctors (Goodman, 2004). The reforms necessary to achieve the improvements identified in these studies are beyond the scope of this report and the work of doctors in training. The pipeline to produce physicians is a long one: a minimum of 4 years of medical school and 3 to 7 or more years of residency. The Association of American Medical Colleges recommends a 30 percent increase in medi- cal school enrollment from 2002 levels by 2015, an increase of 5,000 new positions annually (AAMC, 2008b). The Council on Graduate Medical Education (COGME) has recommended an increase in the number of CMS- funded residency positions by at least 15 percent by 2015, about 3,000 new positions yearly, as well as diversification of training sites based on an
320 RESIDENT DUTY HOURS Institute of Medicine model of care delivery (COGME, 2007; IOM, 2003). This 15 percent increase in residency positions would absorb about half of the expanded number of medical school graduates. The remainder would offset the enrollment of international medical graduates who now compose approximately 27 percent of all residents (Brotherton and Etzel, 2007). The 8,247 additional residents projected by the Nuckols and Escarce economic model, if a larger pool of residents were to cover the excess hours of resident work, is a total incorporating both specialty (7,639) and subspecialty (608) residents (Table 9-5); it is not equivalent to the yearly increase in medical students or residency positions mentioned above. Ad- ditional residency positions are one approach to filling the gap in covered hours. Since the Medicare funding cap on residency positions was put in place, only training programs with access to alternative sources of funding (e.g., private resources) have been able to expand their programs to ad- dress reduced hours. The committee strongly urges that all possible funding mechanisms be considered, including increasing or eliminating the cap on residency positions. At the same time, the committee recognizes that each institution will have to assess its local labor market, educational capacity, and unique circumstances to determine the most effective way to achieve adherence to the proposed requirements for duty hours, workload, and su- pervision. Individual institutions may or may not find additional residents to be the preferred approach. A phased implementation of duty hours, its evaluation, and further research Phase-in of Recommendations To promote safe medical care, improve the education of doctors in training, and increase the safety of residents and the general public, the committee offers its recommendations, which should be implemented with all deliberate speed. The committee believes action is needed urgently be- cause U.S. hospitals still have a too-high error rate and too many patients are harmed during their stay. The committee realizes that its recommenda- tions will not prevent all patient harm and that residents are not responsible for the whole problem. However, resident duty hours and schedules is a risk that can be ameliorated. Fatigue contributes to unsafe conditions and can increase the risk of errors. Fatigue among residents is something that can be reduced through a judicious use of periods for rest and sleep between duty periods and by limits on extended long duty periods. Other benefits are likely to result from the recommendations, including fewer automo- bile accidents caused by tired residents, a better environment for learning
resources to implement improvements 321 and working that enhances the acquisition of needed competencies, and a greater participation by residents in the hospitalâs culture of safety. While some recommendations could and should be implemented im- mediately, changed duty hours, workload, and funding issues might require an integrated phase-in. The recommendations will require additional re- sourcesâboth financial and human. Without the necessary restructuring in resource allocation, attempts to implement certain recommendations will fail to have the desired benefit and could even produce conditions that are less favorable to patient safety. The committee believes that the ACGME and the other organizations charged to implement aspects of the recommendations should undertake their work with urgency and that all institutions with residency programs take action to begin implementation of all recommendations within 24 months. The committee proposes several recommendations that should be con- sidered as a package and implemented in concert. The recommended pa- rameters concerning the reduction in duty hours and fatigue mitigation are particularly intertwined. Although the 80-hour week and the 30-hour limit on extended duty are unchanged from the current rules, the recommended length and scheduling of rest periods during the month are crucial to mak- ing those duty hour limits supportable. The recommended periods for rest during extended duty and periods away from the hospital to allow for rest and sleep recovery are intended to establish safer working conditions, to protect residents from excessive fatigue, and to protect patients from fatigue-induced errors. The recommended duty hour parameters are also closely linked to the recommendations concerning workload, supervision, and funding. Supervision enhancements and workload reductions, however, could be put in place before duty hours are changed. Some institutions will probably be able to implement the recommended changes independently, but others would need outside funds to help support the hiring of additional staff to assume the excess duties (workload and hours) of residents. Not only would the benefits of these duty hour parameters be less likely to materialize if they are implemented piecemeal and in a disjointed fashion, but also unintended and potentially harmful consequences could result without the accompanying committee recommendations on workload and funding. For example, if duty hours are reduced again but workload re- mains at current levels, residents will be under greater stress as they rush to complete work, which increases the likelihood of making errors and could further reduce the amount of time they have for educational activities; this could negatively impact the safety of both current and future patients. The potential benefits to society of well-trained physicians and fewer people suf- fering from PAEs argue for all funders of GME to contribute appropriately to support these GME reforms. Some recommendations should be implemented immediately, such as
322 RESIDENT DUTY HOURS limits on moonlighting, enhanced supervision, improved handovers, error reporting, and the provision of transportation to residents finishing a long duty period and those too tired to drive safely. Also, programs should quickly enhance their formal education efforts on sleep to include the latest scientific research on fatigue, sleep deprivation, and methods to mitigate fatigue. Research and Evaluation Plans The lack of systematic data collection before and after the 2003 rules hampered the committeeâs ability to determine their impact fully and to as- sess how much of the complaints about duty hour reform represent rhetoric and resistance to change rather than valid criticisms. The literature that exists too often comes from single-institution studies with insufficient sta- tistical power to determine effects on patient outcomes and is often specific to one specialty, making findings difficult to generalize. Collecting baseline information now on the current situation concerning residentsâ duty hours and workload would permit evaluations of the impact of this reportâs recommendations once they are implemented. When design- ing evaluation studies, it will be important to include a sufficient number of programs so that the studies can produce a national picture across all programs as well as evaluations targeted to individual specialties. Ongoing data collection at the national level will be useful for moni- toring the full impact of the committeeâs recommendations. Because major policy changes cannot be tested in the laboratory and it is impossible to predict all the effects of those changes on the healthcare system, there could be some unintended and unanticipated reactions to the recommendations. For example, some specialties might find that programs reduce hours and workloads without accompanying redesign of education, causing fewer opportunities for residents to achieve procedural or medical competency or requiring a longer residency. Or the less continuous care provided by residents could increase their detachment and reduce their commitment to their patients if work is not adequately restructured to permit enhanced team coverage. While the committee certainly does not want to increase workload or hours for residents, some programs might try to meet the rec- ommended parameters of rest and work periods by increasing the frequency of overnight duty periods and reducing the current amount of time off duty. When Residency Review Committees (RRCs) implement specialty-specific workload caps, it would be useful for them to monitor the impact of the caps on both residents and their learning as well as on the costs, coverage, and access to services at training institutions. Similarly, when designing and evaluating innovative projects to test creative ways to meet the intent of the committeeâs recommendations while allowing for alternative approaches that might better suit a type of program
resources to implement improvements 323 or specialty, researchers must take care to include a sufficient number of sites to generate statistically powerful findings and a design that will sup- port comparisons of the impact of the recommendations as implemented. While many experts have told the committee that one rule does not fit all and that every specialty has different characteristics and needs, it is difficult to have a realistic understanding of their differences and similarities without reliable data. Some of the ideas for the types of data that would be useful have been detailed in the preceding chapters. Below are some key research topics that the committee recommends for future consideration. These and other research issues have been discussed throughout the report. It is important for all the stakeholders in GME, contributors of ideas as well as funds, to be included in a discussion of an evaluation and research agenda, and in- volved in the necessary priority setting. It will be a challenge to researchers to tease apart the various and overlapping contributing factors to patient safety to determine the extent of their individual impacts, but the results will have widespread benefits and would be of use well beyond the scope of the current study. â¢ The relationship between improved processes for handovers and shorter and longer duty hours. This report has discussed many of the multiple factors in the resident work and learning environment that contribute to error and potential patient harm. Understanding the handover process and the risks and benefits it poses for pa- tients is an important aspect of the issue that is ripe for systematic research, primarily to learn ways to improve on current methods, but also to assess whether the risks to patients from handovers are greater than the risks of being cared for by a fatigued resident who has more familiarity with the patient. â¢ Resident fatigue and patient harm, as well as residentsâ own safety within various scheduling and fatigue mitigation approaches. On the assumption that there will be considerable variation in the way individual programs choose to meet the parameters of the duty hour recommendation, further examination of scheduling effects on PAEs and resident safety for specialties and program sizes is im- portant, along with analysis of the buffering effect of enhanced su- pervision and teamwork. Research could investigate whether duty hours can be further reduced from current recommendations. â¢ Resident workload and its impact on patient safety. Resident work- load has been understudied for its effect on short-term and long- term patient safety. The workload of residents also needs to be assessed for its educational value. â¢ Measuring and achieving competence by specialty within reduced
324 RESIDENT DUTY HOURS duty hours and workload. Although competency-based education is a broader topic than could be covered by this committee and relates to the restructuring of GME more generally, it is of concern to this study as a method for assessing the impact of the reduction of duty hours and workload on resident attainment. More efficient methods of teaching and conveying information and procedural skills, as well as the assessment of residentsâ knowledge and skills would contribute to the positive impact of the committeeâs recom- mendations. Achieving competence also means that residents need sufficient exposure to an appropriate range of patient experiences including, depending on specialty, outpatient exposure. â¢ Opportunities for and limitations to substitution of other provid- ers for residents. Better workforce data are needed to assess the impact of implementing the recommendations on both the existing workforce and future workforce needs nationally and in specific geographic areas. Researchers should also assess whether and the extent to which additional residents are needed. â¢ Impact of the IOM recommendations after implementation. Clearly the impact of implementation on patient safety is the overriding concern; however, other impacts, such as resident safety, are also important. The variations in impacts based on different approaches to implementation by different programs and specialties are also key to the fine-tuning of the rules in the future. The committee believes that it is essential to build an information base to evaluate what happens going forward. With mechanisms to monitor and evaluate the ongoing implementation of the committeeâs recommenda- tions, as well as innovative experiments, it should be possible to adjust the rules periodically as needed. For example, grounds for ACGMEâs granting exceptions might have to be tightened or expanded; or further measures might be necessary after workloads are reduced through the elimination of noneducational activities, if the work remaining is consistently of too high an intensity. Additionally, such information, as previously mentioned, could also help to avoid major problems or unintended consequences, such as an aspect of the duty hour parameters producing insurmountable chal- lenges to certain specialties more than others that cannot be handled on an exceptions basis; or residents not sleeping during the prescribed 5-hour rest period who may choose to catch up on paperwork instead of mitigating their fatigue; or some small residency programs having to close because of insufficient staff to cover the excess hours of residents. Recommendation 9-2: To gather the data necessary to monitor imple- mentation of these recommendations and to prepare for future adjust-
resources to implement improvements 325 ments as needed to achieve the desired objectives, ACGME should convene a meeting of stakeholders and potential funders to set pri- orities for research and evaluation projects. The Centers for Medicare and Medicaid Services, Agency for Healthcare Research and Quality, National Institutes of Health, Department of Defense, Department of Veterans Affairs, and other funders should support this work as a high priority. Because so many individuals and organizations have strong economic and professional interests in GME, and resident duty hours in particular, it will be a challenge to come up with an agenda for research projects. It will be even more challenging to design research projects that can produce sound scientific evidence of use to policy makers. Given the likelihood that research and evaluation funds will be quite limited, it is especially impor- tant that they be spent wisely to focus on priority issues and to do so in a methodologically sound manner that is acceptable to the key organizations that will have to be involved in implementation of future policy changes. The results of these research, monitoring, and evaluation projects should, in the future, indicate areas for further refinement of the rules and alterna- tives that could better enhance the goals of patient safety, resident safety, and training. REFERENCES AAMC (Association of American Medical Colleges). 2006. AAMC statement on the physi- cian workforce, 2006. http://www.aamc.org/workforce/workforceposition.pdf (accessed June 4, 2008). âââ. 2007a. AAMC survey of housestaff stipends, benefits and funding: Autumn 2007 report. Washington, DC: Association of American Medical Colleges. âââ. 2007b. Recent studies and reports on physician shortages in the U.S., 2007. http:// www.aamc.org/workforce/recentworkforcestudies2007.pdf (accessed June 4, 2008). âââ. 2008a. Help wanted: More U.S. doctors projections indicate America will face short- age of M.D.s by 2020. http://www.aamc.org/workforce/helpwanted.pdf (accessed June 4, 2008). âââ. 2008b. Medical school enrollment plans: Analysis of the 2007 AAMC survey. Washington, DC: Association of American Medical Colleges. https:services.aamc.org/Â publications/showfilecfm?file=version111.pdf&prd_id=229&prv_id=279&pdf_id=111 (accessed June 4, 2008). âââ. 2008c. Teaching hospitals: About AAMC-COTH membership. http://www.aamc. org/teachinghospitals.htm (accessed August 12, 2008). âââ. 2008d. U.S. medical school enrollment projected to rise 21 percent by 2012: Both new and existing schools will fuel growth. http://www.aamc.org/newsroom/pressrel/ 2008/080501.htm (accessed May 2, 2008). AMA (American Medical Association) Division of Market Research and Analysis. 2005. Medi- cal students and residents work-hour survey May. Chicago, IL: MemberConnect.
326 RESIDENT DUTY HOURS American Association of Colleges of Nursing. 2004. Nursing faculty shortage fact sheet. http://www.aacn.nche.edu/Media/backgrounders/facultyshortage.htm (accessed June 3, 2008. Andrews, R. M., D. E. Stull, I. Fraser, B. Friedman, and R. L. Houchens. 2007. Serving the uninsured: Safety-net hospitals, 2003. Rockville, MD: Agency for Healthcare Quality and Research. ANSR (Americans for Nursing Shortage Relief). 2008. Testimony regarding FY 2009 appro- priations for Title VIIIâNursing workforce development programs. https://www.ncsbn. org/FY09_Senate_Testimony.pdf (accessed August 26, 2008). Arora, V., C. Dunphy, V. Y. Chang, F. Ahmad, H. J. Humphrey, and D. Meltzer. 2006. The effects of on-duty napping on intern sleep time and fatigue. Annals of Internal Medicine 144(11):792-798. Brotherton, S. E., and S. I. Etzel. 2007. Graduate medical education, 2006-2007. JAMA 298(9):1081-1096. CBO (Congressional Budget Office). 2008. The budget and economic outlook: An update. Washington, DC: Congressional Budget Office. Supplemental data of mandatory out- lays. http://www.cbo.gov/budget/factsheets/2008b/medicare.pdf (accessed September 23, 2008). Chang, B. K. 2007. Presentation to the Committee on Optimizing Graduate Medical Trainee (Resident) Hours and Work Schedules to Improve Patient Safety, December 3, 2007, Washington, DC. COGME (Council on Graduate Medical Education). 2007. 19th report: Enhancing flexibil- ity in graduate medical education. Rockville, MD: Bureau of Health Professions, U.S. Department of Health and Human Services. Colwill, J. M., J. M. Cultice, and R. L. Kruse. 2008. Will generalist physician supply meet demands of an increasing and aging population? Health Affairs 27(3):w232-w241. COTH (Council of Teaching Hospitals and Health Systems). 2008. AAMC teaching hospi- tals and health systems. Washington, DC: Association of American Medical Colleges. Pp. 14-38. Fisher, E. S., D. E. Wennberg, T. A. Stukel, and D. J. Gottlieb. 2004. Variations in the longitu- ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ dinal efficiency of academic medical centers. Health Affairs (Supplement Web Exclusives): VAR19-VAR32. Goodman, D. C. 2004. Twenty-year trends in regional variations in the U.S. physician work- force. Health Affairs (Supplement Web Exclusives):VAR90-VAR97. Henderson, T. M. 2006. Medicaid direct and indirect graduate medical education payments: A 50-state survey. Washington, DC: Association of American Medical Colleges. HRSA (Health Resources and Services Administration). 2008a. Other health professions programs: Childrenâs hospitals graduate medical education payment program. http:// www.hrsa.gov/about/budgetjustification09/childrenshospital.htm (accessed October 16, 2008). âââ. 2008b. Training in primary care medicine and dentistry. http://www.hrsa.gov/about/ budgetjustifcation09/hptraining.htm (accessed October 16, 2008). Iglehart, J. K. 2008. Medicare, graduate medical education, and new policy directions. New England Journal of Medicine 359(6):643-650. IOM (Institute of Medicine). 2003. Health professions education: A bridge to quality, edited by A. C. Greiner, E. Knebel, and the Committee on the Health Professions Education Summit. Washington, DC: The National Academies Press. âââ. 2008. Retooling for an aging America: Building the health care workforce. Washing- ton, DC: The National Academies Press.
resources to implement improvements 327 Jagsi, R., D. F. Weinstein, J. Shapiro, B. T. Kitch, D. Dorer, and J. S. Weissman. 2008. The Accreditation Council for Graduate Medical Educationâs limits on residentsâ work hours and patient safety: A study of resident experiences and perceptions before and after hours reductions. Archives of Internal Medicine 168(5):493-500. Knapp, R. M. 2002. Complexity and uncertainty in financing graduate medical education. Academic Medicine 77(11):1076-1083. Kozak, L. J., C. J. DeFrances, and M. J. Hall. 2006. National hospital discharge survey: 2004 annual summary with detailed diagnosis and procedure data. Hyattsville, MD: National Center for Health Statistics. Landrigan, C. P., L. K. Barger, B. E. Cade, N. T. Ayas, and C. A. Czeisler. 2006. Internsâ com- pliance with Accreditation Council for Graduate Medical Education work-hour limits. JAMA 296(9):1063-1070. Larson, E. H., and L. G. Hart. 2007. Growth and change in the physician assistant workforce in the United States, 1967-2000. Journal of Allied Health 36(3):121-130. MedPAC (Medicare Payment Advisory Commission). 2008. Hospital acute inpatient services payment system. http://www.medpac.gov/publications/other_reports/Sept06_MedPAC_ Payment_basics_hospital.pdf (accessed October 14, 2008). National League of Nursing. 2005. Nurse faculty shortage fact sheet. http://www.nln.org/ governmentaffairs/pdf/NurseFacultyShortage.pdf (accessed June 3, 2008). Nuckols, T. K., and J. J. Escarce. 2005. Residency work-hours reform. A cost analysis includ- ing preventable adverse events. Journal of General Internal Medicine 20(10):873-878. âââ. 2008 (unpublished). Cost analysis for the Institute of Medicine project âOptimizing Graduate Medical Trainee (Resident) Hours and Work Schedules to Improve Patient Safety.â RAND Corporation. www.iom.edu/residenthours or www.iom.edu/hcs. Opas, L. 2008. IOM workshop: Cost impact of duty hour change. Presentation to the Com- mittee on Optimizing Graduate Medical Trainee (Resident) Hours and Work Schedules to Improve Patient Safety, March 4, 2008, Irvine, CA. Petersen, L. A., T. A. Brennan, A. C. OâNeil, E. F. Cook, and T. H. Lee. 1994. Does housestaff discontinuity of care increase the risk for preventable adverse events? Annals of Internal Medicine 121(11):866. Petersen, L. A., E. J. Orav, J. M. Teich, A. C. OâNeil, and T. A. Brennan. 1998. Using a com- puterized sign-out program to improve continuity of inpatient care and prevent adverse events. Joint Commission Journal on Quality Improvement 24(2):77-87. Salsberg, E. 2008. Addressing healthcare workforce issues for the future. Statement before the Committee on Health, Education, Labor, and Pensions (HELP), United States senate, February 12, 2008. Shulkin, D. J. 2008. Like night and dayâShedding light on off-hours care. New England Journal of Medicine 358(20):2091-2093. Siegel, B., M. Regenstein, and P. Shin. 2004. ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ Health reform and the safety net: Big opportuni- ties; major risks. Journal of Law, Medicine & Ethics 32(3):426-432. Stoddard, J. J., D. A. Kindig, and D. Libby. 1994. Graduate medical education reform. Service provision transition costs [erratum appears in JAMA 272(16):1256]. JAMA 272(1):53-58. Thomas, E. J., D. M. Studdert, J. P. Newhouse, B. I. W. Zbar, K. M. Howard, E. J. Williams, and T. A. Brennan. 1999. Costs of medical injuries in Utah and Colorado. Inquiry 36(3):255-264. U.S. Department of Labor: Bureau of Labor Statistics. 2007a. Employer cost for employee compensation, annual data for 2006. http://data.bls.gov (accessed August 12, 2008). âââ. 2007b. National compensation survey: Occupational wages in the United States. http://www.bls.gov/ncs/ocs/sp/ncbl0910.pdf (accessed August 12, 2008).
328 RESIDENT DUTY HOURS Weissman, J. S., J. M. Rothschild, E. Bendavid, P. Sprivulis, E. F. Cook, R. S. Evans, Y. Kaganova, M. Bender, J. David-Kasdan, P. Haug, J. Lloyd, L. G. Selbovitz, H. J. Murff, and D.Â W. Bates. 2007. Hospital workload and adverse events. Medical Care 45(5): 448-455. Wennberg, J. E., E. S. Fisher, T. A. Stukel, and S. M. Sharp. 2004. Use of Medicare claims data to monitor provider-specific performance among patients with severe chronic illness. Health Affairs (Supplement Web Exclusives):VAR5-VAR18. Werner, R. M., L. E. Goldman, and R. A. Dudley. 2008. Comparison of change in quality of care between safety-net and non-safety-net hospitals. JAMA 299(18):2180-2187. Whang, E. E., M. M. Mello, S. W. Ashley, and M. J. Zinner. 2003. Implementing resident work hour limitations: Lessons from the New York State experience. Annals of Surgery 237(4):449-455. Wynn, B., C. Guarino, L. Morse, and M. Cho. 2006 (unpublished). Alternative ways of financ- ing graduate medical education. RAND Corporation, Santa Monica, CA. Zhan, C., and M. R. Miller. 2003. Excess length of stay, charges, and mortality attributable to medical injuries during hospitalization. JAMA 290(14):1868-1874.