by Vikas Parekh, MD,and Scott Flanders, MD, Director,
In July of 2003, the Accreditation Council for Graduate Medical Education (ACGME) implemented new rules that restricted resident work hours to no more than 80 per week and restricted continuous duty to no more than 30 hours (24 hours plus 6 hours for transfer of care, the “24+6” rule). As a result, many major academic medical centers face the problem of handling increasing inpatient volume and ensuring compliance with these new work-hours regulations. The problem has become more pressing as several major academic centers have been cited for work-hours violations by the ACGME, and significant public attention has focused on the impact of excessive work hours on patient safety (1, 2).
Given the success of hospitalists in efficiently managing patents in many non-academic environments, one proposed solution has been the creation of hospitalist services to care for patients independent of residents. These services reduce the volume on resident-based services and therefore reduce resident work hours. We have recently implemented our own non-housestaff service at the University of Michigan and in this article describe the challenges and lessons learned.
The first step for any institution contemplating the creation of a non-resident service is to establish clear goals. Frequently, decisions on the level and scope of uncovered services are made without any rigorous analysis of the data or without a clear idea of what it is that your program should be doing.
The first task for any program is to understand what patient volume must be removed to ensure work-hours compliance without impeding the educational experience of the housestaff . Unfortunately, there is little published opinion on optimal resident workload, and the ACGME is surprisingly silent on this vital issue. While the ACGME does proscribe exceeding theoretical maximum workloads for internal medicine, they cite no minimum or ideal patient census (3). In the absence of firm guidelines, it is important to gather data on both the day-to-day variation of inpatient admissions and volume along with peak admission times (usually early evening). The residency program is likely to offer monthly data or a rough guess at what they think is needed. This can be misleading and does not appreciate the variability of patient flow. It is the “peaks’ that are often remembered, whereas the “troughs” are easily forgotten. Vital data elements that should be obtained include the daily admission volume for each resident-service over the course of the past year. We used this data to calculate average per-intern admission volumes and to project what future volume would be under a variety of possible scenarios, including removing a fixed number of patients per day, creating intern-admissions caps or alternating admissions between residents and hospitalists. We then discussed these models and their projected impact on the residents with residency leadership before settling upon our final model.
Besides the question of volume, there is also the issue of whether the new service will also be used to create other structural changes in the resident services. Some areas that programs may consider include modification of the existing call rotation such as reducing or eliminating short-call, changing the frequency of long-call, or implementing limitations on night-time admissions to the housestaff.
Each of these possibilities comes with its own structural needs, so it is vital to decide whether any of these changes are to be attempted.
There is significant temptation to use established hospitalist workload standards and apply them to non-resident services in academia. To do so is to invite disaster. The complexity of patients on most academic internal medicine services is quite different from the average community service. One big variable to address here is whether or not the new hospitalist service will have a selected patient population (such as low-complexity or “non-teaching” cases). Without specifically selected low-complexity cases, most hospitalist programs will realize that established community work standards do not apply.
Much of what residents do on a day-to-day basis involves pushing their patients through the inefficient and complex maze of an academic medical center. It seems ridiculous to think that one faculty member can replace the work that was previously performed by an attending, a senior resident, and two interns, yet this is what many programs are actually proposing when they suggest that the “established” work load of 15 patients per day per hospitalist could work in academia.
What is an ideal workload in academia? Our answer is based both on our experience and on work-flow analysis of residents, which suggests that less than 20% of their time is actually spent in direct educational activities (4). We suggest that the acceptable workload for a hospitalist in a major academic center managing patients of equivalent complexity as the residents is slightly higher than what a senior resident alone can reasonably handle. In our institution we have had a service without interns, staffed with senior residents and one attending for several years. In institutions without this structure, one could look at what senior residents do on their intern’s days off. In our experience approximately 8-10 patients/day seems to be an acceptable workload that allowed the residents to provide quality care within the confines of a 10 to 12 hour day. This translates into an attending workload of 9-11 patients/day. We acknowledge that with time, an attending may develop more efficient practices than a senior resident but do not think a workload much higher than this is reasonable during the start-up phase.
Many hospitalists rely on physician extenders such as physician assistants and nurse practitioners. In academia, physician extenders have traditionally worked only in specialty areas of inpatient care such as orthopedics, oncology, or cardiology. The great unknown, however, is how extenders perform in an environment where they are asked to work with both complex and varied patients. We have seen that the training of many extenders is often not enough for them to take on the role envisioned for them in this kind of service. Over time they may develop the skill set, but there is much on-the-job learning that requires dedicated physician time. A realistic census for a physician assistant (PA) taking care of complex academic medical patients is likely to be 4 to 6. The incremental impact of extenders on a service’s total work capacity is not entirely additive, given the need for physician oversight and the need to maximize revenue by using shared visit billing. Despite these limitations, however, we believe that extenders are helpful, especially given the inefficiencies of day-to-day patient care in academic centers.
Our own program was designed around an original goal of moving 2000 patients from the resident services. This figure was derived from a per-intern workload target of 25 to 30 admissions per month. Based on our modeling of various ways to share admissions, we ultimately settled on a system that alternates admissions with the resident services after each service admits a “baseline” number of patients. This allowed us to variably offload patients based on day-to-day variation in admission volumes. Our service is staffed 24 hours a day with a total of eight full-time physicians and four physician assistants. We have three physicians and two‑three physician assistants during the day (7 a.m. to 7 p.m.) to coincide with the bulk of the workload. There is one doctor at night (7 p.m. to 7 a.m.) for our entire service, and our hospitalists work an average of 50-55 hours a week during 18 shifts a month. Each hospitalist (working with a PA) averages from 8 to 12 billable encounters a day. We maintain a maximum daily census of 30-35 patients and admit up to 10 patients a day. Given these workloads, we do not come close to financial self-sufficiency, but this is not unique to our program.
For most institutions a non-resident service represents incremental faculty members without any significant incremental professional fee revenue. The billings on the new service really are just a shift in revenue from the resident services. In addition, given the high clinical workload and current market conditions, the salaries of hospitalists hired for such services tend to be on average $15,000 to $20,000 above that of hospitalists hired onto a traditional resident-based service. There is some opportunity for increased revenue capture because of 24-hour attending presence, but the incremental gain is unlikely to be enough to create financial self-sufficiency. In our program there has been an increase in department-wide consultative revenue as specialized patients are now placed on our general medical service where they were previously cared for by residents and a specialty attending. In addition, we have improved our charge capture by a small margin. This extra revenue will not, however, come close to offsetting our overall cost. Many programs therefore require hospital support to be viable. Given the strong incentives for hospitals to ensure compliance with ACGME rules and maintain maximal inpatient occupancy, many hospitals can be convinced to provide funding.
We argue strongly that the creation of programs developed primarily to deal with residency work hours should be viewed separately from the funding of existing or new resident-based hospitalist programs. Similar to how resident salaries are paid for by the hospital (via federal graduate medical education funding), the cost of a new hospitalist service that is created to replace residents should come from the hospital. Programs should exercise caution in using existing paradigms such as reduction in LOS or decrease in cost as a basis for funding. There is little data comparing resident-based care to non-resident-based hospital care in a tertiary center, and what little that exists does not necessarily suggest a cost benefit (5). In addition, there is a significant future risk if such proposed benefits do not become a reality
Once established, many programs will be asked to take on additional tasks that were previously performed by trainees or other faculty. This is especially true of nighttime tasks. Many programs are asked to run code-blue teams, supervise procedures at night, supervise sedation in radiology, triage patients in the ER, provide emergent patient coverage for other services: the list can go on and on. The challenge is accepting some and rejecting others without being seen as non-cooperative.
We strongly believe that taking on some of these tasks provides significant added value for non-resident programs, something that becomes vital in the long-run once the urgency of work-hours compliance has passed. Programs should pick wisely and move slowly when adding additional roles. Whatever roles are added, it is vital that ample consideration is given to the impact on workload and faculty satisfaction. Many of these roles may also present an opportunity to garner additional revenue, whether through billing or direct payment from the hospital.
The greatest challenge that all major academic hospitalist programs will face will be how to create satisfying long-term faculty positions that involve providing direct inpatient care without the assistance of housestaff (6). There is already a growing problem of physician dissatisfaction among clinical-track faculty in many academic centers where the emphasis on clinical productivity has usurped the missions of teaching and research. The challenges faced by academic hospitalists working without residents are even greater than those faced by existing clinical faculty.
The first consideration for academic programs is whether to create two classes of hospitalists within the same program: those that work primarily with residents and those that do not. In our program we had an already established group of classic hospitalist-educators who worked only on resident-staffed services when we were asked to create a non-resident service. Our easiest option, therefore, was to hire new faculty whose sole responsibility is staffing a non-resident service. With this has come a significant struggle on how to ensure faculty satisfaction and avoid creating a split within the hospitalist program. We also struggle with how to administer such a program and whether leadership should have clinical roles on both services (we currently do not).
For many new programs, it may be easier to create one uniform faculty role that mixes non-resident-based and resident-based service duties and avoids the appearance of two classes of hospitalists. For many mature programs, however, the only option may be to hire new faculty who predominantly work on non-resident services. For these groups, we believe that differences in the positions must be addressed. One solution to this problem is creating viable teaching roles for these new faculty. Options that we are examing include medical student teaching, training allied-health professionals, and some involvement in resident education during the night and at regularly scheduled daytime lectures. Each of these roles requires time and will come at the expense of efficiency or work capacity. We also have struggled to create program-level rapport. We have encouraged weekly meetings and have found that clinically oriented collaboration such as case conferences and quality-improvement initiatives seem to provide the best way for the entire faculty to interact. Another solution that has been offered is to create a vigorous inpatient research agenda that uses the non-resident services as the laboratory; we encourage this approach but feel that it may not be a realistic near-term goal for many programs.
In the end, however, while creating these roles will add to faculty satisfaction and long-term viability, there will be ongoing problems similar to those faced by academic primary care faculty who have limited interactions with residents. Our program relies on junior-level faculty who are in transition between residency and further training or faculty who aspire to eventually grow into more traditional academic teaching roles and take on a more hybridized role. There is likely to be value in this variety, and we imagine that large academic programs will have faculty that run the gamut from those who are primarily research focused to those who spend most of their time in direct front-line patient care.
Since the implementation of our non-housestaff service, we have seen dramatic improvements in resident work-hours compliance. Prior to our service, 40% of residents were in violation of the 80-hour week and the “24+6” hour shift limit. After successfully removing 15% of the total inpatient (non-ICU) census from resident-coverage, there have been only sporadic violations during the first 3 months of operation. Therefore, violations of the 80-hour work week rules have been virtually eliminated. Our residents have widely praised the new service and overall morale in the residency program has improved. Yet despite what has been perceived as a significant reduction in resident patient load, there are continued violations of the “24+6”-hour shift rule. In fact many have suggested that violation of the “24+6”-hour rule is a reflection of the competing tension between compliance with external regulation and our residents’ professionalism and dedication to patients. While further reductions in volume might help (although even our residents say that this might jeopardize their education), the more likely solution to this problem is both culture change over time and some re-engineering of the timing of resident shifts.
We envision that in the next few years, non-resident services will exist in almost every major medical center. As our experience highlights, these services can be an effective solution to the resident work-hours problem. We caution, however, that implementation is not an easy task. To be successful, programs should invest significant time in the planning stages and have clear goals in mind. Staffing and finances are likely to remain challenging as is the creation of academically viable roles. Eventually, however, we believe these services will succeed. Their growth will add to the future standing of hospital medicine in academic centers by creating a more diverse group of hospitalist faculty who focus on research, education, and, increasingly, quality patient care.
Dr. Parekh can be contacted at firstname.lastname@example.org.
Dr. Flanders can be contacted at email@example.com.
The Hospitalist newsmagazine reports on issues and trends in hospital medicine. The Hospitalist reaches more than 25,000 hospitalists, physician assistants, nurse practitioners, residents, and medical administrators interested in the practice and business of hospital medicine.