The most rapidly growing specialty in medicine is hospitalist medicine. The demand exceeds the supply and as a consequence, salaries are increasing annually at a remarkable rate. Most hospitals cannot function without hospitalists and most hospitalists cannot earn their entire salary from clinical billings alone. Thus, hospitals and hospitalists have formed a symbiotic relationship over the years with hospitals subsidizing hospitalist salaries and hospitalists directing efforts toward quality and financial metrics that benefit the hospital.
So, from the hospital’s perspective, how do you know how much you should be paying to subsidize your hospitalists? Pay too little and you will be unable to compete with other hospitals for the limited number of hospitalists in your community. Pay too much and you risk losing money on inefficiency and not getting good value for the dollar.
Over the past 10 years, the measure of productivity for hospitalists was primarily the RVU. The advantage of the RVU is that it is a readily available and easily quantifiable measure of productivity and physician work, Many hospitals will subsidize a hospitalist group based on the RVUs that they generate or based on a target of a certain number of RVUs per FTE. I argued in a previous post that you shouldn’t pay the hospitalist by the RVU and that the hospital should instead incentivize the hospitalist based on other measures, such as readmission rates, mortality rates, query responsiveness, etc. I propose that the hospital should subsidize hospitalist groups based on the number of complexity-adjusted admissions. In this post, I will outline the model to calculate this and then lay out the arguments for why it works.
The Complexity-Adjusted Admission Model
This model incorporates several values including the annual number of hospital admissions for the hospitalist group, the case mix index of those patients, the current typical hospital subsidy per hospitalist in the U.S. and the average hospitalist salary. The model can be summarized by:
Calculating the “Subsidy per Adjusted Admission” allows the hospital to compare the cost of hospitalist care from one year to another and between different hospitalist groups that may care for different populations of patients. Lets look at each component:
- Annual Admissions. In this model, the number of admissions that the hospitalist group has per year becomes the foundation of measured volume. In other models, the number of encounters is used as an estimate of volume but there are problems with encounters. If encounters are used, then there is no allowance for the difference in the amount of effort that it takes to do an admission versus a discharge versus a daily inpatient visit. It is much easier for the hospitalist to do 15 return visit encounters than to do 15 admissions or discharges.When encounters are used as the measure of productivity, the incentive to discharge patients promptly is lost and there is actually an incentive to keep patients in the hospital longer so that the hospitalist an get more encounters with relatively less effort. By using annual admissions, the hospitalist would be motivated to discharge patients faster (thus reducing length of stay) in order to make room on his/her census to be able to admit more patients. One question that comes up is should the number of annual admissions include both inpatient admissions and observation admissions? Although arguments can be made to manage inpatient and observation patients separately, the easiest answer is to include both types of admission in the analysis. Another challenge is what to do with consults and co-management encounters, which are commonly done by hospitalists. On the one hand, co-management of the medical conditions of surgical patient provides value to the hospital by reducing the chance of complications that can otherwise increase the length of stay. On the other hand, co-management encounters typically take less time and effort than daily patient management as the primary attending physician of record.
- Case Mix Index (CMI). In this model, the CMI is used as an indicator of complexity in order to adjust the amount of effort that it takes for a hospitalist to care for any given inpatient on any given day. The higher the CMI, the more highly weighted an admission becomes. As an example, this overcomes the difference between a regular medical admission versus an ICU admission. In other models, the number of wRVUs are used as a measure of effort but the problem with wRVUs is that most inpatient medical visits will be coded as either a level II or level III visit and this does not leave much room for the subtleties of patient complexity; for example, most ER admissions will be coded as a level III new patient visit regardless of whether the patient has a medium or high number of co-morbidities. Furthermore, from the hospital’s standpoint, the CMI is very important – documenting all of those co-morbid conditions on admission pushes the CMI up and thus pushes up the amount of money that the insurance company pays the hospital for any given DRG. By using CMI in the model, the hospitalist’s subsidy is aligned with what the hospital values. One pitfall to be aware of is that surgical admissions (especially orthopedic admissions) have a very high CMI that does not really reflect the work for the medical management of that patient and if the hospitalist is co-managing those surgical admissions, then their CMI-adjusted admissions could be inflated – it is therefore better to use the CMI for the patients admitted to the hospitalist’s service rather than the CMI of consult/co-management surgical patients.
- Number of FTEs. It is better to simply use the total FTEs in the hospitalist group rather than admissions/encounters/wRVUs per individual physician because some shifts have inherently fewer admissions/encounters/wRVUs (for example, night and ICU shifts). The ratio of hospitalists providing day coverage to hospitalists providing night coverage can vary considerably depending on the population of patients being cared for and the frequency of admissions during the night. It could be rather low at 2:1 for an ICU-dominated hospitalist practice or as many as 4:1 for a low-acuity hospital practice with few nighttime admissions. Similarly, in a tertiary care hospital, a night shift hospitalist may only be able to care for 50-60 patients but in a smaller community hospital, a night shift hospitalist may be able to care for 90-100 patients. Many hospitalist groups will incorporate advance practice providers (nurse practitioners or physician assistants) but these “APPs” are generally not equal to a physician in the annual amount of work that they can do – one option is to count each APP as a 0.5 FTE.
- Hospital Subsidy. The hospital subsidy per hospitalist can vary considerably depending on the payer mix of the hospital (a higher percentage of uninsured or Medicaid patients will require a higher subsidy), whether the hospitalist is in private practice versus academic practice, the regional average hospitalist salary, etc. An article in the journal The Hospitalist reported that the average hospital subsidy per hospitalist was $157,500 in 2016. This may seem high on first look, however, data from the 2017 MGMA Physician Compensation and Productivity Report (using 2016 salary data) indicated that the median compensation:collection ratio for private hospitalists is 1.402 and for academic hospitalists is 1.386; in other words, a hospitalist’s salary is more than he/she can bring in by professional billing alone.
Why This Model Works
The complexity-adjusted admission model has a number of advantages over other hospitalist productivity models that allow for better alignment of the priorities of the hospitalist group with the priorities of the hospital.
- It encourages co-morbid condition documentation by the hospitalist, thus enhancing hospital reimbursement per DRG.
- It encourages lower length of stay by rewarding the number of admissions to the hospital rather than by the number of encounters or RVUs.
- It facilitates right-sizing hospitalist nighttime coverage. By measuring the total admissions that the entire hospitalist group does per year, rather than the number of encounters/RVUs that each individual hospitalist does, it makes it easier to account for night shifts that have inherently fewer total encounter or RVUs than day shifts.
- It reduces the culture of hospitalists “dodging admissions”. I’ve known residents and hospitalists who will put in more time to get out of an admission than it would actually take to do the admission. The model makes incentivizes the hospitalist to do more admissions.
- It allows equal comparison of hospitalist groups that work longer shifts but fewer shifts per year to hospitalist groups that work shorter shifts but more shifts per year. For example, a hospitalist group where the doctors work 183 12-hour shift per year versus a hospitalist group where the doctors work 210 shifts per year that are a mix of 10 and 12-hour shifts.
Putting It All Together
Next, lets see how the model works given two hypothetical situations.
Hospitalist Group #1. Assume this hospitalist group has 3 daytime physicians and 1 nighttime physician (total of 4 per day). Each shift is 12-hours. Each physician works 182 shifts per year. For full staffing, the hospitalist group needs 8 FTEs. This group only cares for medical floor admissions (not ICU admissions) and has a CMI = 1.1. They have 4,839 admissions per year. The hospital subsidizes the group by $157,500 per hospitalist FTE. Therefore, using our model, the subsidy required is $237 per CMI-adjusted admission:
Hospitalist Group #2. Assume this hospitalist group has 6 daytime physicians and 2 nighttime physicians (total of 8 per day). Each physician works 182 shifts per year and each shift is 12-hours. For full staffing, the hospitalist group needs 16 FTEs. This group cares for ICU patients as well as medical floor patients and has a CMI = 1.3. They have 8,312 admissions per year. The hospital subsidizes the group by $157,500 per hospitalist FTE. Therefore, using our model, the subsidy required is also $237 per CMI-adjusted admission:
Given these two hypothetical hospitalist groups with different CMIs, both come out to a subsidy of $237 per CMI-adjusted admission. These are hypothetical cases; no one knows the average across the country. But from the 2017 MGMA Physician Compensation and Productivity Report (which reports on 2016 data), the average number of encounters per hospitalist in non-academic practice was 2,114 per year. 
If one assumes a 3.5 day average length of stay, then that comes to right around 600 admissions per hospitalist per year; if the length of stay is 4.5 days, then it works out to 470 admissions per hospitalist per year. Assuming a range of annual admissions of 470 (2.5 per shift) to 670 (3.5 per shift) per hospitalist per year, the range of subsidy per adjusted admission should be between $235 and $335. Making allowances for the variation in payer mix, patient volumes, etc. and extending this range, then a reasonable value is probably between $200-$400 per CMI-adjusted admission. The table contains data from the most recent MGMA report (2016 data reported in the 2017 publication). There is a wide-range for each of these values and there is considerable variation from one region of the United States to another.
Structuring The Bonus Incentive
The above model allows the hospital to determine the amount that the hospital will subsidize a hospitalist group, in this case, $237 per CMI-adjusted admission. This should form the base salary but there also needs to be an bonus incentive component to reward the hospitalist for other things that the hospital values such as:
- Readmission rate
- Mortality rate
- Patient satisfaction scores
- Emergency department throughput times
- Discharge time of day
- Pharmacy cost per admission
- Hospital-acquired infection rates
Summary
The hospital will almost always have to subsidize the hospitalist. The challenge is to structure the subsidy in a way that aligns the desires of the hospital with the desires of the hospitalist. Subsidies are like compensation plans, whatever the subsidy rewards is what you get and sometimes there can be unintended consequences if what the subsidy actually rewards is not anticipated. The hospital should start with a target of subsidizing between $200 – $400 per CMI-adjusted admission; if the actual subsidy is either more or less than this range, then further analysis of the reasons why are necessary.
June 13, 2018
