Compensation models are meant primarily to reward physicians for their industry, but also function to assign cost based on usage and the operational needs of the organization. Over time any compensation model will need a revision, ranging from a minor tweak to a complete redesign, for a variety of reasons. Some of these reasons include: change in physician composition or practice style; changes in reimbursement from government and private third party payers (e.g., capitation contracts with per member per month payments); incentives to reward or change physician behavior; reward quality metrics as defined by the organization; recruitment of new providers to the group; and regulatory compliance such as the Stark Law.(1) The multispecialty group met most of these criteria as a reason for the redesign of the compensation formula. In addition, the current compensation formula created silos in the sense that it treated each physician and the various specialties differently. Over time this led to tensions between the multispecialty group physicians and a new formula that created more cohesion was necessary.
Methods
The multispecialty group consisted of medical and surgical specialties, with a slight predominance of primary care providers. In addition, the group had advanced nurse practitioners and ancillary revenue from laboratory and imaging services that needed to be factored into the compensation model.
We first decided to interview all the stakeholders in the multispecialty group to learn what they wanted to see in a compensation formula and also to educate them on the reasons for the redesign. We discovered that the physicians wanted a compensation model that was easy to understand and was fair to everyone. There was a strong sense that the overhead expenses should be shared equitably but not necessarily equally. They did not want a redistribution of funds that could adversely impact the ability of some physicians to continue in the group: no one wanted a model that would deviate significantly from their current compensation.
We also examined the preceding three years of physician compensation using the existing model. We used Microsoft Excel and the R program software(2) to analyze the model. (The choice of software was based on the design team’s preference.) We analyzed physician compensation trends, ancillary contribution to physician compensation, primary care versus specialist compensation, relative value units (RVUs)(3) of the physicians, advanced practice nurse contribution to the compensation of the physicians, and overhead expenses and the range of contribution to those expenses by the individual physicians and specialties (e.g. pediatrics, obstetrics and gynecology, cardiology).
We also had to obtain legal advice about the Stark Law requirements for the laboratory and radiology ancillary revenue and how that could be used legally for physician compensation, because these ancillary units were owned by the multispecialty group.
It was then necessary to decide whether the redesign should be done in-house or contracted out to a consultant. It was going to be more cost effective to design the model internally, and because we had been part of the multispecialty group for several years we felt we would have a better understanding of the group dynamics and history than a consultant. Every organization has a unique culture, and it is important to understand this culture thoroughly before designing a compensation model, especially because a compensation model is one of the key links between members of the organization.
Different compensation models were examined: RVU-based compensation; a combination of RVUs and cash-based production models; cash only models; and a combination cash and quality metric–based compensation. We also needed to factor in the contributions from the laboratory and radiology ancillary departments.
In this instance, RVU-based compensation was going to be unfair to some specialties, especially the primary care physicians. RVUs do not take all the work a physician does into account. Among the things that do not produce RVUs are making referrals to specialists and ancillary units such as the lab and radiology; paperwork, such as prior authorizations and home health certifications; coordination of care with specialists; phone calls; and after-hours call coverage. Also, physician-owned entities are purely cash based, but unlike most corporations that bill and receive an equal payment to what is billed, physician groups are subject to contractual write-offs from insurers and often unpaid balances on patient accounts. This meant that a physician could generate RVUs without this translating into the expected income. Furthermore, because of the discrepancy in payments between government insurers such as CMS(4) and private insurers, RVUs could be the same for different physicians but the revenue could significantly differ. It was decided to have a cash-based model with no RVU input.
It also was difficult to define quality and find metrics that everyone could agree with. However, some insurers, such as Medicare and Medicaid, have quality incentive programs, and it was decided that these would be used to incentivize the primary care specialties to improve their compensation.
The contribution from the laboratory and radiology ancillary units was kept separate because of the need to be in compliance with the Stark Law, and resulted in the creation of a separate ancillary compensation model.
A physician-owned entity is responsible for all its overhead, which translates to high physician overhead. Hence, any compensation model must distribute only the money left over after expenses are paid. We also reviewed the organizational budget and made sure that the physician compensation was in line with budgetary targets.
We also had to decide the best time period to measure physician productivity: a single month or the average of several months. The advantage to choosing the average of several months’ production is that it smoothed out the fluctuations in physician production; however, it also leaves the organization and physician several months behind in physician compensation. This can create interesting conundrums if a physician should leave the organization for any reason and their payout must be calculated. During this phase of the discussion we floated the idea of a base compensation based on the Medical Group Management Association (MGMA)(5) data on physician compensation. There would then be an annual bonus or adjustment based on physician performance. There was resistance to this, because physicians have been used to compensation based on their actual production and not an external standard. This was a strong cultural pattern that they did not wish to give up, mostly because they felt this made each individual physician more responsible for their productivity. After discussions, it was decided that physician compensation would be on a one-month basis and not the average of several months.
The previous compensation model was based on a tiered set of ratios: as an employee produced more revenue, the percentage that person kept increased. For example, the first $1000 would attract a tax ratio of 0.75, the next $1000 a tax ratio of 0.5, the next $1000 a tax ratio of 0.25, and so on. Furthermore each specialty had a different set of tiers. Essentially, each physician could calculate their expected income based on this model. The flaw in this model was that it did not take the organizational expenses into account. As a result, an expected distribution often was not possible because the funds available after expenses were paid did not add up to enough to cover all the calculated distributions for the physicians. This meant that each physician would have to have their compensation prorated to account for the available funds. Obviously this was frustrating for the physicians and led to friction with the administrative staff. The physicians clearly expressed a desire to move away from this model to one that divided the profits without any additional calculations, as noted earlier. It was also felt, as noted previously, that this method of compensation created silos and did not unite the physician body.
The compensation model had to meet the criteria mentioned earlier, but fundamentally it was a vehicle to assign cost to the physicians and then divide the multispecialty group profit to the physicians on a monthly basis. We first tackled how to divide the expense. Most physicians have a good idea of their direct office expenses (e.g., nurses, drugs, supplies), but the indirect expenses (e.g., administrative costs, security, electricity, internet expenses) are less visible. It was the latter category that often proved contentious and was the focus of the compensation model’s attempt to share those costs among the physicians. Medical costs are high, and a simple way to divide the indirect expenses would have been to divide the costs equally, but this would probably prove prohibitively high for some specialties, especially the lower-paying primary care specialties. Based on our analysis of the organizational expenses, we realized that about half of the indirect expenses were about equal (e.g., administrative time, internet, and phone usage.). We decided to share half of those indirect costs equally among the multispecialty group physicians. This served the purpose of having a common burden for cost and uniting the group around this. The remainder of the cost was shared based on a percentage of utilization. Once this was agreed on, it was relatively straightforward to assign compensation from the money set aside for physician compensation.
The advanced practice nurses presented an unusual challenge. They were salaried employees, yet they were working as extenders for physicians. They typically generate less than physicians, because of the scope of their work and lower compensation from insurers when they do not bill incident to the physician.(6) Considerations here included the fact that these advanced practice nurses were still utilizing the same amount of resources as a physician provider, but generating less revenue than a typical provider in the same specialty. The consensus was that full-time advanced practice providers should be treated as physicians and run through the compensation model. Their salary would be subtracted from the guaranteed distribution generated by the compensation formula, and the supervising physician would take the balance as his compensation for supervision. The downside of this decision is that an advanced practice nurse who was going negative in the compensation formula would be costing the supervising physician revenue. This makes the decision to bring on an advanced nurse practitioner a serious financial consideration as well as one that meets a necessary need and not just a convenience.
Government insurers (i.e., Medicare and Medicaid) have incentive payments for quality metrics that they define. This actually brings in extra revenue. We decided to adopt some of these measures for physician compensation, separate from the compensation model. It was easier to agree on those definitions of quality.
After the model was completed, we met individually with each physician and explained the model along with scenarios of how the individual would have been compensated over the previous year. We then ran the model simultaneously with the existing compensation model for several months to make sure that it would behave as designed. Finally it was voted on by all the physicians and adopted for implementation.
Discussion
This compensation model reflects the complexities of designing compensation for a medium-sized physician-owned entity. Each organization that employs physicians will have different considerations based on the employment contracts. A single-specialty group with a few providers may just divide all the overhead equally and share profits. From here on it gets more complex, however, and large groups may need more sophisticated methods. It also depends on ownership. Employed physicians are not concerned about overhead to the same extent as the owner physician. If the physicians are owners, they also own the expenses; these owner-physicians are our target audience. Our experience is that a good first step to the actual model is to identify all the expenses and then decide how to share them in a way that reflects utilization, but it is also important to keep the cohesion of the group. If every expense is assigned, what makes it a group effort?
Physician compensation is very complex, and it is subject to regulatory considerations from the civil authorities. One example is the Stark Law that we have mentioned. Most insurers also align their contractual rates of reimbursement with CMS rates, which also is regulatory.
We struggled with the definition of quality. There are organizations that promote quality metrics, such as the Healthcare Effectiveness Data and information Set (HEDIS)(7) or the National Association for Healthcare Quality (NAHQ)(8) quality programs. Quality programs can be either process or outcome based. Given the shrinking rate of reimbursements for physicians and also the fact that the payer system still rewards volume and procedures performed, physician time is very constrained, and adding these quality metrics is often seen as an imposition on time they do not have. In some cases the work burden also is greatly increased for what some physicians view as paltry compensation. Often the odds are stacked against them, because the metrics often are difficult to achieve, especially the outcome-based ones. It is not surprising that in designing a compensation model there is often some level of resistance to the addition of quality metrics. However, these are extra sources of revenue from insurers such as Medicaid and Medicare.
It is also noteworthy that current reimbursement schemes do not reward the often unseen parts of medical practice which we have noted previously. Other considerations for the future will include physician experience. There is no mechanism where payers factor that into their compensation. Physicians also spend a lot of hours outside of their regular office time on documentation, a problem that has worsened with the advent of electronic records. This issue is not addressed in reimbursements or compensation schemes, but it often involves long hours of actual work, including interpreting and communicating test results to patients and additional actions beyond the office visit.
Multispecialty groups with a large primary care representation also are going to have to contend with the erosion of traditional office-based primary care by telehealth services (at times provided by insurers), concierge services, and the increasing entry of advanced practice nurses into the field. How do you incorporate this into your business model and make sure that the physicians are well compensated?
Furthermore, this article is being written at a time of significant inflation, and to date no insurer has factored that into their payment schemes. Aligning compensation with inflation is another area that will need to be addressed in the future.
Another issue we grappled with was physician time off, whether for illness or vacation. Any time off results in a drop in production and hence a drop in compensation for the time period in question. It is unrealistic to expect a physician to work without a break. How does your compensation model protect physicians from periods of no production? As more women enter the medical field, time off for pregnancy and childcare after that becomes a factor in their job choices. An organization looking to recruit and retain female physicians will have to grapple with this in their distribution formula if they want to retain these physicians.
It is also important for the stakeholders to understand that any compensation model is really a calculation with assumptions and is a quest for the most equitable way to share cost and reward the individual with whatever is left after those costs are paid. Compensation models can be a source of friction and ultimately must be seen as the best compromise across a wide range of specialties. It must be easy to understand and also must be fiscally responsible so the organization is not put at financial risk from guaranteed payments.
We were greatly helped by the availability of software—the Excel and R programs—that made our data analysis less burdensome than it could have been. This allowed us to make our indirect expense utilization more transparent and to run various compensation model scenarios using the most recent past data.
Every distribution model is good only while the original premises under which it was designed prevail. As noted previously, there are many factors that, over time, erode the validity of a compensation model; when this occurs, a redesign is required. Indeed, every organization should regularly check its compensation model to make sure it is doing what it was intended to do. Developing a culture where physicians understand this need and embrace it will go a long way to making future compensation model redesign less arduous in terms of physician buy-in.
A compensation model that works well has many benefits. If physicians understand it and feel their compensation is fair, the group has better retention and motivated physicians. This also translates into better recruitment, because these same physicians will sell an organization they are happy with, and potential recruits also need to understand how they will be compensated. A good compensation model also brings cohesion to a group and stops in-fighting over payment of overhead expenses. In any organization, wages usually are the highest expense, and a good compensation model is part of good financial stewardship, because it controls the amount of wage expenditure on physicians and should be part of budgetary constraints. This will avoid situations where the organization is forced to tap into strategic reserves to fund a compensation formula that is over-paying. It also resolves the issue of underpaying physicians for work they have done.
Lastly, a physician compensation committee made up of some of the group’s physicians is very helpful in getting them to own any compensation model. This group will function to ensure that the model is working as predicted and will be ready to make suggestions about changes. This goes a long way toward obtaining physician ownership of the model. Indeed, after the model is designed and adopted, it is invaluable to have the physicians take control and ownership.
References
Office of Inspector General. Fraud and abuse laws. U.S. Department of Health and Human Services. October 5, 2021. https://oig.hhs.gov/compliance/physician-education/fraud-abuse-laws/ . Accessed January 3, 2024.
The R project for statistical computing. www.r-project.org/ . Accessed January 4, 2024.
AMA. Relative value units. https://cpt-international.ama-assn.org/relative-value-units . Accessed January 4, 2024.
Centers for Medicare & Medicaid Services. CMS.gov. www.cms.gov/ . Accessed January 6, 2024.
Medical Group Management Association. www.mgma.com/ . Accessed January 6, 2024.
Freer E. “Incident-to” billing for nonphysician services requires vigilance. Texas Medical Association. www.texmed.org/TexasMedicineDetail.aspx?id=59312 . Accessed January 6, 2024.
Healthcare Effectiveness Data and Information Set (HEDIS). Healthcare Effectiveness Data and Information Set (HEDIS). Healthy People 2030. https://health.gov/healthypeople/objectives-and-data/data-sources-and-methods/data-sources/healthcare-effectiveness-data-and-information-set-hedis#:~:text=HEDIS%20measures%20address%20a%20range,medication%20management%3B%20immunization%20status%3B%20and . Accessed January 6, 2024.
National Association for Healthcare Quality. https://nahq.org/ . Accessed January 6, 2024.