Population Based Reimbursement: Optimizing A Primary Care Network

Primary Care Networks (PCNs) have been used for many years to provide clinical oversight of patient care and to help control the cost of care and enhance the quality of care. Multiple implementation approaches have been used with varying degrees of success. This document provides valuable suggestion to help optimize the network and its implementation. Perhaps some of the proposed solutions will help resolve your questions before you have them.

Population Based Reimbursement vs. Budget Based Reimbursement

Population based reimbursement is nothing more than establishing a health care budget for a population (i.e., often expressed as a per person per month or PMPM budget) and managing to that budget. The appropriateness of the budget drives the appropriateness of the method. If the budget makes no sense the methodology fails. Therefore, the terminology of a budget-based reimbursement methodology.

For the program to make sense the budget has to be reasonable. This requires the budget to reflect or adapt to the unique risk characteristics of the population, its burden of illness, its intended reimbursement of fee-for-service equivalency level, etc. However, even in the sometimes rare situation of a “perfect” budget it can fail. One of the major causes of this is inadequate risk mitigation.

Budget-Breakers

As mentioned earlier, the primary risk in population based reimbursement is budget adequacy or appropriateness. Any factor or issue that can impact this needs to be monitored, understood, anticipated and managed.

Some good examples of budget-breakers are:

• Statistical fluctuations (i.e., law of small numbers)
• Bias in demographic mix (i.e., older vs. younger populations)
• Bias in risk characteristics of the assigned members
• Shock claims (i.e., catastrophic claims)
• Outlier provider practice styles
• Scope of the provider’s practice
• Excessive referral rates
• Poor referral choices (i.e., specialists and facilities)
• Uncontrollable drug costs

 

There are many others but these will be discussed below.

Risk Mitigation Approaches

The above examples of budget-breakers have plagued implementations for years, yet there are very simple and straight-forward solutions that can be readily applied to minimize their impact.

The negative effects of the Budget-breakers can be mitigated with simple planning processes.”

Let’s look at several of these and show some great solutions to this problem. For each we will explain the problem then provide a solution to the problem.

Too few members in a PCP practice: This is a common complaint of PCPs whose assigned members was small. It is true that the budget was not built on the assumption of the “law of small numbers”! The budget was based upon several key assumptions, but most importantly, a reasonable spread of risk. When the assigned member count is low there are statistical fluctuations, many times significant. In addition, the risk characteristics of the enrolled population is different than assumed in the “average” member budget. How many assigned members are enough. Both statistical analysis and observed practice shows that the most stable results are obtained with a minimum member size of at least 50 commercial members or at least 35 Medicare members. The larger the better but is clear that membership below these levels results in significant statistical fluctuations.

Some have tried to solve the problem by letting providers bill fee-for-service below some threshold and be capitated above the threshold. It may sound good on paper but in practice it rarely works. The biggest risk to the sponsor is with the smallest practice. Unless a narrow network model is chosen, the mere negotiation with a large number of PCPs and the slow start-up of enrollment forces no financial risk being transferred to the PCPs since everyone falls below the reasonable thresholds discussed above. This solution does not work.

The most effective approach utilizes a simple concept called Pools of Doctors or PODs. By financially grouping individual PCPs into convenient operating groups, the collective performance of the POD becomes a very manageable entity. The size limitations often go away immediately and the budgets are stable. We find that grouping 5 – 10 PCPs together is usually the most effective. It is especially effective if they already have some business connection (i.e., same medical group or IPA, collective on-call group, etc.). When the monthly financial results come out then can meet and discuss what happened. One of the most effective ancillary results of this grouping is the peer pressure that comes when the patients are discussed. Some entities encourage the Chief Medical Officer (CMO) to meet with the groups to discuss problem cases and provide solutions to their concerns. This is an ideal situation and has been applied and tested in multiple situations since the mid-1980s[1].

PODs provide a useful solution to inadequate enrollment per PCP.”

Demographic and Risk Bias: When the enrolled population is older / younger than expected or sicker / healthier than expected the budget no longer fits and providers express serious concern. This is easily resolved with the use of both age/gender factors and risk adjustment. The budget is adjusted up/down based upon the variation in mix. Demographic factors have been available forever however market accepted risk factors or risk indices have been available since the late 1990s. The choice of a factor is more an issue of what mechanism is accepted by the provider population. My advice is to choose one that is directionally correct, easy to use and one that has been tested. The precise tool is not as important as the consistent use of it.

Catastrophic claims can disrupt the incentive plan and are hard to predict and manage.”

Catastrophic Claims: Unexpected rare, very expensive claims can create big impacts on incentive mechanisms built into a primary care network. The most common solution for this is stop-loss reinsurance. This works to smooth out the fluctuations and helps avoid discouragement of providers, a very important issue. PCP services, if defined appropriately, rarely have this issue. It is usually limited to those services referred out by the PCP (i.e., specialist, institutional, ancillary, Rx). The most frequent application applies a per person per year threshold to all referred claims. In essence the excess claims are “forgiven” or not included in the financial reporting and are replaced with an average PMPM stop-loss charge. This smooths out the fluctuation from period to period and increases the validity and acceptance of the budget for the provider and their performance measurement.

Scope of a PCP’s Practice: Since the scope of different PCP specialties varies from one PCP to the next (i.e., what is a PCP service and what isn’t) it is critical that this be carefully defined. It is nearly impossible to establish an appropriate budget without specificity. In addition, who should be a PCP? Could a specialist be a PCP in addition to providing their specialist services (e.g., general surgeon, OB/GYN). It is clear that most specialists have some understanding of primary care from medical school and early residence training, but are they best prepared to provide these services? What about self referral? What about capacity to handle it?

What specialties make good PCPs? What about OB/GYNs as PCP? Are self-referrals okay?”

Without clarification and definition, this becomes a serious budget breaker and a hotbed for severe selection bias further breaking budgets.

The solution is fairly simple. First of all, develop a CPT-code list of primary care services that all PCPs are expected to provide. There is no one single appropriate list for all markets but generally a broad consensus. At a minimum survey the local PCP providers, identify those services that at least 80% of them provide and that at least 80% of their revenues are generated from these services. This is an appropriate PCP services list.

Next, have providers choose to be a PCP or a specialist. Avoid at all cost any situation where a PCP could refer to themselves for specialist services. By process of elimination you will find that for the most part the PCP self-selecting roles will be General Practice, Family Medicine, General Internal Medicine (no sub-specialty) and General Pediatrics (no sub-specialty). These are truly the primary care doctors.

The one exception is usually OB/GYN physicians. Ready access to well woman care is statutory right in most states. It is also good medicine. We don’t recommend have OB/GYN physicians on the PCP list since they rarely want to do the rest of PCP services beyond well woman services. Some PCPs do and are willing to provide them if chosen by a PCP.

In the case of some minor office based surgical procedures, internal medicine doctors may not want to provide these. If provided by someone else and these procedures are on the list, there would be a charge back to their PCP services fund or PCP capitation rate when performed by someone else.

Controlling Referrals: The primary care network provides a useful structure to provide and manage health care services. This includes referral services,both who does the service and who should do the service.

How can we control over referral? How do we control referral to the wrong specialist?”

One of the biggest problem plaguing the healthcare system and the cost of it is inappropriate or excessive referrals to specialists. One of the biggest benefits of the primary care network is the framework it provides to manage referrals and help the patient receive the appropriate referral care at the appropriate time.

First of all, the primary care network structure naturally lends itself to recording the referrals, their cost and to whom the patient was referred. The primary care assignment naturally associates these services with the PCP. The comparison of the budget for referral services with the actual services shows one measure of appropriate level of referral (i.e., assuming appropriate budget adjustment for factors previously discussed). Fund balances suggest controlled referrals, deficits suggest over referral.

The primary care network can also help measure appropriate referral. One very successful approach utilizes variable withholds for specialist services. An initial withhold level (i.e., say 10%) is established. In the case of referral providers who provided services to patients associated with a specific PCP experienced a deficit in the referral fund, those providers would forgo some or all of their withhold. In the case where the referral fund experienced a surplus they would receive all of their withhold plus an incentive payment above and beyond their normal fee.

This provides a useful resource to help identify the good referral physicians by calculating the average effective withhold return for each referral physician. Providers with the lowest or negative return are the least desirable providers. Providers with the highest are the one associated with the best performing pools. This can be further enhanced by using “variable withholds”. The referral withhold in each PCP pool experience a loss are increased. The withhold in pools paying bonuses are decreased. Each referral provider’s average withhold represents their association with “good” or “bad” pools. The lowest average withhold level helps define the most attractive referral providers. Although a little complex to manage, this approach helps naturally categorize the providers. The PCP can select specialists from a list of specialists in their area ranked by average withhold. This gives the PCP a “real” tool to make good choices. This also adds a competitive feel to each specialist in accepting the referral from a PCP based upon something that directly impacts them.

This same approach can be applied to institutional care, perhaps with some additional metrics for common types of patients. The hospital with the best outcomes, the lowest length of stay, the fewest readmits, the best mortality, the lowest cost, the highest quality scores, etc. can be easily identified for selection of site of care.

Uncontrollable Drug Costs: Especially in recent years, drug costs have become a more significant concern as extremely expensive drugs have emerged (i.e. Sovaldi). There may be no way to anticipate or predict whether or not a patient with Hepatitis-C would sign up with a particular PCP. Or a specific PCP might have a practice with a cohort of patients that has an excessive high rate of Hepatitis-C or another condition with high drug costs. How can the sponsor be sure that these situations are appropriately handled and the budget-breaker concern is mitigated.

There are two frequently used methods: pooling and carve-outs. Both are similar, differing only by who takes the ultimate risk. Pooling extracts the cost of these expensive drugs and spreads it across all PCP pools using a flat PMPM pooling charge. This way all PCPs pay for the cost knowing they might have a patient tomorrow with this issue and they would be protected down the road. Carve-out has the same effect except the cost is not charged back directly to the other PCP pools but to the sponsor and all pools are adjusted to fund this transferred risk. Any unpredictable budget-breaker needs to be considered to assure providers that budgets are reasonable.

Summary

The primary care model continues to be one of the best ways to control health care, however, specific enhancements will help it be more effective. The risk mitigation principles discussed above are welcome tools in continuing this effort.

[1] First applied in Kitsap County, Washington for Medicaid population.

David Axene, FSA, FCA, CERA, MAAA, is the President and Founding Partner of Axene Health Partners, LLC and is based in AHP’s Murrieta, CA office.