Under the proposed rule, CMS will change how it determines whether evaluation and management (E&M) services were furnished in a skilled nursing facility (SNF), as opposed to a custodial (non-skilled) nursing facility setting, for purposes of beneficiary assignment to an accountable care organization (ACO). While the change will cause minimal impact to physicians who do not primarily practice in a nursing facility setting, there will be important effects for physicians who do:

• Most nursing facility-based physicians will see a material change in the number of assigned beneficiaries, with roughly one-third seeing an increase or decrease of at least 25%. Office-based physicians will see minimal impacts to the number of beneficiaries assigned to them as a result of this rule change.
• ACOs with nursing facility-based physicians will no longer need to focus attention on how to adjust their place of service coding practices to improve their MSSP performances.
• Nursing facility-based physicians who are in ACOs that have chosen retrospective assignment will be less penalized by the regional cost adjustment to their benchmarks under the proposed rule. However, the penalty will continue to be far greater than if the ACO were to choose prospective assignment.

In this paper, we discuss the current and proposed rules related to E&M services in nursing facilities and then present our analyses supporting the findings noted above.

Background

In the MSSP, beneficiaries are assigned to the ACO that provided the plurality of “primary care services” during a defined one-year assignment window. From 2012 to 2016 in the MSSP, E&M services that occurred in an SNF were considered primary care services for purposes of assigning patients to ACOs. During this time, ACOs became concerned that beneficiaries were being assigned to them solely on the basis of care provided in an SNF. These beneficiaries may have seen their assigned physician only during their time in the SNF, which is commonly one to two months, but the volume of E&M visits during that short time was enough to outweigh all community-based primary care visits outside the SNF for the remainder of the year.

To address this concern, CMS modified the definition of primary care services starting in 2017 to exclude nursing facility E&M visits billed with place of service (POS) 31, which indicates the claim occurred in an SNF. E&M visits with POS 32 (Nursing Facility) were still included because providers seeing patients in a non-skilled nursing facility setting are likely to be responsible for the care of these patients on an ongoing basis. However, this rule change did not completely solve the issue, and in fact, it introduced additional unintended consequences. Research by Milliman¹ and others indicated that POS 31 and POS 32 were frequently coded incorrectly, and therefore many E&M services in an SNF were included while others that occurred in a nursing facility were excluded. In addition, changes in coding practices over time created distorted benchmarks if these high-cost beneficiaries were assigned to ACOs at different rates during the historical benchmark years than the performance years.

In the proposed rule, CMS intends to increase the accuracy of the assignment process by changing the way it determines whether E&M services occurred in an SNF. Under the proposed rule, CMS will exclude services for certain nursing facility E&M codes when there is an SNF facility claim with dates of service that overlap with the dates of service for the professional claim. The goal is the same as the rule implemented in 2017: to exclude E&M services that occurred in an SNF from the assignment process.

Anticipated impact of proposed rule change

Using the 2016 Medicare 5% Sample Limited Data Set (LDS), we analyzed the impact of the proposed rule’s approach for determining whether E&M services occurred in an SNF on different types of physicians. In particular, we focused on physicians who primarily see patients in the nursing facility, contrasting them with the large majority of physicians who primarily see patients in an office setting.

Figure 1 shows the change in the number of assigned beneficiaries due to the proposed rule for physicians who primarily work in an office compared to those who primarily work in a nursing facility (or SNF). Although roughly 85% of office-based physicians will see virtually no change in the number of assigned beneficiaries, most nursing facility-based physicians will see material changes. Nearly 10% of nursing facility-based physicians would see an increase of at least 50%. The average change in the number of assigned beneficiaries for nursing facility-based physicians is expected to be an increase of approximately 14% under prospective assignment and an increase of approximately 17% under retrospective assignment.

Figure 1: Percentage change in number of assigned beneficiaries by physician’s primary setting

Note: Values in Figure 1 are based on retrospective assignment. Percentage change in number of assigned beneficiaries is similar under prospective assignment.

In addition to the change in the number of assigned beneficiaries, there is likely to be significant disruption in which beneficiaries are assigned to each physician. For nursing facility-based physicians, we estimate over 20% of assigned beneficiaries under the current rule would not be assigned to the same physician under the proposed rule. Nearly one in five nursing facility-based physicians will have beneficiary turnovers of at least 50%. Office-based physicians will see much less disruption; approximately 99% of assigned beneficiaries under the current rule will also be assigned to the same physicians under the proposed rule.

Average per capita expenditures and the impact of the regional benchmark adjustment for nursing facility-based physicians will also change under the proposed rule. We analyzed this in two ways:

Figure 2 (retrospective assignment) and Figure 3 (prospective assignment) summarize the expected changes in per capita expenditures and regional benchmark adjustments for nursing facility-based physicians compared to office-based and other physicians. End-stage renal disease (ESRD) beneficiaries were excluded for simplicity due to the additional complexity of the ESRD version of the CMS-Hierarchical Condition Category (HCC) risk score model.

• Under retrospective assignment, the average per capita expenditures for nursing facility-based physicians is estimated to be similar under the proposed rule, but the regional benchmark adjustment is estimated to be more favorable by approximately $150 per beneficiary per year (PBPY). This is because the population assigned to nursing facility-based physicians under the proposed rule is expected to have higher risk scores than the population assigned under the current rule. This change should help nursing facility-based physicians to compare more favorably to their region, although they are still expected to generate an unfavorable regional benchmark adjustment.²
• Under prospective assignment, per capita expenditures are estimated to increase by about 2%, but the impact of the regional benchmark adjustment is expected to be relatively similar to the current rule. In aggregate, the regional benchmark adjustment should have only a modest impact on nursing facility-based physicians under prospective assignment.

We want to emphasize that these are averages across the entire Medicare fee-for-service (FFS) population, and individual physicians and ACOs may experience very different impacts.

Although the proposed rule change will have more of an impact to benchmarks under retrospective assignment than prospective assignment, risk-adjusted expenditures for nursing facility physicians are still much higher under retrospective assignment. The reason for this disparity is that when beneficiaries are retrospectively assigned to a nursing facility-based physician, the performance year often includes acute costs (such as a hospitalization) prior to the patient entering the nursing facility. These high costs are not fully accounted for by CMS-HCC risk scores, which are based on diagnosis codes from the prior year and demographic information.

Methodology

Using the 2016 LDS, we assigned beneficiaries to physicians under the current and proposed criteria for identifying E&M services that occurred in an SNF. Under the MSSP, beneficiaries are assigned to ACOs using the tax identification numbers (TINs) on claims; however, because TIN information is not available in the 2016 LDS, we assigned beneficiaries to the National Provider Identifier (NPI) that provided the plurality of primary care services during the assignment window. We then grouped these assigned physicians based on the POS where they most frequently bill. Because of the known issues with billing POS 31 and 32, we combined POS 31 and 32 for purposes of identifying the provider’s primary setting.³  

For simplicity, we used expenditures for the national assignable FFS population to create risk-adjusted regional benchmarks for each physician. For each physician’s assigned beneficiaries, national assignable FFS expenditures in each beneficiary entitlement category were risk-adjusted to the physician’s average risk scores and the beneficiary entitlement categories were weighted together using the physician’s mix of beneficiaries. Risk scores were estimated using the 2017 CMS-HCC risk score model version. We used the 2017 model version to reflect material changes that were made to coefficients for institutionalized beneficiaries relative to the 2016 model version. We renormalized the risk scores using the average risk score of the national assignable population in each entitlement category in our analysis.

Per capita expenditures for each beneficiary were truncated consistent with MSSP specifications, but they do not reflect all other MSSP adjustments, such as removing sequestration, indirect medical education (IME), and disproportionate share hospital (DSH) payments.

In order to isolate the impact of the rule change related to SNFs, we did not reflect other changes to the assignment methodology, including the addition of several Healthcare Common Procedure Coding System (HCPCS) procedures to the list of primary care services, the implementation of new HCPCS codes, and the expansion of voluntary alignment.

Conclusion

CMS has proposed to change how it determines whether E&M services were furnished in an SNF for purposes of beneficiary assignment to an ACO. This rule will have minimal impact on ACOs with physicians who do not primarily practice in nursing facility settings. However, an ACO with physicians who do primarily practice in nursing facility settings may experience a significant change in the number of beneficiaries assigned to it. These ACOs will also typically find that prospective assignment results in more favorable regional cost adjustments to their benchmarks than retrospective assignment. Each ACO should analyze its specific circumstances to understand how the proposed rule change will affect the beneficiaries assigned to the ACO as well as the regional cost adjustment to its benchmark.