Wednesday, September 28, 2016

A Unified Medicare Payment System For Post-Acute Care Is Feasible

An elderly person in a wheelchair

About 40 percent of Medicare beneficiaries discharged from an acute care hospital go on to use post-acute care (PAC) from skilled nursing facilities (SNFs), home health agencies (HHAs), inpatient rehabilitation facilities (IRFs), and long-term care hospitals (LTCHs). Medicare's payments to the more than 29,000 providers of these services totaled $59 billion (or 20 percent of Medicare fee-for-service spending) in 2014.

This care offers important recuperation and rehabilitation services to Medicare beneficiaries, but there are few evidence-based criteria to guide decisions about where beneficiaries should be treated following a hospital stay and how much care they should receive. Congress requested that the Medicare Payment Advisory Commission (MedPAC) recommend features of a unified payment system and consider the effects of moving to such a system. Our work found that a unified payment system is possible and would reduce current distortions.

Medicare's Current Fragmented System For Post-Acute Care

Medicare currently pays for services under each of the four settings using different payment systems that involve different units of payment and payment adjustments, despite overlap in the types of patients treated in the settings. As a result, Medicare pays different prices for similar patients depending on where the patient is treated. While similar patients may receive different services depending on which setting they are in, there is limited evidence that the outcomes for patients in high-cost settings differ from those in lower-cost settings.

For example, prior work showed significant overlap in the patients treated in SNFs and IRFs for 22 health conditions frequently treated in IRFs but for which the majority of patients are treated in SNF. While payments to IRFs were almost always higher than payments to SNFs, to the extent that outcomes could be measured, they were not consistently better in the high-cost setting.

In addition, current payment rules have inherent shortcomings that could be corrected with a unified prospective payment system (PPS) that could span the four PAC settings. Two of the PPSs (HHA and SNF) pay more generously for therapy care (such as physical and occupational therapies) than for medically complex care (such as ventilator care and severe wound care), which drives some providers to focus on therapy patients rather than medically complex patients (such as patients with conditions that involve multiple body systems and patients whose prior hospital stay had a high severity of illness) and to furnish unnecessary services to boost payments.

Given the much higher costs of two other settings (LTCH and IRF), Medicare attempts to differentiate this care from other acute and post-acute care with additional requirements. For example, LTCHs must have a minimum length of stay (25 days) and IRFs must meet certain case-mix thresholds and furnish "intensive therapy" to all patients.

Designing a New, Unified Payment System

Growing concerns over the rapid growth and wide variation in Medicare's PAC spending and the lack of uniform patient assessments to gauge quality prompted the enactment of the Improving Medicare Post-Acute Care Transformation Act of 2014. This law mandated that in June 2016, the Medicare Payment Advisory Commission (MedPAC) recommend features of a unified payment system that would replace the four individual payment systems that Medicare currently uses. Later, the Act also requires the Secretary of Health and Human Services and MedPAC—using two years of data that will begin to be collected in October 2018—to design a prototype PPS and to make recommendations for implementation of a unified system. These reports are expected in 2022 and 2023.

In the course of analyzing potential features of a PAC PPS, we found that a unified payment system is not only feasible, but could be implemented sooner than the timetable outlined in law. As expected, the design would result in major shifts in payments across different types of care and providers — in the directions anticipated (e.g., from providers with high levels of therapy use to those treating a higher share of medically complex patients). We also found that in 2013 (the year analyzed) Medicare payments averaged 19 percent above the costs of PAC, suggesting that program payments for these services could be lowered.

Unified Payments Would Change Incentives

A unified payment system for post-acute care would better match payments to costs based on patient characteristics, and greatly reduce undesirable incentives to treat some types of patients or provide certain services over others. We developed models that accurately predicted the cost of most stays using patient and stay characteristics (including over 60 clinical features) of actual Medicare patients who received PAC services. These models produced estimates of the cost of providing care to these patients that were highly consistent with the actual PAC costs of those stays. This tells us that the models could be used to accurately predict costs for future PAC recipients and could form the basis for setting payments under a consolidated PAC PPS.

We "stress tested" the models by examining how accurate the predictions were for over 40 patient groups defined by the main reason for treatment, patient severity, demographics, and rates of therapy use. The predictions were accurate for the vast majority of the groups, with one main exception. Therapy payments bear little relation to the care needs of patients under the current system, and as a result, the models did not accurately predict the costs of therapy services.

This result was expected and we would expect this to change as providers adjust to a PPS in which payments are no longer based on the amount of therapy services provided. In addition, for the 0.1 percent of PAC stays with the highest acuity, we found that predicted cost was 20 percent less than actual cost. However, under the outlier policy we modeled, additional payments for these cases more than covered the costs of these stays. Inclusion of additional variables in the cost model marking rare but costly health conditions would further improve payment accuracy for this group.

We concluded that it is feasible to develop a common unit of payment (a stay or, in the case of home health care, an episode) and a common set of adjusters to raise or lower payments based on patient and stay characteristics (Exhibit 1). Our results also led to other design recommendations.

First, given the differences in coverage between the settings, two payments should be made for each stay — one for ancillary services other than therapy (mostly drugs) and another for routine and therapy services. Second, because the costs of HHAs are so much lower compared with the other institutional PAC settings, a payment adjustment is needed for stays in HHAs. Third, a short-stay outlier policy would prevent large overpayments and a high-cost outlier policy would prevent large losses by providers and protect beneficiary access to care.

Exhibit 1. Comparison of current and a unified post-acute care payment system

Feature Current PPSs Consolidated PPS
Number of payment systems 4 1
Unit of service Mix of stay- and day-based payments. Per-day payments to SNFs encourage long stays. All payments would be stay-based.
Payment adjusters Adjusters vary by setting, some of which are not empirically supported. SNF and home health payments increase with the amount of therapy provided. Uniform adjusters across settings. Adjustment for home health setting.
Variation in profitability across conditions Large, with higher margins for therapy care and lower margins for medically complex care. More uniform profitability across different conditions.
Selection incentives Incentive to treat therapy patients and avoid medically complex patients. Minimize incentive to selectively admit based on patient condition.
Payments for outlier cases Outlier policies vary for IRF, HHA, and LTCH. No outlier policy for SNF. Short-stay and high-cost outlier policies would apply to all cases.

Unified Payments Would Reduce Distortions

A unified PPS would represent a major change in payments for PAC and would result in large shifts in payments across types of stays and providers. Payments would be redistributed from therapy care to medically complex care and from higher-cost settings and providers to lower-cost settings and providers. Payments would shift from higher-profit stays to lower-profit stays, resulting in more uniform profitability across them. As a result, providers would have less incentive to selectively admit certain types of patients.

For example, based on their current therapy practices and mix of patients, payments would increase to nonprofit SNFs and hospital-based SNFs and would decrease to for-profit and freestanding SNFs. Impacts on any given provider would reflect its relative costliness, the types of stays it treats, and its current therapy practices. Given the objective of a PAC PPS to base payment on patient characteristics rather than setting or the amount of therapy furnished, these impacts are not only expected, they are desired.

If past experience holds true, we expect PAC providers to be responsive to payment incentives in the new PPS. Specifically, we expect high-cost providers to lower their costs to match the PAC PPS payments and all providers to more completely code patient diagnoses. We also expect patient volume to shift to lower cost settings for the types of patients with similar conditions that are currently treated in both higher- and lower-cost settings. A transition period, during which providers are paid a blend of "old" and "new" rates, would give providers time to adjust their costs to match those of lower-cost providers. A high-cost outlier policy that begins with a relatively large outlier pool (but made smaller over time) would help providers adapt and protect beneficiary access.

Unified Payment Could Be Implemented Sooner Than Anticipated

We developed models that could form the basis of a unified PPS using data currently collected by CMS. Although patient assessment data would improve the accuracy of payments for some patient groups, our work shows that average payments would be reasonably accurate for most groups without it. Therefore, it is possible to implement a unified PPS sooner than the legislated timetable, which currently envisions implementation sometime after 2023.

The unified PPS could begin with a design that does not include functional assessment data and transition to one that does when these data become available. During this transition, a larger outlier pool could be established to help compensate for inaccurate payments for high-cost stays and the pool could become smaller as the assessment data were incorporated into the design. As part of the PPS design, the level of total spending for the unified PAC PPS will need to be established. We found that total payments across the four settings exceeded the cost of care by 19 percent, though we expect this difference to decrease over time. That said, payments could be lowered, beginning in a transition period.

As always when a PPS is implemented, the Secretary will need to establish a monitoring program to detect inappropriate provider responses, including: choosing to treat some patients and not others, providing unnecessary PAC stays, stinting on care (which may lower quality and outcomes), and delaying care (which shifts but may not lower program spending). As any unintended consequences of the PPS are documented, the Secretary would need to make refinements.

Further Implications

Our work confirms that a unified PAC PPS is feasible and within reach. A unified PPS that bases payments on patient characteristics will focus providers on each beneficiary's care needs while shifting program spending away from therapy services and towards medically complex care. Because PAC providers would be paid under a single payment system, Medicare will need to re-consider setting-specific policies such as Conditions of Participation rules and cost-sharing requirements [see pages 93-94 of MedPAC June Report 2016 for examples]. Otherwise, providers in different settings would be paid the same for treating similar patients but would face different regulatory requirements, with potentially different regulatory costs. Changes to requirements would also give providers more flexibility in furnishing PAC.



from Health Affairs BlogHealth Affairs Blog http://ift.tt/2dkNJRG

No comments:

Post a Comment