Maryland’s all-inclusive population health payment model continues to show promise, but nonhospital spending poses a concern

An innovative payment model for Maryland healthcare providers has improved utilization, cost and quality thus far, but an increase in nonhospital spending requires further study.

The Center for Medicare & Medicaid Innovation (CMMI) released an analysis of the first three years of Maryland’s Total Cost of Care Model, an effort to improve population health management. Features of the eight-year model, which sets a per capita limit on Medicare costs, include:

• Hospital global budgets across payers
• Incentives to hospitals to reduce total cost of care
• Incentives to improve the efficiency and quality of episodes spanning hospital and nonhospital care
• Waivers allowing hospitals to align incentives with other providers
• Primary care practice transformation
• Funding to hospitals and partners to reduce incidence of diabetes and overdose deaths

From the 2019 launch through 2021, researchers calculated that the model reduced all-cause acute care hospital admissions by 16.1%. Medicare fee-for-service spending dropped by 2.5% overall and by 6.6% among hospitals as a result of the model (the researchers sought to control for the impact of COVID-19 in their calculations, which drew on a group of 10 comparison states, but they note 2020-21 results should be interpreted “with caution”).

The model also reduced outpatient emergency department visits by 3.8% during the three-year period, according to the report, which was produced for CMMI by Mathematica, a research and data analytics consultancy.

A closer look at spending impacts

Standardized hospital spending, an aggregate measure of the intensity of hospital care, declined by $414 per beneficiary per year (PBPY) in 2019-20 as a result of the model (2021 data was not available for that metric).

The model lowered Medicare spending by $348 PBPY, or $781 million overall, during the three-year period relative to the comparison group of states. The driving factor behind the savings was a $510 PBPY reduction in hospital spending. The decrease slowed in 2020-21, compared with 2019, “which might be partly attributable to how global budgets operated during the COVID-19 pandemic.”

Nonhospital spending increased during the three-year period by 2.7%, or $162 PBPY, relative to the comparison group, which largely was to be expected since global budgets give hospitals incentives to shift care to lower-acuity settings.

One concern was that in 2021, the PBPY increase in nonhospital spending surged to $345 relative to the comparison group.

The finding “illustrates a potential risk in the model, and it will be important to examine this outcome in future years,” the report states.

What’s clear is that while global budgets can be used to control hospital spending, no practical levers have been found for non-hospital spending. Incentives can be implemented to limit growth, but those only go so far.

The increase in nonhospital spending can’t be attributed to post-acute care spending, which decreased as a result of the model. But that decrease was significantly smaller in 2021, which could have contributed to that year’s comparative spike in nonhospital spending.

The overall savings to Medicare, relative to the comparison states, was not lessened after factoring in non-claims-based payments such as those for accountable care organizations, primary care programs and physicians participating in advanced alternative payment models.

The influence on quality measures

The model reduced potentially preventable admissions by 7 per 1,000 beneficiaries, or 16.1%, relative to the comparison states. In addition, 30-day unplanned readmissions were 9.5% lower. And the probability of follow-up after acute exacerbation of chronic conditions improved by 2.5%.

Areas not affected by the model included patients’ ratings of their physician and hospital, based on CAHPS surveys in 2019 for physicians and 2019-20 for hospitals.

Utilization of the Medicare Diabetes Prevention Program decreased slightly in Maryland compared with the control group of states. It was hard to derive much insight from that finding because the number of people using the program across states was small.

An evolving effort

Maryland has a long history of trying new payment models and approaches, the researchers noted.

The state began regulating hospital prices for all payers in the 1970s, with CMS offering a waiver that allowed Medicare prices to be included in the rate-setting as long as Medicare payments per hospital stay did not exceed the national average.

“Maryland met this requirement for decades, but hospitals compensated for low price growth by increasing volume, which raised total hospital spending,” the researchers wrote.

In 2014, a predecessor to the TCOC Model was launched. That precursor model, the Maryland All-Payer Model (APM), sought to limit growth in per capita hospital spending and improve hospital quality-focused measures. All-payer global hospital budgets were the primary mechanism.

The TCOC Model builds on the APM in several ways, including by seeking to limit total Medicare FFS spending rather than focusing on hospital spending. Other programs within the model include:

• The Care Redesign Program, allowing hospitals to make incentive payments to nonhospital providers in partnerships to redesign care and improve quality
• The Maryland Primary Care Program, offering incentives for primary care practices and federally qualified health centers to provide advanced primary care services

There’s also more of a focus in the new model on improving overall health via commitments such as increasing follow-up after discharge, decreasing mean body mass index among state residents, and reducing deaths from drug overdose.

The researchers said a noteworthy aspect of the gains seen in the TCOC Model results is that they generally are larger and more favorable than those seen toward the end of the APM. It’s unclear whether the progress stems from the growing impact of global hospital budgets, the additional components of the new program, or a combination.