Keeping people out of the hospital is a primary goal of Medicare accountable care. For the most financially successful ACOs, however, that proved difficult last year, even with money on the line.
That weak performance, policy experts say, raises questions about the accuracy of quality measures in the rapidly expanding program. Proponents of ACOs tout the measures as safeguards for patients from providers who may withhold necessary care to curb spending.
In August, Medicare announced awards for ACOs, including $273 million for 65 Medicare accountable care organizations in their second year under the Affordable Care Act program. The awards are based on how much money ACOs save for Medicare patients under their care. Most ACOs keep half of what they save. But starting in the second year, awards are also adjusted based on the quality of care.
Medicare reduced the financial awards for 65 ACOs by $41 million based on quality performance, according to an analysis by the consulting firm Leavitt Partners. Medicare ties ACO financial performance to zero measures in year one; 19 measures in year two and all 33 measures in the final year.
The biggest contributor to a drop of quality ratings was a measure of avoidable initial hospitalizations for patients with congestive heart failure, which accounted for $5.6 million, or about 13% of the total reduction to ACO awards. Another measure of preventable initial hospitalizations—for patients with chronic obstructive pulmonary disease or asthma—dragged down awards by another $3.6 million.
The top financial performers did so poorly that none ranked in the top 90th percentile nationally, said David Muhlestein, senior director of research and development for consulting firm Leavitt Partners. That was true also for the other roughly 260 ACOs that reported performance on the measure for 2014. With such a large group, more variation in performance would be expected, he said, raising questions about why ACOs uniformly did so poorly.
ACOs must rank in the top 90th percentile to receive full credit for a measure, which is indicated by 2 or 4 points. Points decrease as ACOs performance drops below the 90th percentile. Those that score in the 30th percentile or below get no points. Points are used to calculate an overall score, which Medicare uses to adjust financial awards.
Among ACOs operated jointly by publicly traded insurer Universal American and physicians' groups across multiple states, five of the nine ACOs to earn financial awards in 2014 performed in the bottom 40th percentile or lower for hospital admissions for conditions that may be managed outside the hospital, in this case among heart failure patients.
Jeffery Spight, who oversees Universal American's ACOs, said the performance largely reflected the challenges ACOs faced collecting data and documenting heart failure patients appropriately. The two dozen ACOs under Universal American's joint ventures operate across 340 electronic medical records and all but one are made up entirely of physician practices that range from solo operators to multispecialty groups, he said. That presents significant obstacles gathering necessary data to meet Medicare reporting standards.
The Accountable Care Coalition of Maryland, one of Universal American's ACOs, received no points for prevention of heart failure admissions with performance below the 30th percentile. The ACO saved Medicare roughly $4 million, of which the ACO received $1.4 million after adjustments for quality.
Measures of hospital use may also be significantly affected by differences across ACO patients, said Dr. J. Michael McWilliams, an associate professor of health policy at Harvard University who studies the issue.
The CMS makes some effort to adjust performance to reflect the fact that some hospitals and doctors care for sicker patients than others, but that adjustment is limited, he said.
McWilliams and colleagues reported in JAMA Internal Medicine in September that nearly two dozen social, clinical, financial and demographic characteristics “significantly predicted” hospital readmissions. The characteristics—social support, functional status, income, race and ethnicity, education—are not included in the CMS' formulas to adjust for readmission rates.
Hospitals with high readmission rates were more likely to have patients with characteristics linked to readmissions, for which Medicare now penalizes hospitals based on performance. “The penalties that are levied are due to the patient they serve and not the quality of care they provide,” McWilliams said.
The failure to adjust measures of hospital use appropriately might discourage providers from joining ACOs when financial performance suffers as a result, he said. ACOs already offer relatively weak financial incentives and penalties for poor quality performance weaken it further, he said.
McWilliams said incentives could be designed differently to achieve federal goals to protect patients from stinting and improve quality.
He questioned measures of use, such as admissions, as markers of quality when it's unclear how well that measure works. “The objective of the healthcare system is not to prevent patients from ever going to the hospital,” he said. "Hospitalization can be quite beneficial.”
Instead of using quality measures to monitor and protect against doctors who might withhold needed care for financial gain, the CMS could instead track indicators of stinting separately, he said. That would deter providers from scrimping on care to meet financial targets. Quality measures could then target care for which providers may see little financial benefit, such as preventive care that may not save money for years.
Meanwhile, quality measures should be designed to track care that may not deliver an immediate return on investment. Quality measures could also track historical performance rather than national benchmarks, he said. That might better capture ACO performance over time with less risk of comparisons across diverse patient groups across the country.