Does Hospital Size Impact Value-Based Penalties in CMS Program?

December 22, 2016 - Value-based penalties in the Medicare Hospital-Acquired Condition Reduction Program are disproportionately affected by a participating hospital’s bed size and number of cases, a recent American Journal of Medical Quality study indicated.

The study, funded by the American Hospital Association, showed that large hospitals are more likely to be identified as poor performers on quality measures with low probabilities of complication, increasing a large hospital’s likelihood of receiving value-based penalties.

In contrast, small hospitals tended to score worse on quality measures with higher probabilities of complication.

The study suggested that the program unfairly penalizes hospitals on either end of the size spectrum because of the penalty methodology.

“Even when hospitals have identical underlying quality, the simulations predict that certain providers are much more likely to show poor performance on specific measures,” wrote study authors. “The direction of the bias depends on the expected complication rate and the distribution of eligible cases across hospitals.”

Under the Hospital-Acquired Condition Reduction Program, hospitals in the lowest performing quartile are subject to a 1 percent Medicare reimbursement penalty based on their quality performance. In 2016, CMS reduced Medicare reimbursement to 758 low-performing hospitals in the program by about $364 million in total.

Value-based penalties in the program are based on a hospital’s performance on four measures, including the Agency for Healthcare Research and Quality Patient Safety Indicators (PSI) 90 Composite and three measures from the Centers for Disease Control and Prevention (CDC) National Healthcare Safety Network (NHSN).

CMS determines which low-performing hospitals receive a value-based penalty by using thresholds. Hospitals that fall below the quality performance threshold for the bottom quartile receive penalties, while all others do not.

Researchers noted that the threshold methodology may be flawed because the performance differences just above and below the penalty threshold are usually small and not statistically different. Hospitals may also improve performance, but remain below the penalty threshold if other hospitals improve, too.

When researchers simulated a hospital’s likelihood of receiving value-based penalties, they found that the number of bed and cases in a hospital influenced quality performance.

Larger hospitals were disproportionately placed in the value-based penalty quartile on quality measures with very low complication rates because of more cases. For example, the largest size hospital (1,000 cases) was in the lowest quartile in 40 percent of the simulations when the expected complication rate was 0.001.

The largest size hospital was also in the lowest quartile in 19 percent of simulations when the expected complication rate was 0.01.

The reverse was also true, added researchers. Small case size hospitals tended to perform worse on quality measures with higher expected complication rates.

In addition, number of beds influenced a hospital’s performance on some PSI measures in the Medicare program. Researchers found that the mean simulated chances of being in the value-based penalty quartile increased as bed size grew.

For example, the simulated likelihood of being in the lowest quartile for PSI 03 (Pressure Ulcer) ranged from 1 percent for the smallest ventile of bed size to 56 percent of the largest bed size ventile.

With other PSI measures, however, the trend reversed. For instance, the simulated chance of being in the lowest quartile for PSI 15 (Accidental Puncture or Laceration) was highest among hospitals in the second smallest ventile of bed size and decreased to 11 percent for the largest bed size ventile.

Further, some PSI measures followed a concave down trend, with simulated likelihood rising at first, then decreasing as bed size ventile increased.

As a result, researchers stated some PSI measures, such as PSI 03, PSI 08, and PSI 14, have significant discriminatory power between poor-performing and non-poor-performing hospitals.

For other measures, such as PSI 06, PSI 07, PSI 12, PSI 13, and PSI 15, the discriminately power is less.

“The present study’s results suggest that when hospital quality is gauged using measures based on infrequent events and a performance threshold, variations across hospitals in numbers of cases can bias the results,” wrote study authors. “Even when hospitals have identical underlying quality, the simulations predict that certain providers are much more likely to show poor performance on specific measures.”

To resolve value-based penalty bias in the Hospital-Acquired Condition Reduction Program and other similar quality initiatives, researchers suggested that CMS incorporate some hospital characteristics, such as size, in their risk adjustment models.

CMS could also determine value-based penalties by comparing hospitals of similar size, recommended researchers.

Another improvement to Medicare quality initiatives would be to include a broader “all-harm” measure in assessments. An “all-harm” measure’s higher counts of eligible cases and expected complication rates could offset the bias. A broader measure would also “reduce the importance of random variation and increase the importance of true quality differences in assessing performance.”

Other healthcare stakeholders have also argued that adding socioeconomic adjustments to Medicare quality initiative methodologies would resolve bias. In a September Health Affairs study, researchers found that safety-net hospitals are disproportionately penalized in the Hospital Readmissions Reduction Program because value-based penalties are determined based on national average rate comparisons.

Safety-net hospitals, researchers argued, faced unique challenges compared to other hospitals because their readmission rates were impacted by factors beyond the hospital’s control, such as patient homelessness and lack of family support.

Instead, CMS should separate hospitals in the program into deciles based on the proportion of low-income patients they treat. Then, the federal agency should compare hospitals within each decile to their peers to determine penalties.

While the American Journal of Medical Quality study researchers favored hospital stratification based on size, they also supported more general flexibility with value-based penalty determinations.

“Ultimately, Congress may need to provide CMS with additional flexibility in how it can assess which hospitals to penalize in HAC-RP [Hospital-Acquired Condition Reduction Program],” wrote study authors.

“Whether adapting CMS approaches or using their own, payers should use caution when either constructing value-based purchasing programs or assessing hospital quality so as to not introduce systematic bias by using thresholds of performance when counts of eligible cases vary significantly across providers,” they concluded.

Dig Deeper:

• Are Federal Value-Based Care Programs Truly Promoting Value?

• What Is Value-Based Care, What It Means for Providers?