A study has shown that payments for quality of outcomes in hospitals do not actually improve those outcomes:

Medicare’s quality incentive program for hospitals, which provides bonuses and penalties based on performance, has not led to demonstrated improvements in its first three years, according to a federal report released Thursday.

The Government Accountability Office examined the Hospital Value-Based Purchasing Program, one of the federal health law’s initiatives to tie payment to quality of care. Earlier this year Medicare gave bonuses to 1,700 hospitals and reduced payments to 1,360 hospitals based on their mortality rates, patient reviews, degree of improvement and other measurements.

While the payments to a majority of the nation’s hospitals have been affected each year, the audit found the financial effect has been minimal. Most hospitals saw their Medicare payments increase or drop by less than half a percentage point. In the fiscal year that ended Sept. 30, 74 percent of hospitals fell within that range, with a median bonus of $39,000 and a median penalty of $56,000, according to the analysis.

………

The report said that even before the program began in October 2012, hospitals had been improving in how consistently they followed basic clinical guidelines, such as performing blood cultures before giving patients antibiotics. That improvement continued but did not increase with the advent of the financial incentives. The same was true for patient ratings, on such items as the quality of communication from doctors and nurses, and for mortality rates for heart attack patients. Heart failure and pneumonia death rates stayed roughly the same.

“Our analysis found no apparent shift in … quality measure trends during the initial years of the program, but such shifts could emerge over time as the program implements planned changes,” the GAO wrote.

This is unsurprising.

As these payments become more significant, I expect to see hospitals use the same tactics as the charter schools: Find a way to encourage hard cases to go elsewhere.

These sorts of pay for performance schemes tend to generate efforts to game the metrics, not to improve performance.