Pay-for-performance
Pay for Performance Could Hurt Docs Who Serve Poor, Blacks and Hispanics
May 06, 2010
by Brendon Nafziger, DOTmed News Associate Editor
Popular pay-for-performance schemes that reward physicians for how well their patients comply with preventive medicine programs could hurt the pocketbooks of doctors serving neighborhoods with greater numbers of the poor, blacks, Hispanics and foreigners, according to a RAND Corporation study published in Health Affairs on Tuesday.
A simulation run by RAND based on data from Massachusetts suggests a typical practice serving catchments with higher numbers of the poor and ethnic minorities could earn up to $7,000 less a year under a competitive rewards scheme, possibly further driving away qualified doctors from such neighborhoods.
"These results substantiate the concern that the distribution of performance-based payments may widen sociodemographic disparities in health care resources, potentially exacerbating existing disparities in the quality of care," write the study's authors.
For the study, the researchers at the Santa Monica, Calif.-based think tank checked census data from Massachusetts to determine the socioeconomic makeup of neighborhoods. They also gathered health claims from 2007 on procedures submitted to health plans from close to 500 small, multi-person practices in the state.
They then ran a simulation partly based on a Medicare incentive program, called the Medicare Care Management Performance demonstration, to see what payouts the doctors would get under a competitive rewards scheme that ranks physician practices and pays them for how many of their patients undergo certain routine preventive screenings for cancer or, if they have diabetes, get appropriate tests.
They found that practices in communities with higher percentages of blacks, Hispanics and the poor, or which were classified as community health centers, could earn dramatically less through such a pay-for-performance program. The biggest disparity was seen between those practices that served populations with higher numbers of poor people versus those in richer areas, with doctors in communities that had twice the percentage of unemployed or people living on public assistance getting a median $2,600 per practice through such a program. Doctors' offices in richer areas could get a median additional $9,700 per practice through the rewards schemes.
COMPETITIVE SCHEMES EXAGGERATE DIFFERENCES
The differences in payment were in many cases larger than the differences in quality measured by the payment scheme, which tended to be modest. Only for compliance on colorectal screening, breast and cervical cancer exams, and eye exams for diabetes were there really significant differences between the groups, and they were often small, with compliance on colorectal screening showing the biggest gap. Performance scores were around nine percentage points lower among practices that serve areas with more blacks, Hispanics and non-U.S. citizens.
"The pay-for-performance specifications blew up those actual performance differences into larger dollar differences than you might expect," Mark Friedberg, associate natural scientist at RAND and lead author of the study, told DOTmed News. That's because "even if all providers are within a four percentage point difference, there's still going to be a ranking of one to a hundred" in a competitive payment scheme, he said.
The simulation couldn't look at why these differences came about, although Friedberg speculates that doctors in poorer communities might have fewer resources to devote to efforts like reminding patients to get yearly exams, and that patients in those areas might have a tougher time finding a high-quality provider.
And while poorer patients are often unhealthier, the study investigated process measures, not health outcomes, so Friedberg isn't convinced that patient demographics were the culprit.
"I think you can make a more justifiable case that patient characteristics cause differences for outcomes, but it's hard to make that case with processes of care," he said.
But there were some limitations. The study only investigated patients on health plans, as the RAND researchers didn't have access to data for uninsured patients or those on Medicaid or Medicare, all of whom could be vastly poorer than the patients included in the study and have even worse access to good providers, resulting in more dramatic disparities than those presented here.
"We're looking at a really truncated spectrum of the sociodemographic distribution," acknowledged Friedberg. "If you were to include patients on Medicaid in a pay-for-performance system that also includes commercial patients, you might see bigger effects than what we saw."
Friedberg pointed out that similar results were seen in the UK, where a pay-for-performance scheme was implemented through the National Health Service. Doctors practicing in neighborhoods with ethnic minorities suffered during the first year of the program, earning much smaller incentive payments, Friedberg said. Interestingly, these differences all but vanished after that first year of the project, according to Friedberg, possibly because unlike the plan modeled in the current study, the British program was noncompetitive -- British doctors were simply rewarded for meeting certain performance thresholds, and not according to their performance-based rank in the system.
SUGGESTIONS PUT FORTH
While Friedberg believes matching payments to performance could encourage better patient care, he wants to see it balanced by programs that make it still attractive for good doctors to serve vulnerable populations. He suggests offering grants to physicians in poor areas to make up for the difference in the pay-for-performance scheme, or rewarding practices for how much they improve in performance each year, in addition to achieving high scores.
But the key, he thinks, is to test before you try.
"No matter what pay-for-performance scheme gets rolled out, it's possible to simulate it. That's what we think is a really important step for any payer to take, to simulate what happens to the distribution of dollars, and see if there is this potential from steering dollars away from the care of underserved populations. That's something I don't think any payer wants to do," Friedberg said.
A simulation run by RAND based on data from Massachusetts suggests a typical practice serving catchments with higher numbers of the poor and ethnic minorities could earn up to $7,000 less a year under a competitive rewards scheme, possibly further driving away qualified doctors from such neighborhoods.
"These results substantiate the concern that the distribution of performance-based payments may widen sociodemographic disparities in health care resources, potentially exacerbating existing disparities in the quality of care," write the study's authors.
For the study, the researchers at the Santa Monica, Calif.-based think tank checked census data from Massachusetts to determine the socioeconomic makeup of neighborhoods. They also gathered health claims from 2007 on procedures submitted to health plans from close to 500 small, multi-person practices in the state.
They then ran a simulation partly based on a Medicare incentive program, called the Medicare Care Management Performance demonstration, to see what payouts the doctors would get under a competitive rewards scheme that ranks physician practices and pays them for how many of their patients undergo certain routine preventive screenings for cancer or, if they have diabetes, get appropriate tests.
They found that practices in communities with higher percentages of blacks, Hispanics and the poor, or which were classified as community health centers, could earn dramatically less through such a pay-for-performance program. The biggest disparity was seen between those practices that served populations with higher numbers of poor people versus those in richer areas, with doctors in communities that had twice the percentage of unemployed or people living on public assistance getting a median $2,600 per practice through such a program. Doctors' offices in richer areas could get a median additional $9,700 per practice through the rewards schemes.
COMPETITIVE SCHEMES EXAGGERATE DIFFERENCES
The differences in payment were in many cases larger than the differences in quality measured by the payment scheme, which tended to be modest. Only for compliance on colorectal screening, breast and cervical cancer exams, and eye exams for diabetes were there really significant differences between the groups, and they were often small, with compliance on colorectal screening showing the biggest gap. Performance scores were around nine percentage points lower among practices that serve areas with more blacks, Hispanics and non-U.S. citizens.
"The pay-for-performance specifications blew up those actual performance differences into larger dollar differences than you might expect," Mark Friedberg, associate natural scientist at RAND and lead author of the study, told DOTmed News. That's because "even if all providers are within a four percentage point difference, there's still going to be a ranking of one to a hundred" in a competitive payment scheme, he said.
The simulation couldn't look at why these differences came about, although Friedberg speculates that doctors in poorer communities might have fewer resources to devote to efforts like reminding patients to get yearly exams, and that patients in those areas might have a tougher time finding a high-quality provider.
And while poorer patients are often unhealthier, the study investigated process measures, not health outcomes, so Friedberg isn't convinced that patient demographics were the culprit.
"I think you can make a more justifiable case that patient characteristics cause differences for outcomes, but it's hard to make that case with processes of care," he said.
But there were some limitations. The study only investigated patients on health plans, as the RAND researchers didn't have access to data for uninsured patients or those on Medicaid or Medicare, all of whom could be vastly poorer than the patients included in the study and have even worse access to good providers, resulting in more dramatic disparities than those presented here.
"We're looking at a really truncated spectrum of the sociodemographic distribution," acknowledged Friedberg. "If you were to include patients on Medicaid in a pay-for-performance system that also includes commercial patients, you might see bigger effects than what we saw."
Friedberg pointed out that similar results were seen in the UK, where a pay-for-performance scheme was implemented through the National Health Service. Doctors practicing in neighborhoods with ethnic minorities suffered during the first year of the program, earning much smaller incentive payments, Friedberg said. Interestingly, these differences all but vanished after that first year of the project, according to Friedberg, possibly because unlike the plan modeled in the current study, the British program was noncompetitive -- British doctors were simply rewarded for meeting certain performance thresholds, and not according to their performance-based rank in the system.
SUGGESTIONS PUT FORTH
While Friedberg believes matching payments to performance could encourage better patient care, he wants to see it balanced by programs that make it still attractive for good doctors to serve vulnerable populations. He suggests offering grants to physicians in poor areas to make up for the difference in the pay-for-performance scheme, or rewarding practices for how much they improve in performance each year, in addition to achieving high scores.
But the key, he thinks, is to test before you try.
"No matter what pay-for-performance scheme gets rolled out, it's possible to simulate it. That's what we think is a really important step for any payer to take, to simulate what happens to the distribution of dollars, and see if there is this potential from steering dollars away from the care of underserved populations. That's something I don't think any payer wants to do," Friedberg said.