In response to criticism that their statistical reports on Medicare expenditures failed to take into account regional variations in cost of living expenses, researchers at the Dartmouth Atlas Project of Health Care have released an analysis that adjusts for that variable. The changes in methodology reduced the predominance of outlier locations like New York City and produced a more "fair" comparison of health care utilization across regions, Daniel Gottlieb, co-author and research associate at The Dartmouth Institute for Health Policy and Clinical Practice, said.
The Atlas report, which tracks Medicare expenditures at the "hospital referral region" level for the period between 2003 and 2008, avoids the limitations of previous studies by factoring in price differences between both geographic areas and types of hospitals within a given region, according to Gottlieb.
Researchers have previously conducted comparative studies using numbers collected from the Continuous Medicare History Sample, which is released by the Centers for Medicare and Medicaid Services (CMS), according to the study.
Although the comparisons adjusted for age, gender and race characteristics, they did not include data on density of teaching hospitals, cost of living and disproportionate share hospitals that receive payment for serving low-income patients, Gottlieb said.
"A hospital might get a 25-percent bump because they're a teaching hospital," he said. "They also might get as much as a 25-percent bump because they have [disproportionate share hospital] payments, which help places helping out people."
The changes in methodology had the greatest impact on regions like the New York metropolitan area, which has a large number of training hospitals and low-income patients, as well as relatively high living costs. Medicare reimbursements in Manhattan, for example, fell from almost $13,000 to around $9,000 per beneficiary following the adjustments, according to the report.
While a hospital stay in New York may cost twice that of a stay in Kansas, the adjusted payments are actually similar, according to Gottlieb.
"When we look at it this way, it doesn't look as egregious," he said.
In addition, the June report utilizes data collected from a 20-percent sample of comprehensive claim files, rather than a 5-percent sample of Medicare beneficiaries released by the Continuous Medicare History Sample, according to the study.
The increased sample size means the new study is more accurate, particularly in regions with small populations, Gottlieb said. In addition, the data draw on a random sample of all Medicare beneficiaries, rather than the output of a CMS algorithm.
"The sample that we were using was provided to us directly from CMS," he said. "I've looked at it and I can never get the numbers to line up with what I calculated by myself."
The change in data source will enable researchers both at the Atlas and elsewhere to conduct consistent studies and obtain comparable results, according to Gottlieb.
"When we calculate, what we want to do is compare utilization from one area to another," he said. "That's really the big thing that this report does: It shows what we would think of as closer to pure measures of utilization. The nice thing about this is it's reproducible, by us and by others."
The changes to Atlas researchers' methodology were prompted by criticism that Medicare spending depends heavily on factors missing from previous reports, HealthLeaders Media reported.
When considering changes to health care spending in 2010, officials from President Barack Obama's administration referred to earlier studies produced by Atlas researchers, according to The New York Times. However, the reports assume that "cheaper care is better care," while ignoring patient health and differences in prices, The Times reported.
"Nurses in Houston tend to be paid more than those in North Dakota because the cost of living is higher in Houston," The Times reported in June 2010. "Neither patients' health nor differences in prices are fully considered by the Dartmouth Atlas."
The largest issue remaining with the means of calculation is the need for risk adjustment, according to Gottlieb.
"Is it fair to only adjust for age and gender and perhaps race?" Gottlieb said. "Shouldn't you also adjust for the illness of the population? But that's a harder nut to crack."
Illness is quantified regionally using diagnoses coded by doctors with each patient visit, he said. As a result, "it looks like people in Miami are much sicker than in places like Minnesota, where their visit rates are lower," he said.
More reasonable alternatives may include the utilization of mortality rates, surveys or measures of health like blood pressure and cholesterol among populations, Gottlieb said.
The Atlas report, published on June 21, was written by Gottlieb, economics professor Jonathan Skinner and Donald Carmichael, a statistical research analyst at The Dartmouth Institute.
