The Dartmouth Institute of Health Policy & Clinical Practice, a longtime academic advocate of the notion that Medicare inefficiencies drive geographic cost variations, took aim on a recent study that claimed patients' health is the primary driver of Medicare cost differences.
The counter-report was authored by the Center for Studying Health System Change and published in the journal Medical Care Research and Review. Jonathan Skinner, PhD, economics professor at the institute and Dartmouth College's Geisel School of Medicine, wrote a criticism of the CSHSC piece, specifying four fallacies in the article.
1. Diagnoses on billing documents are unreliable. The CSHSC study slammed Dartmouth for not comparing patients' diagnoses in their research to the magnitude of geographic populations' illness. Dr. Skinner countered that by saying physicians in some regions are far more apt to diagnose patients with more severe illnesses on billing documents in order to justify more aggressive and expensive procedures.
"These biases are severe and will lead to highly misleading conclusions," he wrote, including an example in the threshold for diabetes diagnosis has been shown to be lower in some regions, leading to "reverse causation" for treatment.
2. End-of-life expense risk adjustments don't change cost disparity. End-of-life care cost is particularly prone to large variances across the country. However, Dr. Skinner wrote that a study by Mount Sinai Hospital in New York City showed even when adjusting for cost-risk factors, such as wealth, lifestyle and type of disease, the wide range in Medicare expenditures remained unaffected.
3. "Nondiscretionary" incidences don't explain nationwide variances. CSHSC researchers found nondiscretionary measures of disease shielded from physician bias or subjectivity, such as fractures, were higher in higher-cost regions. However, Dr. Skinner said he found major differences in the two studies' data sets.
CSHSC found the highest-cost Medicare patients were 73 percent more likely to suffer a hip fracture, but Dartmouth found only a 13.5 percent increase in likelihood. He wrote that is likely because CSHSC used physician-based data instead of patient-based data, whereas patients may visit multiple physicians, especially in high-cost regions where patients seek specialists more commonly, he said.
As a result, CSHSC's sample pool of physicians includes a disproportionate share of specialists in high-cost regions, who would naturally see sicker, costlier patients. Dr. Skinner acknowledged that CSHSC had noted this in their report and was exploring fixes to this potential bias, but he claimed the sampling protocol is fundamentally flawed.
4. Automatic billing risk adjustments ignore clinical plausibility. CSHSC used hierarchical condition categories from billing data to determine patients' diagnoses and thus level of illness, Dr. Skinner said. HCCs include predetermined risk adjustments, which he believes CSHSC took wholesale when plugging data into their regressions, automatically assuming such patients cost as much as the HCC formulas adjust for. Since Dartmouth researchers contend that HCC billing data is unreliable across regions, the corresponding risk adjustments further the diagnosis-related explanations for cost variances beyond what Dr. Skinner believes is clinically sound in many cases.
Worth of Remote Patient Monitoring Market May Hit $46B by 2017
Growth is the Answer: 3 Ways to Grow Your Hospital Business
The counter-report was authored by the Center for Studying Health System Change and published in the journal Medical Care Research and Review. Jonathan Skinner, PhD, economics professor at the institute and Dartmouth College's Geisel School of Medicine, wrote a criticism of the CSHSC piece, specifying four fallacies in the article.
1. Diagnoses on billing documents are unreliable. The CSHSC study slammed Dartmouth for not comparing patients' diagnoses in their research to the magnitude of geographic populations' illness. Dr. Skinner countered that by saying physicians in some regions are far more apt to diagnose patients with more severe illnesses on billing documents in order to justify more aggressive and expensive procedures.
"These biases are severe and will lead to highly misleading conclusions," he wrote, including an example in the threshold for diabetes diagnosis has been shown to be lower in some regions, leading to "reverse causation" for treatment.
2. End-of-life expense risk adjustments don't change cost disparity. End-of-life care cost is particularly prone to large variances across the country. However, Dr. Skinner wrote that a study by Mount Sinai Hospital in New York City showed even when adjusting for cost-risk factors, such as wealth, lifestyle and type of disease, the wide range in Medicare expenditures remained unaffected.
3. "Nondiscretionary" incidences don't explain nationwide variances. CSHSC researchers found nondiscretionary measures of disease shielded from physician bias or subjectivity, such as fractures, were higher in higher-cost regions. However, Dr. Skinner said he found major differences in the two studies' data sets.
CSHSC found the highest-cost Medicare patients were 73 percent more likely to suffer a hip fracture, but Dartmouth found only a 13.5 percent increase in likelihood. He wrote that is likely because CSHSC used physician-based data instead of patient-based data, whereas patients may visit multiple physicians, especially in high-cost regions where patients seek specialists more commonly, he said.
As a result, CSHSC's sample pool of physicians includes a disproportionate share of specialists in high-cost regions, who would naturally see sicker, costlier patients. Dr. Skinner acknowledged that CSHSC had noted this in their report and was exploring fixes to this potential bias, but he claimed the sampling protocol is fundamentally flawed.
4. Automatic billing risk adjustments ignore clinical plausibility. CSHSC used hierarchical condition categories from billing data to determine patients' diagnoses and thus level of illness, Dr. Skinner said. HCCs include predetermined risk adjustments, which he believes CSHSC took wholesale when plugging data into their regressions, automatically assuming such patients cost as much as the HCC formulas adjust for. Since Dartmouth researchers contend that HCC billing data is unreliable across regions, the corresponding risk adjustments further the diagnosis-related explanations for cost variances beyond what Dr. Skinner believes is clinically sound in many cases.
More Articles on Healthcare Spending:
Medicare Trustees: Hospital Funds Improve, Now Solvent Through 2026Worth of Remote Patient Monitoring Market May Hit $46B by 2017
Growth is the Answer: 3 Ways to Grow Your Hospital Business