Geographic variations in imaging use do not necessarily mean that these services are being overused, according to a new study published in the Annals of Internal Medicine. Rather, other factors, such as how care is structured in a system and how physicians are paid, may come into play.
The study findings address an issue policymakers have been raising for years: Regional variation in imaging costs is part of an overall increase in healthcare spending, they say, and curbing geographic variations in healthcare service delivery should be a key effort. However, using geographical patterns as a guide -- rather than evaluating the whole picture of how health services are provided across the U.S. -- may not actually make healthcare more efficient.
"While geographic comparisons can be useful for understanding trends in provider behavior, our study demonstrates that geographic variation is not necessarily a reliable indicator of the extent of overuse in a healthcare system," lead author Dr. J. Michael McWilliams, PhD, of Harvard Medical School, told AuntMinnie.com.
To investigate the issue, McWilliams and colleagues chose to study cancer-related imaging, comparing data from the Veterans Affairs (VA) healthcare system with traditional fee-for-service Medicare. Why? In part because over the past 20 years, the VA system has emphasized payment and delivery paradigms that Medicare is currently encouraging to bring healthcare costs to heel, such as integrated care, quality measures and performance incentives, and accountability (Ann Intern Med, December 1, 2014).
"We felt cancer-related imaging would be a useful service for this study, since it's often overused, but also because it tends to be segregated between the VA system and Medicare," McWilliams said.
The researchers used Medicare claims from 2003 to 2005, as well as VA use data for the same time period for men over 65 with primary diagnoses of lung, colorectal, or prostate cancer. Study data covered 40 hospital referral regions across 23 states, which represented 22% of the Medicare population in 2005. The team explored whether the use of cancer-related imaging was lower for VA patients than Medicare beneficiaries, and if it was, whether this lower use was associated with less geographic variation.
The VA study group included 6,835 men, while the Medicare cohort included 34,475 men.
McWilliams and colleagues found that adjusted annual use of cancer-related imaging was lower in the VA cohort than in the Medicare cohort, which they quantified as mean price-weighted utilization of $197 versus $379 per patient, respectively. Lower use of advanced imaging such as CT, PET, and nuclear studies in the VA cohort made up 90% of this difference; not surprisingly, x-ray use was higher in the VA cohort than in the Medicare cohort.
|Adjusted mean use ($ per patient)|
|Study population||VA cohort||Medicare cohort|
|Lung cancer patients||$386||$727|
|Colorectal cancer patients||$323||$396|
|Prostate cancer patients||$102||$240|
Variation across the 40 hospital referral regions in cancer-related imaging use was similar in the VA and Medicare cohorts, at a standard deviation of $78 per patient in the VA group, compared with $60 per patient in the Medicare group.
There are many reasons why the VA system could be associated with lower average use of healthcare services but not geographic variation, including its service delivery structure, the use of global budgets to control spending, and salaried physicians, the researchers wrote.
"VA medical centers operating under budgets may prioritize services differently, leading to heterogeneity in capacity to provide a given type of service," they wrote.
But even if geographic variation isn't the best way to measure healthcare service overuse, the study does suggest there may be waste in the healthcare system, specifically in imaging, that needs to be addressed, McWilliams said.
"The fact that imaging use was almost half as much in the VA system as compared to Medicare suggests that there is room for improvement in cutting healthcare waste," he told AuntMinnie.com. "But the geographic variation paradigm may not be the best way to quantify this."