Getting Savings from Decreasing Variation: Not as Easy as it Might Seem

Today’s Managing Health Care Costs Indicator is 45%

Click on image to enlarge.  Source 

The Dartmouth Atlas  project has been identifying huge unexplained unwarranted variation in health care for decades – and variation as a cause of health care inflation became part of the broader health policy discussion with Atul Gawande’s Cost Conundrum published in the New Yorker in 2009.
Decreasing variation is attractive as a source of medical savings because virtually all providers deeply believe that it’s others who are overutilizing.   However, there are huge problems in harvesting the potential savings from decreasing unwarranted variation.   The Institute of Medicine  has shown that much of the variation among different geographies is simply based on different Medicare fee schedules.    That’s not all – the Dartmouth Atlas database uses Medicare only, and Cooper has shown that there is a reverse relationship between Medicare and private payment rates, which offsets some of this variation). The Dartmouth Atlas also historically has not done any type of risk adjustment.

This month’s Health Affairs has another cautionary note. Makarov and colleagues show that in high utilization regions, low-risk patients with prostate cancer were more likely to get inappropriate imaging, but at the same time high risk patients  were  substantially more likely to get appropriate imaging.  For example, men with low-risk prostate  cancer in New Jersey were 4 times more likely to get inappropriate imaging compared to men  with low-risk prostate cancer in San Francisco.  However, men with high-risk prostate cancer in New Jersey were also three times more likely to get appropriate imaging compared to those in similar circumstances in San Francisco.


This concept, termed the “thermostat model,” posits that each region has a certain proclivity toward resource use that affects both appropriate and inappropriate use. Accordingly, although use varies by region, lower-use regions might not necessarily concentrate their resources on appropriate care.

The likelihood that patients got imaging that was concordant with evidence-based guidelines was disappointing.

More than 45 percent of men with low-risk prostate cancer underwent guideline-inappropriate imaging, and only two-thirds of men with high-risk disease had appropriate imaging evaluations.

 It’s an elegant study, using the SEER (Surveillance, Epidemiology and End Results) Medicare database. This means that the number in the study is large, 29,053, and there are not likely variations in coverage or plan design that could explain these differences.  The authors segment the patients based on National Comprehensive Cancer Network guidelines that were in place during the time of the data collection.
Clearly, variation plays a role in elevated health care costs in high cost regions of the country.  However,  geographies with low utilization have heterogeneous providers – some of whom use resources wisely, and others who simply give too little care.   Further, regions that tend to do a lot of imaging (or surgical procedures, or almost anything else) have more capacity- and use this capacity for both evidence based medicine (decreased undercare) and for unnecessary services (increased overutilization).

We need to be careful not to promote undercare as we push to decrease variation.