Geographic Variation in Breast Cancer Treatment Spending

People are still trying to understand how much, and why, geographic variation in health care treatment patterns exists, and what the effects are on utilization and spending.  Research in Health Services Research uses the case of breast cancer treatment to attempt to further explore the issues.   (HSR Article)   Breast cancer is the most common one for women and accounts for a significant amount of spending.  Because of the emotionality attached to it, breast cancer treatment has often involved low-value, often ineffective therapies, notwithstanding attempts to standardize care protocols.  Variation in treatment exists at a regional as well as local level, as it does for health spending in general.  Researchers continue to attempt to puzzle out how much is due to patient choices and how much to physician practice pattern preferences.  The study used FFS Medicare data from 2009 to 2013, which is unfortunate because private insurance covers a lot of breast cancer treatment and often has different spending patterns, and Medicare Advantage also likely has different treatment patterns.  The primary outcome was total medical spending per patient per month, but didn’t include outpatient pharmacy, which is also unfortunate, as this is a significant portion of spending; with the geographic unit being the usual hospital referral region.  Other outcomes were primarily subsets of care.  Patients were attributed to the medical oncologist or surgeon who delivered most of their care, and a separate attribution was made for the initial phase of treatment and end-of-life care.

The final study cohort included over 20,000 patients.  In the initial year after diagnosis average spending per patient month was $2966 but with a standard deviation of over $4400, so highly variable.   Patient differences appear to explain about 14% of the difference and physician differences, 4%.  In the last year of life, average per month per patient spending was $5768, with a similar very wide range of variation.  Between patient variation accounted for about 11% of the variability and between physician, around 1%.  Patient demographics and disease characteristics explained about 33% of the between patient variability for the first year after diagnosis and around 24% for the last year of life.  Treatment choices were more influential in the first year after diagnosis than in the last year of life.  The demographic and disease characteristics of the patient population treated by a specific doctor accounted for 54% of between doctor variation in the first year after diagnosis and 16% in the last year of life.  Choice of treatment modality explained 31% of this variation in the first year but none in the last year of life.  In the first year after diagnosis, doctors in the top quintile had 70% higher spending than those in the lowest.  In the last year of life, the difference was 230%!!  Trying to understand drivers of greater spending, the researchers found that the highest quintile of physicians by spending had significantly more inpatient days, and with much greater chemotherapy spending.  They also had many more office visits and imaging. Now it could be that these doctors are responding to patient preferences on treatment intensity, but one thing the analysis didn’t look at was practice location.  I strongly suspect that doctors employed by, or whose practices are owned by, hospitals have much higher treatment costs, especially for chemotherapy.  The results also suggest that doctors aren’t functioning as good stewards of health dollars, because unless there is clear evidence of differential outcomes, they should encourage patients to adopt lower-cost treatment options, which lowers out-of-pocket spending.