Care Preferences’ Role in Geographic Spending Variation

Just a few weeks ago we reported on an article in Health Affairs that examined care preferences’ role in geographic health spending variation and now comes a National Bureau of Economic Research paper on the same topic.  (NBER Paper)  Because most analyses find a large percent of variation to be unexplained by demographic, illness burden and other factors, physician practice patterns and patient care preferences have come to be strong suspects for the unexplained portion of variation.  The reasons for variation matter, because different causes may be more or less amenable to change.  In this study, the authors used surveys of Medicare beneficiaries about their preferences regarding aggressive care interventions, particularly at the end-of-life, when much spending occurs, and surveys of cardiologists and primary care physicians seeking to understand how they would care for an elderly patient in four different circumstances.  The survey data was then coupled to last two years of life spending data at the level of the hospital referral region.  The patient surveys asked two questions about desire for items of care that the patient’s physician did not think was needed and three questions about specific end-of-life care services.  Physicians were asked about practice and personal characteristics and several clinical vignettes, which related  to seriously ill patients.

Patient preferences were not associated with spending variation.  Physician beliefs or stated actions were.  In particular, areas with higher percentages of doctors who would practice aggressively had higher spending, while those with higher rates of doctors more oriented toward palliative care had lower spending.  This was true in regard to both cardiologists and primary care doctors.  Combining aggressive care believes with more follow-up visits than recommended by guidelines explained over half of the spending differences.  Going a level of causation deeper to understand why physicians have these different approaches to care, the researchers found that demographic, financial reimbursement arrangements, and practice size did not account for much of the difference, although solo and 2 person practices were much more likely to practice aggressively.  The most prominent factor appears to be the physician’s belief, whether justified or not, about the efficacy of the treatment he or she was recommending.

It should be noted that the results are based on self-reported survey data, but the results seem reasonable.  While physicians say it is not financial concerns that drive their intensive care behaviors, that behavior provides a financial reward to the doctors engaging in it and it is difficult to accept that this financial consequence is not a strong motivator.  It also should be observed that this study dealt only with Medicare, and variation may be different in commercial or Medicaid populations and the reasons for whatever variation is observed might be different.  When there are better all-payer databases, practice patterns across payers can be studied to answer these questions.  And as the authors suggest, getting a thorough understanding of how physician beliefs about treatment efficacy arise could help with early interventions.  It is hard not to belief that medical school, residency and mentoring by other physicians are not key in this process.