The puzzle of why there appears to be such variation in health spending for the same type patients in different US geographies is a long-standing topic in health research. The variation was first noted among Medicare patients but occurs in commercial insurance patients as well. Some of the variation is price but most is utilization–different practice patterns. Even in the same geography this kind of variation can occur. A new paper takes a fresh look, particularly at whether more spending equals better health.
Much of the most recent research has focussed on Medicare beneficiaries who move from one geography to another and whether their spending then increases or decreases in accordance with the average in their new location. The research has found that it generally does and that location factors account for about 50% to 60% of spending variation, with the rest being due to personal characteristics of the patient or other factors. The research also has not noted an increase in various quality outcomes. In examining a potential link to patients’ health, the authors solely use life expectancy as a measure, which seems extremely inadequate. By the time a patient is 65 and Medicare eligible, many events have occurred in their lives which affect life expectancy. It is extremely hard to adjust for all of those adequately, especially since many are not health-specific events.
The analysis also only used Medicare fee-for-service members, which given the prevalence of Medicare Advantage membership seems problematic, since those people in each branch may vary in some systematic way. There is a wide variation in spending and life expectancy across geographies. Despite that variation the authors claim to observe that higher spending is associated with lower life expectancy, after adjusting for health factors. There analysis correlates “place” effects related to health spending levels and “place” effects related to life expectancy, to attempt to avoid confounders. While I tend to think that most good research shows that higher spending on the same patient is driven by provider preferences which vary by geography, I am less convinced that all that spending could be eliminated somehow, and not have an impact on health.
Whatever the reason for high health spending variation, all the efforts to address it and have a more national set of practice patterns have obviously done nothing to reduce total spending and it also is unclear that reducing high spending would improve quality as measured by people’s health status, or even keep that health status the same. (NBER Paper)
At Express Scripts we had a subsidiary called Practice Patterns Science that, among other things, mined pharmacy claims data to identify opportunities for cost savings. The geographical variation in prescribing patterns was dramatic. We adjusted the data for severity of condition and various demographic factors, and excluded diseases the severity of which might have been affected by local environmental conditions (e.g., asthma). We could find no objective explanation for the extreme variability in prescribing patterns among regions. We concluded that the variation was probably the result of physicians in each area talking informally to their peers about how they were prescribing which drugs for which conditions. In other words, prescribing patterns were strongly affected by local medical lore with little or no clinical validity.