Variation matters

One of the most frequently trumpeted “truths” in healthcare policy research is that the system is larded with waste, perhaps as much as 30% of all spending. The Dartmouth Institute for Health Policy & Clinical Practice, now led by Dr. Elliott Fisher, initially developed that insight after finding wide disparities in Medicare spending in different parts of the country. Its Dartmouth Atlas of Health has repeatedly found that variation in provider practice patterns is driving those spending differences, not the level of illness among patients.

McAllen, Texas, became the poster child for overutilization after Dr. Atul Gawande painted an unflattering portrait of the city's physician and hospital practices in a New Yorker magazine profile. The idea eventually received the official imprimatur of the Institute of Medicine, which agreed that overall spending could be slashed by $690 billion a year if every healthcare system delivered care with the efficiency of the lowest-cost providers.

But last week a research team associated with the Washington-based Center for Studying Health System Change launched a frontal assault on the Dartmouth Atlas thesis. Their study, published in Medicare Care Research and Review, found that overall population health, that is, how sick the people were in different regions of the country or even within a region, accounted for as much as 75% to 85% of the difference in spending between high-cost and low-cost areas.

To drive that point home, they looked at diagnoses where physicians had very little discretion in how they treated patients: hip fractures, traumatic amputation, heart attacks and strokes, for instance. They found the high-cost areas had incidences of those acute events that were 73%, 91%, 84% and 74% higher than low-cost areas, respectively.

They took their analysis one step farther by looking at end-of-life care, which the Dartmouth researchers have repeatedly identified as a major driver of higher costs in some areas. No, the study argued. People in high-cost areas tended to have more chronic conditions as they neared the end of life compared to people in low-cost areas, which meant they were being treated for more illnesses and hence were more costly.

The study downplayed the role of income and poverty in driving health status and hence spending. But other skeptics have repeatedly made that point in challenging the validity of the Dartmouth analysis. Want to know where high-spending areas are located? All you have to do is look at the census tracts where the poor and ill-educated are concentrated. They are the ones who enter their Medicare years with multiple chronic conditions, and hence wind up costing a lot more as they move toward end-of-life care.

The Dartmouth researchers fired back, of course, claiming the research couldn't be replicated. But they make a big mistake by not engaging their critics.

Everyone understands there are huge variations in practice patterns. But those differences are just as likely to exist between doctors within a hospital or a physician practice as between regions of the country. Even the researchers behind this latest study admit variations in practice patterns drive costs higher. But it is nowhere near the level that justifies claims of 30% of healthcare spending being wasted.

There's an important lesson in this debate for healthcare providers. What makes those practice variations important is not only the waste they represent, which is probably much less than the ballyhooed 30%, but the likelihood that much of that unnecessary utilization is leading to lower quality care. By eliminating unnecessary variation, the system will save some money, of course. But more importantly, it will lead to better outcomes, which ultimately matter more.

Follow Merrill Goozner on Twitter: @MHgoozner