Maria Polyakova is an assistant professor in the Center for Health Policy at Stanford University School of Medicine. She’s a health economist whose research focuses on public and private health insurance. She is also a member of the 2021 Tradeoffs Research Council.

In addition to a raging pandemic, newly inaugurated President Joe Biden is inheriting a nation where health care spending continues to climb. One commonly cited source contributing to the high spending is intensive, expensive and possibly futile care in the last 12 months of life, which accounts for a quarter of Medicare expenditures.

A recent gloomy, but important, NBER working paper by Dan Zeltzer, Liran Einav, Amy Finkelstein, Tzvi Shir, Salomon Stemmer and Ran Balicer combines an incredibly rich data set and cutting edge analytic tools to try to understand exactly why we spend so much at the end of life.

The authors used detailed claims and electronic medical record data from Israel on more than 160,000 adults who were newly diagnosed with cancer between 2001 and 2013. (Like the U.S., Israel spends a substantial fraction of its health care dollars on end-of-life care.) These adults had an annual mortality rate of 20%, nearly 20 times higher than mortality in the general population. The authors divided those newly diagnosed adults into two groups — those that died within 12 months of their diagnosis and those that did not — and then compared the care received by each within that year. 

The authors found that spending was nearly twice as high among patients who died, when comparing patients with similar initial prognoses. The source of that elevated spending was surprising: The high expenditures among those who died in the year after their cancer diagnosis are not driven by what we would think of as Hollywood-style intensive last-ditch efforts to sustain life or new expensive chemotherapy drugs, but rather by recurring “low-intensity” hospital admissions, which the authors define as being admitted to a department like internal medicine, oncology, rehabilitation or geriatrics, as opposed to the ICU or general surgery. These low-intensity admissions that focus more on evaluation and pain management (instead of surgical procedures or other intensive treatments) are to be expected as patients get sicker and run out of treatment options. Indeed, the start of frequent, unplanned low-intensity admissions is strongly predictive of mortality, and they are responsible for 95% of the higher costs incurred by those who died within 12 months of their initial diagnosis compared to those who did not.  

Yet, the authors also found that few of these low-intensity admissions were followed by death in the next two months, making it hard to say in advance which, if any, are futile and possibly wasteful. In other words, while this study provides new rigorous evidence on the sources of increased end-of-life spending, it does not find an obvious set of interventions that could be eliminated to save dollars without trading off patient health.