Frequently, simplistic answers are given about how to lower the cost of health care in the United States. Much of the energy behind the push for this focus is that we spend more of GDP than any other developed country on health. Within the almost 19% of our GDP spent on healthcare, much is lost when undifferentiated per-person averages are used to frame policy discussions.
What gets lost in the averages is that a small number of individuals drive a good deal of healthcare spending. As example, Axios just reported that among employees getting their coverage from employers, just 1.3% of the employees were responsible for 20% of the overall spending. Of that 20% of spending, 40% of it was for prescription drugs. These figures mirror an early report (NIHCM) using 2009 data, showing 1% of the population accounted for 20% of all health spending and 5% accounted for 50% of all spending. At the other end of limited spending on health, half the population accounted for just 3% of health spending, and 15% of the population reported no spending at all.
Another frequently cited statistic is that one-quarter of all Medicare spending in our country occurs in the last year of life. The inference is that most of this is wasted in this last year. Continuing that theme if we just did a better job of triaging end-of-life care a large amount of savings would accrue. This concept sure fits into a utilitarian model of economics and “living or dying” at the end-of-life. It is another of the simplistic looks at how we can decrease the amount of GDP devoted to health care spending.
Einav and colleagues considered this question and designed a machine-based learning model of annual mortality risk using Medicare claims. In their investigation they sought to better understand a statistical trap involving the simplicity of end-of-life spending: Those who end up dying are the not the same as those who were sure to die. In short, as they state, “spending could appear concentrated on the dead, simply because we spend more on sicker individuals who have a high mortality – even if we never spent money on those certain to die within the year.”
Goldberg from WBUR interviewed one of the investigators, Finkelstein about their study and she said, “In fact, we find there is very little Medicare spending on people with high probability of dying.” She went on to restate what many physicians already know; it is difficult to predict who will die. In their study, if you take all Medicare patients in a given year, and take the top 1% with probability of dying, 44% are going to survive the year. When she went on to share the policy implications with Goldberg, Finkelstein said, first, we do spend an enormous amount of people who die, yet some of that may not be a waste. Second, we need to look at specific interventions that actually add value at the end of life.
Those of us working inside Curadux would add a third point, we need to assure that when the end of life comes, our members are receiving “just the care they desire, not more and not less.” In that way, they and their family will be able to make decisions they will never regret. It is also clear from data that saves money.