On November 7, 2013, Dan Munro published an article titled “The Two Obamacare Charts That No One's Talking About” on Forbes.com. BTW- Dan Munro is excellent. Follow him @danmunro. There were actually three charts in his article and it’s the third that I want to casually address here: A scatter diagram that relates life expectancy of nations’ populations and annual per capita expenditures. I downloaded data from the OECD to recreate Munro’s third graph (top left).

Total Life & Spend US in - no Rsquared.jpg

The US is the yellow dot: clearly an outlier on the X-axis and middle of the pack on the Y. But what does the trend line tell us? The line represents a predictive model, or what a nation should get for its healthcare dollar. By eye-balling the slope of the line, it looks like an OECD nation should get about a year and a half of extra life for every $2000 spent. That yellow dot on the right is clearly not living up to its potential. If American healthcare lived up the predictions of this model, we would be north of 84 years instead of at 78.7.

This model looks at the average members of the populations and controls for almost nothing. Diet, tobacco and alcohol consumption, obesity, education, crime, income, occupational safety standards, environmental protections… all are assumed identical among all 34 nations in this analysis, and none is reflective of a nation’s healthcare delivery system. In an attempt to remedy some of this, we’re going to look at another OECD stat: Years lived past age 65, which is when the healthcare system becomes more important in a population (top right).Linear 65+.jpg

The pattern looks similar. I’ve added the coefficient of determination, or R-squared (R2). R-squared explains how well the model predicts the data. The higher, the better, with perfect prediction being 1.0. This model’s R-squared is 0.2824. Modelers don’t usually get too excited about such a low R-squared. If we assume Years is dependent on Expenditure, then R-squared says, 28.24% of the variation in Years is explained by Expenditure. If you assume causation goes the other way, then re-read that last sentence with the variables swapped. If you make no assumptions about causation, then this graph is not worth much to you.

In this second graph, I separated nations into three natural groups: nations that spend above $2000, nations that spend less than $2000, and the US. Below $2000, no nation gets more than 18.1 additional years; nations above $2000 get at least 18.5 additional years. And the US isn’t the lowest of the $2000+ nations (sorry Denmark.) As separate groups, the variables show no relationship. The 26 nations in the $2000+ group have an R-squared of 0.024. Similarly, the eight nations spending less than $2000 have an R-squared 0.0519.

If you squint at the dots, you might see an arch in the data and if we apply a two degree polynomial model, we get this third graph. I included the model’s equation, conveniently supplied by Excel. This model, with an R-squared of 0.5898 suggests that a whopping 59% of the variation in life expectancy can be explained by healthcare expenditure. More importantly it predicts optimal spending somewhere between $4000 and $6000 per per person-year. It’s tempting to look at this third graph, point to that darn outlier nation, and say, “Yank the Yanks. The US is unduly influencing the model. It’s pulling the curve down.” And it is. With the US out of the picture (fourth graph), R-squared increases to 0.6417 and the predicted optimal range moves left. But I’d also point out that the US represents 25% of the total population being considered in the model, so tossing us out isn’t a trivial analytical decision. Let’s leave it in from here forward.

65+ & Spend US in POLY.jpg65+ & Spend US out POLY.jpg

That parabolic predictive model, with the US in it, looks very much like a notional graph we share at University of Utah Health Care whenever we talk about healthcare value, which is always. We refer to it as the quality-cost curve. The conversation goes like this:

  • Our conceptual healthcare value equation is:
  • To increase value, you can either improve quality or service, or you can reduce costs.
  • But wait— quality and costs are related by the quality-cost curve.
  • Today, American healthcare is at A, with fairly high quality but let no one say we’ve reached our potential.
  • When we focus on value, we move along the curve toward B, optimal value.
  • Thus, we can focus on quality, which is best for the patient, and the cost reduction will follow.

HC Quality Cost Curve.jpg

When I tweeted about this relationship, the responses from healthcare professionals outside our organization were (mostly) supportive. The enthusiastic response is heartening. This principle needs to be understood more broadly. As Dan Munro’s article points out, along with more and more voices, whether the ACA is the best or the worst piece of legislation to come out of Washington, or somewhere in between, the constant chatter about the ACA is a sideshow. The real crisis is about cost. And if we get focused on quality in American healthcare, we can drastically improve the cost crisis.