Q&A: Understanding health care spending in the US

Dr. Joe Dieleman discusses the latest research on health care spending in the US by state and county, type of care, age, sex, and payer.

Video transcript

This transcript has been lightly edited for clarity.

What does the latest research from IHME’s Disease Expenditure project examine?

We need to understand where health care spending is coming from and where it’s going. So that’s really, in the simplest terms, what we try to do – ask the questions, who are the patients? How old are they? What’s their sex? What are the health conditions that the money is being spent on? Is it from Medicare, Medicaid, or private insurance? Is it out-of-pocket? And ultimately, we can track those changes over time. That’s the goal.

So this figure is what we spent four years building, essentially. It’s 3,110 US counties. And the estimates are age- and sex-standardized. And it’s telling us where the spending is occurring. And you can see really remarkable variation across the entire US. But you can also see variation within individual states.

If you look at Washington state, the state where I’m currently based, you see enormous variation within the state, just like you see variation across the country. And so it really highlights the fact that health care spending and health care utilization is a very local phenomenon. And it varies a lot even between two neighboring counties.

How does spending on health care vary according to age?

This is an age pyramid, and that shows us the total per capita spending for each age and sex group. I think the really important observation here is that it’s broken down by the type of care. So you can see that as individuals age, the per capita spending goes up dramatically. And the one place that stands out where that pattern isn’t observed is that first year of life. The first year of life is very expensive. There’s a lot of inpatient spending on those first few days of life. And, of course, any complications, in an inpatient setting, have really enormous costs.

But as people age, they spend more and more, on average, on health. And you can see some types of care, like nursing facility care, really focus on the oldest population.

This is similar to the last figure, in that it’s an age pyramid that shows us where spending is occurring. But the real difference here is that the x-axis is total spending. We see that some of the most spending is on the older populations. We see a big bubble that is, essentially, the baby boomers, between 60 and 70 years old, have the most spending.

We also see this transition from private insurance, which is yellow on this figure, to Medicare, and that huge shock, essentially, to our health care system when people turn 65, are eligible for Medicare, and have a major increase in spending associated with that public insurance program.

How do these papers build upon previous Disease Expenditure project research?

The previous research from the Disease Expenditure project really focused on changes over time at the national level, and looking at what health conditions were the health conditions with the most spending. And we wanted to continue to do that. So we did. But one of the important critiques was that we weren’t getting at a more granular level.

And so, we spent the last four years collecting data and developing new methods to take the Disease Expenditure work and do it at a more granular level.

What variation did you find between the lowest and highest per capita spending on health by county?

We knew there was a lot of variation at the state level, about a two times difference between the states that spend the least and spend the most on health care. But if we look at counties, there’s a five times difference. And that’s really remarkable. The counties that spend the least on health care are right around $3,000 per capita. And some of the counties that spend the most on health care are closer to $12,000 per capita. So a really remarkable difference.

The second paper comes along and says, okay, we’ve observed these differences. Can we try to explain those differences? What are the factors that are contributing to why one county spends so much more per capita than another county?

And so, we’ve started to find that a lot of the variation is driven by some things that you might expect – age and sex of the population, the disease prevalence. But the big thing is utilization. How are the people living in that county using health care services? How often are they going to the doctor? How often are they being admitted to the hospital?

And we found that 65% of the spending variation across these counties was driven by utilization differences.

What factors impact per capita spending on health by county?

Most people think, when they think about how much you spend and how much is changing over time in that health care spending, is it’s all about prices. And this figure highlights that. In this figure, blue is price and intensity of care. And you see that that’s about 25% of the total spending variation. What this figure highlights that’s so novel is how much of the variation across counties is driven by utilization.

We see that 65% of the variation in across-county spending is driven by utilization rates. And what that means is that the biggest factor leading to differences in spending is how often people are going to the doctor. And are they seeking emergency department care, or outpatient ambulatory care? Are they being admitted to the doctor or going to a nursing facility? That variation is really what’s driving the huge amount of variation that we see across the US counties.

What comes next for the Disease Expenditure project?

The next steps for the Disease Expenditure project are to start to ask the question, what’s the right amount of utilization? What these papers have done is said, look, there’s a lot of variation. And that variation is much larger than we knew beforehand.

Some of that variation is related to social economic correlates, like income or education or insurance rates. But now the question really is, and the place we want to go, is what’s the right amount of utilization? If we compare these estimates to health outcomes, can we start to tell a story about what is the right amount? And that’s really the direction we want to go next.