Understanding health care spending in the United States

Transcript

This transcript has been lightly edited for clarity

Rhonda Stewart: Welcome to Global Health Insights, a podcast from IHME, the Institute for Health Metrics and Evaluation. Hi, I’m Rhonda Stewart. In this episode, we’ll hear from IHME Associate Professor Joe Dieleman as he discusses US health care spending by county and health condition.
The new study is the most comprehensive to date in its analysis, which found significant variation in health care spending across the country.

Understanding these trends is critical to identifying ways to reduce costs and increase access to care.
Researchers examined out-of-pocket spending, private insurance, Medicaid, and Medicare for different types of health care. They studied more than 3,000 counties in the US and looked at the data by age and sex.

Ambulatory care, hospital inpatient care, and prescribed retail pharmaceuticals were the types of care with the highest levels of spending with respect to health conditions. The highest level of spending was on type 2 diabetes at $143 billion. The conditions with the next highest levels of spending were musculoskeletal disorders – including joint pain and osteoporosis – oral disorders, and ischemic heart disease.

Health care spending in the US is expected to reach more than $7 trillion by 2031. The study will be published in the Journal of the American Medical Association [JAMA].

Joe, the research examines variations in health spending across locations over time by health condition and county. What were some of the key findings?

Joe Dieleman: Yeah, thanks for asking, Rhonda. The main takeaways from the paper are about how much variation in spending really exists in the United States. I think we tend to think of health care systems as federal or state systems. But what we found when we looked at 3,110 US counties was that spending levels and the types of care that are being accessed by different people really varied dramatically, both across the US and within states.

And that takeaway is something really novel and new. Most analyses in the past have focused on state-level analyses, or if they do dig deeper, have really focused on just one payer like Medicare.
In this analysis, we are able to look at all major payers – Medicare, Medicaid, private insurance – and look at variation across all of the counties.

Rhonda Stewart: And why is it important to understand variations in health spending?

Joe Dieleman: So, variation in health spending sounds kind of wonky, but what it really means is where are resources going for the health care sector? And if there’s large variation that can’t be explained by other factors, then it suggests that some places are getting more resources than they may need. Or, on the contrary, some places are getting fewer resources than they need. So the trick is to try to understand why that variation exists.

In this study, we looked at things like age and disease prevalence. You’d expect places that have older people, like Florida, or have less healthy populations with maybe more disease prevalence of key diseases like diabetes or cancers would have more spending. And we see that to be true. But what we really find is that there’s a lot of unexplained variation.

Places that spend more, that are maybe the average age and the average health of the population, but are spending substantially more, more resources are devoted toward them and their health care. Whereas we see the exact opposite in other places. Again, take a county that has a very common kind of average age profile, maybe disease prevalence levels, but has substantially less spending or more spending that’s out-of-pocket. That sort of variation is really important to try to understand further and to understand why there are some places with more resources than others.

Rhonda Stewart: And you mentioned that this study takes a really new and novel approach. Talk a little bit about the kind of data that you analyze to produce the findings.

Joe Dieleman: So we’ve been working on this study for about four years. It’s been a large team, dedicated and hard at work in making these really granular spending estimates. To get down to looking at payer-specific, county-specific, year-specific spending estimates, you really need to have a lot of data.

I mentioned before, there are 3,110 US counties. And of course some of them are very large with large populations, but others are really small. And so you need to have a lot of data, even in the smaller counties with very few people.

So we used in this study over 40 billion insurance claims. Those claims come from Medicare and Medicaid, but also a lot of private insurance claims. And then we also use administrative records from hospitals, so emergency department visits or admissions. A lot of that information comes specifically from hospital records. And then finally, we had to use survey data as well.

There are not a lot of great estimates about the uninsured population because of course they don’t show up in claims data. And so for the uninsured population, we use survey data as well.
So really what this study was was a big data exercise of trying to pull together a broad variety, a lot of data, and try to make sense of it and really be able to compare across time and across counties.

Rhonda Stewart: Which health conditions are associated with the highest levels of spending?

Joe Dieleman: Yeah, we’ve been talking mostly about spending variation, but another thing that we spent a lot of time on was just understanding what diseases health expenditure is going toward, what diseases drive the most utilization of health care.

And so I think there’s the big four that I like to look at of health conditions with the most spending, and those are diabetes, something we call other musculoskeletal disorders, oral disorders, and ischemic heart disease – they’re the top four. Diabetes, for example, we spend over $143 billion on each year.

But each one of these diseases has kind of an interesting story. So if you take diabetes type 2 and other musculoskeletal disorders, something that they have in common, other than both having over $100 billion of spending each year, is enormous growth rates.

Both of them have a year-over-year growth that is above 5%. And even after you take out things like inflation or an aging population, they’re still growing very, very quickly. One thing that’s different about them is that a lot of the diabetes spending is on the oldest population, whereas other musculoskeletal disorders, which includes a lot of things like joint pain or osteoporosis, is people who are working-age.
And so 57% of the spending on other musculoskeletal disorders is the working-age population, which of course has really big impact on the economy as well as the individuals with these diseases.

The third largest health condition is oral disorders, which is just a fancy way of saying essentially cavities and orthodontia, you know, braces and whatnot. And the spending was just under $100 billion a year on that category. But what’s super interesting about it is over 50% of the spending is out-of-pocket, which is a really big deal because it means people are essentially shelling out resources to pay for that in the moment where they need the resources. And of course, we know from studying out-of-pocket spending that that can lead to people avoiding the care that they actually need. And so oral care and dentists in general really make up a lot of out-of-pocket spending.

And then the fourth of the big four is ischemic heart disease, which, if you go back in time, ischemic heart disease was the health condition we used to spend the most on. And because the growth rate for ischemic heart disease is actually quite low, it’s kind of fallen away from being the biggest health condition to now the fourth biggest. I think that really speaks to a lot of progress in controlling, managing, preventing ischemic heart disease and other cardiovascular diseases.

Rhonda Stewart: That’s incredibly interesting, especially when you think about the out-of-pocket spending on some of those oral conditions.

Joe Dieleman: Yeah, absolutely. It really is where it hits the bottom line for a lot of households. You can imagine having a high schooler who needs orthodontia, and that’s just an expense that many households need to pay up front.
And again, it does lend itself to this question of do we need this, or can we avoid this? Which we know is one of the biggest impacts of out-of-pocket spending – that people oftentimes will avoid the care that maybe they really do need.

Rhonda Stewart: Which types of care are associated with the highest levels of health spending?

Joe Dieleman: So in our categorization, we have about seven different categories, and, and the biggest of those is what we call ambulatory care. So ambulatory care is a fairly large category. It’s 42% of all the health care spending that we could track. And ambulatory care includes primary care, but also outpatient specialty visits and urgent care, any sort of day rehabilitation programs, that sort of thing. Really, anything that’s outpatient, and not in an emergency department, falls into this ambulatory care bucket.

And of course, it’s one of the buckets that’s growing the fastest. Not only is it 42% of total spending, but it’s growing tremendously fast as well. And that’s because – and this is probably a good thing from a health care spending perspective – we continue to substitute or remove different procedures from the hospital and put them in ambulatory surgical centers or outpatient surgery. And so that, as far as a health care spending perspective and kind of constraining growth, is really a good thing, that ambulatory care is growing, and growing faster than the spending on inpatient care.

Rhonda Stewart: Let’s talk also about specific locations. Tell us about the places where you saw the highest and lowest levels of health spending.

Joe Dieleman: So this is really interesting. I think we’ve always known some states stand out as spending dramatically more on health than other states. So think Massachusetts and New York spend substantially more than Utah and Idaho. And we found that to be true. Of course, most of the counties in Massachusetts and New York spend a great deal.

But as I said before, one of the interesting findings is that across many states, there’s a lot of variation. So take Florida: it’s the oldest state out of the 50, and there’s a lot of variation within Florida.
Or take the state of Washington, where I’m based. Again, across the state, looking at the counties, there’s really a huge amount of variation or heterogeneity in spending levels. And so when we dug down deep into trying to understand why that variation exists, we started to see some real patterns that struck us as really important.

We first found that utilization is explaining a lot of the spending. And we’ve always known, and it’s been talked about for a long time, that the price and intensity of care is one of the things that sets the US health care system apart from, say, other comparing countries like in Europe, where we know we spend a lot more and most of that difference is because of our prices and higher intensity of care.
It’s not so much about utilization, but when we looked across the US just comparing spending in county A to spending in county B, at a single time point, we found that 65% of the variation was because of utilization differences.

And you know, we think that’s really important because it suggests that there are different ways from a policy perspective that you’d want to consider that if spending is going up just because of price and intensity of care, you want to think of certainly policy interventions that could protect prices from going up higher. But if it’s about utilization, then you really want to start asking questions about where is this spending happening and is it the right amount? Is there a way to approximate how much the population in a specific county needs versus how much they’re actually getting?

Rhonda Stewart: And so obviously understanding what’s behind these changes is going to be really important and interesting to people. So as you said, utilization and pricing seem to be some of the key factors in answering that question.

Joe Dieleman: Yeah, 65% of the variation across the US when looking at counties was about utilization, which of course then lends itself to the question of why are some places using more resources than other places?

And so we looked at a bunch of correlates, covariates, to try to understand what is correlated with more utilization. And some of the things that popped out most clearly with very strong relationships were things like wealth, insurance rate, education rates. Those are places where there’s the highest utilization rates, especially outpatient ambulatory care.

The other thing that we found is that places where there was more managed care or Medicare Advantage had less utilization. So from a policy perspective, that’s an important observation that’s been shown elsewhere. But our study really substantiates that and highlights that managed care, and specifically in this case Medicare Advantage, is one way to curtail or prevent utilization. Which of course comes back, I think, to the million-dollar question of what’s the right amount of utilization.

We see this amount of variation across the United States, and we see that it’s being driven by utilization. What our study can’t do necessarily is tell us exactly how much is the right amount.
But it does suggest, because there’s so much unexplained variation, that either some places are having more than they need or other places are not getting as much as they need.

Rhonda Stewart: And Joe, you’ve touched on the policy implications of this research. How can policymakers and other decision-makers use the information?

Joe Dieleman: So I think there are a couple ways. The first is simply getting data and a new dataset out to be able to be used, and that’s for policymakers specifically to maybe direct where additional research is needed. And that’s especially for academia and other researchers to say, hey, here’s a novel, very exciting dataset that has a lot of patterns that can be explored, way more patterns and analyses than my team could do.

And so our hope really is that other researchers begin using it. But as far as policymakers specifically go, we hope that what this study will do is sound the alarm that there is all this unexplained variation and that it is especially true in out-of-pocket spending. It’s especially true in emergency department care. It’s in these places that really impact the type of care and the amount of resources people are spending. And so from a policymaker’s perspective, I think it lends itself to just asking, or in many ways demanding, more questions be answered about why this variation exists.

Rhonda Stewart: And you’ve covered many fascinating things about the study, but what would you say is the main thing that people should take away from the research?

Joe Dieleman: So when I think about this study, I think it highlights that there are essentially 3,110 very unique health systems in the US. The people that live in the counties are very distinct.
The things that they’re asking for as far as health care are distinct, and the spending levels, in some cases for very good reasons, and other times and reasons we don’t really understand, vary dramatically across those counties.

And I think it really opens the door, this new dataset, to explore even more what is the right amount of health care and what needs to change as far as making sure everyone has the right amount of access to care.

Rhonda Stewart: Great. Thanks so much, Joe.

Joe Dieleman: My pleasure. Thank you.

Rhonda Stewart: Details about the US health care spending study can be found at healthdata.org.