transcript

Speaker 1:
[00:08] Welcome to the Cato Podcast. I'm Ryan Bourne, Cato's R. Evan Schaaf Chair for the Public Understanding of Economics. Autism diagnoses in the United States have risen sharply over recent decades, and so has Medicaid spending on therapies tied to them. But how much of that actually reflects a growth in the number of children with severe, genuinely disabling autism, and how much reflects looser diagnostic boundaries, distorted payment incentives, and weak oversight that invite waste and abuse? To discuss this and whether this is just the tip of the iceberg of medical misdiagnosis or overdiagnosis more generally, today I'm joined by Jeffrey Singer, a Senior Fellow in Cato's Department of Health Policy Studies.

Speaker 2:
[00:49] Good morning.

Speaker 1:
[00:50] And Adam Omary, making his, I believe his debut on the podcast today, a research fellow at the Cato Institute Center for Global Liberty and Prosperity.

Speaker 3:
[00:59] Debut Cato Podcast. Thanks for having me.

Speaker 1:
[01:02] So Adam, let's start with you, given this is your debut. If somebody were to just briefly ask you, what actually is autism or what is the kind of medical definition of autism that we're trying to get at? I mean, how would you describe it?

Speaker 3:
[01:18] Autism is a rare genetic psychiatric disorder, characterized by social impairments, excessive, repetitive behaviors. And now we frame it as autism spectrum disorder. That we see everything from, in severe cases, nonverbal children or adults who need constant care to what we would now consider more mild autism spectrum cases, as in high functioning but exhibits social difficulties.

Speaker 1:
[01:50] So there's been a well-documented increase in autism diagnosis numbers in recent decades, and that's got plenty of people worried. In a piece that you wrote for The Washington Post, you said, blame for this has been placed at the door of maternal Tylenol use, food dyes and additives, chemical manufacturing agents and other possible stresses. And yet, you argue that these concerns really start from a faulty premise, that the whole idea of an autism epidemic is a myth. So what exactly is the myth?

Speaker 3:
[02:20] Well, as I mentioned, autism is a genetic disorder. It's up to 90 percent heritable, up to 90 percent of the variation that we see in the population is attributable to genes. So it doesn't make sense to talk about it as an epidemic that's spreading throughout the environment. And if you look at the known causes of autism, the single greatest risk factor is having a parent or other family members who have the condition. None of these other factors that have been looked into, whether it's vaccines or environmental toxins, none of those associations hold up. And if you think about it from first principles, it doesn't make sense. Again, the only factors that we know of that replicate are the genetic risk factors.

Speaker 1:
[03:06] So give us a sense of the scale of that increase in diagnoses.

Speaker 3:
[03:09] The scale of the increase. Well, according to the CDC, in the last two decades or so, autism diagnoses have increased more than 400%. Now, knowing that this is a genetic disorder, we would expect the population prevalence to be stable. The genes that contribute to autism are stable in the population. So if the diagnostic rates are changing, that means one of two things. It either means that we were under diagnosing it in the past and we're getting better screening, which some people argue, or it means we're over diagnosing. Over the many decades we've been studying autism, we've certainly seen both. In the past, it was under diagnosed and now we're very likely over diagnosing.

Speaker 1:
[03:56] So what's the best evidence or framework for suggesting that the more recent rise is simply not better identification of kids who might have previously been missed, might have needed help, but actually for whatever reason, there was just not enough awareness and not enough screening for autism.

Speaker 3:
[04:14] We have to look at the screening tools themselves. So according to the CDC, much of their own data that contributes to this rise is not gold standard in-person assessments by a psychiatrist doing in-person testing, but it's surveys, not just surveys completed by the child, but by the parent. And these survey items address things like, my child struggles with eye contact, my child is shy, my child prefers to be alone than with others. All of these are behaviors that are associated with autism, but they're also associated with social anxiety or just ordinary shyness. And the most dramatic increase that we've seen has been post-COVID. Again, if you're talking about a rare genetic disorder, COVID should have nothing to do with that. But if you're really measuring social behaviors that can easily generate false positives, when you're really just measuring social difficulties, that could arise from autism, but could also arise from social isolation during the pandemic or social anxiety or other factors, that seems to be what the CDC's screening instruments are picking up.

Speaker 2:
[05:23] I think it's also important to mention that the criteria have changed dramatically. When I was a medical student way back when, we had just gone from calling what used to call childhood schizophrenia to autism. So my understanding as a medical student of autism were these extreme cases where a child was non-communicative, oftentimes rocking in a corner. And then after I got out of medical school, Asperger and others started saying, you know, there are variations in this. It doesn't just have to be the extreme cases. So I think it was, Adam could correct me if I'm wrong here, but around 2013, when the criteria in the diagnostic statistic manual put out by the American Psychiatric Association, that they went from to autism spectrum disorder, ASD, instead of autism. And that also had a lot to do with people being diagnosed with autism who previously hadn't been. Am I correct?

Speaker 3:
[06:26] Correct. So as a very rough heuristic, I believe it was 1993 when in the DSM-4, the psychiatric manual's fourth edition, Asperger's syndrome was introduced. And that characterized a lot of these subtler, high-functioning cases. And then in 2013, as Dr. Singer just mentioned, it was reframed again in the fifth edition to autism spectrum disorder. Now, my sense is that over the previous few decades, when we were first improving awareness of these subtler cases like Asperger's, these high-functioning cases that didn't match previous characterizations of severe autism, that was genuine recognition of subtler cases that was missed. That was previously we had under diagnosed and we were catching up. But post 2013, after the broad spectrum reframing, we're almost certainly over diagnosing. And post 2010 or so, with mental health parity provisions introduced in the Affordable Care Act, where insurance now has to cover psychiatric conditions at the same rates as physical health conditions, we also saw an explosion in diagnoses, most particularly autism because the therapy provided for autism, it's called applied behavior analysis. It's effectively social skills training. And because autism is so uniquely broad, both due to the spectrum labeling within the psychiatric manual itself, and just the breadth of social difficulties, where it could include all the way from these extreme disabled nonverbal cases to subtle impairments in social cognitive ability or empathy, emotional intelligence, it's something that anyone can improve on. And when you have financial incentives to help these cases, well, it's a never ending game.

Speaker 1:
[08:31] Yeah. So, I mean, Jeff, I'm an economist, so I'm actually very tempted by the idea that Adam sets out there, that these types of incentives drive behavior. And this matters to us as libertarians, not just because we care about the welfare, well-being of the patients involved, but with these diagnoses comes a stream of federal taxpayer dollars as well. How does the kind of structure of programs like Medicaid encourage this type of overdiagnosis?

Speaker 2:
[09:01] Yeah, that's the other reason why we're seeing an increase in diagnoses. For example, if you are diagnosed with ASD, then you become eligible for coverage under Medicaid for Applied Behavior Analysis. And it basically opens the door for all sorts of people to be diagnosed with it. In fact, Adam and his research assistant found that a great number of people who are getting ABA treatment are mild cases of ASD. They're on kind of the low end of the spectrum. And ABA was primarily designed for more severe cases. So generally speaking, if you subsidize something, you get more of it. So if you're subsidizing autism diagnoses, look to get more of them. And it's not just Medicaid. The Affordable Care Act requires mental health parity. So there are no caps on billing for treatment for autism or any other mental health disorder, just like with physical conditions. And so we're seeing this proliferation of treatment programs for children with autism spectrum disorder. And Adam has uncovered numerous cases of Medicaid fraud, places that are saying they're providing ABA to children with autism and not even doing so.

Speaker 1:
[10:41] I'll come on to the fraud point in a minute. But just to be clear, because I know that a lot of parents with kids who have been diagnosed and are very sensitive to the fact. So it's worth us saying that we aren't saying that autism isn't real. We're not saying it's not a serious condition for many children and that the provision of services for those with particularly bad cases, non-verbal children, these things are kind of understandable. What we're trying to get out here is the delta, it's the change over time and how we've seen this big increase in diagnosis.

Speaker 2:
[11:15] Right, but it's also important because, at least at the Department of HHS level, there's this new drive to do more research to once again confirm that there's no evidence that MMR vaccine is responsible for the increase in autism cases or Tylenol during pregnancy. There are only so many times you can come up with studies showing no evidence and say, okay, it's time to stop repeating these studies. What we're doing is providing explanations that should be satisfactory to explain why we've seen this increase in autism diagnoses other than vaccines and Tylenol.

Speaker 1:
[11:58] I guess when it comes to programs like Medicaid, though, my question would be those programs have been relatively long-standing. So how do they explain the change in recent decades? I could understand, you could make the argument that structurally, these lead in courage of perhaps an over-diagnosis of certain conditions. I guess you kind of alluded to it in one of your previous answers. There were genuine changes with the ACA. But in terms of Medicaid, surely this is something that's more structural rather than something that has changed in recent decades.

Speaker 3:
[12:29] In terms of Medicaid, the biggest structural change would be the introduction of mental health parity requirements. And it's also worth noting that this applies to psychiatry as a field, which compared to most of medicine is more inherently subjective. So that's not to dismiss the real mental suffering that people experience, but it relies on a clinician's judgment to determine what is the boundary between ordinary worry and clinical anxiety. And there's no objective boundary. The psychiatric manual that we've been referring to is expert committee consensus, but it changes across editions. It's certainly changed over time. Each successive edition has gotten more broad, and it's a fine line between increasing the number of cases and presentations that you can see in terms of helping people. We're catching subtler cases versus over pathologizing ordinary struggles and ordinary behavior. And when you consider the financial incentives that go into it, that clinicians get paid more when they diagnose more, it can explain a lot of these changes.

Speaker 1:
[13:42] You also argued that the pandemic emergency waivers and telehealth policies and things like that further loosened restrictions on how these services could be delivered and reimbursed. So kind of all of these things just exacerbate the cycle, right?

Speaker 2:
[13:56] Right. And this is not like Adam was saying, it's not just limited to autism, it's psychiatric, but even physical disorders. You get paid for diagnosing things. So now it's harder, it's harder to be subjective with a lot of organic disorders. You can't, whether or not you have cancer is pretty objective. Whether or not you have an intestinal obstruction, that's pretty objective. You can't have differences of opinion on it. But there are a lot of ways in which you could move the diagnostic goalposts. So over the years, by the way, Gilbert Welch wrote a great book a few years ago called Overdiagnosed that goes into this in detail. So over the years, we've changed the criteria of what we're calling high blood pressure or what we call diabetes or pre-diabetes. Back in the 90s, we made it where the blood sugar level had to be lower than previously was recommended to be diagnosed with diabetes. And overnight, we had this epidemic of diabetes. So, a lot of people today, for example, we've lowered the thresholds for LDL cholesterol levels so that now a lot more people are being treated for hypercholesterole.

Speaker 1:
[15:10] Including me.

Speaker 2:
[15:12] That weren't in the past or chronic kidney disease, also called CKD. A whole lot of people who their kidney function as measured by certain blood tests, that was considered, well, normal for age. You're in your 70s, so your kidney function isn't what it used to be. Now, all of a sudden, they've changed the thresholds for that, so that people now who weren't, who were just considered to have old kidneys, now they have CKD. Also, sometimes they're being put on medication for that to help keep the kidneys functioning better. So, there are these, we see the same problem in physical diseases that we see in mental health disorders. Although, like I say, it's more difficult to be subjective with certain physical conditions.

Speaker 1:
[16:05] One would imagine if the same incentives are there, the type of work that you've done on autism is just the tip of the iceberg. But I want to pivot back just a little bit before we explore that a bit further to talk about the scale of spending on autism diagnoses. Why don't you give us a sense of how much spending is increased in certain states because of this? Because I think as a kind of laboratory of democracy, the US has a lot of variation across states in provision of various services. Is there evidence when you look across states to suggest that these financial incentives influence diagnosis rates between them? Or is this just a whole kind of national issue?

Speaker 3:
[16:51] We just released a Cato report looking at autism Medicaid spending across eight states with publicly available data from the last decade. And we saw spending across these eight states increase from about $300 million around 2017 to over $2 billion in 2024. So if you scale that up to 50 states, we're talking about billions of dollars being spent on autism therapy alone. And the percent increase that that represents massively outpaces even the CDC's 500% increase that we've seen over the last couple decades, which I already mentioned I'm skeptical about because they're using excessively broad criteria. And the vast majority of these increases are mild cases. And as Jeff mentioned, it's the mild cases that are least in need of therapy. And in this report, we saw that much of the excess spending and borderline fraud was coming from cases where children were receiving therapy, but they were receiving excess of 30 hours per week, which historically is unprecedented, only for these severely disabled children that require constant care. Among the worst offenders we saw in Minnesota in 2017, they spent less than a million dollars per year on autism therapy. And by 2024, over 300 million, 50,000% increase.

Speaker 1:
[18:30] Well, and I know that many of us are sceptical that the official inflation measures fully account for the increase in the cost of living. But for those keeping score, these increases in spending vastly exceed the increase in the general price level that we've seen. Now, the federal government itself has undertaken an audit of autism therapy payments in Medicaid across four states. You guys wrote about this in one of your blog posts, and they found vast amounts of improper and potentially improper payments. Now, I know from some of my own research in other areas, that's not necessarily the same thing as fraud. Means a payment that perhaps shouldn't have been made, was made in the wrong amount, cannot be verified. But I think it's fair to say that the findings of that report, there were serious red flags here.

Speaker 3:
[19:25] Hundreds of millions of dollars. And yes, as you mentioned, it's not necessarily fraud. In many cases, the services were being provided. They were just cutting corners in terms of not filling out the proper documentation. At the very least, there are some accountability issues. But when we see billions of dollars of spending increasing just in the span of a few years for a condition that is uniquely broadly defined despite knowing that it's actually a stable genetic disorder that we shouldn't expect to see many hundreds of percent increases on the order of just a few years. And you look at the broadening financial incentives and the billions of dollars that states are raking in providing this therapy. It's also worth mentioning that states aren't paying for most of this. The federal government is reimbursing 50 to 83 cents on the dollar. Some states are paying as little as 17 cents per dollar of Medicaid spending.

Speaker 2:
[20:33] And there's a disincentive for states to police this, to audit and keep an eye on this kind of behavior for that very reason, because the federal government is really picking up most of the bill, not the state government. So, for example, if instead of that, the federal government block granted Medicaid to the states and said, this is our contribution for the year, you make it work for you, and if you need more money than raise taxes on your citizens, that would probably change the behavior of the states that are administering their Medicaid programs.

Speaker 1:
[21:10] Yeah, that was the next question I was going to come on to really. Given that we know this and we suspect some of the underlying causes, what could we do about it? Obviously, Jeff has highlighted one issue there on the funding side, but are there any other policy recommendations that if you were advising state governments, that you'd be suggesting, or even the federal government that you'd be suggesting to try and curb this over diagnosis?

Speaker 3:
[21:37] I just think it traces back to the incentive structures. As it's currently laid out, clinicians are paid more for diagnosing more rather than better for these mild cases, especially mild cases of psychiatric conditions where, and we can nitpick about social difficulties, introversion, shyness, you can seek to optimize any sort of social behavior or psychological behavior. But there's such a fuzzy border between ordinary struggle and clinical pathology. And it's not helpful to diagnose and pathologize ordinary struggles. In many of these cases, children that might be shy or introverted but can work on their social skills, if they instead internalize it as a label, and I have this rigid condition called autism, which we know is genetic and something that you can't control, it might demotivate them to actually work on their social skills. And you can say the same about other mental health conditions. So when it comes to policy recommendations, I think before we even get there, we need to have a discussion within psychiatry as a field of making sure that we're defining these conditions in a way that's helpful.

Speaker 1:
[23:06] Yeah. And I was going to suggest to Jeff that there are really other non-fiscal costs of over-diagnosis, including to the well-being of the patients, but also the opportunity costs, the time of resources, the time of the administration of these programs. There's a lot of potential waste here.

Speaker 2:
[23:25] So another factor is that the Mental Health Parity Act actually discourages private insurers from auditing not just autism treatment programs, but mental health programs, drug rehab programs, and doing such things as pre-authorization and to see if it's necessary, because they risk being sued for violating the Mental Health Parity Act. So the Mental Health Parity Act is a large part of the problem. I'd like to mention also that about a week ago or so, Adam and I wrote a blog post making the case against social media addiction. And I think it's interesting to note after we just discussed all of the incentives to diagnose mental health and physical health disorders, the DSM-5, the latest manual by the American Psychiatric Association, does not categorize so-called social media addiction as addiction. It's an area they designate as an area for study. So with all the incentives in place to come up with diagnosis for another addiction or disorder, the fact that the people who get compensated financially for treating that disorder are saying as the consensus is right now that they're not convinced there is such a thing, as social media addiction should carry a lot of weight.

Speaker 3:
[24:51] And it's a good example of this dance between financial incentives towards overdiagnosis and the clinical responsibility to hold back and make sure that the labels are actually serving the patients. We saw a similar thing play out a few years ago in the ICD, that's International Classification of Diseases, where they did introduce video game addiction as a psychiatric condition. And the exact same debate was playing out then as what we see with social media now. And as Jeff mentioned and as we discussed in this blog post, there's a strong difference between habit, behavioral addiction, and the physiological dependence that you get in substance use addiction. And conflating these two has all sorts of negative consequences, and it's just part of this broader trend towards psychiatric overdiagnosis.

Speaker 1:
[25:47] Well, and it also tends to be a justification for blaming the providers of the services quite often, right? You've seen a lot of moral panic at the moment about sports betting in the United States. Many other countries have big open sports betting markets, and they seem to work reasonably well. You're always going to have extremely difficult cases where people become addicted to actually playing the games. They get so stimulated by engaging them. So you have problem gamblers. But we as libertarians would say that's not a reason to curb the whole industry for everybody and those that are able to engage in these activities.

Speaker 2:
[26:21] And when it comes to social media addiction, if all of a sudden everybody's going to say it is an addiction, now you're going to see lawsuits take place. It's already happening against the social platform holders, which is going to make them want to curtail what they're going to allow because they don't want to risk the liability risk. And so even without Congress or state legislatures intervening and putting all sorts of restrictions on freedom of speech and freedom of expression, the litigation is going to lead a lot of that to develop on behalf of the purveyors of social media platforms. So that's why it's very important to be precise when we're discussing addiction.

Speaker 1:
[27:05] I mean, clearly, Adam, there are kind of dangers to extensive social media use for some young people. As a forum, as a kind of way of meeting people, just like anywhere in life, there's going to be bad actors on those platforms. The access to people of very young ages to certain forms of content, which they might be able to access, that tells them how to undertake self harm and things. I can see why many parents are worried about it, but you've argued that these types of very crude restrictions, like bans on social media, kind of age-related are not the way to go about mitigating those harms that certain people do face.

Speaker 3:
[27:48] Yes, this is something I've written a lot about. So first, the associations between social media use and mental health outcomes in youth are real, but they're not causal. A lot of people frame it as though, because people who use more social media are more depressed and anxious and lonely, they frame it as though this was an otherwise healthy child who starts using social media and then they experience all this harmful outcomes. And in the majority of cases, what you have is children who are already depressed, anxious and lonely, seeking out social media connection. And in some cases, particularly for those who are homeschooled or might lack opportunities for in-person connection, it might be very enriching. And then to the point about attempts to regulate, it often doesn't work. It often backfires. Children are tech savvy. They know how to install a VPN and route their internet traffic through an area that doesn't have the same legal restrictions. It often means in practice that they're gravitating towards platforms that have less content moderation and more risk of exposure to the very types of harmful content that people are trying to avoid.

Speaker 1:
[29:05] Yeah, that's certainly true. It's like governments that engage in this. It's like a game of whack-a-mole, right? As soon as you put restrictions or age gating on certain websites or whatever, VPN use goes through the roof and then governments are looking to put restrictions on the use of VPN.

Speaker 2:
[29:20] Yeah, the UK is showing in Australia now.

Speaker 1:
[29:22] Yeah, the UK as well. So let's just finally return to our kind of through line of the conversation. How worried are you, Jeff? And it sounds like you are pretty worried from some of your previous answers that this autism example case study is really just the kind of tip of the iceberg of an overdiagnosis problem more generally in the US health system.

Speaker 2:
[29:48] I'm very worried, because I think it's among other things, it's going to be economically unsustainable. I suspect that a whole host of other mental health disorders, ADHD, is another one that comes to mind. We've seen so many children diagnosed with ADHD in recent years. There's only so much money to pay for these things. Plus, I think it's not a good idea to medicalize every single feature of human conduct. People, it takes away their agency and their autonomy to make every single possibly negative behavioral activity, a medical condition, subject to treatment.

Speaker 3:
[30:30] I share Jeff's same concerns, but I work on Cato's Human Progress Team, we're documenting increases in global prosperity and lifespan, decreases in disease rates and so forth. And as much as these are real problems within psychiatry, I think it's remarkable that we're even able to have this discussion about overdiagnosis. It means that we're wealthy and secure enough to attend to mental health problems, which historically were neglected. They were considered just perennial problems of human existence, that you're going to experience mental suffering. And now we're able to have a conversation about mental health is perhaps something that can be optimized. Perhaps we're over-correcting, but I view that as part of this long trend of human progress. I'm quite optimistic.

Speaker 1:
[31:19] That's that great philosopher Dolly Parton once said, if you want the rainbow, you got to take the rain. That's all we've got time for today. The real issue here is not whether autism is real or whether some children benefit from therapy. It's whether a badly designed public finance system has really turned what is a real condition for many children into an open-ended reimbursement category. Thank you to Jeff, thank you to Adam for being here today. That's all we've got time for. We'll see you back on the Cato Podcast next Tuesday.