Adam Omary and Jeffrey A. Singer
Psychiatric diagnoses in the United States are rising across virtually every category, in every age group. According to the National Institutes of Mental Health, more than one in five US adults—59.3 million people—lived with a mental illness in 2022. An estimated 49.5 percent of American adolescents have met diagnostic criteria for at least one mental disorder at some point in their lifetime. Additionally, about 31 percent of adults will experience an anxiety disorder at some point in their lives, and 21 percent a mood disorder.
By these numbers, mental illness is not a rare affliction but a near-universal feature of American life, prompting some, including former US Surgeon General Vivek Murthy, MD, to declare a mental health epidemic.
The rise is evident across specific conditions as well. The Centers for Disease Control and Prevention (CDC) now places autism prevalence at 1 in 31, a 381 percent increase since 2000. Attention-Deficit/Hyperactivity Disorder (ADHD) diagnoses among American children nearly doubled from 6.1 to 11.4 percent between 1997 and 2022. Among adults, self-reported ADHD diagnosis among working-age adults has more than tripled since 2012, from 4.25 to 13.9 percent. Diagnosed anxiety among children aged 3 to 17 rose from 6.9 to 10.6 percent between 2016 and 2022—a 54 percent increase in just six years. Diagnosed depression among the same age group climbed from 3.1 to 4.6 percent, a 48 percent increase, in the same time period. Among adults, past-year prevalence of any mental illness rose to 23.1 percent in 2022, with young adults aged 18 to 25 reporting the highest rate of 36.2 percent.
Genetics, environmental factors, and improved screening alone are unlikely to explain these numbers. A less-examined driver is the US health care financing system, which pays more when providers diagnose more.
The Subjectivity Problem
Psychiatric diagnoses differ from most of medicine because they rely on subjective mental phenomena and behavioral symptoms instead of physical symptoms or biomarkers. There is no blood test for autism, no imaging scan that confirms ADHD, and no objective test that differentiates clinical anxiety from ordinary worry. Diagnosis depends on clinical judgment about whether a person’s behavior exceeds a threshold established by committee consensus in the Diagnostic and Statistical Manual of Mental Disorders (DSM).
The DSM has progressively broadened the boundaries of major psychiatric categories over successive revisions. The DSM‑5, published in 2013, collapsed previously distinct autism categories into a single spectrum, making “on the spectrum” a label elastic enough to encompass both nonverbal children requiring constant care and socially awkward adolescents who prefer solitude. The same revision loosened ADHD criteria, allowing symptoms to appear as late as age 12 rather than requiring onset by age 7, and reducing the symptom threshold for adults. Generalized anxiety disorder requires only that worry be “excessive” and cause “clinically significant distress or impairment,” judgments that depend entirely on a clinician’s interpretation of where normal worry ends, and the disorder begins.
Defenders of modern psychiatry often claim that expanding diagnostic criteria reflect better screening, capturing subtler presentations, and that rising diagnoses reflect more accurate assessments of the true population prevalence of mental illness. But aside from the grim forecasts of living in a world where half of all young people have experienced mental illness, there is reason to believe that psychiatric diagnoses have become less precise, not more.
Broad diagnostic criteria often interact with screening instruments that cannot reliably distinguish clinical conditions from normal variation. The CDC’s autism prevalence estimates, for instance, rely on surveys such as the Social Responsiveness Scale, which asks parents to rate statements like “Would rather be alone than with others,” “Has difficulty making friends,” and “Is regarded by other children as odd or weird.” These items describe behavioral traits common to social anxiety, introversion, and ordinary shyness and cannot reliably distinguish autism. Yet researchers routinely use high scores on such instruments as proxies for clinical diagnosis in prevalence studies, including in the CDC’s own data.
The limitations of this approach became especially apparent after the COVID-19 pandemic. CDC autism prevalence surged an additional 40 percent in just four years, from 2018 to 2022—a period during which millions of children experienced prolonged social isolation, disrupted routines, and reduced peer interaction that would predictably elevate scores on parent-reported behavioral surveys measuring social difficulties, whether or not the underlying rate of autism had changed.
None of this diminishes the reality of autism, ADHD, or anxiety disorders for individuals with significant functional impairment. But when the boundaries of diagnosis are inherently subjective, and when diagnosis is the key that unlocks streams of taxpayer-funded services, the system will predictably expand those boundaries.
How Medicaid and the ACA Reward Diagnostic Expansion
When diagnosis is subjective, and payment depends on diagnosis, the system will reward expanding the definition of illness.
Incentives drive behavior. Psychiatric overdiagnoses would matter less if the diagnosis were merely a label. But in the American healthcare system, diagnoses serve as keys that unlock streams of taxpayer dollars.
The Mental Health Parity and Addiction Equity Act of 2008, extended by the Affordable Care Act (ACA), requires health plans, including Medicaid managed care plans, to cover behavioral health services at parity with medical and surgical services. Parity addressed a real problem: mental health conditions were historically under-covered. But parity also limits the tools that plans can use to manage utilization. Prior authorization requirements, visit caps, and annual spending ceilings can all be challenged on parity grounds. Plans that wish to avoid litigation or regulatory action have strong reason to approve rather than deny.
Under the fee-for-service payment model in Medicaid, which 2008 parity provisions dramatically expanded, providers submit claims to the state Medicaid agency. The state then pays the provider in accordance with the predetermined price of the service, otherwise known as the fee schedule. The fee schedule, in theory, serves to regulate providers’ room for maneuver with respect to payment claims, thereby preventing undue financial gain. The reimbursement structure underlying the fee-for-service model is designed to mitigate abuse by binding providers to a prearranged sum.
However, the fee schedule only governs the prices to which providers are entitled for their services. It introduces no effective mechanism for governing the legitimacy of the services themselves. This empowers providers to profit by inflating the frequency of services, knowing that the fee-for-service model fixes only the pricing and not the services. This creates the conditions for supplier-induced demand.
In practice, therefore, providers have the freedom to manipulate demand by lowering the diagnostic threshold for services. Across states, weak spending constraints further subsidize this demand. This serves to distort natural market forces by enabling providers to expand mental health services beyond the point at which their cost would be acceptable to recipients, especially those with minimal diagnostic eligibility.
Similar risks persist in managed care, which pays per patient rather than per service. While this model improves cost predictability, it does little to ensure services are necessary. Providers still control enrollment, and expanding the number of patients can drive spending just as effectively as increasing the number of services. Changing the payment mechanism does not eliminate the incentive—it simply shifts how it is exploited.
Additionally, under Medicaid’s Early and Periodic Screening, Diagnostic, and Treatment (EPSDT) benefit, states must cover all medically necessary services for children under 21, even services not otherwise included in the state’s Medicaid plan, including mental health services.
When diagnoses rest on subjective behavioral criteria, and when coverage means open-ended reimbursement for services billed by the hour, the connection between spending and genuine clinical need begins to erode.
Then there is the federal matching structure. Medicaid’s open-ended Federal Medical Assistance Percentage reimburses states for 50 to 83 percent of Medicaid expenditures. When a state spends a dollar on autism services, it pays 17 to 50 cents. Federal taxpayers cover the rest. And because the match is open-ended, more spending automatically brings in more federal dollars. States bear only a fraction of the cost, weakening the fiscal discipline that comes with spending their own money.
Once therapy became mandatory, states used Medicaid waivers to circumvent standard rules and expand services and eligibility with federal funds. These waivers—and similar authorities—opened the door for providers to significantly increase Medicaid billing.
Enhanced federal matching rates during the COVID-19 public health emergency further reduced the state share, especially during the period when mental health spending grew the fastest. The pandemic significantly increased both the supply of and demand for psychiatric services. Telehealth services for mental health conditions surged 16- to 20-fold during the first year of the pandemic, according to a RAND study of over 5 million commercially insured adults, more than compensating for the drop in in-person care. By August 2022, overall mental health service utilization was 38.8 percent higher than before the pandemic. Mental health and substance use diagnoses grew from 11 percent of telehealth visits in early 2019 to 39 percent by mid-2021. The share of all outpatient visits carrying a mental health or substance use diagnosis doubled from 4 to 8 percent.
Pandemic emergency waivers and telehealth policies further loosened restrictions on how services could be delivered and reimbursed. States such as Massachusetts, North Carolina, Indiana, and Colorado expanded telehealth eligibility (including audio-only services) and adopted payment parity for telehealth, effectively turning remote services into scalable, high-volume billing opportunities. The result was not just a shift in how care was delivered, but a notable increase in utilization and spending, often in the tens of millions of dollars per state annually, consistent with policy changes that reduced the marginal cost of delivering and billing for services.
A substantial body of research suggests that financial incentives can influence psychiatric diagnosis rates. In the United States, eligibility for school services and insurance coverage often depends on specific diagnostic categories. For example, states offering more autism-specific services tend to report higher autism prevalence, while classifications of other developmental disabilities decline—a pattern consistent with diagnostic substitution. A 2009 study estimated that at least 26 percent of the increase in autism diagnoses in California between 1992 and 2005 could be explained by diagnostic substitution, primarily from children previously classified as having intellectual disability.
Reform Incentive Structures
In economic terms, this is supplier-induced demand operating within a system that lacks the price signals, utilization controls, and outcome accountability mechanisms that would normally constrain it. The therapy industry has expanded to absorb the available reimbursement, exactly as economic theory would predict in a fee-for-service system with elastic diagnostic criteria, open-ended coverage mandates, and absent oversight.
When eligibility for large, open-ended public benefits relies on diagnoses based on inherently subjective criteria, the system will inevitably expand the scope of those diagnoses. This creates fertile ground for overdiagnosis, misclassification, and sometimes outright fraud.
Federal and state policymakers should prohibit the structural conflict of interest in which the same entity diagnoses a mental disorder and then delivers and bills for the treatment, consistent with principles already applied to Medicaid home- and community-based services. States should also require independent reassessments of medical necessity at least every 12 months, conducted by qualified professionals with no financial ties to the treating provider. But these are damage-control measures. The underlying system still rewards diagnosing more and treating more.
The most effective reforms will change the incentives facing providers, states, and families. Under the current system, states oversee spending for which the federal government pays 50 to 83 cents of every dollar. When states spend other people’s money, accountability erodes. A more flexible financing structure would put families—not distant payers—in charge of deciding which interventions are worth pursuing, rather than steering them toward heavily subsidized, one-size-fits-all services.
Block-granting Medicaid behavioral health funding, or implementing per capita spending caps with state flexibility in program design, would require states to bear the full cost of their spending choices. Block grants or per capita caps would link responsibility with decision-making, giving states both the flexibility and motivation to create programs that prioritize actual value over federally subsidized volume. States that decide to fund mental health services would compete with other uses of the same limited resources, establishing the cost-benefit discipline that the current system completely lacks. States should be allowed to use competitive bidding, value-based contracting, and outcome-linked reimbursement instead of open-ended fee-for-service billing. Indiana’s 2024 rate reform, which cut autism spending by 27 percent without reducing the number of children served, demonstrates that this approach works.
Psychiatric overdiagnosis is the predictable result of financial incentives that shield all participants from the true cost of care. The outcome is exactly what economic theory would predict: steady growth in diagnoses, services, and public spending, with little ability to assess whether patients are better off. Those most harmed are people with genuinely impairing conditions, whose resources are diluted across an ever-expanding pool of recipients. Reforming these incentives is not about cutting care. It is about making sure care reaches those who truly need it.