The $366 Billion Question: Why Untreated Psychosis Is a Budget Line, Not Just a Clinical One
- Schizophrenia cost the United States an estimated 366.8 billion dollars in 2024 across roughly 3.07 million adults, a per-person societal burden of 119,436 dollars.
- Indirect costs of 291.8 billion dollars dwarfed direct medical spending: lost caregiver time alone reached 104.6 billion dollars, more than direct health care at 36.7 billion dollars.
- Only 68.4 percent of affected adults lived in independent households; the rest were in supportive housing, long-term care, prisons and jails, or unhoused settings, marking where the system already pays.
- The figures were briefed to congressional staff in February 2026 as the empirical case for funding early intervention and continuity of psychosis care.
A new cost-of-illness model published in JAMA Psychiatry puts a precise number on a problem clinicians describe every day in qualitative terms: psychosis that is detected late and treated discontinuously is ruinously expensive, and the system pays for it whether or not it funds early care. The Schizophrenia & Psychosis Action Alliance team, led by Holly B. Krasa, estimated the 2024 societal cost of schizophrenia in the United States at 366.8 billion dollars for 3,070,739 adults (1.17 percent of the adult population). Per person, that is 119,436 dollars a year, with state-level estimates ranging from 110,975 dollars in Utah to 126,225 dollars in Alaska.
The structure of the bill is the clinically interesting part. Direct costs accounted for only 75.0 billion dollars of the total. Within that, health care proper was 36.7 billion, while supportive housing and homelessness reached 35.2 billion, justice-system interactions 11.9 billion, and disability benefits 5.1 billion. In other words, the state already spends nearly as much housing and incarcerating people with schizophrenia as it spends treating them.
The far larger share was indirect: 291.8 billion dollars. Lost wages reached 55.4 billion, reduced quality of life 41.4 billion, and lost life expectancy 47.5 billion. The single largest line item in the entire model was caregiver burden: 104.6 billion dollars in unpaid time plus 60.5 billion in damage to caregivers' own health, productivity, and out-of-pocket spending. The illness, in fiscal terms, is substantially carried by families.
Where the money is already going
The settings breakdown is a map of where care has failed to reach. Only 68.4 percent of affected adults lived in independent households; 18.6 percent were in supportive housing, 5.0 percent in long-term or skilled nursing care, 4.7 percent incarcerated, and 3.3 percent unhoused. Each of those non-household categories is, in part, a downstream cost of late detection and dropped continuity. The model was the centrepiece of a February 2026 congressional briefing precisely because it reframes early intervention not as discretionary spending but as cost avoidance.
What it does and does not say
A cost-of-illness model is an accounting exercise, not a treatment trial. It cannot prove that earlier or better-coordinated care would reclaim a given fraction of the 366.8 billion dollars; it only shows the scale of what is currently being absorbed elsewhere in the public ledger. For practitioners, the value is rhetorical and strategic: when a referral for first-episode psychosis stalls, or a young patient ages out of a coordinated programme into fragmented community care, the cost does not vanish. It migrates to housing, to families, and to the justice system, where it is larger and harder to recover.
Schizophrenia is not an expensive illness because it is treated; it is expensive because, too often, it is not.
Cost-of-illness models depend on prevalence assumptions and on cost inputs drawn from prior literature, so the headline figure carries wide uncertainty. The model estimates current burden, not the savings achievable by any specific intervention. It is a United States analysis and does not transfer directly to health systems with different financing or housing structures.