Something is not adding up

In the United States, spending on mental health treatment has increased steadily for decades. The number of people receiving psychiatric medication has grown significantly, access to therapy has expanded, and public awareness of mental health conditions is higher than at any point in recorded history. Stigma, while still real, has measurably decreased.

Yet the epidemic keeps growing. Rates of depression, anxiety, and disability attributed to mental illness have not declined. Suicide rates have not meaningfully improved. The Global Burden of Disease studies consistently rank mental and substance use disorders among the leading causes of disability worldwide. By any serious population-level outcome measure, the system is treating more people than ever, and the epidemic is worsening.

There are several explanations for this. Social media, economic precarity, climate anxiety, the long fallout of the pandemic, and the ongoing erosion of community belonging are all real drivers of population-level distress. None of these can be solved by psychiatry alone. Structural forces produce structural suffering, and no clinical model, however complete, can substitute for a just society.

But there is a clinical explanation that the field has been slow to name honestly, and it deserves direct examination: a significant part of the epidemic is being sustained, and in some populations worsened, by the very model deployed to address it.

This is not an argument that psychiatry causes net harm; the system's genuine contributions are real. It is an argument about scale and fit. An incomplete model, applied universally to populations it was never designed to serve, produces predictable failures at a population level. These failures compound over time. Understanding them is necessary if the goal is to actually bend the epidemic's curve.

When Treatment Produces Chronicity

Thomas Insel, who spent thirteen years as director of the National Institute of Mental Health overseeing approximately twenty billion dollars in research investment, made an assessment in 2022 that deserves to be read carefully. Despite the scale of investment in neuroscience and genetics, he wrote, the field had not moved the needle on morbidity and mortality. Suicide rates had not declined, disability from mental illness had not improved, and the biological targets being pursued had not translated into better outcomes.

This is a self-assessment from one of the most powerful figures in American psychiatric research. It is a reckoning with the limits of a framework, delivered by someone who built much of it.

One of the clearest mechanisms producing chronicity is treating adaptation-based suffering as a primary biological illness. When a person's anxiety, depression, or addiction is rooted in developmental trauma, nervous system dysregulation, or the chronic stress of navigating a world never built for their particular neurology or social reality, medication acts on the downstream physiological state without touching the upstream cause. The person experiences partial relief, but the adaptive pattern generating the symptoms continues unchanged. When medication is reduced, or when life circumstances change and the underlying vulnerability is activated, the symptoms return. A second episode is treated, and then a third.

Each episode in conditions like major depressive disorder increases the likelihood of a subsequent episode. It also increases the probability that the next episode will be more severe and more difficult to treat. This is documented in the literature and built into our clinical understanding of treatment-resistant depression. What is less often discussed is the proportion of people cycling through these episodes whose root cause was never addressed, not because it was untreatable, but because the system treating them had no framework to see it.

The Non-Compliance Cycle

Non-compliance with psychiatric medication runs at 40 to 60 percent in schizophrenia and is significant across virtually every diagnostic category. Each cycle of discontinuation and relapse carries a heavy neurobiological cost, leading to progressive treatment resistance, longer recovery times, and higher rates of incomplete remission.

The standard clinical response to non-compliance is enhanced adherence monitoring, long-acting injectable formulations, and psychoeducation about the risks of stopping medication. These responses are not wrong, but they address the symptom of non-compliance without examining its causes. These frequently include side effects the therapeutic relationship has not created space to discuss, a treatment model the patient does not experience as addressing what is actually wrong, and a justified skepticism about a system that has offered management rather than recovery.

Non-compliance is a feedback signal. It is patients communicating through behavior that the treatment they are receiving is not meeting them where they are. A system primarily organized around increasing adherence rather than understanding its causes is optimizing for the wrong variable.

Tachyphylaxis and the Limits of Long-Term Pharmacotherapy

A clinical phenomenon that receives insufficient attention in both professional and public discussions of psychiatric medication is tachyphylaxis: the gradual loss of medication effectiveness over time. Estimates of its prevalence in long-term antidepressant use vary considerably, with some studies placing the rate of antidepressant tolerance between 9 and 57 percent depending on definition and duration. The range itself is revealing because this phenomenon has not been adequately studied or tracked.

For a patient who achieves remission on an antidepressant and then finds, two or three years later, that the medication is no longer working, the clinical path typically involves dose increases, augmentation with a second or third agent, or switching to a different medication. Each of these steps introduces additional pharmacological complexity, additional side effect burdens, and the psychological weight of a treatment course that keeps changing without producing stable recovery. The question the system rarely asks at this juncture is whether the medication was ever the right primary intervention, or whether it produced initial relief by acting on symptoms whose roots were always elsewhere.

Polypharmacy, the co-prescription of multiple psychiatric medications, is partly a structural response to this problem, and it creates its own harms. Polypharmacy falls heaviest on Medicaid and low-income populations, including disproportionate numbers of BIPOC patients, who carry the highest burden of complex regimens with the least institutional power to question them.

The CBT Ceiling

Cognitive behavioral therapy is the most extensively researched psychotherapy in the literature. Its evidence base for specific conditions, including certain anxiety disorders and mild to moderate depression, is real. For many patients, it produces meaningful relief.

For a significant portion of the populations this series describes, it does not. This is not because the evidence base is fabricated, but because the model has a structural limit: it operates primarily at the level of thought and behavior. It addresses what a person thinks and does, while leaving the deeper nervous system state, the developmental history, and the adaptive patterns generating those thoughts and behaviors largely untouched.

A person whose anxiety is rooted in a nervous system organized around chronic threat does not become regulated by learning to identify cognitive distortions. The distortions are real, but they are outputs. Addressing the output without the substrate is like straightening a picture on a crooked wall. The wall continues to tilt, and the picture continues to hang wrong. Coping skills, applied to suffering whose cause is unaddressed, are management tools. They are not healing. They can be valuable, but they are not sufficient. In clinical practice, they are frequently offered in place of something more fundamental rather than alongside it.

The research literature on treatment-resistant depression increasingly reflects this ceiling. A meaningful proportion of people who do not respond to sequential antidepressant trials and standard psychotherapy are not biologically refractory. They are carrying suffering whose roots the available treatments were simply not designed to reach.

The Structural Drivers No Model Can Ignore

Any honest account of why the epidemic is growing must include the structural forces clinical psychiatry did not produce and cannot address alone.

Poverty is one of the most potent predictors of poor mental health outcomes in research literature. Housing instability, food insecurity, chronic economic precarity, and community violence produce nervous system dysregulation with the same physiological signatures as individual trauma because they are forms of chronic threat. A person living in conditions of material deprivation cannot regulate their nervous system in therapy if they return from every session to the exact conditions generating the dysregulation. The window of tolerance within which therapeutic work is possible narrows severely under chronic stress.

Racism functions as a chronic physiological stressor with measurable health consequences. The ongoing stress of navigating systemic racism, microaggressions, and structural disadvantage produces HPA axis dysregulation, elevated inflammatory markers, and cardiovascular effects that are physical and biological. They are not addressed by any clinical model that treats the person's nervous system in isolation from the social conditions shaping it.

The WHO Commission on Social Determinants of Health was explicit on this point: health outcomes, including mental health outcomes, cannot be meaningfully improved without addressing the structural conditions that produce them. A clinical model that ignores this is operating at the wrong level of analysis for a significant portion of the epidemic it is attempting to address.

The epidemic is rising most steeply in the communities the current system is least equipped to serve. That is not a coincidence; it is the predictable output of a model applied without regard to the conditions that make it insufficient.

What This Means

The mental health epidemic is a complex phenomenon with multiple drivers. Attributing it to any single cause, including the limits of the psychiatric model, would be an oversimplification. But a model that consistently treats symptoms rather than causes, that applies biological interventions to populations whose suffering has developmental, relational, and structural roots, that produces chronicity in a significant proportion of the people it intends to help, and that distributes its failures most heavily on the populations with the least power to seek alternatives, contributes to the epidemic it was built to address.

Naming this honestly is not a counsel of despair. It is the beginning of a more accurate diagnosis, which is the beginning of more useful treatment.

The question this series has been building toward is not only what the current model gets wrong, but what a more complete model would need to get right: what root-cause healing actually requires, what research across developmental neuroscience, trauma treatment, somatic approaches, and the science of memory reconsolidation tells us about how human beings actually change, and what it would take to make that knowledge available at the scale the epidemic demands.

That is the subject of the next essay.

Dr. Yasin Choudry is a board-certified psychiatrist with nearly thirty years of clinical experience. His work focuses on the populations mainstream psychiatry consistently misses, including highly sensitive people, complex trauma survivors, neurodivergent adults, and those whose suffering has roots deeper than a diagnostic checklist can reach. He is the author of Radical Recovery: A Holistic Approach to Mental Health.

Books and Resources

If you want to study the population data, systemic outcomes, and limits of current biomedical and behavioral interventions, these resources cover the primary literature:

• Population Outcomes and Biomedical Limitations: Thomas Insel’s Healing: Our Path from Mental Illness to Mental Health offers a systemic evaluation of modern psychiatric research and clinical outcomes. The long-term global impact metrics are updated regularly in the Global Burden of Disease studies published by the GBD Mental Disorders Collaborators (2022).

• Adherence and Relapse Architecture: For foundational metrics on long-term treatment adherence and recovery dynamics in major psychiatric conditions, see the clinical data compiled by Lacro and colleagues (2002) and Leucht and colleagues (2012).

• Pharmacological Limitations and Tachyphylaxis: The mechanics of antidepressant tolerance and long-term treatment dynamics are analyzed in the clinical work of Byrne and Rothschild (1998). Giovanni Fava’s research (2003, 2015) explores the long-term course of depressive disorders and oppositional tolerance concepts.

• Behavioral Limitations and the CBT Ceiling: The historical trends and outcome ceilings of standard cognitive behavioral interventions are tracked in the meta-analyses of Johnsen and Friborg (2015). Bruce Wampold and Zach Imel’s The Great Psychotherapy Debate breaks down the contextual factors that drive actual therapeutic change over specific techniques.

• Social Determinants and Structural Stressors: The physiological and biological footprints of systemic discrimination are documented in the structural health research of David Williams and Shauna Mohammed (2009). The comprehensive socioeconomic frameworks for health outcomes are detailed in the World Health Organization Commission on Social Determinants of Health Report (2008), and compiled for clinical training in Compton and Shim's The Social Determinants of Mental Health.

• Long-Term Systemic Treatment Trials: The landmark outcomes regarding sequential pharmacotherapy steps and remission limits are documented in the NIMH-funded STAR*D Report by Rush and colleagues (2006).