The Household Buying Bottled Water

A household that buys bottled water every morning while the pipe beneath its kitchen sink corrodes is not solving a water problem. It is financing one. The invoice arrives twice: at the checkout, and later when the plumber is called.

Mental health policy has drifted toward that arrangement. Capital flows toward the bottle — the subscription, the breathing module, the nineteen-minute session — while the mains stay under pressure.

The World Health Organization's World Mental Health Report, published in June 2022, recorded a 25% rise in the global prevalence of anxiety and depressive disorders in 2020. The U.S. Surgeon General's advisory of May 2023 then reframed adolescent social media exposure as a population-level exposure rather than a collection of individual failings. Between them, those documents describe the system this sector now operates inside.

Five Anchors, One Balance Sheet

Read together, the following are not five unrelated developments. They are a single supply chain: exposure, detection, access, treatment, and the bill.

• June 2022 — WHO, World Mental Health Report. The prevalence and financing baseline for the post-pandemic period.
• July 2022 — SAMHSA, 988 Suicide & Crisis Lifeline. A fragmented call network consolidated behind three digits.
• May 2023 — U.S. Surgeon General, advisory on social media and youth. Adolescence reclassified as a public health population.
• 2016 — Chisholm et al., The Lancet Psychiatry. The return-on-investment model now circulating in procurement rooms.
• 2019 — Riehm et al., JAMA Psychiatry. The observational dose-response study underneath much of the screen-time debate.

The Funnel That Backs Up

An access layer is not a treatment system. Consolidating crisis care behind three digits compressed the first mile of help-seeking into one national funnel. Every contact, however, terminates somewhere: a clinician's calendar, an emergency department, a psychiatric bed, a family able to supervise. If those downstream slots do not exist, the funnel does not disappear. It backs up. Crisis centers absorb the overflow and begin performing triage for conditions they were never staffed to treat. The reporting metric quietly shifts from completed care to contacts handled — a throughput statistic masquerading as an outcome.

When the Ledger Changes Rooms

The second migration is accounting. When an employer purchases a mental health platform, the intervention moves from the medical ledger to the human resources ledger, and the definition of success travels with it. Remission becomes retention; diagnostic improvement becomes absenteeism. Vendor dashboards optimize for the buyer's problem, which is rarely the patient's. Aggregated workforce data then raises a question most contracts leave unanswered: who can infer an employee's clinical trajectory from utilization patterns, and on what basis? The technology is not the risk. Ownership of the resulting ledger is.

That objection, carried too far, becomes its own neglect. A person on a four-month waitlist in a rural catchment does not experience a digital service as a compromise; they experience it as their first contact with care. The economics reinforce the point. Chisholm and colleagues, modeling scale-up of depression and anxiety treatment in The Lancet Psychiatry in 2016, estimated roughly $5 returned in health and work-related output for every $1 invested. Rejecting scalable delivery because it is imperfect forfeits the sector's strongest efficiency argument. The defensible position is conditional adoption, not purity.

The Evidence That Survives Attrition

The third blind spot is evidential. Commercial applications are typically validated in short trials, among self-selecting users, with attrition that would be conspicuous in a pharmaceutical dossier. Surviving completers are disproportionately the people likely to improve regardless of assignment. This is a problem of external validity, not merely sample size: an effect demonstrated in a motivated cohort does not transfer automatically to a population referred by a distressed employer or an overstretched school.

The screen-time literature is the clearest case, and the correction cuts against the industry's critics too. Riehm et al. reported associations between heavier social media use and internalizing problems among U.S. adolescents in JAMA Psychiatry in 2019; the Surgeon General's advisory summarized evidence linking more than three hours daily with approximately twice the likelihood of poor mental health outcomes. Those are observational findings. Offline adversity, pre-existing distress and reverse causation remain in the model. A blanket device prohibition is a weak instrument; safer design, age-appropriate defaults and literacy are stronger ones.

The Receiving Network That Never Arrived

The United States financed this mistake before. The Community Mental Health Act of 1963 envisaged community centers to receive people leaving state hospitals. Closures proceeded; the receiving network did not arrive at anything like the required scale. What followed was not deinstitutionalization in the clean sense the legislation promised, but a relocation of clinical burden into emergency departments, shelters and correctional facilities. Historians dispute the relative weights of asylum closure, housing policy and welfare retrenchment in that outcome. The narrower finding is durable: an access reform unmatched by funded downstream capacity transfers the problem to whoever is standing nearest.

Buy the Pathway, Not the License

Employers should stop buying licenses and start buying pathways. Demand time-to-first-human-appointment, completed-referral rates and a named clinical escalation route in the contract; require that workforce utilization data be excluded from employment decisions. Local providers should publish conversion rather than volume — how many crisis contacts became an assessment within seventy-two hours. Citizens should apply three mundane tests to any app: is a licensed clinician behind it, what happens when it judges a user unsafe, and can the record leave with you. None requires technical expertise. All are currently answered in marketing copy.

Six Months of Procurement, Not Breakthrough

The base case over six months is a procurement correction rather than a clinical breakthrough. Buyers will shift from per-seat licensing toward contracted care pathways; weaker vendors whose evidence was largely attrition will consolidate or vanish. Debates over school devices will intensify while the workforce shortage goes unfunded. The binding constraint remains the one that constrained the community centers of the 1960s: too few trained people, in the right places, at a price a public budget can carry. Everything else is plumbing.

Primary-source ledger: WHO · U.S. Surgeon General · SAMHSA 988