The societal and economic cost of research exclusion

The urgency for a paradigm shift is illustrated by the economic and individual toll of long-term schizophrenia. The burden of this illness is driven not only by recurrent psychotic episodes, but also by persistent disability, spanning decades. A recent comprehensive economic evaluation estimated the societal cost of schizophrenia in the USA alone reached $366.8 billion (£270 billion) in 2024 for an estimated 3.07 million adults. ^{Reference Krasa, Baumgardner, Brewer, Chou, Flottemesch and Markowitz2} The direct costs of the disease (such as hospitalisations and medications) constitute only a fraction of this burden. Health care spending makes up less than half of the direct costs and represents ∼9% of the total societal burden. Most of the financial burden, totalling $291 billion (£215 billion) for the year, is driven by indirect costs. These include productivity losses, reductions in quality of life and shortened life expectancy. Furthermore, the hidden cost of long-term care for people with schizophrenia (e.g. unpaid caregivers), represents an often invisible burden, with the economic impact of uncompensated caregiver labour, lost wages and productivity losses accounting for $104.6 billion (£77 billion) per year. Substantial, ambitious research investment, and novel trial design and treatments to prevent and treat TRS will be critical in reducing this burden and provide urgently needed solutions for unmet needs.

Single randomised controlled trials and the problem of representativeness

Traditional fixed two-arm clinical trials are costly, slow to generate results ^{Reference Drazen, Harrington, McMurray, Ware, Woodcock and Woodcock3} and, in populations such as TRS, often struggle to recruit sufficient participants to yield definitive evidence. Innovative trial designs, already widely implemented in oncology and infectious diseases, could offer more efficient approaches by enabling simultaneous evaluation of multiple interventions, adaptive decision-making and more efficient use of participants.

The excluded majority

A recent registry-based study looked at what proportion of patients would meet the eligibility criteria used in a typical randomised controlled trial (RCT) in schizophrenia. They found that approximately 80% of individuals with schizophrenia would be ineligible to participate. ^{Reference Taipale, Schneider-Thoma, Pinzón-Espinosa, Radua, Efthimiou and Vinkers4} The most frequent reasons for ineligibility included serious physical comorbidities, concomitant use of mood stabilisers or antidepressants, lack of mental capacity, a history of substance use and a high risk of suicide. We are thus excluding the very patients who most desperately need effective therapeutic innovations. This poor representativeness is not only an external validity issue; it likely undermines the internal validity of existing trials.

Over past decades, placebo responses in antipsychotic clinical trials have increased. ^{Reference Huneke, Amin, Baldwin, Bellato, Brandt and Chamberlain5} These two issues are conceivably related – by systematically excluding patients with the most enduring, complex and genuinely refractory illness, researchers might inadvertently select for non-representative cohorts with better outcomes (often individuals with milder, transient symptomatic exacerbations). These poorly representative samples are more susceptible to rater biases, regression to the mean and non-specific therapeutic effects, driving up the placebo response. Consequently, identifying the effectiveness of a novel compound requires inflated sample sizes, increasing costs of development to unsustainable levels and discouraging investment in further trials.

A new vision: next-generation trial designs

To break this deadlock, we must move beyond the constraints of traditional, two-arm, placebo-controlled RCTs. The standard paradigm of testing a single drug against a placebo over a short time period is ill-equipped to answer the questions that matter most to patients, caregivers and society for a chronic, and potentially lifelong illness. The complexity of schizophrenia demands equally sophisticated, dynamic and patient-centric trial methodologies.

A modernised, fit-for-purpose trial agenda must establish the following framework:

1. (a) Multi-domain evaluation: Chronic illness cannot be adequately characterised by the reduction of positive symptoms alone. For decades, the gold standard for efficacy has been short-term reductions in total symptom severity or positive symptom subscales. While the stabilisation of hallucinations and delusions remains a clinical imperative, these symptoms alone do not drive the substantial indirect societal burden ^{Reference Krasa, Baumgardner, Brewer, Chou, Flottemesch and Markowitz2} or functional impairment for the individual. Rather, occupational exclusion and loss of independence are rooted in enduring negative symptoms and cognitive impairment. Future trials must position meaningful, clinically relevant multi-domain evaluation at their core. We must design trials, prospectively powered and structured, to evaluate parallel outcomes across these neglected domains. We should incorporate low burden, cognitive batteries, novel scales for primary negative symptoms and standardised, ecologically valid measures of real-world functioning and quality of life. We must design trials that measure what matters to patients, families and society, and co-design such measures. Below are some examples of novel trial designs that could address some of these limitations:

   1. (i) Platform trials and master protocols: Traditional approaches of testing single treatments in isolated trials are slow, financially inefficient and exacerbate placebo response effect. Platform trials such as PUMA (Early Psychosis Multi-arm Multi-stage Platform Trial) could offer an infrastructure-level solution. ^{Reference Park, Siden, Zoratti, Dron, Harari and Singer6} With a shared master protocol and shared control groups, platform trials allow for the simultaneous and sequential evaluation of multiple investigational and repurposed interventions. This reduces the total number of participants exposed to placebo compared with traditional trial designs, accelerates the testing of pipeline agents, allows ineffective arms to be dropped early and provides a permanent infrastructure for clinical discovery. ⁷ While these complex clinical trial designs require sophisticated coordination and demanding logistics, the future integration of Digital Research Environments (DREs) and FAIR (Findable, Accessible, Interoperable and Reusable) data principles are likely to further enable their application in routine clinical settings.

   2. (ii) Sequential multiple assignment randomised trials (SMART): Routine clinical care for chronic schizophrenia is inherently sequential. Initial treatments are followed by reassessments and modifications, such as medication augmentation, or switching and the addition of psychological and other non-pharmacological interventions. SMART designs explicitly mirror this clinical reality: ^{Reference Almirall, Nahum-Shani, Sherwood and Murphy8} rather than testing a single intervention in isolation, SMART trials repeatedly randomise participants at multiple decision points based on their real-time response or tolerability. This approach is particularly powerful for overcoming the limitations of short-term placebo trials, as it allows researchers to systematically evaluate adaptive treatment sequences and identify optimal pathways for patients who fail first-line therapies.

   3. (iii) Adaptive designs and sample size reassessments: Rigidity in trial design often leads to underpowered studies, premature terminations and wasted resources. Adaptive methods should be adopted in psychiatric research. ^{Reference Pallmann, Bedding, Choodari-Oskooei, Dimairo, Flight and Hampson9} Pre-planned adaptations based on accumulating interim data – such as dropping non-effective arms, adjusting patient allocation to favour better-performing treatments, or seamlessly transitioning from Phase 2 to Phase 3 – enhance both ethical acceptability and statistical efficiency. Adaptive sample size re-estimation can rescue trials that encounter unexpected fluctuations in the placebo response, ensuring that studies remain adequately powered despite baseline volatility. There is also the potential use of Bayesian statistical approaches, which formally incorporate prior evidence into analyses, allowing posterior estimates to be updated as new data accrue. In settings where recruitment is challenging and sample sizes are small, such approaches can improve statistical efficiency and enable more informative inference compared with traditional methods that rely solely on data from the current trial.

   4. (iv) Targeted and stratified approaches: One of the key advances needed in mental health conditions are objective markers able to select patient groups most likely to respond to new treatments. The recent success of biomarkers in Alzheimer’s has fuelled significant investment and growth in trials, driven by ‘de-risking’ outcomes, and novel treatments by investigator-led and industry studies – and will change the landscape considerably. The hope of such biomarkers in schizophrenia remains, and it should be noted that Alzheimer’s success has come after decades of sustained investment – which should be matched in schizophrenia and other areas of mental illness. However, there are key areas where advance is within reach: genomics; proteomics, immunology and neurophysiology offer accessible bedside tests with recent potential for use in stratified trials.

   5. (v) More effective support for recruitment to clinical trials and proactive treatment of medication side-effects: Poor recruitment to studies is a major issue. Better infrastructure support to recruitment to clinical trials is needed. Once in trials we also need to make the patient experience a good one. A new approach emphasising patient comfort and the proactive treatment of medication side-effects could reduce withdrawal and drop-out rates.

A shift in mindset

Delivering adequately powered platform and adaptive trials requires more than better methods; it requires a different way of thinking. We need to move from single-site, bespoke studies to standing, research-ready services and multi-country consortia where recruitment is continuous rather than episodic. In parallel, we should mine routinely collected clinical data-sets to support efficient screening, follow-up, pragmatic outcomes and to study long-term trajectories that short trials cannot capture. This depends on investment in data quality, linkage, interoperability and analytic capacity, alongside clear governance and safeguards to maintain trust.

National research infrastructures can catalyse this shift, but funding must match the burden and complexity of severe mental illness. Options include ring-fenced mechanisms or a dedicated mental health funding structure to sustain permanent trial infrastructure embedded in community teams and specialist clinics. Critically, these infrastructures should be designed to connect with large international collaborations. The initial steps signalled by the HTA international schemes (Australia–UK platform studies and Swiss–UK investigator-led trials) are welcome, and they need to be continued and expanded into wider multi-country programmes with shared protocols, aligned approvals and common data standards. This will increase sample size, diversity and generalisability, while improving the speed and reliability of translation into routine care.

In conclusion, schizophrenia research needs to evolve at all levels. Until now we have been restricted to small, exclusionary and short-duration trials heavily compromised by lack of representation of the more severely ill, exaggerating placebo response and often failing to reflect the patients whose illnesses are most severe. Instead, we should seize the opportunities afforded by novel treatments and research methodology and commit to a bold new era of collaborative, adequately powered and methodologically sophisticated clinical trials.

The community is prepared to engage and drive this paradigm shift. We have identified the clinical targets, generated national and international coalitions and are willing to embrace the necessary methodological innovations. It is now time for governments, policymakers and funding bodies to match the staggering societal cost of schizophrenia with an equally ambitious structural investment. Embracing multi-domain evaluations and adaptive trial infrastructure is not merely a scientific aspiration; it is an urgent moral and economic necessity.