Cost-consequence analysis of a criterion-based screening and triage pathway compared with full ADHD assessment for adults referred to NHS secondary care
TL;DR
The CASQ-first pathway may reduce assessment costs and redirect a substantial proportion of referrals away from a direct full ADHD assessment, but the present validation study is not adequately powered to support a definitive economic conclusion.
Abstract
Background National evidence indicates that attention-deficit hyperactivity disorder (ADHD) referrals, recorded diagnoses and waiting lists have increased substantially. The Comprehensive ADHD Screening Questionnaire (CASQ), a criterion-based tool delivered by trained Band 7 clinicians, achieved 100% sensitivity, with a 95% confidence interval of 61.0–100.0%, while screening out 39.6% of referrals in a prospective NHS validation study of 48 participants. This analysis compares the costs and diagnostic consequences of the CASQ-first pathway with full ADHD assessment from the NHS perspective. Methods A prospective observational study-based cost-consequence analysis was conducted from the NHS perspective over a one-year time horizon, following the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) 2022. A non-inferiority margin of −10 percentage points, corresponding to sensitivity of at least 90%, was specified to assess diagnostic safety. The CASQ assessment-and-triage cost was £120.27 per referral, derived bottom-up from Personal Social Services Research Unit (PSSRU) 2024 unit costs. The standard or low-complexity full ADHD assessment cost was £850, taken from the final NHS England 2026/27 Adult ADHD Assessment, Face-to-Face guide price. The weighted-mean full ADHD assessment cost was calculated using South West Yorkshire Partnership Teaching NHS Foundation Trust (SWYPFT) case-mix proportions from a previous explainable clustering analysis, giving £1, 011.28 per referral. Sensitivity analyses examined alternative cost assumptions, proportion screened out, CASQ duration and incremental clinical time for more complex full ADHD assessment cases. An expanded structural probabilistic sensitivity analysis with 10, 000 Monte Carlo simulations separately propagated uncertainty in complexity-cluster proportions, incremental Band 7 clinical hours, CASQ assessment-and-triage cost, standard full ADHD assessment cost, Band 7 hourly cost and diagnostic accuracy parameters. Results The CASQ achieved 100% sensitivity and 100% negative predictive value in the validation study. In the SWYPFT case-mix weighted-mean base case, the CASQ-first pathway cost £731.08 per referral compared with £1, 011.28 for universal full ADHD assessment, a saving of £280.20 per referral and £598, 507 annually at 2, 136 referrals. In commissioning terms, this corresponds to an estimated saving of approximately £280, 200 for every 1, 000 referrals managed through the CASQ-first pathway under the SWYPFT case-mix weighted base case. In the expanded structural probabilistic sensitivity analysis, the CASQ-first pathway was cost-saving in 100.0% of 10, 000 simulations, with a mean saving of £285.18 per referral and £609, 143 annually at 2, 136 referrals. CASQ sensitivity met the specified 90% sensitivity threshold in 75.0% of simulations. The mean projected false-negative rate was 9.4 per 1, 000 referrals. Conclusion The CASQ-first pathway may reduce assessment costs and redirect a substantial proportion of referrals away from a direct full ADHD assessment, but the present validation study is not adequately powered to support a definitive economic conclusion. Because the economic result rests on a sensitivity estimate derived from only six confirmed ADHD cases, the projected per-referral and national savings are illustrative planning estimates and not established savings. The economic estimates presented here are preliminary and conditional and should be regarded as a commissioning planning aid that becomes interpretable only when independent multi-site validation has established a reliable diagnostic safety margin. Decision-makers should weigh the estimated per-referral cost reduction against diagnostic accuracy uncertainty and potential harm to missed cases. Independent multi-site replication with formal non-inferiority testing is required before unconditional cost-minimisation conclusions can be drawn.