Background The effectiveness of exercise interventions to improve activities of daily living function in people with dementia is inconclusive. This study aimed to assess the long-term cost-effectiveness of the PrAISED intervention from a National Health Service (NHS) perspective.

Method This novel robust economic analysis used a Markov model to evaluate the incremental cost-effectiveness ratio (ICER) over a lifetime horizon of 15 years for a cohort of patients. Sensitivity analyses were conducted to investigate the uncertainty and robustness of high-impacting parameters and results.

Results This study included 365 adults, aged 65 years and above with 183 and 182 randomised to the PrAISED and standard care groups respectively. The PrAISED intervention had mean per-patient cost of £60,465 for the PrAISED arm and £54,604 for standard care. The Praised intervention gained an incremental QALYs of 0.05 resulting in an ICER of £129,614 per QALY. The sensitivity analysis of the intervention cost varied the ICER value between £68,173 and £191,054/QALY. To achieve the recommended NICE willingness to pay threshold value of less than £30,000/QALYs would require the intervention cost to be reduced from £1,236 (current cost) to £263 to break even and be cost-effective. The sensitivity analyses revealed that there was a 40% probability of standard care dominating the PrAISED treatment.

Conclusions Although the PrAISED intervention was a low-cost intervention, it did not produce a cost-effective intervention in this analysis. The flexibility of the PrAISED program to adapt to government policy during the COVID-19 pandemic was positive.

Trial registration ISRCTN15320670