Effectiveness and cost-effectiveness of a cardiovascular risk prediction algorithm for people with severe mental illness (PRIMROSE)

OBJECTIVES: To determine the cost-effectiveness of two bespoke severe mental illness (SMI)-specific risk algorithms compared with standard risk algorithms for primary cardiovascular disease (CVD) prevention in those with SMI. SETTING: Primary care setting in the UK. The analysis was from the National Health Service perspective. PARTICIPANTS: 1000 individuals with SMI from The Health Improvement Network Database, aged 30–74 years and without existing CVD, populated the model. INTERVENTIONS: Four cardiovascular risk algorithms were assessed: (1) general population lipid, (2) general population body mass index (BMI), (3) SMI-specific lipid and (4) SMI-specific BMI, compared against no algorithm. At baseline, each cardiovascular risk algorithm was applied and those considered high risk (> 10%) were assumed to be prescribed statin therapy while others received usual care. PRIMARY AND SECONDARY OUTCOME MEASURES: Quality-adjusted life years (QALYs) and costs were accrued for each algorithm including no algorithm, and cost-effectiveness was calculated using the net monetary benefit (NMB) approach. Deterministic and probabilistic sensitivity analyses were performed to test assumptions made and uncertainty around parameter estimates. RESULTS: The SMI-specific BMI algorithm had the highest NMB resulting in 15 additional QALYs and a cost saving of approximately £53 000 per 1000 patients with SMI over 10 years, followed by the general population lipid algorithm (13 additional QALYs and a cost saving of £46 000). CONCLUSIONS: The general population lipid and SMI-specific BMI algorithms performed equally well. The ease and acceptability of use of an SMI-specific BMI algorithm (blood tests not required) makes it an attractive algorithm to implement in clinical settings.