The cost-effectiveness of prostate cancer screening using the Stockholm3 test

Abstract

Objectives: The European Randomized Study of Screening for Prostate Cancer found that prostate-specific antigen (PSA) screening reduced prostate cancer mortality, however the costs and harms from screening may outweigh any mortality reduction. Compared with screening using the PSA test alone, using the Stockholm3 Model (S3M) as a reflex test for PSA ≥ 1 ng/mL has the same sensitivity for Gleason score ≥ 7 cancers while the relative positive fractions for Gleason score 6 cancers and no cancer were 0.83 and 0.56, respectively. The cost-effectiveness of the S3M test has not previously been assessed.

Methods: We undertook a cost-effectiveness analysis from a lifetime societal perspective. Using a microsimulation model, we simulated for: (i) no prostate cancer screening; (ii) screening using the PSA test; and (iii) screening using the S3M test as a reflex test for PSA values ≥ 1, 1.5 and 2 ng/mL. Screening strategies included quadrennial re-testing for ages 55-69 years performed by a general practitioner. Discounted costs, quality-adjusted life-years (QALYs) and incremental cost-effectiveness ratios (ICERs) were calculated.

Results: Comparing S3M with a reflex threshold of 2 ng/mL with screening using the PSA test, S3M had increased effectiveness, reduced lifetime biopsies by 30%, and increased societal costs by 0.4%. Relative to the PSA test, the S3M reflex thresholds of 1, 1.5 and 2 ng/mL had ICERs of 170,000, 60,000 and 6,000 EUR/QALY, respectively. The S3M test was more cost-effective at higher biopsy costs.

Conclusions: Prostate cancer screening using the S3M test for men with an initial PSA ≥ 2.0 ng/mL was cost-effective compared with screening using the PSA test alone.

Fig 1. Prostate cancer screening interventions using PSA test and the S3M reflex test.

Fig 2. Schematic of the prostate cancer natural history model.

Individuals are assumed to be disease-free at age 35 years. They may progress to preclinical cancer states with a fixed Gleason score, with progression by T-stage and to metastatic cancer. Preclinical cancers may be diagnosed from nine different states, with survival from prostate cancer death evaluated from the possibly counterfactual time of clinical diagnosis. Death due to other causes is represented as a competing event.

Fig 3

Panel (A) shows the cost-effectiveness plane comparing no screening, PSA screening and S3M screening with reflex thresholds at 1 ng/mL, 1.5 ng/mL and 2 ng/mL. The effectiveness and costs are incremental relative to PSA screening and discounted at 3% per annum. Panel (B) shows the ICER (€/QALY) for S3M as a reflex for PSA above 1, 1.5 and 2 ng/mL as functions of the unit cost of S3M. The cost-effectiveness ratios are incremental to PSA screening alone and are discounted at 3% per annum. The current S3M unit cost, of €196, is shown as the grey line. Finally, Panel (C) shows the probability that an intervention is more cost-effective than PSA screening under parameter uncertainty for a specific willingness to pay threshold.

Fig 4. One-way sensitivity analysis showing the effect of no and high discounting rates, the S3M test performance, high and low biopsy costs, 20% variation in all costs and health state value decrements on the cost-effectiveness.

From the left, the dashed lines show the limits for low (less than €8,300), moderate (€8,300–41,600), high (€41,600–83,300), and very high costs (over €83,300) for Sweden. The bounds for the biopsy costs, together with the urology assessment, varied between €330 and €880.