Cost-Effectiveness of Hepatocellular Carcinoma Surveillance: An Assessment of Benefits and Harms

Abstract

INTRODUCTION: The value of hepatocellular carcinoma (HCC) surveillance is defined by the balance of benefits, i.e., early tumor detection, and potential harms, related to false positive and indeterminate results. Although physical harms can be observed in 15%-20% of patients with cirrhosis undergoing HCC surveillance, previous cost-effectiveness analyses have not incorporated costs of harms. We aimed to evaluate the cost-effectiveness of HCC surveillance including both benefits and harms. DESIGN: We constructed a Markov model to compare surveillance strategies of ultrasound (US) alone, US and alpha fetoprotein (AFP), and no surveillance in 1 million simulated patients with compensated cirrhosis. Harms included imaging and biopsy in patients undergoing surveillance for HCC. Model inputs were based on literature review, and costs were derived from the Medicare fee schedule, with all costs inflated to 2018 dollars. The primary outcome was the incremental cost-effectiveness ratio per incremental quality-adjusted life-year. RESULTS: In the base case analysis, US with AFP was the dominant strategy over both US alone and no surveillance. In a probabilistic sensitivity analysis, US with AFP was the most cost-effective strategy in 80.1% of simulations at a willingness-to-pay threshold of $100,000 per quality-adjusted life-year. In our threshold analyses, an HCC incidence >0.4% per year and surveillance adherence >19.5% biannually were necessary for US with AFP to be cost-effective compared with no surveillance. DISCUSSION: Accounting for both surveillance-related benefits and harms, US and AFP is more cost-effective for HCC surveillance than US alone or no surveillance in patients with compensated cirrhosis.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 161 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.