Putting Lung-cancer Screening Through the Actuarial Wringer

An article by Milliman actuaries¹ in the April 2012 issue of Health Affairs details an interesting accounting exercise that is likely to cause private insurers to take notice. Using a method employed to evaluate new insurance features, the researchers created an actuarial model designed to estimate the cost (and cost benefit) of lung-cancer screening for US smokers and former smokers, aged 50–64, with at least 30 pack-years of smoking. The results were surprising from a cost perspective, particularly when viewed in the context of what other screening tools cost. The baseline scenario modeled the cost per life-year saved at $18,862, which compares favorably with the costs (in 2012 dollars) of cervical-cancer screening ($50,162–$75,181), colorectal-cancer screening ($18,705–$28,958), and breast-cancer screening ($31,309–$51,274). To calculate the cost of screening for this population—an estimated 18 million US residents (30% of the total US population aged 50–64)—the researchers used published protocols for initial and follow-up screening. They assumed that initial low-dose spiral-CT screenings of patients would initiate diagnostic evaluation about 31% of the time, and that repeated annual screening exams would initiate diagnostic evaluation about 9% of the time (somewhat more often than indicated in other published data). Based on results of the New York Early Lung Cancer Action Project,² they assumed that actual cancers would be detected in 0.6% of initial screenings and 0.2% of repeat screenings. They built the cost of an annual episode of care methodically (estimating the cost of a low-dose CT lung-cancer screening exam based on the ratio of cost differences between screening and diagnostic mammography) and then applied this to a thoracic CT exam. The annual cost per screened patient was converted into a per-member, per-month cost. For the purpose of pricing the rider, the authors assumed that 50% of the target population would actually use the screening program; this is consistent with colorectal-cancer–screening compliance. The cost of the program was spread across the entire commercially insured population. At the center of the cost-benefit calculation was a stage-shift model with a two-year offset, whereby an intervention (screening) shifted the distribution of stages of cancer—in this case, the earlier detection of lung cancer, leading to earlier treatment and lower treatment costs, as well as improved survival. Multiple scenarios were created, including a status-quo scenario (based on what currently happens without screening); a baseline scenario (to show what happens with screening); and several other scenarios that featured different key assumptions, including the number of people screened, the percentage of early-stage cancers detected, and the cost of screening. All of these scenarios assumed that 100% of the target population would be screened. This is unrealistic, the authors acknowledged (and is not the assumption used to price the insurance rider), but a 100% assumption made it easier to calculate costs and benefits, as well as to compare results for alternative protocols and across diseases. The analysis did not consider societal effects such as productivity, tax contribution, disability, life-insurance costs, or the cost of additional survivors entering the Social Security and Medicare programs. Costs and Benefits The team estimated the average annual cost of lung cancer screening to be $247, assuming that 75% of screenings would be repeat screenings and that the insurer cost (spread over a commercial population) would be $0.76 per member per month. The baseline scenario resulted in an additional 130,000 lung-cancer survivors in 2012. Reasons that the cost of lung-cancer screening was lower than that of breast-, cervical-, and colorectal-cancer screening were that much of the evaluation of suspicious nodules occurs without biopsy and that the target screening population is smaller. Another reason for the lower cost of lung-cancer screening is that symptomatically detected lung cancer often is more quickly fatal than other cancers, resulting in a higher number of life-years saved by screening. The team had already completed its research when the National Lung Screening Trial (NLST) Research Team³ published results showing that the use of three annual CT screenings in a high-risk population, aged 55–74, resulted in a 20% reduction in cancer-related mortality (compared with the use of three annual screening chest radiographs). The Milliman team had actually used a higher proportion of early-stage cancers detected, as well as a higher mortality reduction. They speculated that the NLST results might have been even higher if the design of the trial had not required it to be terminated when the 20% improvement threshold was reached. As care is integrated and incentives are realigned, one hopes that reports of such long-view research on where our health-care dollars can yield high returns in health and wellness will not fall on deaf ears. Currently, however, insurance-company actuaries are likely to be most interested in two numbers: the researchers’ estimates of the annual cost of care without screening ($11 billion) and with screening ($15.4 billion). In the minus-zero-sum game moving forward, radiology will be called on to answer the hard questions: Where is the waste that will free up dollars for expenditures that can yield high health-care returns? Cheryl Proval cproval@imagingbiz.com