Colorectal cancer (CRC) is a major cause of morbidity and mortality in the US and worldwide. Data from the American Cancer Society suggest that, in 2005, over 145,000 people in the US will develop CRC and over 56,000 will die from this disease. Furthermore, the WHO recently estimated that 945,000 new cases of CRC and 492,000 CRC-related deaths will occur annually worldwide. A person's lifetime risk of CRC is 56%. The majority (>60%) of CRC is detected at the nonlocalized stage, and the 5-year survival rate is only 64% in the US and even lower in other countries.
Fortunately, CRC and CRC-related mortality are preventable. Randomized trials suggest that screening with fecal occult blood testing (FOBT) in adults over 50 years of age can reduce the incidence of CRC by 1720% and CRC mortality by 1633%. Case-control studies suggest that other screening methods, such as sigmoidoscopy or colonoscopy, are also effective; however, these methods have not yet been compared in head-to-head studies. Moreover, screening is currently underutilized in the US: up-to-date screening has reportedly been carried out in only ~50% of age-eligible adults.
Because CRC screening has been shown to be effective, many healthcare systems worldwide are considering implementing organized screening programs. As policy makers consider whether to implement widespread screening, they must consider the value of CRC screening relative to other health-related and non-health-related needs. One method for doing so is cost-effectiveness analysis. Several high-quality studies have examined whether CRC screening is cost-effective compared with no screening. A critical examination of these analyses can help elucidate the relative value of CRC screening compared with other health-related programs. This article highlights some of the key issues involved in studying the cost-effectiveness of CRC screening and suggests areas for further research. I focus mainly on studies conducted in the US, but analyses conducted in Europe have generated similar results.
Cost-effectiveness analyses have found that screening for CRC with any of several common screening tests (annual FOBT, sigmoidoscopy every 5 years, colonoscopy every 10 years, or annual FOBT combined with sigmoidoscopy) is both effective and cost-effective compared with no screening. Most cost-effectiveness analyses of CRC screening have found the cost per life-year gained to be less than US$35,000, which compares favorably with many other commonly recommended health interventions, such as mammography for women aged over 50 or treatment of moderate hypertension.
These findings seem to stand up to reasonably pessimistic assumptions about the effectiveness of CRC screening. Existing cost-effectiveness analyses, however, have not reached consistent conclusions with respect to which testing strategy is the most effective or cost-effective. Data limitations and differences in the modeling of CRC's natural history, test accuracy, effectiveness of screening tests for preventing CRC and CRC-related mortality, and patients' long-term adherence, limit the ability to make precise and accurate comparisons between testing strategies. The limitations come in three main forms. First, many screening models fail to include a potentially important parameter (e.g. most existing models don't account for the increased costs and risks associated with the removal of nonadenomatous [usually hyperplastic] polyps, which could increase costs considerably). Second, analyses might include the same type of parameter but differ in the value assigned to it (e.g. the cost of colonoscopy varies widely across different cost-effectiveness analyses). Third, differences in the model structure (e.g. microsimulation versus Markov analysis) or outcomes considered (e.g. models that examine the effects of 10 years of screening compared with models examining effects over 20 years), can make comparisons across models difficult.
Only one study has attempted to quality-adjust its outcomes. Providing outcomes in terms of quality-adjusted life-years accounts for the non-mortality-related benefits and adverse effects, and allows better comparison across health conditions.
A further limitation of existing cost-effectiveness analyses is that none have adequately measured the true costs associated with performing a screening program. An exhaustive analysis would include infrastructure costs (required to assure adequate adherence and follow-up) and the cost of a person's time devoted to screening and follow-up. For example, the cost of screening using FOBT is often estimated using the amount of money reimbursed by the medical insurer for the test, a value that might not accurately represent the true costs in an analysis conducted from the societal perspective. A better cost estimate of FOBT would include the cost of identifying eligible patients, test materials, labor required to carry out and explain the testing process, labor required for the laboratory development of the test, and costs related to reporting the results, arranging follow-up for positive tests, and assuring that the person completes the examination.
Finally, existing cost-effectiveness analyses do not adequately account for the inaccuracies that are inherent within them. Although extensive one-way sensitivity analyses are often reported in the existing cost-effectiveness analyses, the inability to represent the higher-order uncertaintycaused by combining multiple uncertain variables together in a modelruns the risk of creating 'pseudocertainty' for consumers and policymakers. For example, if policymakers compare a cost-effectiveness ratio of $35,700 per quality-adjusted life-year saved with one technology, against a cost-effectiveness ratio of $40,000 per quality-adjusted life-year saved with another intervention for a different condition, they might be tempted to consider the first intervention to be superior. If, however, these estimates are imprecise, as they are likely to be for CRC screening, such a conclusion might be in error. Bayesian probabilistic sensitivity analysis techniques, which utilize simulations to account for CREDIBLE INTERVALS, can help estimate quantitatively this type of uncertainty.
In summary, current evidence provides support for the cost-effectiveness of CRC screening compared with no screening, but does not provide sufficiently strong evidence to determine which testing strategy should be preferred. When interpreting these data, it is important to remember that existing analyses have not accounted for certain costs that should at least be considered when evaluating the implementation of a screening program; in particular the costs of an individual's time and the costs needed to set up, support and sustain a screening program in practice. Existing cost-effectiveness analyses also presume that clinicians will follow up patients with diagnostic and treatment measures to produce the reductions in disease incidence and mortality that account for the benefits of screening. If, for example, positive FOBT is not followed by accurate and safe colonoscopy, the screening program will not be effective; conversely, if screening leads to overuse of surveillance colonoscopy, as some data from the US suggest, screening costs might be higher than predicted.
At present several developed countries are considering whether to implement CRC screening programs. Although the available evidence regarding cost-effectiveness compares well with other commonly utilized services, other factors, such as total program costs, start-up requirements, and the availability of enough trained practitioners, must also be considered. These factors might vary across different healthcare systems; developing appropriate cost-effectiveness data for a given healthcare system might require adjustment of key parameters, particularly cost data.
Click here for subscription information about this journal.