As health-care costs escalate, health-care planners must determine how the allocation of health-care dollars should be prioritized. One approach is to assess the cost of achieving a quality-adjusted year of life and then allocating the dollars in descending order, from least to most expensive, until all available money has been expended. Of course, calculating the cost per life-year is the real challenge because it is usually determined from mathematical decision models, which include many assumptions that may be subject to criticism.
Such is the case for the prostate cancer screening cost per life-year as determined by Benoit and Naslund. They suggest that the cost may be under $5,000, which would make it the most cost-effective medical intervention that money can buy! By comparison, screening for cervical cancer every 4 years is estimated to cost $17,000 for a quality-adjusted life-year, compared to almost $50,000 per life-year for breast cancer screening.
Detecting Insignificant Disease
In support of their argument, the authors conclude that insignificant prostate cancers are not increasingly detected by screening. Unfortunately, no known histologic feature of prostate cancer can predict the natural history of that tumor or the remaining life expectancy of any patient. Over the last 6 years of prostate-specific antigen (PSA) screening, almost one million additional cases have been detected. Based on an average annual mortality of 40,000, this would mean that either deaths from prostate cancer should almost be eliminated over the next 25 years or that a large number of nonlethal cases have been diagnosed. On what basis do the authors justify their statement that these men would have eventually required treatment?
The authors criticize previous estimates of screening costs because not all men will be screened, undergo follow-up tests, or even undergo treatment once a cancer is detected. Actually, these events increase the cost per quality-adjusted life year because some men who were at risk of dying will not be prevented from doing so.
Another argument offered in support of the cost-effectiveness of prostate cancer screening stems from a critique of previously published decision models and a reanalysis using new data. Unfortunately, these reanalyses overlook the fact that the Markov models are based on the mortality from conservative management. Lacking reliable data, however, the rate of progression was used to calculate mortality.
Uncertain Benefit from Screening
Subsequently, two meta-analyses were published that actually provide mortality data, and, based on these results, the anticipated benefit from screening is much reduced. Another problem is the principal assumption made by the authors in estimating the benefit of screeningthey assume that men with confined cancer benefit from screening even though 10% to 15% will still ultimately die of their disease. This fact, combined with the unlikelihood that no insignificant cancers will be detected, means that the probable number of years regained will be less than the number that would have been lost.
Finally, any cost analysis that fails to include time lost from work and lost productivity is not truly estimating the total cost impact for society. Granted that such costs are difficult to ascertain, nevertheless, without them, any comparison with other health-care endeavors is simply unreliable.
In summary, as the authors acknowledge, screening for prostate cancer will be a costly endeavorsubstantially more than what they estimated. Until the cost can be more accurately determined, however, allocating health-care dollars to an endeavor that seems to provide uncertain benefit is imprudent.