Sunday, March 15, 2009

Intuition and Cancer Screening


The latest issue of the Annual Review of Medicine has this interesting article on the role of science and intuition in cancer screening. Biases can often mislead clinicians when determining what the efficacy of screening tests alone are. This nicely illustrates the many diverse and complex challenges we face with respect to the duty to prevent harm. Here is a sample:

.... In 1924, the New York Times reported a call to public action by Johns Hopkins surgeon Dr. Joseph Colt Bloodgood, who asserted, “Deaths from cancer would be practically eliminated…if persons afflicted sought medical aid immediately upon the discovery of a foreign growth in any part of the body” (1). He obviously overestimated the potential impact of available early-detection strategies. Nevertheless, the development and application of new medical technologies have accelerated the actualization of this concept.

.... Despite the strength of the messages transmitted to the public about the value of cancer screening, cancer mortality statistics remain sobering. Cancer remains the second most common cause of death in the United States, accounting for 23% of all deaths and dwarfing the third most common cause, stroke, at 6% (see Reference 4, table C). There has been clear progress, but it has been incremental. There has been a decrease of about 10 deaths due to cancer per 100,000 persons per year between 1950 and 2005 (194 versus 184 deaths per 100,000 persons) (5). Population trends reflect a mix of changes in exposures, treatment advances, and screening, so they cannot be used to draw any definitive conclusions about the contribution of early-detection strategies. Nevertheless, the population trends contrast with many public perceptions and clinical intuitions about the magnitude of efficacy of cancer screening.
A core question is, how could Dr. Bloodgood's clinical intuitions and observations have been so misleading? And how can we use scientific methodology to protect us from our strong intuitions? We propose that large-scale randomized trials come to the rescue.

.... Early detection advances what would have been the original date of diagnosis to an earlier point in time, but it does not necessarily follow that the patient's time of death will be delayed. For example, if a particular disease has no known treatment, earlier detection can have no impact on the lifespan of an affected person. In Figure 2a, the two lines represent two lifespans. In the one case, cancer is detected through symptoms, and the person dies at a set point. In the second case, the cancer is detected while asymptomatic through screening. The proportion of the lifespan affected by disease has been extended—that is, the person is a patient longer—but total years of life remain exactly the same.

.... Although clinical intuition is a fundamental “art” of medicine, in the field of cancer screening it is easy to be misled. Powerful, pervasive biases make reliance on experience alone a dangerous strategy. Successful evaluation of early-detection efforts requires strict adherence to the scientific method to protect us from simply ratifying our desires. As Roman playwright Terence noted, “One easily believes what one earnestly hopes for.” We should harness this passion to generate evidence as strong as our messages. At the very least, we should be aware that soundbites can do injustice to complex trade-offs when proposing cancer screening tests to a healthy population.

Cheers,
Colin