Lead-time bias occurs when there appears to be a decrease in mortality because of screening and “early detection”, when in fact the clock just started ticking earlier.

Take 2 identical smokers. One has a CT today which shows a lesion, he undergoes treatment, and dies in five years. The other has no CT and develops a bloody cough in 3 years, dying two years later. If lead-time bias is ignored, the CT-screened patient had a 5 year survival, while the non-screened patient had only a 2 year survival. In reality, they are identical, and died on the same day, but the CT-screened patient underwent three additional years of invasive surgery and chemotherapy.

This has been shown on several studies of radiologic screening of smokers (better called “case-finiding” since the patients are being selected).

Even though it may be counter-intuitive, it is what it is. Also, “highly suggestive” data is just that, suggestive, not conclusive or even significant. I believe there was highly suggestive data that Vioxx did not increase the risk of cardiovascular events, either.

There are lots of reasons prostate antigen testing is of questionable usefulness, form the lack of specifictiy of the test to the natural history of prostate cancer – and CT screening of a select population of heavy (former or current) smokers is a horse of a different color. The signifigance of a lung lesion in a heavy smoker is not the same as that of an elevated PSA in the general population of males.

Kevin’s flat statement that there is no data that routine CT screening improves mortality really ought to be qualified. There is data highly suggestive that there is improvement in mortality in that subset of the population that has smoked, or smokes, heavily, and demonstates an early lung lesion detected by a screening CT.

If managed properly a screening program such as this would be a possible way to GET evidence, in a willing population.

Whether a company that actively encouraged people to minimize smoking harms should pay for such tests is obviously up for debate, and it’s reasonable to argue that foolish people should pay for their own foolish choices. But as far as the intrinsic use of screeing, it would without any doubt be of use in this select population. My argument is essentially that improved mortality is not the only measure of whether this testing is of benefit to heavy smokers. There are other benefits to be gained that are worthy of consideration.

So damn the evidence? The harm is not to those who have disease, it is to those who do not. How many “healthy” people should have open lung biopsies because of false-negative exams in order to find one true-positive lesion, treatment of which has not been shown to prolong life?