Volume 6  Number 1  January 24, 2009
Second Opinions

Screening for Disease

Don't speak through the screen door, you'll strain your voice.

Anon.

The idea of screening for disease sounds great: Discover the disease before symptoms appear, thus allowing a dangerous condition-cancer, for example-to be treated early. But does the act of screening save countless lives by revealing unsuspected disease? Certainly, in the case of breast, cervical, and (probably) colon cancer, this appears to be true. But considering the costs and consequences of screening, we must remain skeptical unless conclusions can be verified scrupulously. What we do know is that screening, though it may reveal serious disease, frequently discloses unexpected but benign conditions, a result that often leads to unnecessary, expensive, and sometimes, dangerous testing.

Do we always achieve what we hope for, earlier diagnosis? Even if earlier diagnosis is made, when and how often, in the present state of knowledge, does it change the patient outlook for survival? This would seem to be especially pertinent in the diagnosis, for example of lung, esophageal, pancreatic and possibly colon cancers. Yet no one argues the value of screening for cervical, breast, or skin cancer and a host of other conditions, such as diabetes, hypertension, etc.

It turns out that "lead time bias" is one of the most pervasive problems in proving that screening does what it promises. This simply means we cannot routinely assume that the patient would have died earlier if screening had not been done. Looking at raw statistics and without long term follow-up of control patients who had not been screened, we often draw the wrong conclusions: We have no consistently reliable way of knowing whether patients discovered at screening live longer than unscreened patients who would have died at the same time. In this case no additional life would have been gained through a screening test which did nothing except advance diagnosis. Indeed, there may be added cost in the form of anxiety as the patient must live with knowledge of the disease for longer.

An interesting example of this problem comes out of disputed research on screening for lung cancer, recently reported in the New York Times. Researchers at Weill Cornell Medical did spiral CT screening of 35,000 people with a history of smoking or occupational exposure. 484 were found to have lung cancer, and most of the tumors were removed. The researchers estimated without proof of that assertion, that 92% with early stage tumors would be alive 10 years later, an amazing survival rate compared to the 10% who survive that long after diagnosis discovered without screening. This sounds like a medical "slam-dunk" case for screening, but unfortunately there was no control group of patients who did not receive CT scanning. There was no proof that people who are screened die less frequently or live longer than people who are not screened.

To underscore this point, another study published in 2007 analyzing the results of CT screening in over 3,000 patients showed an increased number of small tumors found and the number of surgeries to remove them, but it did not reduce the total number of lung cancer deaths. A possible explanation is that many tumors would not have killed even if left alone, while the truly lethal tumors were not actually caught earlier. As of now, no major organization recommends widespread use of spiral CT screening for lung cancer, although individual patients may choose to have it.

The best hope of showing the effectiveness of lung cancer screening lies in a large federal trial of 50,000 current and former smokers comparing spiral CT screening with standard chest X-rays to see which "saves" more lives. This will take years to complete. The real issue remains whether it will ever be easy to demonstrate that early diagnosis in certain deadly conditions really prolongs survival unless or until effective new treatments for these conditions are developed.

Martin F. Sturman, MD, FACP

Copyright 2009, Mathemedics, Inc.

EasyDiagnosis is an automated online service that analyses existing medical symptoms and predicts likely causes and conditions. Click here to find out more about this unique service and click here to try one of our modules for free.