Earlier this month, the House of Representatives unanimously passed a resolution asking Medicare and the Veterans Administration to pay for expensive lung cancer screening tests using CT scans. Local hospitals near my home are already jumping on the bandwagon. I suspect that if you check your local hospitals, you'll find that they are championing the tests, too.
The images invoked in the advertising by these entrepreneurial hospitals is enticing, especially if you are among the 45 million Americans who smoke or the millions more who are routinely exposed to second-hand smoke. Over 160,000 Americans die each year from lung cancer, making it the leading cancer killer by far. Moreover, it's an odds-on death sentence. Only 15 percent of people diagnosed with lung cancer live five or more years, one-quarter the rate of other leading cancers. And only one in ten lives for a decade after diagnosis. If you only catch it early, the hospital websites suggest, it should increase your chances of survival because you're catching the cancer when it is "most treatable."
And that's precisely what a major study that appeared in the New England Journal of Medicine last fall claimed to show. Researchers associated with New York Presbyterian Hospital-Weill Medical College, who've been working with financial support from General Electric, Kodak and assorted foundations for decades to offer experimental CT scans to screen for lung cancer, totaled up the results from over 31,000 tests and follow-up treatments in asymptomatic smokers. They found that their screening program raised the 10-year survival rate among those in the group who were identified with cancer to a stunning 88 percent. "Annual spiral CT screening can detect lung cancer that is curable," they concluded in the nation's most prestigious peer-reviewed journal.
The media had a field day. Stories touting the new technology appeared in seven of the nation's ten largest papers, including front page coverage in the New York Times. Four of the five major television networks devoted their precious on-air minutes to this latest medical breakthrough. A patient advocacy group called the Lung Cancer Alliance, which lobbied for the resolution, launched a national advertising campaign featuring legendary quarterback Troy Aikman throwing, not a Hail Mary pass, but a pitch for widespread lung cancer screening. Trial lawyers have brought a number of suits seeking to force the tobacco industry to pay for the pricey procedures.
The campaign suffered a minor setback in March when a government-funded study in the Journal of the American Medical Association appeared to throw cold water on the earlier NEJM study. It claimed screening failed to save lives. But it received much less prominent coverage. The main difference between the two studies, according to stories in the New York Times (pg. 18) and USA Today (pg. 7D), was whether survival time after diagnosis mattered as much as mortality, for which the earlier study provided no evidence.
Does screening actually save any lives? We won't know the answer to that question until the results of a controlled clinical trial sponsored by the National Institutes of Health, which involves 50,000 smokers given either routine CT scans or usual medical care, are in. That won't be for another three years at least.
In the meantime, VA-backed researchers led by Gilbert Welch, Steven Woloshin and Lisa Schwartz at the Dartmouth Medical School are fighting an uphill struggle to educate medical professionals and reporters about the flaws in the original study. I say uphill because their new report, which appeared in this week's Archives of Internal Medicine, received nary a word of coverage in the nation's press - at least not yet.
Survival, their common sense report points out, is always prolonged by early detection. And the only way to determine if screening works is through a randomized trial. Here's their explanation for why the composite results of the screening tests conducted by the New York Presbyterian researchers are essentially meaningless.
First, there is lead-time bias, "a mathematical certainty associated with any successful effort to detect disease early." Whether people live or die, if the screening works, they'll know about it sooner.
Then there's overdiagnosis bias, caused by "screen-detected abnormalities that meet the pathologic definition of lung cancer but will never progress to cause symptoms." Imagine two groups of 100,000 smokers over age 60 who are followed for ten years. One group is screened with CT scans every year. The other group isn't. Each group has 1,000 patients who get lung cancer, 900 of whom die by the end of the decade. But the screened group identifies 5,000 people with lung cancer, what the Dartmouth researchers call "pseudodisease."
"Although this concept may seem implausible to physicians, basic scientists have begun to uncover biological mechanisms that halt the progression of cancer," they write. Tests are also wrong, generating what are in essence false positives.
In our example, that's 4,000 people who automatically live the full ten years. So what are our comparative survival rates? In the group without CT scans, 1,000 contracted lung cancer, 900 died and 100 survived, a 10 percent survival rate. In the CT scanned group, 5,000 people contracted (or were identified as having contracted) lung cancer, 900 died and 4,100 survived, an 82 percent survival rate. Voila! A miracle wrought by screening.
Is this a plausible explanation for the New York Presbyterian study? The Dartmouth researchers offered one more scary piece of evidence, this time from Japan, where researchers used CT screens in more than 5,000 people, about evenly divided between smokers and non-smokers. They found almost ten times as much lung cancer as had been found in the same population when they were given chest X-rays, and, more significantly, they found just as much lung cancer in non-smokers as in smokers! "The Japanese data provide powerful evidence that overdiagnosis can be a substantial problem with spiral CT screening," they conclude.
The Dartmouth researchers spend the rest of their paper pointing out the harms associated with overdiagnosis: the increased biopsies, the dangerous surgical procedures for the mostly elderly population with lung cancer (the death rate during lung cancer operations is 3.6 percent to 6.1 percent for people over 65), and the unnecessary grief associated with false diagnoses.
They conclude with a stern admonition to the New York radiologists, whose financial support from GE and Kodak was buried in a long list of credits, for using a mountain of individual case data to conclude that CT screening can detect cancer that is curable. "While technically true, the conclusion failed to highlight two fundamental unresolved questions: Would screening 'cure' the 160,000 cancers that people die from now? Or would it 'cure' cancers that never needed to be cured?"
For that answer, we'll have to wait for the results of the scientifically-controlled, government-funded study.