Annual Lung CT Scans for Smokers: Good or Bad?

When it comes to medical studies, I can't help but fall back upon that quote often attributed to Benjamin Disraeli by Mark Twain about "lies, damned lies, and statistics".  If you've been following some of my recent posts about number needed to treat, absolute vs relative risk reduction, and clinical outcome vs physiologic markers, it's pretty clear why the American public doesn't trust physicians & scientists very much since our research conclusions tend to vacillate with the wind, based upon how the results are presented.

Two months ago, a study was published in the New England Journal of Medicine that followed 53,454 persons at high risk for lung cancer who were randomized to either annual low-dose CT scans or single view chest xrays and followed for 5 years.  When all was said and done, the authors concluded that annual low-dose CT scans would reduce lung cancer deaths by 20%.

Of course, this news was immediately trumpeted around the world back in November 2010 when the trial was originally stopped as the next great thing to save all the smokers from their self-induced lung cancers.  At the time, all the statistics weren't available as the researchers were still crunching the numbers.  However, that didn't stopped hospitals and others from advertising low cost lung CT scans for heavy smokers, which some critics are now calling a marketing ploy despite one leading specialist claiming that "the data is pretty compelling" in terms of saving lives.

However, when you look at the numbers, 99.5% of those who received annual CT scans received no benefit while just 0.5% (absolute risk reduction) were helped by preventing death (20% relative risk reduction).  In other words, we would need to scan 217 persons to save one person from dying.  On the other hand, 23% of those scanned would have a false positive in terms of a cancer scare; 3.5% would have to undergo an otherwise unnecessary surgical procedure to clarify the false positive cancer scare; and 0.6% of those scanned would suffer a surgical complication.

From the NNT perspective, one in four scanned would receive a false cancer scare, most of whom would should benign changes upon repeat scan.  However, one in thirty would have to undergo an otherwise unnecessary surgery to prove that they didn't have lung cancer.  And finally, one in one hundred sixty one would suffer an otherwise unnecessary surgical complication.  Is this risk really worth it?

And of course, this doesn't even take into account who's going to pay for all these scans . . .