Universal screening with computed tomography for lung cancer? Finally a randomized trial…but what to do??
11 Jul, 11 | by Dr Richard Saitz, Editor of Evidence-Based Medicine
Expert opinion and observational studies have favored lung cancer screening but trials (of plain xrays) have not shown benefit, until now. For years, many have been asking for a randomized trial. Now that the results of the (US) National Lung Screening Trial (NSLT) have been published, it reminds me of the admonition to “be careful what you ask for, you just might get it.”
The randomized trial compared three screenings, either low dose CT scans or plain chest radiographs and adherence to the screening protocol was >90%. How to address a positive test was left to clinicians outside the trial. It enrolled 53 454 participants age 55-74, who smoked at least 30 pack-years currently or who had quit in the past 15 years, and had not had lung cancer or a recent chest CT scan, hemoptysis or unexplained weight loss. The authors estimate only 7% of US current or former smokers would meet these criteria. So the study results do NOT apply to 93% of smokers. One wonders though, to whom they will be applied (paying customers?).
What were the results? Amazingly, despite the publication source (New England Journal of Medicine, http://bit.ly/qKs6Lq) and widely agreed upon reporting guidelines, the main results are presented as relative reductions, making them seem large (a 20% decrease in death from lung cancer). I don’t mean to minimize though, just to be clear—it is important that there was a reduction, and more lung cancers were diagnosed in the (low dose) CT group, which had fewer lung cancer deaths. The absolute reduction was the difference between 356 lung cancer deaths in 144,103 person-years in the CT group and 443/143,368 in the plain x-ray group. The absolute risk of lung cancer death among those screened at least once was 1.3% in the CT group, 1.6% in the x-ray group, an absolute risk reduction of 0.312%, for a number needed to screen with 3 tests of 320 to prevent 1 lung cancer death over 7 years. Overall mortality was also reduced (by 0.5%). Complications of evaluation of a positive test were 1.4% in the CT group, 1.6% in the x-ray group.
The vast majority of positive tests were false positives (96% in the CT group, 95% in the x-ray group).
None of this addresses long term cumulative population harms from radiation exposure or any impact on likelihood of quitting smoking.
What to do? Many may reach different conclusions depending on the presentation of the results. Others will take these results and then add patient values and preferences. Others still may wish to wait to see the cost effectiveness analyses.
What will you do? You now have the evidence…