“I Smoke and I Am Not Going To Quit. My Physician Says I Need a CT scan. Do I?”

Screen Shot 2015-01-09 at 8.51.43 AM


I get asked by many who smoke or who have just quit smoking for help making the decision to have/not have a CT scan to screen for lung cancer. The man responsible for the question above had been smoking his entire life, and, at age 62, he raised the question.

Screening is the term used when tests are done for patients without symptoms. The hope of screening is that a test will find lung cancer (or any other clinical situation for which screening tests are considered) early in the course of the condition so treatment may be beneficial. In the study above, 87 fewer people of the nearly 54,000 in the study died of lung cancer in the LDCT arm. In addition, the number of people dying of other conditions beside lung cancer was fewer (1526 died of other conditions in the LDCT group and 1557 in the CXR group). I did not present this data on the figure as the difference is small and it is unclear why LDCT would reduce other reasons for dying.

The harm of screening, as discussed in earlier blog posts, is that some people will have a positive test and not have cancer. This can cause worry, but in this clinical situation, additionally, the abnormalities found by the test are located in the lung. Getting to these lesions to provide assurance that cancer is not present may be dangerous and costly, and in this study, more people did die early in the LDCT arm.

Since I believe only patients can decide for their tests/treatments, a person would have to trade-off the potential 0.4% added chance of not dying of lung cancer in the future against the potential 40 fold greater chance of a false positive finding and a potential 0.24% added chance dying early or having a major complication by following a LDCT strategy.


  • Benefit is the absolute difference between compared options for care in the percent of people suffering the adverse outcomes associated with a disease . The benefit of screening for lung cancer with LDCT compared to CXR (added percent of people who did not die of lung cancer) over nearly 7 years of the study and follow-up was 0.4% (1.7% died in CXR group and 1.3% died in the LDCT group, 1.7%-1.3% = 0.4%). This means that 250 people would have to undergo the LDCT rather than a CXR to benefit an additional person (249 people will have a LDCT and not benefit). The difference in the number of people who died of lung cancer by undergoing a LDCT rather than CXR was 87 people out of the nearly 54,000 studied (443 in the CXR group; 356 in LDCT group died of lung cancer)


  • Harm is the absolute difference between compared options for care in the percent of people suffering the adverse outcomes associated with the test. Screening may find cancer but may also falsely identify individuals as potentially having cancer when they do not have cancer. The LDCT group, while finding more cancer, also found more false positive findings. The difference, harm, was 16%; 23% in the LDCT group and 7% in the CXR group, (23%-7% = 16%). In other words, for every 6 people undergoing a LDCT strategy rather than a CXR strategy, one additional person would have a false positive finding.
  • People with a false positive finding may require further testing, and, sometimes, a surgical procedure is needed to document if cancer is present or not. In fact, more people died or had a major complication of the work-up for the test findings in the first 60 days following the LDCT than died or had major complications following the CXR strategy. The difference was 0.24%; 0.36% in the LDCT group and 0.12% in the CXR group, (0.36% – 0.12% = 0.24%). Hence, another harm of the LDCT versus CXR strategy for screening is that for every about 420 people having the LDCT, 1 additional person will die or have a major complication in the first 60 days following the test.


  • The harm: benefit ratio for the false positive findings is 40: 1 (LDCT leads to 16% more false positive results than a CXR and this harm must be balanced against the benefit of 0.4% less chance of dying of lung cancer over about 7 years; 16%: 0.4% = 40: 1). Hence, harm is 40 times more likely. You would have to gain at least 40 times the value from not dying of lung cancer in 7 years than lose from the worry and work-up of a false positive finding near the time of the test.
  • The harm: benefit ratio for dying or having a major complication within 60 days of the screening test is 1: 2 (0.24%: 0.4% = 1: 2). Hence, benefit is 2 times greater relative to the harm of dying or having a major complication within 60 days of the screening test. However, the difference in this harm: benefit ratio is that, while a lower percent chance of dying early during a work-up for a positive test finding (harm) occurs than dying later in life from lung cancer (benefit), dying earlier in your life rather than later must be considered. If the value to gain from not dying early rather than later is at least 2 times greater than not dying later in life, then this harm must also be considered in your decision-making.


The study presented above began in 2002 and follow-up was to 2009. Thirty-three medical centers and many physicians at those centers contributed patients to the study. 53, 454 people ages 55-74 were randomized to LDCT versus CXR. A review of this study and others on screening for lung cancer is included. http://jama.jamanetwork.com/article.aspx?articleid=1163892.

Making medical decisions is difficult when the absolute differences in outcomes are small. Small differences  in outcomes are hard to wrap our heads around, and small differences increase our concerns about the veracity of any study.  In studies with small differences in outcomes, it is crucial that only the absolute differences are communicated. Unfortunately, even academic and government leaders communicate the results of this study in relative number terms (20% relative reduction in the chance of dying of lung cancer and 6.7% relative reduction in the chance of dying). These numbers are not helpful, only the absolute differences can inform you of the trade-off you must make.

One last comment; the comparator test in this study was the CXR. That test has been shown to be no better for patients than having no test (randomized trials in over 150,000 patients failed to show a benefit to CXR). So, if you decide against having a LDCT scan for screening for lung cancer, having a CXR is not an alternative.

Robert McNutt


Categories: Uncategorized

5 replies »

  1. A paper recently published online shows the benefits of smoking cessation AND screening (yes it is a post-hoc analysis of the NLST dataset and not a true prospective trial), also the benefits of smoking cessation even in the CXR group. A low-dose CT is easy to order, it is challenging to get people to change their behavior, to make abstract statistics visceral for an individual. http://www.ncbi.nlm.nih.gov/pubmed/26502000 (Full disclosure: I was involved in this work)

  2. My lastp ost was a rant, an accurate one if I may say so myself, on the politics, philosophy and economics of screening.

    McNutt does bring an interesting point.

    That about early demise versus prolongation of life. It’s an interesting trade-off since one can’t enjoy both – i.e. die sooner and live longer.

    My only quibble is the imprecision surrounding the effect, or bad effect size. As we are talking about transparency surely patients must be informed of the p values and confidence intervals surrounding the point estimates.

    No? That’s just paternalism.

  3. Before ordering ANY test, you as the physician need to know why you are ordering it and what you are going to do differently depending upon the results. (this is the biggest issue I have with mid-levels, who are good at putting pen to paper and ordering CT’s of head/neck/chest/abd/pelvis both with and w/out contrast, then want me to interpret what the results mean, but I digress . . .)

    Bottom line, maybe have a talk with your patient about what they want/expect IF the CT shows a mass. Are they willing to have a biopsy, which is moderately invasive, depending upon location of the lesion?

    Are they willing to undergo chemotherapy and/or surgery, depending upon the pathology? If your pugnacious patient says, “hell no, not quitting my cigs, and I don’t want no damn chemo or radiation,” then why exactly are you ordering a CT??

    As the article pointed out, “screening” sounds like mom & apple pie. How can anyone be against “screening”?? But screening should be individualized, as the article suggests. My father, a 3-4 pack a day smoker for 30 years, died of lung cancer, diagnosed (@ stage IV) when he was 59. Getting a CT at 60 wouldn’t have helped him. A physician friend had prostate cancer in his mid 40’s. (He’s alive and well, sans prostate) The “Choosing Wisely” campaign would have told him not to get a PSA.

  4. I doubt the irreverent courageous Christopher Hitchens would have opted for screening CT for lung cancer. He was a defiant smoker who smoked through his esophageal cancer.

    The lure of screening is a sign of fatalism being replaced by scientific determinism. No don’t blame the medical-industrial complex. We are chicken littles. Scared of antibacterial soap. Scared of power lines. Scared of environmental toxins. Scared like never before.

    We want to live, live, live. Live at any cost. So death has become expensive. We are medicalized in sickness. We have become medicalized in health.

    In such circumstances, highlighting the false positives or overdiagnosis as one of the harms of lung cancer screening is an academic exercise. For the frightened it will do little to dissuade them from being screened.

    For the live free or die brigade, who are the biggest myth since Big foot, if they will be perturbed by FP, they will be even more concerned by the disease they may be falsely accused of harboring.

    Good reading though.