In a rather unusual action, the National Cancer Institute (NCI), a division of the National Institutes of Health (NIH), issued a press release on November 4, 2010 announcing preliminary findings from a research study that, at that time had yet to be published (it since has been). This bulletin, Lung cancer trial results show mortality benefit with low-dose CT, announces that a large, multi-center, randomized controlled trial (RCT), called, the National Lung Screening Trial (NLST), has found that regular screening of current and former heavy smokers with low-dose chest computed tomography (CT) scanning aged 55-74, compared to screening by regular chest x-ray, led to 20% fewer deaths from lung cancer. It is unusual in that it is there is no associated published study in a journal describing these results (the statement says that it is being “prepared for publication in a peer-reviewed journal within the next few months”); the only concurrently published article is a description of the methods of the NLST study, with discussion of previous screening studies for lung cancer, is the “National lung screening trial: overview and study design” published in the November issue of Radiology and made available November 2, 2010.
The study comparing these two screening tests – x-ray and CT – appears very strong. Because it is randomized, there is no significant difference between the pre-existing characteristics of those assigned to CT versus chest x-ray screening, and because the end point is death, it largely eliminates one of the most important confounding issues in prior studies called “lead time bias”. This means that if a more sensitive test identifies cancer earlier in its course, the time between diagnosis and death will be longer even if the death itself is not forestalled. (E.g., you have cancer and one test finds it at 55 and you die at 60; another test could find it at 50 and you die at 60; finding the cancer earlier didn’t make you live longer.) It is also a very large study (53,000 people) and well designed in many other ways, so that a 20% reduction in death is important. There is an excellent FAQ for this study, including graphics of lead-time and length-time bias, at the NCI website
So is there any problem? Will lung cancer, the biggest killer among cancers, become like breast cancer, where a screening test can find the cancer earlier, lead to earlier and effective treatment, and decrease mortality? Not exactly. In addition to a 20% reduction in mortality being far less than the reduction in breast cancer mortality from mammography, lung cancer is not breast cancer; we know the cause of the vast majority of cases: smoking. The authors, and the NCI, emphasize that such screening, even if widely adopted, is no substitute for stopping smoking or increasing efforts to get people to stop smoking. The real question is what is the benefit of spending huge amounts of money (while there is no statement of cost of screening in either the NCI brief or the Radiology article, estimates are as much as $12 billion a year -- 30 million screened at $400 per CT screening with interpretation, including follow-up exams -- to screen people who smoke, or smoked, heavily for cancer in pursuit of a significant, but relatively small, reduction in mortality? Moreover, there is no estimate of the potential risk of repeated CT scans (even low-dose, such as studied) and the degree to which the existence of a screening test might decrease the interest of smokers in stopping. (If this seems perverse, it is almost certain to happen; it happens every time news of a possible preventive intervention is announced: some people decide there is no need to stop their risky behavior.) Thus, to save 1 in 300 lives (about 100,000 of the 30 million screened, or 0.039% of the US population), not even considering quality of life (generally low for long-term smokers with cancer who have other conditions such as chronic lung disease), will cost about $40 per every person in the US per year.
How do we evaluate cost-benefit? In the current political environment, the popular theme is “don’t spend public money”, but there is always the implicit caveat “except if it benefits me” – and in this country we have over 300 million “me’s”. Dr. Robert Bowman, who has previously contributed to this blog, describes for us the potential alternative uses of not only the ongoing cost of screening, but even the cost of the study itself:
· The $250 million for this one study involving CT screening for lung cancer is about what the United States spent for all Agency for Healthcare Research and Quality (AHRQ) health care cost, quality, and outcomes research in 2008. (AHRQ is the main government agency looking at these issues, including “outcomes”, particularly important as I have previously discussed; it is obviously funded at a lot less than the $30 billion for NIH.)
· $250 million is the entire sum that the Health Resources and Services Administration (HRSA, the government agency that funds workforce research, training programs in primary care, dentistry, physicians assistants, pipeline programs, etc.) could scrape together to address emergent needs for primary care workforce this year.
· $250 million, if used to train family doctors at about $30,000 cost per Standard Primary Care provider, would produce 8333 Standard primary care years of workforce in family medicine graduates, or about 333 FM physicians serving their entire careers and improving cost, quality, and access where it is most needed.
And the $12 billion?
· The $12 billion a year spent on CT screening for 30 million current or former smokers could graduate 16,000 family physicians a year.
· $12 billion a year, expended each year for the 30 years required to actually build any workforce (i.e., a generation), if applied to family medicine would supply the entire nation enough primary care for all locations and populations in need of primary care. Sufficient primary care for over 90% of Americans in all needed locations would begin 30 years after reaching 16,000 annual graduates and would be maintained with continued funding of 16,000 annual graduates. (Indeed, compared to less-efficient spending on training programs, such as internal medicine, that yield far fewer primary care years per dollar spent, this $12 billion is actually is a savings of a few billion dollars!)
Or, if we are concerned about lung cancer, using this for tobacco control campaigns, both the "stop smoking" kind and the legislating non-smoking venues, cities and states.Dr. Bowman continues:
"The US continues to fail, time after time, in the most basic choices regarding care for Americans most in need of care. The US can focus on the health care needed for nearly all people nearly all of the years of their lives in nearly all locations or the US can continue to spend its $7000 per person ($2.5 trillion) on the health care needs for only some of its people for a only a few years of their lives, with health care delivery services concentrated in only a limited number of locations (4% of the land area).
There is little point to research about rural workforce, health access, or primary care until the nation makes a decision to quit sending health care spending to locations with top concentrations for the care of Americans already with the most care."
Any economist – or wise investor or businessperson – can tell you about “opportunity cost”. This means “if you spend money on one thing, you can’t spend it on something else.” Therefore, the benefit of what you spend money on needs to be looked at not only for its intrinsic value (“will spending $12 billion a year on lung cancer screening with CT save lives?”) or against a very limited range of options (“Is it more cost effective than screening with chest x-rays?”) but weighed against reasonable alternative strategies to improving the health of people – all people.
Dr. Bowman gives strong arguments for the benefit of investing in primary care workforce development, and particularly in family medicine. Maybe there are other strategies for most improving the health of most of our people. But looking at new scientific advances in isolation is clearly a flawed approach.
 For a detailed description both of this measure – standard primary care (SPC) years – and the reasons that Family Medicine, as opposed to other physician and non-physician primary care training (internal medicine, pediatrics, physician’s assistants, nurse practitioners) is the most efficient producer of SPC years, see Ten Biggest Myths Regarding Primary Care in the Future, January 15, 2009.