Skip to content

H. Gilbert Welch has written extensively about the issue of "overdiagnosis" and resulting  "overtreatment" of cancers. Cancer screening can cause the problem of overdiagnosis (finding small tumors that may never cause problems) and lead to overtreatment (treating unnecessarily, which can cause harm).

But now Welch and coauthor Otis Brawley discuss the issue of how too much screening and diagnostic testing of people thought to be "high risk"  for certain cancers results in more being found - thus the risk factors are "self-fulfilling". And it occurs the most in "scrutiny dependent cancers" - which are cancers that the more you look, the more you find, and the more of what you find is harmless. Many are referred to as slow-growing, indolent, subclinical, or even as precancerous. Prostate cancer, thyroid cancer, breast cancer, melanoma, and  lung cancer are  examples of "scrutiny-dependent" cancers.

Looking so hard and then finding cancer gives a false impression of an increased incidence of some cancers. The authors also said that risk factors in determining  who should be screened should not be cancer diagnosis (e.g. in a family member), but death from cancer. From STAT News:

Too much screening has misled us about real cancer risk factors, experts say

The best-known downside of cancer screening, such as PSA tests for prostate cancer and mammograms for breast cancer, is that they often flag cancers that pose no risk, leading to overdiagnosis and unnecessary, even harmful, treatment. But widespread screening for “scrutiny-dependent” cancers — those for which the harder you look the more you find, and the more of what you find is harmlesscauses another problem, two leading cancer experts argue in a paper published on Monday: increasing the apparent incidence of some cancers. That in turn is misleading doctors and the public about what increases people’s risk of developing cancers — or at least the types of cancer that matter.

“Detecting cancers that would never become apparent is screwing up our understanding of risk factors,” said Dr. H. Gilbert Welch of the Dartmouth Institute for Health Policy and Clinical Practice, co-author of the analysis in Annals of Internal Medicine. The problem is especially clear in prostate, breast, and thyroid cancers, all of which are scrutiny dependent.

Men whose relatives developed prostate cancer are more likely to get PSA and other screening tests, either because they request them or because their physicians, noting their family histories, order them. Men with no such family history are less likely to be screened. .... (More than half of such cancers are so slow-growing that they don’t affect health or longevity.) Men who don’t get screened are less likely to have biopsies and so are less likely to be diagnosed with prostate cancer — not because they develop the disease at a lower rate but because they get screened at a lower rate. What you don’t look for, you don’t find.

“If we biopsied men without a family history of prostate cancer at the same rate that we biopsy men with a family history, we’d find more prostate cancer in them as well,” Welch said. “Family history influences how hard we look for prostate cancer and therefore how much we find. The risk factor becomes a self-fulfilling prophecy.”

2016 study of increased prostate cancer screening in men with a family history of the disease concluded that the risk due to family history has been overestimated by nearly half. “The risk factor of family history is spuriously strengthened because men with a family history are exposed to greater scrutiny,” write Welch and Dr. Otis Brawley, chief medical officer of the American Cancer Society, in the Annals report.

Wealthier, better educated women are, however, more connected to the health care system and therefore get more mammograms, breast ultrasounds, and MRIs. The more scrutiny, the more likely that harmless cases of breast cancer are found. (The idea of “harmless” breast cancer sounds like an oxymoron, but an estimated one-half of breast cancers detected by screening would never cause problems even if undetected and untreated.)

Breast tumors found by imaging are much more likely to be harmless than those discovered by women or their physicians finding a breast lump. Income and education are therefore less likely to be a true risk factor for breast cancer and more likely to be a “risk factor” for undergoing screening. If poorer, less educated women were screened for breast cancer at the same rate as wealthier, better educated women, the socioeconomic risk factor would likely vanish.

Thyroid cancers are also scrutiny dependent, which is why when countries launch screening programs the incidence of the disease skyrockets (but death rates don’t, showing that what’s being found is a false epidemic). 

Welch and Brawley call for less focus on risk factors for developing cancers, since those numbers both determine and reflect who gets screened, and more on risk factors for death from cancer.

The following is an excellent commentary by Dr. John Mandrola regarding an important British Medical Journal article that I posted about earlier (see Rethinking Cancer Screening). He has a highly regarded web-site and also frequently posts on Medscape. His view is that cancer screening "may be one of Medicine’s largest reversals. A reversal happens when something (testing or treatment) doctors did, and patients accepted, turned out to be non-beneficial." (www.drjohnm.org)

I can't overstate how big a reversal this is in medicine - it's huge, a paradigm change in the making. The reason for this is that studies show that overall death rates are basically the same in screened vs non-screened for mammography, colon, prostate, and lung cancer screening. This means our view of how cancer grows and spreads may have to be reexamined and changed. One possibility suggested by Dr. H. Gilbert Welch is that aggressive cancer is already "a systemic disease by the time it's detectable" (Oct. 28, 2015 post).  From Medscape:

In Cancer Screening, Why Not Tell the Truth?

The problem: cancer screening has not worked. Recent reviews of the evidence show that current-day screening techniques do not save lives. Worse, in many cases, these good-intentioned searches bring harm to previously healthy people.

I realize this sounds shocking. It did to me, too. Millions of women and men have had their breasts squished, veins poked, lungs irradiated, and bowels invaded in the name of "health" maintenance. I've been scolded for forgoing PSA tests and colonoscopy — "you should know better, John."....Anecdotes, however compelling, are not evidence. When you pull up a chair, open your computer, take a breath, suspend past beliefs, and look for the evidence that screening saves lives, it simply isn't there.

One reason that this many people (doctors and patients alike) have been misled about screening has been our collective attachment to the belief that if screening lowers disease-specific death rates, that would translate to lower overall mortality. That is: breast, lung, and colon cancer are bad diseases, so it makes sense that lowering death from those three types of cancer would extend life. It is not so.

In a comprehensive review of the literature[1] published in the BMJ, Drs Vinay Prasad (Oregon Health Sciences University, Portland) and David Newman (School of Medicine at Mount Sinai, New York), along with journalist Jeanne Lenzer, find that disease-specific mortality is a lousy surrogate for overall mortality. They report that when a screening technique does lower disease-specific death rates, which is both uncommon and of modest degree, there are no differences in overall mortality.

The authors cite three reasons why cancer screening might not reduce overall mortality:

  • Screening trials were underpowered to detect differences. I'm no statistician, but doesn't the fact that a trial requires millions of subjects to show a difference, mean there is little, if any, difference?
  • "Downstream effects of screening may negate any disease-specific gains." My translation: harm. Dr Peter Gøtzsche (Nordic Cochrane Center, Copenhagen) wrote in a commentary[2] that "screening always causes harm. Sometimes it also leads to benefits, and sometimes these benefits outweigh the harms." To understand harm resulting from screening, one need only to consider that a prostate biopsy entails sticking a needle through the rectum, or that some drugs used to treat breast cancer damage the heart.
  • Screening might not reduce overall mortality because of "off-target deaths." An illustration of this point is provided by a cohort study[3]that found a possible increased risk of suicide and cardiovascular death in men in the year after being diagnosed with prostate cancer. People die — of all sorts of causes, not just cancer.

Let's also be clear that this one paper is not an outlier. A group of Stanford researchers performed a systematic review and meta-analyses[4] of randomized trials of screening tests for 19 diseases (39 tests) where mortality is a common outcome. They found reductions in disease-specific mortality were uncommon and reductions in overall mortality were rare or nonexistent.

Drs Archie Bleyer and H Gilbert Welch (St Charles Health System, Central Oregon, Portland) reviewed Surveillance, Epidemiology, and End Results (SEER) data from 1976 through 2008 and concluded that "screening mammography has only marginally reduced the rate at which women present with advanced cancer and that overdiagnosis may account for nearly a third of all new breast cancer cases."[5] Likewise, a Cochrane Database Systematic review[6] of eight trials and 600,000 women did not find an effect of screening on either breast cancer mortality or all-cause mortality. This evidence caused the Swiss medical board to abolish screening mammography.[7]

These are the data. It's now clear to me that mass cancer screening does not save lives. But I'm still trying to understand how this practice became entrenched as public-health gospel. It has to be more than fear. Dr Gerd Gigerenzer (Max Planck Institute, Berlin, Germany)...He pointed to language and the ability of words to persuade. Instead of saying "early detection," advocates might use the term "prevention." This, Dr Gigerenzer says, wrongly suggests screening reduces the odds of getting cancer. Doesn't looking for cancer increase the odds of getting the diagnosis of cancer?

Gigerenzer noted two other ways language is used to emphasize screening benefits over harms: -The reporting of benefits in relative, not absolute terms. - The equating of increases in 5-year survival rates with decreases in mortality. I would add to this list of word misuse, the practice of referring to women sent to mammography screening as patients. They are not patients; they are well people. Dr Gigerenzer agreed with the commonsense notion that overall mortality should be reported along with cancer-specific mortality. His editorial included a fact box on breast cancer early detection using mammography provided by the Harding Center for Risk Literacy. I challenge you to tell me why such text boxes should not be shown to people before they undergo screening,

The first action healthcare experts should take is to spread the word that there is nothing about the mass screening of healthy people for cancer that equates to health maintenance. Embrace clear language. Saying or implying that screening saves lives when there are no data to support it and lots to refute it undermines trust in the medical profession.

The second action healthcare experts should take is to stop wasting money on screening. If the evidence shows no difference in overall mortality, why pay for it? I'm not naive to the fact that use of clear language will decrease the number of billable procedures. I am not saying this will be easy. One first move that would be less painful would be to get rid of quality measures or incentives that promote screening.

I want to be clear; I'm not saying all cancer screening is worthless. People at higher baseline risk for cancer, such as those with a family history of cancer or environmental exposures, might derive more benefit than harm from screening. Prasad, Lenzer, and Newman say this group of patients would be a good place to spend future research dollars. That sounds reasonable. I also acknowledge that some people, even when presented with the evidence, will want to proceed with screening. We can argue about who should pay for non–evidence-based medical procedures.

 A provocative and thought-provoking article in which the title says it all: Cancer screening has not been shown to 'save lives'. The following is from the Medscape analysis/reporting of the original British Medical Journal article and accompanying editorial ( BMJ. January 2016, Article, Editorial), and both the original and Medscape analysis are well worth reading. From Medscape:

Cancer Screening Has Not Been Shown to 'Save Lives'

Debates about cancer screening programs tend to focus on when to start, who to screen, and the frequency of screening. But some investigators are asking a far more provocative question: Do screening programs actually save lives?

We do not know the answer to that question, and would need to conduct massive clinical trials to find out, said Vinay K. Prasad, MD, MPH, assistant professor of medicine at the Oregon Health Sciences University in Portland."Proponents of cancer screening say that screening tests have been shown to save lives. What we're trying to show is that, strictly speaking, that's sort of an overstatement," he told Medscape Medical News.

In an analysis published online January 6 in the BMJ, Dr Prasad and his colleagues argue that although cancer screening might reduce cancer-specific mortality, it has not conclusively been shown to have an effect on overall mortality. The researchers go on to suggest that the harms of screening might actually contribute to overall mortality rates. These potential harms include false-positive results that lead to unnecessary biopsies or therapeutic interventions and overdiagnosis, in which treatment is delivered for a condition that is unlikely to affect patients during their natural lifespans.

"There are two chief reasons why cancer screening might reduce disease-specific mortality without significantly reducing overall mortality," the researchers write. "Firstly, studies may be underpowered to detect a small overall mortality benefit. Secondly, disease-specific mortality reductions may be offset by deaths due to the downstream effects of screening." "The bar to say that screening saves lives should be overall mortality, and we haven't met that bar," Dr Prasad told Medscape Medical News.

The rationale for cancer screening is based on the assumptions that screening will reduce deaths from cancer and that lowering cancer-specific deaths will decrease overall mortality. These assumptions remain unsupported by facts, Dr Prasad's team contends.

They illustrate this point with data from the National Lung Cancer Screening Trial (NLST). Although there was a 20% relative reduction in lung cancer deaths with low-dose CT screening, compared with chest x-ray, and a 6.7% relative reduction in overall mortality, the absolute reduction in risk for overall mortality was just 0.46%....The team also notes that "the benefit in lung cancer mortality of CT screening (estimated to avert over 12,000 lung cancer deaths in the United States annually) must be set against the 27,034 major complications (such as lung collapse, heart attack, stroke, and death) that follow a positive screening test."

The decision to undergo screening should be part of an informed discussion between the patient and clinician that takes into account personal preferences and the risks and benefits of screening. Dr Prasad explained. "Declining screening may be a reasonable and prudent choice for many people," the researchers write. "Doctors should be comfortable with whatever choice people make when they hear about all the potential benefits and the known harms," Dr Prasad added.

However, in an accompanying editorial, Gerd Gigerenzer, PhD, from the Max Planck Institute for Human Development in Berlin, argues that "rather than pouring resources into 'megatrials' with a small chance of detecting a minimal overall mortality reduction, at the additional cost of harming large numbers of patients, we should invest in transparent information in the first place." He explains that information about screening should be presented in a "fact box" that lays out the benefits and risks of mammography with numbers for how many women would be affected."It is time to change communication about cancer screening from dodgy persuasion into something straightforward," he concludes.

Richard L. Schilsky, MD, chief medical officer for the American Society of Clinical Oncology (ASCO), said that although, in general, ASCO supports cancer screening, "it's a very imperfect process....The often high variability in the natural history of many cancers has been the source of particular confusion and uncertainty surrounding screening, he noted. For example, there is little value in screening for aggressive cancers for which interventions are unlikely to make a difference in outcomes, no matter how early the disease can be detected. Conversely, "if the cancer is never going to kill you, no matter what the doctors do, then screening won't help either," he said. Additionally, there are some cancers for which treatments are sufficiently effective that they can be successfully managed whether they are diagnosed at an early or later stage. "When you consider all these factors, the number of individuals who will actually benefit from detecting a screen-detected cancer is very small," Dr Schilsky said.

 The following article supported what I have been reading over the past few years: that medical tests and treatments also have downsides, that it is possible to "know too much", that more harm than benefits can occur from certain tests, procedures, and medicines, and lifestyle changes (eat a less processed more plant-based diet, move more, and don't smoke) can be better than some medicines or certain procedures. The doctor mentioned in this article (Dr. H. Gilbert Welch) recently published a book aimed at the general public which I just read and highly recommend: Less Medicine, More Health. Dr. Welch is an academic physician, a professor at Dartmouth Medical School, and a nationally recognized expert on the effects of medical testing. In 2012 he published the well regarded and more technical and in-depth book on this issue: Overdiagnosed: Making People Sick in the Pursuit of Health. From The Atlantic:

The Downside of Medical Screening

If you had a disease, and you could find out sooner rather than later, why wouldn’t you?Medicine has long focused on early detection of diseases as part of a move toward preventive care. But imperfect tests, false positives, and overdiagnosis mean that sometimes the tests do more harm than good, and in recent years, there have been more recommendations to reduce some kinds of screening, including pap smears, colonoscopies, mammograms, and even annual pelvic exams.

“This is something we all need to understand, the two sides of early detection. It does help people, but it’s almost guaranteed to harm others,” said H. Gilbert Welch, a professor of medicine, public policy, and business administration at Dartmouth College, and author of the book Should I Be Tested for Cancer? (He reveals his answer in the book’s subtitle: “Maybe not.”)

The more you look for disease, the more you find it. And in the case of cancer, it’s hard for doctors to know if what they find is dangerous and needs to be addressed, or if it’s just a small tumor that won’t grow and poses no threat. “We can’t be sure which is which, so we treat everybody,” Welch explained at the Aspen Ideas Festival’s Spotlight Health session. “That means we’re treating people who will never experience problems from their disease.”

But they may experience problems from the treatment.The panel gave the example of prostate cancer, which is very common in men—one in seven American men will be diagnosed with it in their lifetimes. “But it turns out a lot of these cancers are very indolent,” said Jessica Herzstein, a preventive-medicine consultant and member of the U.S. Preventive Services Task Force. Around 30 to 40 percent of men who’ve been treated for prostate cancer likely had “slow-growing tumors that would never have become a threat to the man’s lifespan or health,” according to the Prostate Cancer Foundation.

In other words, “you’re going to die with them, not of them,” Herzstein said, “and the treatments are very very harmful.” Radiation therapy, for example, can cause incontinence and erectile dysfunction, and hormone therapy can cause osteoporosis and depression.

The possibility of a false positive is another downside. Not only could it lead to more invasive follow-up tests or treatments that aren’t needed, but it can also give patients unnecessary anxiety.“If we resolve the test by saying ‘The test was wrong, you’re fine!’, that’s one thing,” Welch said. “But most false alarms aren’t resolved that way. [It’s more like] ‘You don’t have cancer, but you have some abnormality that possibly puts you at a higher risk for cancer, but we’re not going to do anything about it. I think that’s where there can be [mental] harm.”

Ultimately, it comes down to a weighing of the benefits and the harms, and, in the absence of clear evidence, the preferences of the patient. The U.S. Preventive Services Task Force helps identify which tests are beneficial by evaluating and grading them. It gives tests an A if there’s a high certainty of substantial benefit, a B if there’s moderate certainty of substantial benefit, a C if there’s moderate certainty of a small benefit, a D if there’s moderate or high certainty of no benefit, and an I if the evidence is just too insufficient to say.

The task force gave prostate cancer screening a D. HIV screening got an A. For breast cancer screening, an always-controversial topic, the results vary. Breast self-exams got a D. Mammograms got a B, but only for women between 50 and 74 years old. For women in their 40s, the grade is a C, meaning the task force recommends patients and physicians discuss and decide together.

Before getting a screening test, patients should think about what would happen if they get a positive result, and if they’d be ready for it, Welch advised. “If I were to go through this, and have this diagnosis, would I want to have this surgery?” Herzstein asked, posing a hypothetical. Would you want to undergo the biopsy, the chemo, whatever treatments come next? “Maybe you don’t even want to go there if there is no treatment for the disease,” she added. Welch gives an example. “With Alzheimer’s disease that’s a fundamental question: What are you going to do with a positive result?” he asked.