Total Pageviews

Tuesday, April 24, 2012




     Have you ever been annoyed by your car's 'check engine' light? I mean, have you responded to this light and taken the car in to be checked, only to find there  was nothing really wrong and nothing to be done. [This analogy is provided by Dr. H. Gilbert Welch in his book "Overdiagnosis: Making People Sick in the Pursuit of Health".] The problem now is that your car is so computerized that it even has sensors to detect the function of other sensors. At a certain degree of complexity something has to be at least slightly askew a large fraction of the time. The point that Dr. Welch makes is that getting frequent recommended medical check-ups produces essentially the same result. Your doctor will likely tell you on most of these occasions that "you're 'check engine' light is on" and there's something you have to do to turn it off (despite the often overwhelming odds that everything is just fine). Examples of such situations include the mammogram or chest x-ray with a "suspicious spot," the Pap smear with minor abnormalities like "ASCUS," the stool blood test that is positive (due to hemorrhoids, which are much more common than colon cancer), the blood pressure that is elevated 10 mm Hg more than usual, the freckle that looks just slightly darker than all the other ones, etc. The question for you the patient should be, How often are you metaphorically willing to have your engine checked? It can get to be quite a nuisance. Even more than that, as Dr. Welch points out, it can at times be frankly hazardous to your health.

     Consider this table that he provides for us [his table 2.1]. In it he shows the effects when, hopefully, well-meaning "experts" decide to change the criteria for diagnosis of common conditions. Their imputed motivation for doing this is to catch more people with this condition [i.e., their 'check engine' light is on], so that we might prevent more bad disease outcomes like blindness from diabetes, stroke from hypertension, a heart attack for high cholesterol, and a compression fracture from osteoporosis. The problem is that at the lower range of these physiological variables there are actually a few bad outcomes, but only a very few. As you can see in the table, as you lower the threshold for diagnosing diabetes, hypertension, high cholesterol, and osteoporosis, you are instantly creating literally millions of patients with a "new" disease condition. They are now "sick" and are expected to get "treatment." Only a small percent of such newly diagnosed patients (typically 1-8%) will benefit from this treatment; essentially, a minimum of 92% of the new diagnoses are being subjected to medical scrutiny, testing, and treatment with absolutely no benefit to show for it.


Old criterion
New Criterion

Diabetes: the criterion for diagnosis was changed from a sugar level of 140 down to 126 mg%.
Hypertension: the criterion for diagnosis was changed from 160/100 to 140/90 mm Hg.
High cholesterol: the criterion for  diagnosis was changed from a total cholesterol of 240 mg to 200 mg/dL.
Osteoporosis: The criterion for diagnosis was changed from a “T score” of 2.5 to 2.0

The other problem is that the motivation for these new diagnostic criteria among the "experts" is not always beneficent. Often it is frankly commercial. "The head of the diabetes cutoff panel was a paid consultant to Aventis Pharmaceuticals, Bristol-Myers Squibb, Eli Lilly, GlaxoSmithKline, Novartis, Merck, and Pfizer--all of which make diabetes drugs. Nine of the eleven authors of recent high blood pressure guidelines had some kind of financial ties--as paid consultants, paid speakers, or grant recipients--to drug companies that made high blood pressure drugs. Similarly, eight of the nine experts who lowered the cholesterol cutoff were paid consultants to drug companies making cholesterol drugs. and the first cutoff for osteoporosis was established by a World health Organization panel in partnership with the International Osteoporosis Foundation--an organization whose corporate advisory board consisted of thirty-one drug and medical equipment companies."
     Dr. Welch summarizes the outcomes for persons affected by these new guidelines in this table: 

“If 100 patients are diagnosed on the basis of elevated but near normal levels of a variable and treated for a lifetime, there will be…”
(avoided a heart attack)
TREATED FOR NAUGHT (had a bad outcome despite treatment)
(not helped because they were never going to have a heart attack anyway)
Mildly elevated cholesterol
Mildly low T score for osteoporosis

     This problem also exists with all our fancy new imaging technologies like CT scans and MRIs, as well as for old technologies like plain x-rays and ultrasounds. Here are some typical examples:
  • "Gallstones: In people without any symptoms of gallbladder disease (pain, nausea, or problems with fatty foods, for example), about 10% have gallstones when scanned by ultrasound.
  • "Damaged knee cartilage: In people without knee pain or a history of knee injury, about 40% have meniscal damage in their knees when scanned by MRI.
  • "Bulging discs in the back: In people without any back pain, over 50% have bulging lumbar discs when scanned by MRI."
     And, of course, as I have said in prior blogs, this problem applies to cancer screening. I have already covered prostate cancer, cervical cancer, colon cancer, and breast cancer. Let's take a look at 3 other relatively common cancers--thyroid cancer, melanoma, and lung cancer. 

Thyroid cancer: Thyroid cancer is responsible for about 1,600 deaths each year in the US. The number of cases of thyroid cancer diagnosed each year in the US is much greater       --37,000 cases. This discrepancy is explained by two basic factors:  (1) treatment for the most common thyroid cancers is pretty good; and (2) there is a large portion of thyroid cancers that exist in asymptomatic patients and which is never going to cause any symptoms. In  one autopsy study of 101 older patients who had died of other causes in a hospital, pathologists found clusters of thyroid cancer cells in a full 33% of patients. The researchers in this study concluded that "the smallest forms of thyroid cancer were so common that they should be regarded as normal findings." The moral of this story is that you don't really want your doctor examining your neck (thyroid gland) for lumps because nodules are extremely common (now that we use ultrasound routinely; prevalence is 30-65%), yet death from thyroid cancer is very rare. The United States Preventive Services Task Force recommended against any form of routine thyroid cancer screening in 1996.

Melanoma: There are about 8,400 deaths in the US each year from malignant melanoma. The incidence of melanoma, however, is much higher--around 116,000 cases per year. In recent years there have been many efforts directed at increasing early diagnosis of melanoma and these have resulted in a doubling of the number of cases diagnosed between 1975 and 2005. It would seem like that should be a good thing. But when you look at the annual death rate from melanoma in the same period there has been absolutely no change. This means that all of these extra diagnoses have not reduced the death rate. We have just labelled more people with a diagnosis of melanoma that was going to have no effect on the overall death rate. What we are finding with our increased tendency to biopsy pigmented lesions is that there are many people who have microscopic clusters of abnormal melanin-producing cells who were never going to have any clinical disease from it. Increased screening does not appear to be helping at all. 

Lung cancer: "From a public health perspective, lung cancer is the cancer that warrants the most attention. It's responsible for 162,000 deaths annually in the US. That's more than breast cancer, prostate cancer, melanoma, thyroid cancer, and colon cancer combined. About 215,000 Americans get the diagnosis annually. That means most people diagnosed with lung cancer die from it." Since we know that people who smoke or used to smoke are the ones who have the highest risk for lung cancer, you would think it might make sense to screen this group of patients for lung cancer. If so, you would be wrong.
     Three randomized trials completed in the 1990s showed the screening chest x-rays did not lead to a reduction in lung cancer deaths. "In fact, in two of the studies, screening appeared to cause more deaths." Virtually all cases identified by screening underwent surgery, and the surgery has a significant mortality rate. But this is not the only problem. The other problem is that, after 20 years of follow-up, it appears that approximately 50% of all cancers detected by screening were lung cancers that were small and were never going to cause a problem, yet almost all of these "overdiagnosed" patients underwent surgery.

     Finally, there is the problem of "incidentalomas". This is a new term in medical jargon that refers to abnormal findings on an imaging procedure that was carried out for a purpose unrelated to cancer screening. For example, if a chest CT scan is done for a pulmonary problem, it may also unexpectedly find a "spot" in the liver or kidney. If an ultrasound is done on a carotid artery to assess circulation, it may also see a cyst or nodule in the thyroid gland. 

% of people with an incidentaloma on CT scan
10-year risk of cancer death
Maximal chance that the incidentaloma is a lethal cancer
Chance that the incidentaloma is NOT a lethal cancer
Lung (smokers)
(never smokers)
Thyroid (ultrasound)
< 0.01%
Ø  99.99%

When you reflect on the data in this table, it should be clear to the average person that, if an "incidentaloma" is discovered while some imaging test is done for another purpose, this result should simply be ignored. Otherwise you risk biopsy or surgery for no reasonable expectation of benefit.

COMMENT: I have been citing examples of these kinds of problems for years, but Dr. Welch puts them all together in one easy to read book with tables and diagrams that makes this otherwise dense information very accessible. While he makes a clear case that our traditional (and rather frivolous) approach to screening and diagnosis for these diseases poses significant hazards to your health, he provides less guidance on another approach to health promotion and disease avoidance other than better consumer information. Here is where my approach to prevention through the "Formula for Health" really shines.

    Adopting the 5 simple health habits identified in the "Formula for Health" has been shown in 13 major observational studies and 1 randomized controlled trial to be associated with benefits of the magnitude illustrated in the second poster. This approach requires no screening test, no blood testing or imaging, and creates no risk of overdiagnosis. In fact, it represents the maximum that you can do to improve your overall health. The problem with our modern medical technology is that "overdiagnosis" (with subsequent treatment without benefit) has become a bigger problem than the diseases they were designed to avoid. We have to find another way. A healthy lifestyle (The Formula for Health) is this way.


Product Details

Wednesday, April 11, 2012




     Last week I wrote about the limited knowledge that most physicians have about cancer incidence, survival rates, and the effects of screening tests. Numerous studies indicate that a majority of patients base their decisions to be screened on their doctors' advice. Unfortunately, if doctors don't know much, your decision won't be very good. The only way to correct this is to try to know a little bit more about cancer yourself. So what do you need to know to make good cancer risk assessments and screening decisions?
     You need to know some basic epidemiologic facts about each type of cancer. In the old days, this information used to be hard to come by, but now it's all available instantly with a simple Google search. Let's consider several of the major cancers that people often get screened for. Here is a table that lists the core information you would like to know in order to make a decision on testing. 

Type of cancer
Test: % of negative test results that are true
Test: % of positive test results that are false
Reduction in relative risk of dying of this cancer after testing.
Effect of testing on chances of being alive in 10 years.
Breast cancer
3 in 100 for each decade after 50 in average risk women
Mammogram: 90%
Mammogram: 2% (each time)
16% with serial testing every 2 years
Colon  cancer
57/100,000 in men at age 50; 150/100,000 at age 80.
Stool blood test: 30%
15% with annual testing
Prostate cancer
8 in 100 at age 50;  37 in 100 at age 70
PSA:  80% (for result > 4)

     If you are considering having a test for cancer, you want to know, first, How common is this type of cancer in persons like me? You also want to know what test are they using and how good is this test. The effectiveness of such tests are rated and compared by telling you how reliable a finding of a negative result is (i.e., How much confidence can you place in a negative test result indicating that you do not have this kind of cancer?) and the chances of having a positive test result that is not due to cancer. The simple fact is that no test rules out your chance of having cancer by 100%, and all tests have a number of false positive results that cause much anxiety but no disease. Finally, you want to know that there have been studies done to assess this screening technique and that it does in fact, after a number of years, reduce your chance of getting one of these cancers.
      In the table above we see that for an average woman, her risk of being diagnosed with breast cancer over any 10-year period after age 50 is about 3%. If she gets a mammogram regularly (at least every 2 years), she can reduce this risk by about 16%;  that is, after testing, her risk for a 10-year period will be 2.5%. This has to be compared to the risk of a false positive test result, which occurs in 2% for every mammogram performed; if you get several mammograms (e.g., the screening program calls for at least 10), then your risk of a false positive result is multiplied, and is 20% over 20 years if you get 10 mammograms. Each woman has to decide for herself whether she can accept this relatively high risk of a false-positive result in comparison to the relatively small benefit (2.5% absolute risk reduction over 20 years). Finally, every one undergoing any cancer screening test has to be aware that getting all the recommended tests is likely (but not guaranteed--note the result for prostate cancer screening with a PSA test) to offer a lower risk of dying of this type of cancer, but will, in fact, not insure that you live any longer overall than a person who does not get the screening test. In technical language, this means that there is no reduction in "all-cause mortality" for any of the common cancer screening tests.
     For women considering having a mammogram they should be aware that very recent studies have suggest that that absolute reduction in deaths from breast cancer is reduced by only 1 person per 2000 by screening, and the chance of having a cancer incorrectly diagnosed and treated is around 20-25%
     A group of widely respected medical experts (the Cochrane Collaboration folks) have put together a nice summary of the basic facts needed to make the breast cancer screening decision--"What you always wanted to know about breast screening". 

     Colon cancer screening poses the problems that the disease is considerably rarer than breast cancer (only about 50 cases per 100,000 persons at age 50, increasing to about 150 cases per 100,000 by age 80), that the test (stool testing for blood) is not very good (it reliably detects only about 30% of actual cancers, and gives a false positive result in 2% of persons tested every time it is used; this is a lot when you screen a large population. For example, if 100,000 persons were screened at age 50, there would be 57 true positive tests, and 2000 false positive tests). All of these false positive results have to be evaluated with additional testing (colonoscopy). The final disadvantage is that people just don't like having the necessary follow-up test for a positive result (the colonoscopy), which requires a full day of bowel preparation (total bowel wash-out) and a full-day lost from work or home on the day of the test, since sedation is used.

     Prostate cancer screening is the most curious case since there really is no good evidence that the test reduces even your relative risk of dying of prostate cancer because this disease is so common in men and often just sits there for many years without causing any problem at all.

     So this is the summary of the kind of basic information you should know before undergoing any cancer screening test. Because there are so many studies demonstrating that most physicians do not actually know or understand these facts, I am afraid I have to suggest that you look them up yourself.


     Above I have discussed the real risks of getting these cancers and your chances of living longer, if you are diagnosed with one. The truth, however, is that most people (including both physicians and consumers) are not driven primarily by the real risks. They are driven by perceived risk--something tantamount to a gut feeling that one is likely to get such a cancer and this risk can be abated or avoided by a screening test. To illustrate how much of a problem the overperception of risk is, consider this chart from a 2010 publication.(1)

     For purposes of this graphic, you are a "participant." What the chart shows is the risk of dying of each cancer and the risk of being diagnosed with each cancer as estimated by the "participants" in a screening program compared to the actual risks as demonstrated by epidemiological studies (SEER). The dark gray bar shows participants estimates of their chance of dying of each cancer, if they have it, and the black bar shows their estimate of their chances of getting each cancer. Compared to the actual risks demonstrated in studies, participants tend to dramatically overestimate their risk both of getting these diseases and of dying from the disease, if they do get it. This fact then induces them to undergo testing whether or not the testing is in fact very effective.
     As I showed in my last blog, physicians are prone to very similar biases. They are inappropriately influenced in favor of testing by reported increases in "5-year survival" even though this means nothing about the actual effectiveness of a test. More than half of physicians are not aware that screening will always lead to an apparent increase in "5-year survival" (because cancers are detected earlier) that does not necessarily imply any reduction in mortality at all. While "5-year survival" is a critically important measure for different treatments for diagnosed cancers, it has no relevance to the assessment of screening effectiveness; a majority of physicians do not know this and will provide inappropriate endorsements of screening tests when told that 5-year survival was noted to increase after starting a screening program. In the referenced study below, only 1 of 65 physicians was aware that screening consistently leads to "overdiagnosis"--imputing a diagnosis of cancer to someone who was never at risk of dying from it.

     So what's the bottom line in cancer screening? There are really only two facts you want to know. First, what is the absolute reduction in your risk of dying of a cancer for which you have been screened?

For some common cancers, here are these numbers:

Breast cancer: The absolute reduction in risk of dying from breast cancer through screening is 1 per 2000.

Colon cancer: The absolute reduction in the risk of dying of colon cancer through screening at age 50 is about 10 cases per 100,000 persons screened. For persons being screened at age 70, the absolute reduction in the risk of dying is about 25 per 100,000 persons screened.

Prostate cancer: The absolute reduction in the risk of dying of prostate cancer through screening is 0.

Cervical cancer: The absolute reduction in the risk of dying of cervical cancer through screening with Pap smears is 3 per 100,000.

Melanoma: There is no evidence that screening programs reduce the absolute risk of dying from melanoma.

The other fact you want to know is, What is your chance of actually living longer due to undergoing these screening tests? Here are the relevant numbers:

Breast cancer: There is no increase in your total expected lifespan from breast cancer screening with mammography.

Colon cancer: There is no increase in your total expected lifespan from colon cancer screening with either the fecal blood test or colonoscopy.

Prostate cancer: There is no increase in your total expected lifespan from prostate cancer screening with the PSA test.

Cervical cancer: There is no increase in your total expected lifespan from cervical cancer screening with Pap smears.

Melanoma: There is no increase in your total expected lifespan from melanoma screening of the skin.

     Well, those are the facts. Now you can make your decision.

References:  Hoffman RM et al. Decision-making processes for breast, colorectal, and prostate cancer screening: results from the DECISIONS study. Med Decis Making 2010; 30(5 Suppl(): 53S-64S. doi: 10.1177/0272989X10378701