Articles

True Informed Consent Is Elusive

Most of us would agree that doctors should not treat patients without their consent, except in special cases like emergency care for an unconscious patient.  It’s not enough for doctors to ask “Is it OK with you if I do this?” They should get informed consent from patients who understand the facts, the odds of success, and the risk/benefit ratio of treatments. The ethical principle of autonomy requires that they accept or reject treatment based on a true understanding of their situation and on their personal philosophy. Numerous studies have suggested that patients are giving consent based on misconceptions. There is a failure of communication: doctors are not doing a good job of providing accurate information and/or patients are failing to process that information. I suspect it is a combination of both.

An article in The New England Journal of Medicine reports that while the great majority of patients with advanced lung cancer and colorectal cancer agree to chemotherapy, most of them have unreasonable expectations about its benefits. For some cancers chemotherapy can be curative, but for metastatic lung or colorectal cancer it can’t. For these patients, chemotherapy is only used to prolong life by a modest amount or to provide palliation of symptoms. Patients were asked questions like “After talking with your doctors about chemotherapy, how likely did you think it was that chemotherapy would… help you live longer, cure your cancer, or help you with problems you were having because of your cancer?” A whopping 69% of lung cancer patients and 81% of colorectal cancer patients believed it was likely to cure their cancer, and most of these thought it was very likely.

False beliefs were:

  • More common in colorectal cancer than in lung cancer
  • Three times as high for non-white and Hispanic patients
  • Twice as high for patients who rated their communication with their physician very favorably.

There was no correlation between inaccurate beliefs and educational level, income, functional status, or the patient’s role in decision-making.

It’s ironic in a way. These patients thought chemotherapy was better than it really is, but generally chemotherapy has had a very bad press. It has been so demonized that some patients reject it out of hand, and those who accept it often find it nowhere near as bad as they had been led to believe. An e-mail correspondent wanted to label chemotherapy “quackery” because he said it only offered a 2-5% benefit for cancer survival. I had to explain to him that “quackery” refers to treatments with zero benefit, that it is meaningless to speak of cancer as a single entity, that the benefits of chemotherapy depend on the type and stage of cancer, that chemotherapy is curative for some types of cancer, and that it is also used for its adjuvant and palliative effects.

He also asked, “Surgery for cancer has a much greater success rate than 2-5% right?” I explained that it is meaningless to talk about the success rate of “surgery for cancer.” You have to look at specific types and stages of cancer. The success rate for breast cancer surgery in ductal carcinoma in situ stage 0 is 98% for 10-year survival. For pancreatic cancer, virtually all patients are dead within 7 years of surgery. It makes a huge difference whether metastasis has already occurred at the time of surgery. And some cancers, like blood cancers (leukemia, etc.), are not treatable with surgery.

I don’t think oncologists are deliberately lying to patients about chemotherapy. But I suspect they may be carefully choosing their words to put it in a more favorable light, since they naturally want to do something and to offer the patient hope. This may not be operating on a conscious level. And the very fact that they are offering chemotherapy to a patient gives it credibility, no matter what they say.

Whether or not the oncologist offers subtly biased information, the patients’ own biases contribute to poor communication. They want to survive. They want to have hope. It is only natural for them to put a positive slant on what they are told; they may not try to understand the negatives or they may minimize their importance. In some cases, they simply refuse to hear the facts and continue to believe what they want to believe.  This study suggests that patients perceive physicians as better communicators when they convey a more optimistic view of chemotherapy.

Should we accept these misunderstandings because they give patients hope? Should we be concerned that they have not met the standard for giving informed consent for their treatment? Other studies have shown that patients with advanced cancer would accept toxic treatment for even a 1% chance of cure but would be unwilling to accept the same treatment for a substantial increase in life expectancy without cure. So if they are accepting chemotherapy on the basis of these misunderstandings, they are not doing what they say they want to do. The misunderstandings might even interfere with end-of-life planning and care.  Previous studies have shown substantial discrepancies between patients’ and doctors’ estimates of their life expectancy.

Another recent study published in the Annals of Family Medicine showed patients have similar misunderstandings about the value of screening tests and preventive treatments.

  • 90% overestimated the effect of breast cancer screening
  • 94% overestimated the effect of bowel cancer screening
  • 82% overestimated the effect of hip fracture preventive medication
  • 69% overestimated the effect of preventive medication for cardiovascular disease.

Doctors also have some serious misunderstandings in these areas. Ina study by Gigerenzer et al. researchers did a survey of 160 gynecologists attending a continuing education session in 2007. They described a patient who has a positive mammogram and who asks her gynecologist whether that means she has cancer for sure, or what the chances are. What would they tell her? They were given the pertinent information (1% prevalence in that population, 90% of women with cancer test positive, 9% of women without cancer test positive). They were given a multiple-choice question with 4 answers: 81%, 90%, 10%, and 1%. The correct answer, which could be easily calculated from the statistics provided, was 9%. The other answers were wrong by an order of magnitude. They could derive the answer from the statistics or they could simply recall what they should have known anyhow. The gynecologists gave answers ranging from 1% to 90%; the majority of them grossly overestimated the probability of cancer as 90% or 81%, and only 21% answered correctly (not even as good as chance). I don’t know about you, but I find that positively frightening.

Conclusion

Doctors and patients have misconceptions about the value of chemotherapy, screening tests, and preventive measures. And probably a lot of other things! If patients are to give truly informed consent, doctors must have accurate science-based information and must find ways of effectively communicating their knowledge to patients without losing their patients’ trust and regard and without destroying hope.

 

Posted in: Cancer, Medical Ethics, Pharmaceuticals

Leave a Comment (42) ↓

42 thoughts on “True Informed Consent Is Elusive

  1. nybgrus says:

    The problem is that most medical students don’t understand or like statistics. EBM and ethics are the two areas of med school that are traditionally trudged through just to barely pass so we can do the “other stuff.” I cannot tell you how many med students, residents, and even fellows I have met who just throw up their hands and say with exasperation that they just don’t “get” math and just to give them the conclusion of a study so they can practice however it says. I’ve even had a heme-onc attending once comment that a Bayesian framework is “hard” and she never bothers with that and a resident with a PhD misunderstand the implication of a Bayesian prior changing the p-value and thus the significance of a study.

    If med students don’t know it, they don’t care to learn it as residents, how can we expect attendings to convey it well to patients? The true mark of understanding something is being able to teach it to someone else.

    My “just-so” hypothesis is that to go to medicine you need a lot less math than any other rigorous scientific field. Physics, engineering, chemistry, even evolutionary biology is a lot of math. So otherwise smart kids with deficiences in math and no desire to make up for it go into medicine.

    I myself only had to take one upper division calculus course to complete my degree. I had taken stats and lower division calculus as AP courses in high school and received credit for them. I continued to learn anyway, and ended up playing poker for almost 4 years every Tuesday with PhD’s in statistics and economics. Then I played semi-pro for 6 months after graduating. I can tell you that a lot of the lessons learned and principles of being a succesful poker player translate to medicine. Statistical knowledge, making best decisions with incomplete information, pot-odds (aka odds ratios), positive and negative predictive value of various plays and actions, all translate very well into practicing medicine.

  2. BillyJoe says:

    Harriet,

    Quick, correct this and delete my post before anyone notices…

    ” They described a patient who has a positive mammogram and who asks her gynecologist whether that means she has cancer for sure, or what the chances are. What would they tell her? They were given the pertinent information (1% prevalence in that population, 90% true positives, 9% false positives). They were given a multiple-choice question with 4 answers: 81%, 90%, 10%, and 1%. The correct answer, which could be easily calculated from the statistics provided, was 9%”

    The information you provided here is wrong.
    There cannot be 90% TP and 9% FP, because then TP + FP = 99%. But TP + FP must = 100%
    I think you mean: 1% prevalence in that population, 90% specificity, 9% false positives.

    Also your answer is wrong (especially as it is not even one of the multiple choices).
    The answer is actually closer to 10%.

    1% prevalence in that population: 1000 women -> 10 have cancer + 990 do not have cancer
    90% specificity: 10 have cancer -> 9 have cancer and test positive
    9% false positives: 990 women do not have cancer -> 89 do not have cancer but test positive.
    So, a total of 98 test positive but only 9 actually have cancer = ~10%

    Of course, if I am wrong….delete my post anyway :D

  3. Janet says:

    @Billy Joe

    I am terribly math-challenged, but does 1% make a lot of difference in this case?

    @HH

    The longer I read this blog, the less confidence I have in my doctors–not in the CAM way, but rather a developing lack-of-confidence in medical education. Should I start asking them more about their pre med and medical education?

    Nybgrus reports disturbing trends–I was one of those “smart kids” who was very bad at math, but I had the sense to stay out of the hard sciences. Fortunately, I got an excellent grounding in the history of science which in some ways may put me ahead of some of these “smart kids”. What is going on with med school entrance qualification anyway? I have heard about the movement to exclude science in general, which is hopefully a joke, of sorts!

  4. rork says:

    “But TP + FP must = 100%” Not at all, thank goodness too. For a good test TP+FP approaches zero.
    Imagine 100 people. We expect positives 99*.1 + 1*.91 = 9.9+.91 = 10.81 times where the first term (9.9) are false positives and .91 are true positives. .91/10.81 = 9.9 which is nearly 10%.

  5. rork says:

    Sorry, my first sentence sucks. TP +FP here is 10.8%. I was thinking FN+FP.

  6. stanmrak says:

    Keep in mind that disease care is a for-profit industry. The people providing the technology and facilities to treat disease are interested in making more profit, so they will paint as rosy a picture as possible to encourage doctors to promote their services and patients to buy more of them. It’s just like selling anything else. Even tho we’re talking about people’s lives and health, it’s a business first to corporate execs.

  7. rork says:

    That is, I think it’s just an english problem. Take “true positives” to be the true positive rate – the % of people with cancer who test positive. “false positive” to be false positive rate – the % of people without cancer who test positive. It’s how I read it, but yeah, perhaps not perfectly clear.

    I was surprised by no correlation to education, and that leads to a possible hypothesis that it’s mostly about what the docs tell patients. How can it be that a doc can fail to say what the sizes of the various benefits are in clear terms? That’s “inconceivable”, in the Princess Bride definition of the term.

  8. jpmd says:

    Some of the problem lies with terminology. There is really no such thing as a “positive mammogram” in that there are gradations of suspicion as to abnormality. I think the more valid statistic to present to a patient is that only about 10 % of biopsies are positive when the mammogram leads to a biopsy. However, everyone knows that there are mammograms which you can bet money will be cancer, and those that while needing to be biopsied, are probably benign.

  9. Harriet Hall says:

    @Billy Joe,

    I have revised the statement to make it clear. 90% of women with cancer test positive; 10% of women without cancer test positive.

    From the numbers they used, the authors of the study calculated 9%. See their figure 3. Nine out of 98 is 9.18%, closer to 9% than to 10%. Their multiple choice question phrased the option as “about” one out of ten, which is perfectly reasonable.

    See how difficult this is? I understand the statistics, but even I failed to communicate clearly what I meant. No wonder the average clinician and patient get confused! The “natural frequencies” approach on the right side of figure 3 is the easiest to understand.

  10. Harriet Hall says:

    @jpmd,

    I think the common understanding of a “Positive” mammogram is one where the patient is told “something doesn’t look right, we need to investigate further.”

  11. WilliamLawrenceUtridge says:

    Keep in mind that disease, chiropractic, acupuncture, homeopathy and herbal care is a for-profit industry. The people providing the technology and facilities to treat disease are interested in making more profit, so they will paint as rosy a picture as possible to encourage doctors to promote their services and patients to buy more of them. It’s just like selling anything else. Even tho we’re talking about people’s lives and health, it’s a business first to corporate execs and individual practitioners.

    Fixed that for you. Unless chiropractors, acupuncturists, herbalists and homeopaths all provide their goods and services free of charge?

  12. Harriet Hall says:

    @WLU,

    More than that needs fixing. He is talking about the people providing the technology and facilities painting a rosy picture to encourage doctors. That has nothing to do with the subject of my post.

  13. gears says:

    Are those the reals numbers for mammogram specificity and selectivity? If so, that’s kind of frightening to me, too.

    If I get into medical school, first thing on my to do list is to get a good handle on statistics. I have never really understood why other pre-med types that I knew were so math-averse (and many were physics and chemistry averse, as well).

    I don’t want to stereotype, though. We should do a study.

  14. gears says:

    @Janet,

    I had a organic chemistry professor who joked that he always asked his doctors what grade they got in organic chemistry. Not sure how well that went over, though.

  15. Quill says:

    The article and this post explore the thinking behind consent but I think the emotional factor needs to be considered otherwise solutions won’t be possible. Fear is a big thing and fear of death, that inevitable happening few want to talk about much less think about, clouds all of these discussions between doctor and patient. I would logically suspect that this fear is a factor in this overestimation of chemotherapy benefits.

  16. geo says:

    “I don’t think oncologists are deliberately lying to patients about chemotherapy. But I suspect they may be carefully choosing their words to put it in a more favorable light, since they naturally want to do something and to offer the patient hope. This may not be operating on a conscious level. And the very fact that they are offering chemotherapy to a patient gives it credibility, no matter what they say.”

    I think that this could be fairly seen as a form of quackery. Homoeopathy may be able to provide some relief to patients by inducing a sense of being cared for, and so long as this is done with informed consent, it could be viewed as an acceptable treatment (I am not saying that the ritualistic aspects of homoeopathy are useful to patients once it’s nature has been explained to them, but just using this as a hypothetical). When it is sold as doing anything more than this, then it becomes quackery. In the same way, I think that over-sold science based treatments can be quackery.

  17. elburto says:

    Stan – You are aware that America is merely one country, and that a whole world exists outside it’s borders, aren’t you?

    The US healthcare system is an outlier. NHS medical staff, for instance, are salaried. So a GP could potentially diagnose every patient with a lifelong condition that requires multiple medications and treatments, or tell every patient “You have a virus. It’s self-limiting, and you’ll feel much better within the week”. In either scenario the GP will take home the same amount of money.*

    An NHS pharmacist will not take home more money dispensing novel, branded drugs to patients, than if they were dispensing generic paracetamol (acetaminophen) worth pennies.

    A consultant oncologist will make the same basic salary whether they tell every patient that they have cancer, or tell everyone that they don’t.

    It’s not piecework like assembling toys in a factory, it’s not commission-based like selling cars. But then again, even American healthcare professionals are not so morally bankrupt that they view their work that way.

    Your worldview is so childlike and unsophisticated that I have secondhand embarrassment for you. The world is grey, not black or white.

    *There are GP bonuses for reducing the number of patients who smoke, properly detecting dementia, properly detecting and treating hypertension, providing weight-loss clinics that educate. about diet and exercise etc.

  18. BillyJoe says:

    Harriet,

    “1% prevalence in that population, 90% true positives, 9% false positives”

    “I have revised the statement to make it clear. [1% prevalence in that population,] 90% of women with cancer test positive; 10% of women without cancer test positive.”

    My point was that “90% of women with cancer test positive” means “90% specificity” not “90% true positives”
    Also “9% false positives” means “9% of women without cancer test positive”, not 10% as you have above.
    I did get the math wrong at the end though, goddammit, 9 out of 98 is closer to 9% than 10%

  19. windriven says:

    @stanmrak

    “Keep in mind that disease care is a for-profit industry.”

    Do you have a job? Do you work for free?

    How could, say, an oncologist do her work without compensation? Why would corporations build hospitals without compensation?

    For profit does not necessarily mean without ethics.

  20. BillyJoe says:

    rork,

    ” Take “true positives” to be the true positive rate – the % of people with cancer who test positive. “false positive” to be false positive rate – the % of people without cancer who test positive. It’s how I read it, but yeah, perhaps not perfectly clear.”

    Yes, you are correct, “true positives” usually means “the number of true postives”, not “the true positive rate”. When Harriet said “true positives” she meant “the true postive rate”.
    I was correct when I said “TP + FP = 100%”, but “TPR + FPR =\= 100%” as you pointed out.

  21. WilliamLawrenceUtridge says:

    “Without ethics” doesn’t necessarily mean “worse for patients” either. Arguably, previous standards for informed consent (don’t tell the patient what they have, particularly if they are a woman – because they might have vapours) could produce superior outcomes because patients would do what their doctors told them without questioning it. In some cases (i.e. any patient who thinks CAM is a treatment for cancer) this would produce superior outcomes. Increasing patient informed consent has definite benefits and from an ethical standpoint is very good practice – but itself has consequences that affect outcomes. I’m not arguing for the good ol’ days of paternalistic care (which almost certainly let terrible doctors keep practicing, and would be an even greater problem given the current complexity of modern medicine). But informed consent definitely makes research and practice slower, harder and takes some of the control out of the hands of a genuine expert. If I ever get sick, if I ever get cancer, I plan on basically asking my doctor what s/he thinks is the best option and giving that answer a lot of weight.

    Plus, as a drone of Big Pharma, I must blindly obey any option that feeds more profits into the hands of our all-might overlords.

  22. lilady says:

    I’m going to agree with “Quill” about the emotional factor and its impact on patients’ unrealistic “hopes” about the palliative chemotherapeutic treatments for Stage IV cancers…

    The NEJM article that Dr. Hall linked to, states this about the patient profiles,

    Methods

    We studied 1193 patients participating in the Cancer Care Outcomes Research and Surveillance (CanCORS) study (a national, prospective, observational cohort study) who were alive 4 months after diagnosis and received chemotherapy for newly diagnosed metastatic (stage IV) lung or colorectal cancer. We sought to characterize the prevalence of the expectation that chemotherapy might be curative and to identify the clinical, sociodemographic, and health-system factors associated with this expectation. Data were obtained from a patient survey by professional interviewers in addition to a comprehensive review of medical records.

    Would patients who were diagnosed with early stage cancers, who did not respond to multiple chemotherapy regimens, have these unrealistic expectations/”hopes” about palliative chemotherapy?

    Based on nothing, but my experience with a few friends who were first diagnosed with late stage cancers…or who had close relatives who received that diagnosis, IMO, patients and their loved ones, do have unrealistic “hopes” for treatments that are palliative in nature.

    (Warning-Anecdotal). Two close friends have confused “remission” with cure. In both instances my close relationship with the sister of a terminally ill breast cancer patient and with my friend who was diagnosed with multiple myeloma five years ago and who is still alive, have had a positive effect on their lives.

  23. Pemphipug says:

    Memories of being a medical student. I would have to draw it out as a chart. Take 1000 women.

    ——- | cancer | no cancer
    test +| A | B
    test – | C | D

    On prevalence of 1%, A+C = 10. A+B+C+D = 1000. So B+D = 990.
    If 90% of people with cancer test positive, A = 90% * (A+C) = 9, which means C = 1.
    If 9% of women without cancer test positive then C = 9% * (B+D) = 0.09 * 990 = 89.1.
    Therefore D = 990-89.1 = 900.9.

    ——- | cancer | no cancer
    test +| 9 | 89.1
    test – | 1 | 900.9

    SENSITIVITY, defined as the proportion of people who have the disease and score positive on the test is A/(A+C) = 90%. (No surprise as this was defined by the question).

    SPECIFICITY, defined as the proportion of people who do not have the disease and score negative on the test, is D/(D+B) or 901/990 = 91%

    But the question is actually asking is what is the positive predictive value of the test – the chance that a person with a positive test actually has the disease. This is (A/A+B) and is 9.1%.

    Betcha on the quiz they were getting confused as to the difference between sensitivity of the test and positive predictive value, because it almost confused me too…

  24. Woody says:

    “See how difficult this is? I understand the statistics, but even I failed to communicate clearly what I meant. No wonder the average clinician and patient get confused! The “natural frequencies” approach on the right side of figure 3 is the easiest to understand.”

    This encapsulates the problem. It is difficult to walk that fine line between giving enough information to help guide patients to an informed decision versus overwhelming them with too much information. For example, I just finished counseling a family about a dementia diagnosis I had made for an elderly family member. A relatively straightforward question about anticipated rate of progression could not be truthfully answered in a black and white manner because there is obviously a range. This led to discussions about genetic risk, lifestyle interventions, support services, etc. Thirty minutes later, we hadn’t even tackled practical questions like medication management, and this was ~15 minutes beyond when the appointment was supposed to be over, at least according to my clinic calendar! Quite frankly, if I don’t steer the conversation somewhat (paternalistic, I know), every clinic day would be a disaster in terms of running behind, which certainly isn’t fair to the patients scheduled at the end of the day.

    This issue reminds me of a Bertrand Russell quote:

    “To be perfectly intelligible one must be inaccurate, and to be perfectly accurate, one must be unintelligible.”

  25. pmoran says:

    The gynecologists gave answers ranging from 1% to 90%; the majority of them grossly overestimated the probability of cancer as 90% or 81%, and only 21% answered correctly (not even as good as chance). I don’t know about you, but I find that positively frightening.

    I also suggest that they were simply responding to the question as clinicians, not recognizing that they were being asked a hypothetical statistical question that would never arise in such a simple form in medical practice. As others have pointed out a “positive mammogram” has no fixed clinical meaning. The correct answer to any individual patient’s question concerning the likelihood of cancer therefore would range from 1% to 99%, depending on numerous factors..

    Those answering 1% would obviously be thinking of the likelihood that “any mammographic abnormality” represented cancer. Those giving very high figures would be placing a different interpretation on the word “positive” or perhaps thinking of the diagnostic accuracy now possible within the most sophisticated screening clinics after multimodal assessment..

    It is reassuring that once the doctors were wised up as to the statistician’s train of thought, they performed much better. Whether the statisticians understood clinicians better, so as to most fairly evaluate such findings as these is another matter.

  26. weing says:

    @pmoran,

    I trust your THR went well.

  27. Harriet Hall says:

    @pmoran,
    “I also suggest that they were simply responding to the question as clinicians, not recognizing that they were being asked a hypothetical statistical question that would never arise in such a simple form in medical practice.”

    I disagree. The question does arise in such a simple form in practice, maybe not to the doctor, but certainly to the patient. I had cluster of calcifications on a mammogram and as a patient my first question was “What is the likelihood that I actually have cancer? How worried should I be prior to the biopsy?” I looked up the numbers and was reassured to find that the overall likelihood was 10% or less. This greatly reduced my worries prior to the excisional biopsy, which turned out to be benign. If I had asked my doctor and he had said 81% or 90%, I might have had sleepless nights. If I had asked him about my individual likelihood, considering the specific number and type and location of my findings, he would not have had any reliable statistics to guide him to an answer. If he had said between 1% and 99%, I would have thought him worse than useless; and if he had tried to be more precise he would have been guessing.

    I don’t think I’m an anomaly. I think most women would like to have an estimate and would realize (or should realize) that it was a rough ballpark guideline. Similarly, most terminal patients want to know how long they have to live. The doctor can’t make any promises; but he should be able to tell them the average life expectancy for a group of patients with similar cancers in a similar stage, and he should be able to explain what that means.

    As a surgeon, when patients asked you about the risks of a proposed operation, I hope you could provide some overall estimates of the rate of complications. Or did you say “Some patients have no complications, others have a lot of complications depending on many factors.”? Or did you try to give them an estimate of complications for someone who was exactly like them: same age, weight, on the same medications, with the same other medical problems, etc.”? (Which would be an informed guess based on your experience rather than an evidence-based answer.)

  28. Harriet Hall says:

    @Woody,

    You bring up some excellent (and very important) points. A surgery consent form can’t list every possibility (the surgeon could have a massive heart attack and collapse into your open abdomen). Patients want enough information to make a decision, but they don’t want to be burdened with all the minutiae of precise details. Time constraints are a real problem. Some patients are unable to understand and you can’t be expected to give them an entire education. Some don’t want to understand. We’ve all had patients who wanted us to decide for them or who asked “What would you do if you were in my shoes?”. This is part of what is meant by the art of medicine, and we could talk about it forever. The point remains: we need to understand the available evidence and be able to communicate to patients what they need to know without leaving them with misconceptions.

  29. Sawyer says:

    Thank you so much for this post, especially for referencing the Gigerenzer study. I’ve heard a lot of these issues discussed separately on Science Based Medicine, but it’s nice to be able to refer to a single paper that covers so much ground. It’s a huge boon anytime I can learn about these topics from a statistics or risk vs. benefit perspective since my medical knowledge is limited. Obviously not every patient (or doctor) shares my appreciation of Bayesian analysis. Still, I think it’s a lot easier to get someone up to speed on how to talk to their doctor using statistics than to try to teach patients 3 years worth of anatomy, biochemistry, etc.

    Major props to Dr. Hall for this one.

  30. pmoran says:

    I disagree. The question does arise in such a simple form in practice, maybe not to the doctor, but certainly to the patient. I had cluster of calcifications on a mammogram and as a patient my first question was “What is the likelihood that I actually have cancer? How worried should I be prior to the biopsy?”

    I stand by what I have said. A cluster of microcalcifications is a far cry from the question posed to the gynaecologists regarding a “positive mammogram”. I am sure no surgeon dealing with breast cancer would be unable to give a rough idea of the odds of cancer in your case using such factors as number, size, shape, association with a mass lesion and other factors, and without reference to pooled statistics based upon unstated presumptions as to what a “positive mammogram” is.

    When I retired from medical practice experienced screening clinics were already able to achieve a positive to negative biopsy rate of cancer of one in three or less, based purely on the evidence available from breast imaging and with very low rates of missed cancers. I am not sure what it is now.

    I am not against informing patients or defending poor use of available data by doctors. I just think this is a poor example, being a little unfair on the participants.

  31. Harriet Hall says:

    @pmoran,

    You are moving the goalposts by talking about the rate of positive biopsies. That was not the question.

    The example of mammograms and gynecologists was a pretest in a course designed to educate doctors about statistics. They were asked to respond to a patient’s simplistic question about overall odds from an unspecified “positive” mammogram, a question that I think commonly comes up in practice. The question is asked before other tests are done: repeat mammograms, ultrasounds, needle biopsy, etc. Not all “positive” mammograms end up with biopsy. The gynecologists were asked about the odds before any further investigation, which is very different from asking a surgeon the odds of a biopsy being positive.

    You say a surgeon could give me a rough idea of the odds of cancer in my specific case, but that would be his educated guess, not an accurate science-based answer. I think the most honest answer would be to give the available evidence-based odds and then say he would guess that my odds would be better than that average because factors x, y, and z put me at the low end of the spectrum.

    I would be interested in updated information on the positive/negative rates for biopsy and the rates of missed cancer, if anyone can provide it.

  32. pmoran says:

    I was not on thhis occasion spoiling for an argument. I was giving a very likely explanation as to why these researchers were able to produce results that you found “positively frightening”.

    I personally also found them counterintuitive, since even non-specialist doctors will have some idea of the likelihood of cancer in any particular breast lesion in any particular patient and can convey that to the patient if asked. So I thought about the methodology. It seemed clear, to me at least, what was going on.

    The statistician’s question was also of scant relevance to any individual patient’s question in clinical practice. Even in your considerably more precise example of a cluster of microcalcifications the true likelihood of cancer in any particular patient will range from very low to very high simply from the radiological features.

    I personally would avoid the term “positive mammogram”. Apart from its lack of clear meaning, most patients would interpret it meaning “positive for cancer”.

  33. Harriet Hall says:

    @pmoran,

    “I personally would avoid the term “positive mammogram”.”

    I would too. It is better to say there were findings on the mammogram indicating that further investigation is needed. Nevertheless, many patients will hear that as “my mammogram was positive; I probably have cancer.” There is much to be said for direct doctor-patient communication of mammogram results rather than a routine postcard in the mail with blocks checked to say either the mammogram was normal or the mammogram showed findings that require followup, contact your doctor.

    Even if “the statistician’s question was of scant relevance to patients’ questions,” the point of the example was to find out how well doctors understand statistics. It clearly showed that most of them don’t. For a doctor to convey the likelihood of cancer to an individual patient, he surely needs to have a better grasp of how to interpret published evidence. If they can’t answer this simple question based on the facts provided, how can we expect them to answer the patient’s questions in a more complicated situation when there are fewer data? Don’t you find it frightening that these gynecologists were not able to answer a simple test question about statistics?

  34. pmoran says:

    Don’t you find it frightening that these gynecologists were not able to answer a simple test question about statistics?

    Not very. Not yet. I would expect the majority of doctors in current practice to be as weak on statistics as I admit to be, but not nearly so badly as this study is being held to show.

    I would like the study performed again using a better-designed question and with less ambushing of the ill-prepared. The clinician’s minds may still be locked into “this patient, here” mode and thus all at sea when being asked to offer an opinion with none of the normally available clinical information. I don’t agree that this is reflective of daily medical practice. When have you ever see a mammogram report that includes, let alone stops at, the word “positive”?

  35. Harriet Hall says:

    Mammogram reports don’t say “positive,” but that is what patients often hear. The idea of a “positive” mammogram is firmly ensconced in the public mind. As an example, here is a typical popular article about “when your mammogram is positive.” http://www.blufftontoday.com/bluffton-news/2011-10-05/when-your-mammogram-positive#.UK27p2nuUzE
    I don’t see the question as “ambushing the ill-prepared.” Doctors should be prepared to answer questions like this based on the kind of information they were given in the study. Whatever the actual facts about breast cancer, situations constantly arise in medical practice where doctors need to know how to express the odds to patients, and how to determine the odds from specificity and sensitivity data. I hope you’re not advocating that doctors “inform” their patients by giving opinions based on their clinical experience rather than on statistical data. If you object to the term “positive mammogram,” at least you must understand the concept of true positive and false positive tests.

  36. Amalthea says:

    It looks getting the correct balance while trying to inform patients is an art that will never be perfected as each situation and the people involved will differ.

    As far as cancer and informed consent go I know a family that is puzzled about the “why” of a doctor’s choice. An adolescent male was diagnosed with AML a few years ago. He received a bone marrow transplant in Cincinnati and now needs to be sent back for a second one. This time the doctors told the family that they are specifically looking for a young girl as the donor but didn’t explain the reason for their preference.
    Since part of the aging process is accumulated errors in cell division I can see that as possibly being one of the reasons for preferring a young donor for a young patient but the rest has me confused to. Does anyone here know what the logic behind this choice might be?

  37. BillyJoe says:

    Amalthea,

    The information came to you third hand, so it might have become distorted.
    There is a shortage of male donors so there may not have been a preference for a young female donor, so much as a likelihood that the donor would be a young female rather than a young male. They might have been looking for a reaction by the male recipient to having a bone marrow transplant from a female donor.

  38. Harriet Hall says:

    “They might have been looking for a reaction by the male recipient to having a bone marrow transplant from a female donor.”

    My cousin had a kidney transplant from a female donor; he jokes that now he gets the urge to pee sitting down.

  39. Amalthea says:

    Thank you, that does make sense.
    I suspect that, if things work out, he won’t mind the gender of the donor, though his brothers might tease him about it.

  40. David Gorski says:

    When have you ever see a mammogram report that includes, let alone stops at, the word “positive”?

    A nit to pick: Actually, mammograms are not read as “positive” anymore, if they ever were. The reading of mammograms has been largely standardized in a way that they weren’t 25 years ago when I first started my surgical residency. They are now interpreted according to the BIRADS classification and assigned to one of six categories:

    BIRADS 0 = Needs additional imaging; imaging incomplete.
    BIRADS 1 = Normal
    BIRADS 2 = Benign finding
    BIRADS 3 = Probably benign finding. (BIRADS 3 studies generally have less than a 1-2% chance of malignancy. Usually a short interval followup mammogram is recommended six months later.)
    BIRADS 4 = Suspicious. (BIRADS 4 studies as a group have an approximately 15-25% chance of malignancy. Biopsy is almost always recommended.)
    BIRADS 5 = Highly suspicious. (BIRADS 5 studies portend a 70-95% chance of malignancy. Biopsy is always required.)
    BIRADS 6 = Known malignancy. (BIRADS 6 is usually applied to imaging taken after a biopsy that reveals a malignancy. This is usually done as part of surgical staging or in order to localize the clip that is left behind after a needle biopsy.)

  41. Harriet Hall says:

    The doctors in the study were not given BIRADS results; they were asked a question that the researchers, and I think most people, would interpret as including every category but BIRADS 1. Of course, it would have been a better study if it had defined what it meant by a “positive” mammogram, or if it had simply made it clear that it wanted doctors to consider a hypothetical scenario based only on the numbers given in the example; but the study as carried out served its purpose: it demonstrated that doctors could not interpret the data they were given about prevalence, sensitivity, and specificity.

  42. etatro says:

    This conundrum comes up frequently with incidental MRI findings in research studies. There are many MRI or fMRI research studies and in normal, healthy, young adults, there is a rate of incidental abnormalities. How do (or should) you report this to the subject? What if they are a minor? It’s an interesting ethical question, one that I think should be discussed more often than it is.

Comments are closed.