Doctor says The New York Times killed his patient

In a Mediaite column, Dr. George Lombardi blames what he believes to be misleading science journalism in the New York Times for the death of one of his patients. The patient declined a PSA test, and in so doing, failed to learn that he had prostate cancer until it had progressed to a very advanced state.


  1. What?  The patient was following the advice of the U.S. Preventive Services Task Force concerning PSA tests:

    In October 2011, the USPSTF posted for public comment the draft of its recommendation regarding prostate cancer screening. Since then, Task Force members have read the many comments received and reviewed the most up-to-date evidence.  Based on this work, the Task Force concludes that many men are harmed as a result of prostate cancer screening and few, if any, benefit.  A better test and better treatment options are needed. Until these are available, the USPSTF has recommended against screening for prostate cancer.

    There wasn’t anything “misleading” about the New York Times article, as far as I can tell from the article.  The current recommendations are that men should *not* have PSA tests, since the harms (unnecessary surgeries and procedures from false positives) can outweight the benefits.

    This means that for some men, early cancer unfortunately will not be detected (as with the subject of this post).  This also means for many other men, they fortunately will not be subjected to unnecessary medical procedures, which can have a significant adverse effect on their health and well-being.  Anecdotes != evidence.

    1. The accusation that the article was ‘misleading’ does seem troubled (also, if your patient is taking health advice from the NYT, rather than doctors, or journals… um… they might have other problems).

      However, it is perfectly possible for the article to be based on the best available epidemiology and have killed his patient. At a population level, the test is deemed to be a bad idea because there are too many false positives who undergo too many side effects to make up for the true positives who are saved by treatment. If you, personally(since no one of us is a population), were in the ‘would have been saved by treatment’ category, the advice is bad. Unfortunately, we just don’t know how to identify you without giving more bad advice to the false positive contingent.

      1. Well, there is also the fact that treatment doesn’t seem to change mortality rate any. In the US, where prostate cancer is treated aggressively, the number of 60 year old white men who die of prostate cancer is pretty much identical to the number of 60 year old white men in Britain who die of prostate cancer, in a nation where the Cult of Prostate Testing never happened. So how many lives got saved, if the same number of men die? The reality is that roughly 40% of 60 year old men autopsied after death have prostate cancer, and most 60 year old men will die of something *other* than prostate cancer, because *in general* prostate cancer is not a killer. Of course, we have famous counterexamples like Frank Zappa where it *is* a killer, but would Frank Zappa have been saved by early and aggressive treatment? The numbers suggest not.

  2. The doctor who wrote the article doesn’t understand that his patient could have died because he didn’t have a PSA test, AND cutting back on the use of the PSA test is the right thing to do.   It’s basic statistical literacy.   If the test causes too many false positives, the overall effect could be negative and cause more deaths than it prevents.   His patient’s death would be countered by the lives saved due to the absence of those false positives.   The problem is with the accuracy of the test, not the USPTF. If he wants to complain, he should complain about the availability of a better, more accurate screening test.

  3. “Dont read the newspaper or try to understand statistics; I had a friend who died of cancer that way”.

    Christ, what an asshole.

  4. Of course,  the patient might have had the test, and STILL DIED. Amazingly enough earlier diagnosis does not always lead to better outcomes even among those who were properly diagnosed. 

    Of course the Doctor saved Mother Theresa in the late 1980’s! Oh wait,I guess even if you’ve been saved by this guy, death gets us all. 

  5. All I see in the article is pain and remorse. A doctor who genuinely watched his patient health. He is in pain for the loss, he is looking for the reasons that took the life of a friend. I can totally relate.
    Have none of you ever thought “Could I have done anything to prevent this outcome?”? 

  6. Statistics are great for looking at large populations.

    Statistics are terrible for making individual decisions.

    If an expert recommends something in a specific case, it’s foolish to prefer the generalized, statistical recommendation instead.

    1. What makes the expert an expert? Where is the judgment coming from? What overrules the general statistical evidence? The only reason to do so is if you have some evidence, beforehand, to reach a different conclusion. Is this person already more likely than average to get prostate cancer, and therefore a positive test result for him is more likely to be real than false? If not, then it is foolish to think that because someone has a degree (in medicine, not statistics or epistemology) they can make such a judgment. Intuitions are not magic, they come from some sort of thought process based on some sort of evidence.

      1. FTFA: “…since the DRE was normal no PSA blood test was sent.” So he WAS prescreened and the result of that showed no reason to do a PSA. QED.

        1. Correct. At at the time, based on the available medical evidence for this specific patient and the statistical evidence for the test, that was the best call to make. Maybe today we have better info on subpopulations for who should and shouldn’t get a PSA test. Maybe we are developing less dangerous tests and procedures to more accurately distinguish real from false positives and indolent from truly dangerous cancers. If so, great. If not, then there is no QED about it. It still sounds like in this case the doctor had no evidence that this person should receive different treatment than the general screening recommendation.

          There are no currently available strategies that result in zero deaths. But if you ignore the data that is available and perform treatments and tests that are known to cause more harm than good on average, you’re guaranteed to do worse and get more total deaths and suffering than you need to.

      2. Lots of juicy rhetorical questions, but I won’t fall into that trap.

        Statistics are good as guides, but fair consideration of what an expert says is important.

        For example, if a contractor says “Most homes in this area didn’t use asbestos when built, but I’d recommend a test in advance just in case, so we can avoid the problems that would occur if we find it during construction.”

        The statistics say “Don’t test”, but experience and/or the negative results of not testing can outweigh the choice.

        Experience of good and bad decisions make experts.  The patient disregarded the advice of his doctor due to what he had read in the newspaper, to his personal detriment.

        1 in 3 men get prostate cancer. That statistic alone should encourage men to be proactive in testing.  

        1. The salient point to me is that it’s just a blood test. He’s almost certainly having other things tested, so there’s not even any extra mini-invasiveness. The argument against testing is that it leads to other invasive procedures, but that’s a slippery slope argument.

          1. Correct me if I’m wrong, but I think the slope is only slippery until you include the numbers. When the experts study large populations, they find that the test will result in x real and y false positives, which will produce p saved lives of cancer victims and q unnecessary deaths of prostate-cancer-less people and r deaths from invasive procedures of people-who-have-prostate-cancer-but-wouldn’t-have-died-from-it. If p>(q+r), then perform the test on everyone. If p<(q+r), then only perform it when there is more-than-average reason (like a result from a DRE) that the person is likely to have prostate cancer.

        2. They weren’t rhetorical questions. I answered them myself after asking; you are free to do the same. In your example of the contractor, the only negative effect of testing is loss of money.The potential effect of not testing is life. In all likelihood the contractor is just trying to fleece you into paying for something you don’t need, but the worst thing you can lose is money.

          “The negative results of not testing can outweigh the choice.” The point – what the statistics tell us – is that no, they don’t, unless you have specific evidence otherwise in a particular case. The experts measured it. Your claim, in this particular case of the PSA test, is incorrect, based on the nature of medical technology at the time the data was collected.
          Ok, 1 in 3 men get prostate cancer. About 10% (3% of the total population) of them die from it. What fraction of men would test positive at least once if they got the test every year from middle age onwards? How many would die because of the negative consequences of false positives? And what of the 90% that have it but die from something else? In most of their cases they likely had it for a decade or more with little to no health effects.

          Yes, expertise comes from the experience of good and bad judgement *and feedback.* How many doctors carefully analyze what worked and what didn’t over every patient they treat, when a disease can take many years to run its course and no one doctor treats a large enough sample to get a representative population across all variables? What the statistics do is *aggregate* the experience of many experts to identify general principles that lead to good judgments.

    2. The trouble is that (for practical purposes, I leave the question of whether statistical judgements represent uncertainty about deterministic phenomena or genuinely stochastic phenomena to the physicists and philosophers) you are only an individual in hindsight.

      If I am about to roll a die, the best possible information you can have is that there is a 1/6 probability of a number 1-6, inclusive, coming up. Whatever your purpose in caring about the roll’s outcome, you don’t have better data than that, despite the fact that we all know that, once I roll it, the actual result will be 100% occurrence of a number 1-6, and 0% occurrence of the remaining numbers.

      It is true that using a non-representative population can lead to considerably worse results(as in trying to predict the behavior of a loaded die by analyzing the behavior of a fair one, or assuming that ‘psych 101 students doing studies for course credit’ are representative of humans generally); but pending a way to import data directly from the future, you don’t really have a choice about using statistical data to inform individual decisions.

  7. I’m the wife of a dying man who was diagnosed by PSA test at age 50. He had surgery, he had radiation, there is no chemo that provably kills prostate cancer faster than it kills the patient. So now we’re left with hormones and no idea how long they will continue to work. We keep checking for clinical trials, of course.

    Fortunately, except for the side effects of the treatment, he is asymptomatic at this point.

    Because his reoccurrence after radiation was within 18 months, he’s estimated (50/50) to have another five years-ish, of which half will be good years. Of course, it could be longer. It could be shorter.

    The one thing that’s certain is that having a PSA test: a) caught the cancer and gave him the best chance of early treatment; b) prolonged his life — even in this worst-case scenario where surgery and radiation did not, ultimately, cure the cancer. So far, that’s been one year of treatment and three asymptomatic years.

    The problem is that treatment for suspected prostate cancer is very tied up in male identity, and a big part of the “don’t test” movement is about the side effects of overtreating suspected cases. I’m guessing there’s also been more than one lawsuit over the issue.

    When my husband was told that intermittent hormone therapy had just been shown to be less effective than constant, they suggested he go home and think about it. He said no, he wanted the hormone shot now. Why give the cancer cells more of a chance?

    I cannot tell you how much it sucks not to know how much longer one will  have the love of one’s life, especially as someone who’d already been widowed once. True, we never know how long we’ll have, but why not ensure we get the best shot at life possible?

Comments are closed.