Commentary on medical news by a practicing physician.

  • Epocrates MedSearch Drug Lookup


    "When many cures are offered for a disease, it means the disease is not curable" -Anton Chekhov

    ''Once you tell people there's a cure for something, the more likely they are to pressure doctors to prescribe it.''
    -Robert Ehrlich, drug advertising executive.

    "Opinions are like sphincters, everyone has one." - Chris Rangel

    email: medpundit-at-ameritech.net

    or if that doesn't work try:


    Medpundit RSS

    Quirky Museums and Fun Stuff

    Who is medpundit?

    Tech Central Station Columns

    Book Reviews:
    Read the Review

    Read the Review

    Read the Review

    More Reviews

    Second Hand Book Reviews


    Medical Blogs


    DB's Medical Rants

    Family Medicine Notes

    Grunt Doc




    Code Blog: Tales of a Nurse

    Feet First

    Tales of Hoffman

    The Eyes Have It


    SOAP Notes


    Cut-to -Cure

    Black Triangle



    Kevin, M.D

    The Lingual Nerve

    Galen's Log



    Doctor Mental



    Finestkind Clinic and Fish Market

    The Examining Room of Dr. Charles

    Chronicles of a Medical Mad House



    Health Facts and Fears

    Health Policy Blogs

    The Health Care Blog

    HealthLawProf Blog

    Facts & Fears

    Personal Favorites

    The Glittering Eye

    Day by Day


    The Business Word Inc.

    Point of Law

    In the Pipeline


    Tim Blair

    Jane Galt

    The Truth Laid Bear

    Jim Miller

    No Watermelons Allowed

    Winds of Change

    Science Blog

    A Chequer-Board of Night and Days

    Arts & Letters Daily

    Tech Central Station





    The Skeptic's Dictionary

    Recommended Reading

    The Doctor Stories by William Carlos Williams

    Pox Americana: The Great Smallpox Epidemic of 1775-82 by Elizabeth Fenn

    Intoxicated by My Illness by Anatole Broyard

    Raising the Dead by Richard Selzer

    Autobiography of a Face by Lucy Grealy

    The Man Who Mistook His Wife for a Hat by Oliver Sacks

    The Sea and Poison by Shusaku Endo

    A Midwife's Tale by Laurel Thatcher Ulrich



    American Academy of Pediatrics

    General Health Info

    Travel Advice from the CDC

    NIH Medical Library Info



    Thursday, November 01, 2007

    Lies, Damned Lies, and, Well, You Know: The New York Times is parsing Rudy Guiliani's prostate cancer statitistics:

    "I had prostate cancer five, six years ago," Mr. Giuliani, a Republican presidential candidate, said in a speech that has been turned into the radio commercial. "My chance of surviving prostate cancer — and, thank God, I was cured of it — in the United States? Eighty-two percent. My chance of surviving prostate cancer in England? Only 44 percent under socialized medicine."

    .... The Office for National Statistics in Britain says the five-year survival rate from prostate cancer there is 74.4 percent. And doctors also say it is unfair to compare prostate cancer statistics in Britain with those in the United States because in the United States the cancer is more likely to be diagnosed in its early stages.

    "Certainly, if you intensively screen for prostate cancer, you will find early disease,” said Dr. Ian M. Thompson, chairman of the department of urology at the University of Texas at San Antonio. "And simply because you find it earlier, you will always have longer survival after the disease is diagnosed."

    One reason that prostate cancer is diagnosed earlier in the United States than in Britain is that they don't screen for it at all in Britain - at least not at the expense of the NHS. (Which is one of the reasons they spend less on healthcare than the United States. They don't indulge in as much screening as we do.) At any rate, his statistics don't appear to be all that far off the mark, at least for men in their 80's. But even the NHS admits that prostate cancer survival is increasing because more people are starting to have their PSA checked - meaning that slow growing early cancers are being added to the mix, just as happens here in the US. As it happens, even back in 2002, the five year survival rate for prostate cancer in the US was 99% - still a much better figure than the UK's 71%.

    But, as the astute bloggers point out, prostate cancer isn't the best example of the benefits of screening. Prostate cancer is, in most cases, slow growing - and although our screening policies detect many early cancers that would never do harm if left undetected, we also end up spending a lot more money treating these same cancers. When given the choice between watchful waiting and removal, many choose removal. (Another reason why we end up spending more and being less healthy in surveys like this.)

    But the Astutes take a closer look at cancer in England and cancer in the US:

    See this report, entitled "Cancer Survival Rates Improving Across Europe, But Still Lagging Behind United States" (and remember that England's rates, not broken out, are among the worst in Europe).

    Taking recent figures, female five-year cancer survival rates are 62.9 per cent on average in the US and 52.7 per cent in England. To compare America's privately insured with England's NHS patients, you'd need to bump up that American survival rate a bit (the uninsured most likely have lower survival rates--otherwise why worry about universal coverage) and bump down the English one (because some Brits have private insurance, and so buy better care).

    Nationally, American cancer survival rates are significantly better. Certainly not by the 40-point margin Giuliani implied, but still.

    Looks like the truth is somewhere between Rudy and the Times.

    posted by Sydney on 11/01/2007 09:43:00 PM 5 comments


    A more telling statistic is death rates of those diagnosed with prostate cancer.

    In the US, only 13% of those diagnosed with prostate cancer die of that cancer. In the UK, the death rate is 57%, France is 49% and Canada is 25%.

    You can see why the most demanded perk from UK employers is private health insurance.

    Also you will find no truth in New York Times medical coverage. Their goal is to bring socialized medicine to the US, and they will distort any or all facts to achieve that goal.

    By Blogger Jake, at 10:42 PM  

    Jake, as usual you show your total lack of understanding of basic epidemiology. Look up overdiagnosis and lead-time bias, also the effect of screening on the incidence of the desease. It is a known fact that screening that detects cancer early increases incidence of the desease This happens regardless of whether screening reduces mortality or not.

    There is no evidence if PSA results in reduction in mortality. Everyone knows that it results in overdiagnosis.

    Since men are screened with PSA in the US, there are more people in the US than in the UK diagnosed with cancer. This is expected since men are screened in the US and not in the UK. Some of these extra cases are due to overdiagnosis. The magnitude of overdiagnosis is difficult to estimate since you don't know how many of these additional cases will progress. These overdiagnosed cases are very easy to cure since they wouldn't have spread anyway.

    This is the reason why the mortality numbers per population is the only valid measure when looking at screened vs non-screened population, not mortality among people diagnosed. Incidentally, in randomized controlled trials designed to evaluate the effectiveness of screening only total number of people who die of cancer in both groups is considered a valid evidence. Survival is meaningless, as screening always increases the number of people diagnosed and it always increases the 5-year survival rate because of lead-time bias. If you are a doctor, than you really should know it.

    Here is a really simple examples why death rate of those diagnosed is meaningless. Let's say you have 8 people diagnosed without overdiagnosis, and 4 of them die. You get survival rate of 50%. Let's add a couple of cases of everdiagnosis, which are not life-threatening, so they are easy to cure. You get 10 people diagnosed, and still same 4 dead. 60% survival rate yet 4 people who really has cancer are still dead. Now, explain to me again how the death rate of those diagnosed is more telling than numbers per population?

    Given that we don't know if all extra cases of cancer in the US is due to overdiagnosis, we cannot really deduce anything at all from the survival rates. However, the fact that the death rate from prostate cancer per population in the US and the UK is close is pretty telling.

    Oh, here is a simple example of lead-time bias, since you obviously haven't heard of it either. Say two people die of cancer at the age of 50, but in one the cancer was diagnosed at 43 because of screening and in another at 47. The first person survives longer after diagnosis and is counted toward numbers of people who survive over 5 years. Another only lives 3 years after diagnosis. Do you think the first person is better off?

    I am amazed that even some doctors (like the guy who defended Giuliani's data) don't understand these simple concepts.

    For the record - I am not a proponents of NHS, I am sure it has a lot of problems. But I cannot stand use of an abviously misleading statistics.

    By Anonymous Diora, at 7:14 PM  

    Taking recent figures, female five-year cancer survival rates are 62.9 per cent on average in the US and 52.7 per cent in England.
    The recommendations for mammography are different between the US and the UK. So again you have more women screened in the US and in the UK, and there is likely to be more overdiagnosis in the US. Additionally, because of legal fears the US doctors may be likely to call questionable or borderline cases cancer. This is only an assumption, but it is a logical assumption, wouldn't you say?

    The population numbers between the US and the UK would be more interesting. Of course, you could say that there are lifestyle differences that affect incidence as well. Still, these numbers would be better than the totally misleading statistics above.

    God, why doctors fail to understand it? Do you guys study statistics?

    By Anonymous Diora, at 7:24 PM  


    Your ignorance of prostate cancer is astounding. I hope to God you are not in the medical profession.

    Due to equipment breakthroughs in the last five years, the treatment of choice for prostate cancer has to be radiation. If the cancer is caught soon enough, the cure rates are in the high 90s with little or no side effects.

    Contrast that with surgery where the cure rate varies from 90% for the best surgeons and 68% for the worst surgeons. Plus surgery has a high incidence of horrendous side effects. Also surgery costs twice as much as radiation.

    Go to the nanogram at the Sloan Kettering web site. Vary the amount of cancer found in a stick and vary the amount of PSA, you will see that early diagnosis is key to high survival rates.

    You should also know that slow growing prostate cancer can turn into rapid fire cancer overnight and there is no way of predicting when that will happen. You should also know that prostate cancer is the most painful way a man can die. So watch full waiting is a fool's game forced on men by government health program that don't have the resources to treat the disease.

    Rudy's stats are valid given the fact that Clinton is trying to force us into NHS structure that will force all men to play the fools game.

    By Blogger Jake, at 12:21 PM  

    "Your ignorance of prostate cancer is astounding"
    Jake, similarly is your ignorance of epidemiology which is what you attempt to argue about when you try to claim that cure rate is a more telling statistics. By the way, do you even know statistics?

    You are also changing the subject from which data is more telling to the subject of PSA screening and treatment, and then use survival data (not considered reliable because of lead-time and length biases) to claim that your formula is valid.

    You agree that there is no way to say which cancer will progress. So you admit, that there are some that do not. You don't know exactly which percentage of cancers progress. Do you realize that these cases that don't progress make cure rate (and survival rate) look better as my simple example showed?

    Forget PSA for a moment. If cured/detected is a valid measure, it should work for all cancers, right? Because you cannot really apply the statistics you like when it is convenient and ignore it when it is not.

    Keeping it in mind, let's look at a screening that we know (now) doesn't work - screening babies for neuroblastoma. This is considered an example of a screening that caused more harm than good. It is also a great example of why the cure rate is a bad indicator of screening's success.

    Briefly. Some years ago, in Japan, there was a program to screen kids under 1 for a potentially fatal and rare cancer. There was a test that could detect this cancer early when it was more curable. After this screening program was implemented, the number of children diagnosed with the desease more than doubled. Given that large number of additional cases all of which were cured, the cure rate looked much better. More diagnosed kids survived, so the survival rate looked much better as well. But there was one little problem. The same number of kids died. Additionally, the number of advanced cases diagnosed in older children wasn't reduced. These results were seen in other countries where this screening was attempted - more cases diagnosed, no difference in a number of kids that died in screened vs non-screened population. After a few years, the screening program was halted as it was clear that it was not saving kids' lives, but was harming healthy kids. But guess what - the cure rate for these early cancers looked great; after all most of them wouldn't have progressed anyway.

    It is very easy to show using simple algebra that even if screening for some desease doesn't save lives but results in overdiagnosis, it still increases the ratio of cured/detected as well as survival data. If a value can "look" better even in the absense of benefit, it cannot be used to demonstrate that benefit exists. It cannot be used to demonstrate the opposite either - it is just unreliable. To demonstrate that benefit exists, you need to use a formula that produces positive result when there is benefit and negative result when there is no benefit. Are you still with me?

    "Rudy's stats are valid given the fact that Clinton is trying to force us into NHS structure that will force all men to play the fools game."
    What???? Are you saying that we can use statistics in any way we want as long as it is for "the greater good"? Why did I even bother.... .

    By Anonymous Diora, at 12:09 AM  

    Post a Comment

    This page is powered by Blogger, the easy way to update your web site.

    Main Page


    Home   |   Archives

    Copyright 2006