Breast cancer mortality and screening: results of a randomized trial approved by the School of Public Health, University of Toronto, Toronto, Ontario M5T 3M7, Canada;

15 Feb

In 1980 the School of Public Health, University of Toronto, Toronto, Ontario M5T 3M7, Canada approved a randomized controlled trial: 89 835 women, aged 40 to 59, were affected at random to either an annual mammogram during five years or an annual physical examination without any mammogram during the same period of time. Now 25 years after the results are publicly available here:

And they are astonishing, so astonishing that we feel compelled to quote them:

“Conclusion Annual mammography in women aged 40-59 does not reduce mortality from breast cancer beyond that of physical examination or usual care when adjuvant therapy for breast cancer is freely available. Overall, 22% (106/484) of screen detected invasive breast cancers were over-diagnosed, representing one over-diagnosed breast cancer for every 424 women who received mammography screening in the trial.”

Meaning: not only searchers did not find any evidence of difference in the instantaneous risk of death from breast cancer between the two groups (hazard ratio not significantly different from one) but moreover what they found was over-diagnosis of breast cancers (ie 106 more cancers in the mammogram group even 15 years after the screening that is even when all the cancers of the non screening group should have been detected due to their natural evolution).

Those chilling facts have to be discussed, that is the least that we should do in the health services community given the budget allocated to breast cancer screening by mammograms.

Perhaps the mammographies in the eighties were not as sophisticated as those which are offered now? And in the contrary physicians in the eighties were perhaps more efficient than the 21century doctors at the physical examination of the breast?

Thanks to the incidental economist who gave me the news:

The effects of Expanding health coverage

8 Feb

The Affordable Care Act (aka Obamacare) is practically a laboratory experiment at the scale of a continent that allows health economists to observe the effects of expending the health coverage to a whole population (a thing that Europeans have done and that they call modestly Universal Disease Coverage, in French couverture maladie universelle or CMU). Starting from his reading of a
Congressional Budget Office (CBO) report the health economist Austin Frakt lists the incentives and disincentives to work that a mandatory health coverage creates. But in my view the point is: does the labor market need workers anymore, with or without health coverage? If it really needs workers then it would be better that they could afford care and rehabilitation, it is the interest of both the employer and the employee. If it doesn’t, the labor market will always consider that the costs are to high.
The blog the incidental economist is about health economics, below is the post with a link to the CBO report:


6 Feb


4 Feb

Medicine and particularly health care is always a matter of time: time needed for recovery, time until the cure is completed, time elapsed until relapse, survival time . Time questions the searchers in health services. What if an event has occurred at a time when nobody was present to attest the exact moment of its appearance? What if we know when a peculiar health condition has ended but not exactly when it has begun?  If a health condition (disease or good health depending of what the searcher is studying) is interrupted during a short lapse of time and then resumes, how to handle the interruption interval? Last but not least when a searcher has not enough time to devote to wait for the final result should he eliminate the entire observation? How mathematicians apprehend this curious entity which we name time?

Their mathematical answer is: interval censored data.

Many thanks to SAS and it’s programmers!

Happy are the Buddhists with their here and now philosophy ;-)


Enhanced by Zemanta

Flawed evidence based medicine

26 Jan

James Coynes is a professor of psychology who has dedicated an important part of his research field in denouncing the flawed evidence based medicine applied to the psychology practice. His writings in PLOs blog are an example of the critical appraisal we all should exercercise when we read such advertisements as “evidence supported”.
Here bellows is a link toward his posts in PLOs blogs:

Convenience sample

25 Jan

Convenience samples allow the researchers to easily have a first approach of what happens in a given population. But researcher has always to keep in mind that such a sampling method in addition of being easy is also always exposed to bias. Once a first approach have been made, it is mandatory that the results obtained by mean of a convenience sample be confirmed with a random sample or a clustered sample or a stratified sample. The two YouTube videos here bellows are indeed very helpful to apprehend the concept.

The unexpected reader

18 Jan

When writing a medical article the author has in mind that it will be read by his (her) colleagues or supervisors or more widely by other health professionals. But more and more the patients, with internet, have access to the scientific medical writing. Thus authors have to be careful about how they describes the subjects of their study. A respectful description of the study participants and of the people to which the authors aim to apply the results of their study is the less that is required for an ethical writing.

Professor Patt Thomson warns the authors about this sensitive topic in his blog here:

And Tessa Richards, a BMJ editor, plans to include representatives of patients in the team of BMJ reviewers here:

Have a good Week End.

conclusion mise-en-place. christmas present six

2 Jan


For an author the conclusion is a tricky section to write, be it in a thesis or an article. The two pitfalls that one has to avoid are to repeat what has already been said in the preceding sections and not to respond to the questions “so what ?” and “now what?”. The conclusion section is the place where one must describe what our findings imply for policy and practices at an operational level. At this stage one must not forget either to situate our article in the extant literatures. In conclusion the pillars of the conclusion section should be:
practice and policy implications,
place in the literature ,
weakness and further research to be done.
The two consecutive posts in the blog “patter” on the subject are worth to be read.

Originally posted on patter:

Any of you who watch cooking programmes will know the cheffy talk about mise-en-place. It’s a term used to describe all the various kinds of preparation that need to be done in order to whip up something that can be described as “freshly cooked to order”. In reality many restaurant meals have components that are precooked and cut into the right portion sizes – they need only to be added, heated, stirred and assembled, with a minimum of actual cooking time between order and service. That you don’t have to wait too long for your food is down to lots of mise-en-place.

The notion of mise-en-place is also helpful in thesis writing. There is a lot of preparation than can be done before a draft text is begun. And just as in cooking, the more preparation you do, the quicker and less painful the actual writing time involved.


View original 714 more words


25 Dec

There are numerous ratios in epidemiology, some of them being called the probability ratio, the proportion ratio, or in studies of existing disease, the prevalence ratio. Others can be called Odds ratio or relative risk or prevalence rate ratios or also risk ratio and last but not least hazard ratio. All those ratios aim to assess the risk difference between two sets of binomial health data. Below I selected a few articles with the aim to understand who is who among all those ratios.

Ten years after

19 Dec

Even ten years after, sound data are worth to be published. Below is an example. Actually a great Christmas story.


Get every new post delivered to your Inbox.

Join 82 other followers

%d bloggers like this: