Does evidence based medicine do more good than harm?BMJ 2004; 329 doi: https://doi.org/10.1136/bmj.329.7473.1051 (Published 28 October 2004) Cite this as: BMJ 2004;329:1051
All rapid responses
Critical appraisal at the expense of critical thinking.
A 15 year old girl from an all-girls boarding-school was found to be
pregnant. She denied a history of sexual intercourse. We did this study to
see if there is any relationship between pregnancy and sexual intercourse.
Structured clinical question:
In girls (patient) is sexual intercourse (intervention) and pregnancy
Search strategy and outcome:
Medline 1966-February 2004 using PubMed : ‘sexual intercourse’ AND
‘pregnancy’. Limits: Unlimited and All languages
No relevant or irrelevant paper were found
Apparent Clinical Bottom Line:
No evidence was found to support the popular belief that pregnancy
results from sexual intercourse.
Our search strategy was similar to that used by Parikh and
Maconochie (1) who found no evidence to support the ‘glass test’ in the
diagnosis of petechiae . We found there was no evidence to substantiate
the belief that sexual intercourse is a prelude to pregnancy.
Reductio ad absurdum is a mode of argument that seeks to establish a
contention by deriving an absurdity from its denial, thus arguing that the
thesis must be accepted because its rejection would be untenable. We
contend that the EBM technique used for testing the glass test is
There is a small chance that a good double blind randomized placebo
controlled study was indeed done, looking at sexual intercourse and
pregnancy but it did not get past the peer review process and so we could
not retrieve it in our search of literature after 1966.
It is possible that the EBM penchant for excessive structure and
oversimplification (take for example the ‘structured clinical question’:
In a child with a rash (patient) does a positive glass tumbler test (test)
really pick up petechiae (outcome); It is worded as if feeble-minded
readers would confuse the patient and the test, unless identified in
parentheses) leads to use of ‘search terms’ that are inappropriate. To
test if changing the search terms can makes a difference to the yield in
this form of arm-chair-research, we decided to look for the term ‘non-
blanching rash’ AND ‘purpura’ in a search limited to the archives of the
Archives of Disease in Childhood after 2000, and it yielded the paper by
Wells et al (2) where the authors specifically studied ‘non-blanching rash
(defined by them as a rash that does not blanch when pressed with a glass)
There is little evidence that EBM improves patient-satisfaction with
treatment but there is emerging evidence that it can do harm. The New
England Journal of Medicine reports about Merck’s drug Rofecoxib that was
withdrawn (after more than 80 million patients had taken this medicine and
annual sales had topped $2.5 billion,) because of the risk of myocardial
infarctions and strokes(4). Initially, peer-reviewed-literature was
flooded by papers from the employees of Merck and their consultants. Meta-
analysis does not discount papers, with a declared conflict-of-interest.
It removed the label of bias and facilitated this fudge. Many
epidemiological studies showed concerns about myocardial infarction with
Rofecoxib, but Merck (like a good evidence-based-practitioner), claimed
that only randomized-controlled-trials were suitable for determining
whether there was any risk! The editorial concludes, “Given the
finding.... of an excess of 16 myocardial infarctions or strokes per 1000
patients receiving the drug (and that 80 million received the drug,) there
may be tens of thousands of patients who have had major adverse events
attributable to Rofecoxib”. We submit that, had it not been for EBM’s
farcical ‘hierarchy of evidence’ (4) cognizance of the epidemiological
evidence would have been taken, and the harm could have been avoided.
Indeed we see here, that EBM can perhaps do more harm than good.
It is unfortunate that medical science has allowed one system of
literature-search, to abrogate to itself the eponym ‘Evidence Based
Medicine’ to the exclusion of other sensible methods of finding the
relative harms and benefits of medical interventions. It is our
contention that this exclusive usage of the term EBM is pejorative of all
other methods and is therefore the stumbling block to evaluating the real
evidence. Critical thinking, rather than mindless critical appraisals, is
what is needed. Like Professor Bastion, we too are diehard-evidence-
enthusiasts (5) but we are not enamoured by EBM.
1. Parikh A, Maconochie I. What is the use of the glass test?
Arch. Dis. Child. 2003; 88: 1135.
2. Wells LC, Smith JC, Weston VC, Collier J, and Rutter N
The child with a non-blanching rash: how likely is meningococcal disease?
Arch. Dis. Child. 2001; 85: 218 - 222.
3. Topol EJ. Failing the public health -- Rofecoxib, Merck, and the
FDA. N Engl J Med 2004; 351:1707-9.
4. Hunink MGM Does evidence based medicine do more good than harm
5. Bastian H Learning from evidence based mistakes BMJ 2004;329:1053
Competing interests: No competing interests