Bayes theorem can only be applied properly when considering a single
diagnosis or its absence e.g. the presence or absence of asymptomatic
diabetes mellitus, hypothyroidism, hyperlipidaemia, cervical cancer,
breast cancer, etc. These asymptomatic situations are very important in
the clinical setting and also in the community. It is in these situations
that the reasoning described in this paper can be...
Bayes theorem can only be applied properly when considering a single
diagnosis or its absence e.g. the presence or absence of asymptomatic
diabetes mellitus, hypothyroidism, hyperlipidaemia, cervical cancer,
breast cancer, etc. These asymptomatic situations are very important in
the clinical setting and also in the community. It is in these situations
that the reasoning described in this paper can be used. However, much of
our time is taken up with differential diagnosis, which cannot be modeled
by Bayes theorem (which explains why it is not often used as pointed out
by the authors).
When dealing with differential diagnoses, we must first choose a good
diagnostic lead [1, 2, 3] (or pivot [4]) e.g. acute abdominal pain, which
has a limited differential diagnosis. Tim de Dombal and his colleagues
did something similar [5]. We then choose a diagnosis from the list (e.g.
appendicitis) and look for findings (e.g. guarding) that occur commonly in
that chosen diagnosis and rarely (or never) in at least one other
diagnosis in the list (e.g. non-specific abdominal pain). This 'other'
diagnosis thus becomes less probable (or is ruled out if the finding
cannot by definition occur in people with that diagnosis). The chosen
diagnosis (e.g. appendicitis) correspondingly becomes more probable. A
'definitive' or 'sufficient' criterion (e.g. a red swollen appendix at
surgery would confirm a diagnosis.
A different 'reasoning by elimination theorem' is thus required to
model this reasoning process, which uses Bayes theorem as the starting
point of its proof [1, 2]. The author's qualitative approach in this
paper could also be applied to this 'reasoning by elimination theorem'
that models differential diagnostic reasoning.
References
1. Llewelyn DEH. Mathematical analysis of the diagnostic relevance of
clinical findings. Clin Sci 1979;57(5):477-479.
2. Llewelyn H, Ang HA, Lewis K, et al. The Oxford Handbook of
Clinical Diagnosis, 2nd ed. Oxford University Press, Oxford 2009;754-
772.
3. Lipschick GY, Von Feldt JM, Frame L, et al. The Oxford American Handbook of Clinical Diagnosis. Oxford University
Press, New York 2009;18-27.
4. Eddy DM, Clanton CH. The art of diagnosis: solving the clinico-
pathological conference. N Eng J Med. 1982;306:1263-1268.
5. Horrocks JC, de Dombal T. Computer-aided diagnosis:
description of an adaptable system, and operational experience with 2,034
cases. Br Med J 1972; 2: 5-9.
Different individuals have adapted and adopted the principles of EBM
to different degrees and in vastly different ways [1]. There is yet
another approach to EBM, which goes back a long way in terms of verbal
justification on ward rounds but which has not been put in writing and
taught in that way until recently [2]. It involves specifying the
'evidence' obtained from the patient that was used for ea...
Different individuals have adapted and adopted the principles of EBM
to different degrees and in vastly different ways [1]. There is yet
another approach to EBM, which goes back a long way in terms of verbal
justification on ward rounds but which has not been put in writing and
taught in that way until recently [2]. It involves specifying the
'evidence' obtained from the patient that was used for each diagnosis and
decision.
Instead of simply listing diagnoses and actions after the history and
examination, the 'patient's evidence' for each diagnosis can be placed in
parentheses after each diagnosis (e.g. "Diagnosis 1: Acute bronchitis
[Cough with yellow sputum for 5 days, sweats, no wheeze, no crackles on
20/11/11, all findings resolved by 25/11/11]". After each treatment or
test, the diagnostic indication is placed in parentheses (e.g.
"Amoxicillin 250mg tds for 5 days for diagnosis 1". The Oxford Handbook
of Clinical Diagnosis [2] provides thousands of such examples for students
and young doctors and suggests different formats for presenting the
'patient's evidence'.
Many 'problem oriented' computer systems allow these links to be made
automatically for each episode of care. If the patient has many diagnoses
then they can all be listed with their 'diagnostic evidence' and
management [2]. These can be updated (ideally automatically by computer)
after each clinical episode in hospital and general practice.
When 'transparent evidence-based summaries' of this kind were written
on a cohort of 76 patients admitted to a hospital under my care, an audit
showed that 45 (59.2%) of patients went home within 3 days compared with
161 out of 376 (42.8%) patients going home within 3 days when such
summaries were not written on admission (a difference of 16.4%, P= 0.008).
This appeared to happen because the nurses and doctors taking over care
could understand what had been done quickly and easily and continue the
care or change it if necessary without delay.
This approach of linking in writing individual patients' findings to
diagnoses and management for patients and others to read can be termed
'transparent medicine'. When the patient's findings match those in
evidence-based guidelines (or a supporting publication discovered during
or after a ward round) then the process can be described as 'transparent
EBM'. I doubt whether anyone would argue with this way of practising
medicine, which combines written transparency with EBM.
If the public become widely aware of this approach, it might become
ethically unjustifiable to practice 'non-transparent EBM' unless some RCT
showed that non-transparent EBM was as safe, efficient and effective as
'transparent EBM'!
References
1. Glasziou P. Editorial: What is EBM? Evid. Based Med. 2011 16:129-
130; doi:10.1136/ebm-2011-100099
2. Llewelyn H, Ang HA, Lewis K, et al. The Oxford Handbook of
Clinical Diagnosis, 2nd ed. Oxford University Press, Oxford 2009.
I agree with Dr. Tomedi's assertion that one might question the
credibility of a commentary written by someone who has received
pharmaceutical industry
support, or for a number of other reasons. It is for this reason that BMJ
Group policies ask authors to acknowledge and openly state any competing
interests (1), and as a result of this policy, Dr. Tomedi could become
aware
of them and consider them in reading the commenta...
I agree with Dr. Tomedi's assertion that one might question the
credibility of a commentary written by someone who has received
pharmaceutical industry
support, or for a number of other reasons. It is for this reason that BMJ
Group policies ask authors to acknowledge and openly state any competing
interests (1), and as a result of this policy, Dr. Tomedi could become
aware
of them and consider them in reading the commentary. An alternative
policy
could be to prohibit any person who has received pharmaceutical company
funding from writing a commentary. However, in some fields, many experts
have received some such funding. In the specific case of this commentary
(2), the commentator has raised a number of cautions about the results of
the study he critically appraised. And the competing interests appear to
span a number of sources (companies). It is certainly appropriate to
question bias, and then use one's judgment based on the available
information. I hope readers will continue to do this as they evaluate not
only the methodology of articles in the medical literature, but also the
context and other factors such as the qualifications of the authors, to
the
extent they can be evaluated.
2. Thomas, Mike. Cross-over randomised controlled trial: Long-
acting ?-agonist step-up therapy is more likely to provide best response,
compared to inhaled corticosteroid or leukotriene-receptor antagonist step
-up in children with uncontrolled asthma receiving inhaled
corticosteroids. Evid Based Med 2010;15:167-168 Published Online First: 16
August 2010
doi:10.1136/ebm1117
The important influence of pharmaceutical manufacturers on the medical literature, and the positive "spin" placed on results and conclusions, has been well documented. Given this pervasive problem, readers may question the credibility of the commentary of an EBM reviewer who has "received consulting fees from MSD, Schering, Novartis and GSK and received honoraria from Altana, Astra Zeneca, Boehringer Inglehi...
The important influence of pharmaceutical manufacturers on the medical literature, and the positive "spin" placed on results and conclusions, has been well documented. Given this pervasive problem, readers may question the credibility of the commentary of an EBM reviewer who has "received consulting fees from MSD, Schering, Novartis and GSK and received honoraria from Altana, Astra Zeneca, Boehringer Inglehiem, GSK, MSD, Merck Respiratory, Schering-Plough and Teva."
I was surprised to see the assertion by Philip Home that
rosiglitazone is a "particularly safe drug." Given its withdrawal from public use (the current prescribing ban recommended by MHRA in the UK, the European Medicines Agency and the FDA in the United States) due to concerns over increased cardiovascular risks, it would seem that our
health authorities see it as anything but a safe drug.
We are making some changes going forward to the Commentaries such that they are more structured and contain key features related to critical
appraisal of the evidence: http://ebm.bmj.com/content/15/4/103.full.
Some of what Dr. Bossano refers to (e.g. the use of absolute versus relative risks will be addressed by such changes.
We are making some changes going forward to the Commentaries such that they are more structured and contain key features related to critical
appraisal of the evidence: http://ebm.bmj.com/content/15/4/103.full.
Some of what Dr. Bossano refers to (e.g. the use of absolute versus relative risks will be addressed by such changes.
It's quite a while since I read EBM but I received an eTOC today and was pleased to discover I could read it via my NHS Athens login.
I couple of abstracts caught my including this one, because as a GP, LUTS and concern about PSA are relatively commonly seen.
The commentary seemed to have a lot of detail about prostate disease,
which was interesting but perhaps not entirely relevant to the paper it
was comm...
It's quite a while since I read EBM but I received an eTOC today and was pleased to discover I could read it via my NHS Athens login.
I couple of abstracts caught my including this one, because as a GP, LUTS and concern about PSA are relatively commonly seen.
The commentary seemed to have a lot of detail about prostate disease,
which was interesting but perhaps not entirely relevant to the paper it
was commenting upon.
I was very surprised to see RRRs and no ARRs in the summary of the
results, surely a mainstay of EBM is to provide both otherwise the
perception of the results is almost inevitably biased.
I turned to the abstract of the article (after perusing the full text
article re finasteride), unfortunaly the full text was not available.
What I discovered from the abstract was that the ARR seems to be about 5%
for all biopsy detected prostate cancer (about 25% placebo and 20%
dutasteride) which I guess is an NNT of 20 for 4 yrs, to prevent a
positive biopsy, so debatable to what extent this is a patient oriented
outcome.
What was puzzling is that the abstract also mentions that Gleason 7-10
biopsies were lower in years 1-4 of the study and the Gleason 8-10
biopsies were higher in years 3-4 (12 in dutasteride, 1 in placebo). I am
not really clear why the 2 different intervals and why the 2 different
Gleason ranges, but I would be concerned about the apparently higher
incidence of high grade biopsies with worst prognosis at 3 - 4 years.
The net result is I am not clear whether dutasteride has any patient
benefit or possibly even makes things worse.
It would have been nice if the commentary could have addressed this
question in more detail, because the tone of commentary is that
dutasteride is beneficial and I am unclear that the data shows this.
This is indeed an important finding on several levels, yet it remains
difficult in
translating this into clinical practice. I have found myself even more
ambivalent about suggesting SMBG to patients reasonably well-controlled on oral anti-diabetes medications.
In an effort to translate these findings, I propose the following
practical
suggestions.
1. For patients struggling to comply with health care...
This is indeed an important finding on several levels, yet it remains
difficult in
translating this into clinical practice. I have found myself even more
ambivalent about suggesting SMBG to patients reasonably well-controlled on oral anti-diabetes medications.
In an effort to translate these findings, I propose the following
practical
suggestions.
1. For patients struggling to comply with health care
recommendations, we
now know that SMBG does not need to be as high a priority in our
counseling.
2. For patients struggling to afford all the components necessary to
actually
perform SMBG (e.g., glucometer, strips, lancets), this now becomes an area
of
potential cost savings. Freeing up their limited discretionary income
might
allow them to afford a nutrition consultation, important medications or
other
expensive interventions with better evidentiary support.
3. For patients fully complying with SMBG, this might be a time to
allow them
to check their BG less frequently (e.g., Tuesday and Friday, fasting and
post-prandial, and prn for signs and symptoms of hypoglycemia and/or
hyperglycemia) instead of at their current rate.
Finally, since there is potential to have a critical study in the
future overturn
these recommendations, I am choosing to use caution both in how I explain
this study and in discontinuing anyone from SMBG.
Re: editorial BMJ 5th April 2008 Volume 336 pages 729-30: Improving uptake of MMR vaccine - Recognising and targeting between population groups are the
priorities
Dear Editor,
In the editorial on improving uptake of MMR vaccine, no mention was made
of the parents who decline MMR vaccination on ethical grounds.
The rubella vaccine component of MMR is derived from an aborted human
fetal cell line. The Takah...
Re: editorial BMJ 5th April 2008 Volume 336 pages 729-30: Improving uptake of MMR vaccine - Recognising and targeting between population groups are the
priorities
Dear Editor,
In the editorial on improving uptake of MMR vaccine, no mention was made
of the parents who decline MMR vaccination on ethical grounds.
The rubella vaccine component of MMR is derived from an aborted human
fetal cell line. The Takahasi rubella vaccine is a safe and effective
alternative. It is extremely difficult to access the Takahasi rubella
vaccine, grown on animal cell lines, due to import licence
restrictions. (Measles and mumps vaccines are not grown on aborted human
fetal cell lines. For those who can afford them, it is still possible to
have separate immunisations against measles and mumps.)
To achieve the goal of 95% immunisation several strategies should be
employed which also recognise diversity in the population. In 1994,Kenneth
Calman, then Health Minister, promised the Joint Faith Committee that an
ethical vaccine would be made available. However,as noted previously,
access to an ethical rubella vaccine has been made even more difficult.
Only coercive strategies to improve
completion of immunisation were mentioned in the article and the term
'..require legislative action..' (Ref 12) was used.
Are we now facing compulsory vaccination?
Ref 12 SalmonDA, TeretSP, MacIntyreCR, SalisburyD, BurgessMA,
Halsey NA. Compulsory vaccination and conscientious or philosophical exemptions:past present and future. Lancet 2006;367:436-42.
Dear Editor,
Bayes theorem can only be applied properly when considering a single diagnosis or its absence e.g. the presence or absence of asymptomatic diabetes mellitus, hypothyroidism, hyperlipidaemia, cervical cancer, breast cancer, etc. These asymptomatic situations are very important in the clinical setting and also in the community. It is in these situations that the reasoning described in this paper can be...
Dear Editor,
Different individuals have adapted and adopted the principles of EBM to different degrees and in vastly different ways [1]. There is yet another approach to EBM, which goes back a long way in terms of verbal justification on ward rounds but which has not been put in writing and taught in that way until recently [2]. It involves specifying the 'evidence' obtained from the patient that was used for ea...
I agree with Dr. Tomedi's assertion that one might question the credibility of a commentary written by someone who has received pharmaceutical industry support, or for a number of other reasons. It is for this reason that BMJ Group policies ask authors to acknowledge and openly state any competing interests (1), and as a result of this policy, Dr. Tomedi could become aware of them and consider them in reading the commenta...
Dear editor,
The important influence of pharmaceutical manufacturers on the medical literature, and the positive "spin" placed on results and conclusions, has been well documented. Given this pervasive problem, readers may question the credibility of the commentary of an EBM reviewer who has "received consulting fees from MSD, Schering, Novartis and GSK and received honoraria from Altana, Astra Zeneca, Boehringer Inglehi...
Dear Editor,
I was surprised to see the assertion by Philip Home that rosiglitazone is a "particularly safe drug." Given its withdrawal from public use (the current prescribing ban recommended by MHRA in the UK, the European Medicines Agency and the FDA in the United States) due to concerns over increased cardiovascular risks, it would seem that our health authorities see it as anything but a safe drug.
We are making some changes going forward to the Commentaries such that they are more structured and contain key features related to critical appraisal of the evidence: http://ebm.bmj.com/content/15/4/103.full. Some of what Dr. Bossano refers to (e.g. the use of absolute versus relative risks will be addressed by such changes.
Thanks
Richard Saitz, Editor.
Conflict of Interest:
...It's quite a while since I read EBM but I received an eTOC today and was pleased to discover I could read it via my NHS Athens login. I couple of abstracts caught my including this one, because as a GP, LUTS and concern about PSA are relatively commonly seen. The commentary seemed to have a lot of detail about prostate disease, which was interesting but perhaps not entirely relevant to the paper it was comm...
This is indeed an important finding on several levels, yet it remains difficult in translating this into clinical practice. I have found myself even more ambivalent about suggesting SMBG to patients reasonably well-controlled on oral anti-diabetes medications.
In an effort to translate these findings, I propose the following practical suggestions.
1. For patients struggling to comply with health care...
Re: editorial BMJ 5th April 2008 Volume 336 pages 729-30: Improving uptake of MMR vaccine - Recognising and targeting between population groups are the priorities
Dear Editor,
In the editorial on improving uptake of MMR vaccine, no mention was made of the parents who decline MMR vaccination on ethical grounds. The rubella vaccine component of MMR is derived from an aborted human fetal cell line. The Takah...