Diagnosis—the next frontier
BMJ 2006; 333 doi: https://doi.org/10.1136/bmj.333.7565.0-f (Published 24 August 2006) Cite this as: BMJ 2006;333:0-fAll rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
I have extensive experience with superfluous testing, both as a
patient; and as a hospital based physician, where against my judgement I
had to order tests as part of the hospital's policy routine. Yet one
specific example is foremost in my mind.
A man was admitted to my psychiatric unit, and was found to be
significantly anemic. As we could not find any cause for this, we sent him
to the teaching hospital for a workup. He returned with the diagnosis of
"anemia secondary to extensive venipuncture". He had had a previous
admission to that same hospital, for some other diagnostic workup, which
had resulted in no significant finding. In the academic zeal for blood
tests, he was bled to the point of anemia.
Competing interests:
None declared
Competing interests: No competing interests
If “evidence based diagnosis is still in the dark ages”, then so is
evidence based treatment – they are inextricably linked. The doctor’s job
is to choose the right treatment. If the diagnosis is wrong then the
treatment will be wrong. Inaccurate diagnoses will also affect clinical
trials. A treatment well may be ‘evidence based’ because it has worked in
a published study. However some patients who would have responded might
be left out because of diagnostic inaccuracy. Also, some patients who
have no prospect of responding can be included incorrectly in a trial [1].
If these non-responders had been left out the result of the trial might
have been better. It is therefore important to assess the performance of
tests as ‘gold standards’ for use in diagnostic and treatment criteria, as
well as their ability to predict other ‘gold standard target results’. If
not, published clinical trial evidence may result in patients being
deprived of treatment or given it inappropriately in the name of evidence
based medicine.
Evidence based diagnosis is about convincing others using shared
rules of evidence that a diagnosis (and its implications in terms of
treatment) should be accepted by others. There are two types of such
evidence. There is the evidence gathered from the individual e.g. “this
little girl is aged 2 years 6 months and is 85 cm tall” [2] and facts
gathered from groups of patients e.g. “in this survey 2/98 little girls
aged 2 years 6 months were 90 cm tall or less”. Evidence based diagnosis
means specifying the individual’s facts in addition to pointing to facts
relating to that diagnosis in the literature [3].
Bayes Theorem uses ‘unconditional’ initial prior probabilities. This
means that it cannot be used to interpret diagnostic leads e.g ‘short
stature’ or hyperpyrexia. Diagnostic leads are based on ‘conditional’
probabilities and are used to initiate diagnostic thought processes [3].
However, there are closely related theorems which can be used to interpret
diagnostic leads [4] as well as the likelihood ratios also used in Bayes
Theorem. These other theorems allow doctors to reason with diagnostic
evidence in a more familiar way [3] thus reducing misunderstandings which
appear to be common [5]. So to improve evidence based diagnosis we also
need to collect better data on diagnostic leads, e.g. short stature,
abnormal liver function tests in children, hyperpyrexia, etc.
The published evidence that will be given for a diagnosis and any
related actions cannot realistically be assembled when actually seeing a
patient. A draft evidence-based rationale might be prepared in advance
and kept in a personal or published handbook or easily useable computer.
It would have to be capable of being accessed within seconds to provide
evidence in support of a suspected diagnosis and decision arrived at by
using kindness, imagination and commonsense. It could be put into context
by inserting the patient’s details into the draft evidence summoned up
from a computer [6]. If we are to make progress and allow evidence based
diagnosis to emerge from the “dark ages” then in addition to doing more of
the same, we may also have to be receptive to new ideas.
References:
1. Llewelyn D E H, Garcia-Puig J. How different urinary albumin
excretion rates can predict progression to nephropathy and the effect of
treatment in hypertensive diabetics. JRAAS 2004, 5; 141-5.
2. Dunkelberg S. A patient's journey: our special girl. BMJ 2006
333: 430-431.
3. Llewelyn H, Ang H, Lewis K, Al-Abdullah A. The Oxford Handbook of
Clinical Diagnosis, Oxford: Oxford University Press, 2006.
4. Llewelyn D E H. Assessing the validity of diagnostic tests and
clinical decisions. MD thesis, University of London, 1988.
5. Bianchi M T, Alexander B M. Evidence based diagnosis: does the
language reflect the theory? BMJ 2006; 333: 442-445.
6. Llewelyn D E H; Ewins D L; Horn J, Evans T G R; McGregor A M.
Computerised updating of clinical summaries: new opportunities for
clinical practice and research? BMJ 1988, 297, 1504-1506.
Competing interests:
None declared
Competing interests: No competing interests
"Testing can become almost an end in itself." This is a scourge,
especially for GPs whose work is often underminded by inapporpriate
requests for more tests. Patients mistake that approach as better medicine
when it is often just a way of getting out of a difficult consultation. Of
course in out patient clincis this can mean that you may never see the
patient again and therefore there are rewards for this approach as you can
feel as if you are doing something and there is a chance you will not have
to face the problem again.
An experienced GP with excellent clincal acumen can conclude and plan
without testing. The patient is seen by another doctor who orders lots of
tests and the patient thinks the GP has been sloppy. Only time will show
the GP to be right but by then the damage to the relationship is often
done.
It is also important to talk down investigations. With MRis scans in
backs, I will often say, "it is unlikely that the scan will show anything
of signficance and even if it does it is unlikely we will do much more
than we are doing now." Or with tiredness, "most blood tests in tiredness
come back as normal and we need to talk about lifestyle and possible
psychological reasons.."
It comes down to making sure the buck stays with you.
Competing interests:
None declared
Competing interests: No competing interests
How to act decisively in the absence of certainty.
Dear Editor,
David Eddy, the former cardiovascular surgeon at Stanford turned Duke
University mathematics PhD, who has devised a new computer model called
ARCHIMEDES, has elegantly shown in retrospect that most, if not all,
treatments for chronic diseases like diabetes, high blood pressure and
raised cholesterol might not do much good; might even do harm. Herman and
colleagues showed in a prospective study of IGT that it is better and much
cheaper to change patients’ life style than treat them with drugs as per
the evidence based medicine protocol to postpone the onset of diabetes.(1)
Delay in onset of diabetes in IGT individuals-(life style change 11 yrs
and drugs only 3 years.Reduced incidence(life style 20%, drugs only 8%
and, cost (life style $1,100 and drugs $ 31,000).That much for the gold
standard of evidence based therapeutics.
Now comes the plea for Bayesian theorem in diagnostic methods which
are said to be in the dark ages.(2) Having spent nearly half a century on
the patient’s bedside trying to make diagnoses, to me this looks like a
child playing with esoteric mathematical tools on the beach sands while
the large ocean of hidden imponderables in diagnosis lie in the dark
unfathomed caves of the ocean. There are very few, if any, black and white
regions in the field of diagnoses, most of it falling into the grey zone.
One would shudder to think of applying the Bayes’s theorem in the
emergency room to make a diagnosis. Even with a ready made computer soft
ware it might be too late in the day by the time we calculate all the
probabilities for all signs and symptoms before we arrive at evidence
based diagnosis. Then there is the enigma called the mind of the patient
in the overall picture. Look at what the mind can do without the
mathematical jugglery.
Harvard Medical School associate professor of medicine Dr. Roger J.
Laham reported on follow-up results of a randomized trial looking at laser
surgery to improve blood flow. Patients who got the surgery had
significantly less pain and improved heart function. But so did patients
who had a sham operation -- the equivalent of a placebo. After 30 months
the placebo effect was still there. Scans and other tests showed
physiological gains in blood flow among only those who thought they had
been operated on. This is where the patient’s mind works better than the
procedure.
The Reverend Thomas Bayes is by far the most enigmatic figure in
mathematical history. Almost nothing is known of Bayes's life, and very
few of his manuscripts survived. The exact date of Thomas Bayes's birth
is not known for certain. In 1742 Bayes was elected a Fellow of the Royal
Society of London, the most prestigious scientific body of its day,
despite Bayes having published no scientific or mathematical works at that
time.
”Bayes's sole publication during his known lifetime was allegedly a
mystical book entitled Divine Benevolence, laying forth the original
causation and ultimate purpose of the universe. Most mysterious of all,
Bayes' Theorem itself appears in a Bayes manuscript presented to the Royal
Society of London in 1764, three years after Bayes's supposed death in
1761!”
Despite the shocking circumstances of its presentation, Bayes'
Theorem was soon forgotten, to be revived later by the efforts of the
great mathematician Pierre-Simon Laplace. Laplace himself is almost as
enigmatic as Bayes. Laplace's papers are said to have contained a design
for a model capable of predicting all future events, the so-called
"Laplacian superintelligence". Unusual effects in experiments even today
are attributed to a "Laplacian Operator" intervening.
”Bayes' Theorem describes what makes something "evidence" and how
much evidence it is. Statistical models are judged by comparison to the
Bayesian method because, in statistics, the Bayesian method is as good as
it gets - the Bayesian method defines the maximum amount of mileage you
can get out of a given piece of evidence, in the same way that
thermodynamics defines the maximum amount of work you can get out of a
temperature differential.”
Previously, the most popular philosophy of science was probably Karl
Popper's falsificationism - this is the old philosophy that the Bayesian
revolution is trying to currently dethrone. Karl Popper's idea that
theories can be definitely falsified, but never definitely confirmed, is
another special case of the Bayesian rules. “So there's a limit on how
much mileage you can get from successful predictions (3); there's a limit
on how high the likelihood ratio goes for confirmatory evidence.” Let us
not make doctors’ life more miserable on the bed side.
Yours ever,
bmhegde
References:
1) Herman WH, Hoerger TJ, Brandle M et. al. Cost effectiveness of life
style modification and Metformin treatment in preventing Type II diabetes
in people with impaired glucose tolerance. Ann. Intern. Med 2005; 142: 323
-332,
2) Matt T Bianchi and Brian M Alexander Evidence based diagnosis:
does the language reflect the theory? BMJ 2006 333: 442-445.
3) Smith R. The Screening Industry. BMJ 2003; 326: 889-890
Competing interests:
None declared
Competing interests: No competing interests