Content uploaded by Nicholas John Matheson
Author content
All content in this area was uploaded by Nicholas John Matheson
Content may be subject to copyright.
Evidence based diagnosis
We may need to be open to new ideas
Editor
—
If evidence based diagnosis is still
in the dark ages, as Delamothe writes,1then
so is evidence based treatment. 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 may be “evidence
based” because it has worked in a published
study, but some patients who would have
responded might have been left out because
of diagnostic inaccuracy while some patients
with no prospect of responding might have
been included incorrectly.2
Evidence based diagnosis is about
convincing others using shared rules of evi-
dence that a diagnosis (and its implications
in terms of treatment) should be accepted by
others. Evidence is gathered from the
individual and from groups of patients.3Evi-
dence based diagnosis means specifying the
individual’s facts in addition to pointing to
facts relating to that diagnosis in the
literature.4
Bayes’s theorem uses unconditional ini-
tial prior probabilities. Diagnostic leads are
based on conditional probabilities and are
used to initiate diagnostic thought processes.4
However, closely related theorems can be
used to interpret diagnostic leads, which
allow doctors to reason with diagnostic
evidence in a more familiar way,3thus reduc-
ing misunderstandings.5So to improve
evidence based diagnosis we also need to col-
lect better data on diagnostic leads.
The published evidence given for a diag-
nosis and any related actions cannot realisti-
cally be assembled when actually seeing a
patient. A draft evidence based rationale
might be prepared in advance. It would have
to be capable of being accessed in seconds to
provide evidence in support of a suspected
diagnosis and decision arrived at by using
kindness, imagination, and common sense. It
could be put into context by inserting the
patient’s details into the draft evidence
summoned up from a computer. 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.
Huw Llewelyn consultant physician
Department of Endocrinology and Diabetes, Great
Western Hospital, Swindon SN3 3BB
deh.llewelyn@orange.net
Competing interests: None declared.
1 Delamothe T. Diagnosis
—
the next frontier [Editor’s
choice]. BMJ 3006;333:0-f. (26 August.)
2 Llewelyn DEH, Garcia-Puig J. How different urinary albu-
min excretion rates can predict progression to nephro-
pathy and the effect of treatment in hypertensive diabetics.
JRAAS 2004;5:141-5.
3 Dunkelberg S. A patient’s journey: our special girl. BMJ
2006;333:430-1. (26 August.)
4 Llewelyn H, Ang H, Lewis K, Al-Abdullah A. The Oxford
handbook of clinical diagnosis. Oxford: Oxford University
Press, 2006.
5 Bianchi MT, Alexander BM. Evidence based diagnosis:
does the language reflect the theory? BMJ 2006;333:442-5.
(26 August.)
Multiple tests with multiple responses are
important
Editor
—
Diagnostic reasoning is never car-
ried out by using a single test alone: doctors
should take a history first, do an examination,
then do the tests.1Each stage adds variables to
a multivariable rather than a
univariate prediction process,
already well recognised in
prognostic studies.2
Using log10 likelihood
ratios and assuming inde-
pendence of the predictor
variables is appealing to me.3
Computer scientists would
use the log2unit, the informa-
tion bit. This method seems
to be equivalent to the naive
bayesian classifier, used for
filtering out spam mail, and is
relatively easy to program.
The tests often have more
than just a single purpose
—
for example, some recent
electronic responses have argued about
performing lumbar puncture for scan nega-
tive, rapid onset headache.4It is not just for
diagnosing or excluding subarachnoid haem-
orrhage but relevant to diagnosing meningi-
tis. Binary outcome logistic models do not
reflect clinical reality. That is why clinical
medicine is harder than mathematics.
You might have to go back to the patient
to clarify the history and examination in light
of unusual test findings. The real diagnostic
process is nowhere near as linear and
directed as implied in the research
protocols.
Ram Kumar paediatric neurology specialist registrar
Royal Manchester Children’s Hospital, Manchester
M27 4HA
ram16k@yahoo.com.au
Competing interests: None declared.
1 Bianchi MT, Alexander BM. Evidence based diagnosis:
does the language reflect the theory? BMJ 2006;333:442-5.
(26 August.)
2 Moons KG, van Es GA, Michel BC, Buller HR, Habbema
JD, Grobbee DE. Redundancy of single diagnostic test
evaluation. Epidemiology 1999;10:276-81.
3 Van den Ende J,Basinga P, Moreira J, Bisoffi Z.The trouble
with likelihood ratios. Lancet 2005;366:548.
4 Al-Shahi R, White PM, Davenport RJ, Lindsay KW.
Subarachnoid haemorrhage. BMJ 2006;333:235-40. (29
July.)
Terminology is unsatisfactory
Editor
—
Undoubtedly the failure of diag-
nostic theory to catch on is partly due to
unsatisfactory terminology.1“Sensitivity,”
and to a lesser extent “specificity,” are words
of multiple meanings that can be confusing
in the context of test evaluation. The terms
“true positive rate,” “true negative rate,”
“false positive rate,” and “false negative rate”
are much less ambiguous. Likelihood ratios
are also more easily understood, comparing
as they do the proportions of positives or
negatives in the diseased with the reference
population. These are the “weights of
evidence” defined by Pierce.2
It can be helpful too to recall
that odds ratios are the ratio
of the positive and negative
likelihood ratios
—
ratio-
cination with a vengeance.
“Positive predictive value”
is another term best dis-
carded. “Posterior probability
(or odds)” emphasises the
process as well as the out-
come. Nevertheless, the
probabilists would do well to
remember that many clinical
rules of thumb include a
dimension of utility. “Never
diagnose a condition you
can’t treat” is an irritating
remark that is difficult to confute.
G H Hall retired physician
Exeter EX1 2HW
h.2@which.net
Competing interests: None declared.
1 Bianchi MT, Alexander BM. Evidence based diagnosis:
does the language reflect the theory? BMJ 2006;333:442-5.
(26 August.)
2 Fitelson B. A bayesian account of independent evidence
with applications. Philosophy of Science (Proceedings)
2001;S123-140.
Prior probability saves money, time, and
possibly lives
Editor
—
Bayesian concepts of prior prob-
ability come to the aid of clinicians who aim
to expedite diagnosis and treatment.1In one
series comprising 63 patients in whom the
final diagnosis was choledocholithiasis, four
patients in whom this diagnosis was
subsequently validated by endoscopic retro-
grade cholangiopancreatography bypassed
Letters
549BMJ VOLUME 333 9 SEPTEMBER 2006 bmj.com
ultrasonography purely on the strength of
the index of clinical suspicion
—
that is, prior
probability
—
on the basis of their clinical and
biochemical stigmata.2Rightly so, given that
ultrasonography may itself be falsely nega-
tive for stigmata such as dilatation of the
common bile duct or calculi in the common
bile duct.34 Potentially life saving treatment
is also expedited by bypassing potentially
redundant “routine” investigations
—
an
important issue for patients with ascending
cholangitis.5
Oscar M Jolobe retired geriatrician
Didsbury M20 2RN
oscarjolobe@yahoo.co.uk
Competing interests: None declared.
1 Bianchi MT, Alexander BM. Evidence based diagnosis:
does the language reflect the theory? BMJ 2006;333:442-5.
(26 August.)
2 Jolobe OMP. Endoscopic retrograde cholangiopancrea-
tography in elderly patients (letter). Age Ageing
1999:28:498
3 Welbourn CRB, Haworth JM, Leaper DJ, Thompson MH.
Prospective evaluation of ultrasonography and liver
function tests for preoperative assessment of the bile duct.
Br J Surgery 1995:82:1371-3.
4 Gross BH, Harter LP, Gore RM. Ultrasonic evaluation of
common bile duct stones: prospective comparison with
retrograde cholangiopancreatography. Radiology
1983:146:471-4
5 Lai ECS, Mok FPT, Tan ESY,Lo CM, Fan ST, You KT, et al.
Endoscopic biliary drainage for severe acute cholangitis. N
Engl J Med 1992:326:1582-6.
What about the patients?
Editor
—
Test results are not just of interest
to clinicians.1Patients are commonly told,
“Your blood results were absolutely normal.”
Although clinicians may know exactly what
this phrase is intended to mean, patients are
likely to interpret it differently. The test may
have been a full blood count with urea and
electrolytes, but the “your blood is normal”
message may be interpreted as meaning that
everything in the blood is normal
—
possibly
up to and including HIV status.
Such a message may have profound
healthcare implications. In an ideal world,
this phrase would never be used. In reality, it
is used all the time. The timely focus that
Bianchi and Alexander have put on diagno-
sis and investigations may be an opportunity
to deal with this use of terminology.
David A Haslam general practitioner
Health Centre, Ramsey, Huntingdon PE26 1BP
davidhaslam@hotmail.com
Competing interests: DAH is national clinical
adviser, Healthcare Commission.
1 Bianchi MT, Alexander BM. Evidence based diagnosis:
does the language reflect the theory? BMJ 2006;333:442-5.
(26 August.)
Subarachnoid haemorrhage:
lumbar puncture for every
negative scan? Authors’ reply
Editor
—
Coats suggests that a lumbar
puncture should not be undertaken after a
negative computed tomogram for every
patient with “query subarachnoid haemor-
rhage.”1We robustly disagree.
Firstly, in his bayesian calculations,
Coats used the sensitivity of computed
tomography for identifying subarachnoid
blood at 12 hours after onset of headache
(98%). Sensitivity decays rapidly within days,
so patients who do not present immediately,
or who have to wait for computed tomogra-
phy, are less likely to be identified, and the
importance of a subsequent lumbar punc-
ture cannot be overstated. Furthermore,
computed tomography is often interpreted
by junior, non-specialist radiologists, who
are more likely to miss subtle signs of
subarachnoid blood than senior specialist
neuroradiologists. We see a steady trickle of
patients whose subarachnoid haemorrhage
was identified in a district general hospital
on the basis of a lumbar puncture, after a
normal report of their scan, but review of
the scan confirms the presence of subarach-
noid blood. These patients would not have
been diagnosed, and might subsequently
have had a fatal re-bleed had the clinicians
not done a lumbar puncture.
Secondly, Coats assumes lumbar punc-
ture is performed only to exclude subarach-
noid haemorrhage, yet it may be the key to
diagnosing other causes of sudden head-
ache, such as meningoencephalitis and
intracranial venous thrombosis (box 2 of our
review).
Lastly, we are intrigued by the notion
that doctors might enter into a bayesian
debate with a frightened, distressed, vomit-
ing patient in the hectic environment of the
emergency department. Without wishing to
sound paternalistic, patients are likely to
prefer their doctors to quickly and accu-
rately diagnose what is wrong with them,
rather than debate the merits of not
conducting a low risk diagnostic procedure
for a potentially life threatening neurologi-
cal disease.
In a perfect world, all patients with
sudden onset headache would present
immediately for senior specialist medical
attention and lie completely still in a
modern generation scanner, the scan would
be performed within 30 minutes of their first
assessment and immediately interpreted by
an experienced consultant neuroradiologist.
Acknowledging that such a state does not
exist, we stand by our recommendation that
all patients with a headache of maximal
intensity either immediately or within
minutes, lasting longer than an hour, and
who have received a normal computed
tomogram report, should have a lumbar
puncture.
Rustam Al-Shahi MRC clinician scientist
Rustam.Al-Shahi@ed.ac.uk
Philip M White consultant neuroradiologist
Richard J Davenport consultant neurologist
Division of Clinical Neurosciences, University of
Edinburgh, Western General Hospital, Edinburgh
EH4 2XU
Kenneth W Lindsay consultant neurosurgeon
Institute of Neurological Sciences, Southern
General Hospital, Glasgow G51 4TF
Competing interests: RJD and KWL have
acted as expert witnesses in cases involving
subarachnoid haemorrhage. PMW has received
reimbursement for expenses in attending
international conferences from Siemens,
Cordis, Boston Scientific, UK Medical, and
Microvention; has been reimbursed by Pyramed
UK for running an educational programme; and
holds a research grant from Microvention fund-
ing a randomised controlled trial (hydrocoil
endovascular aneurysm occlusion and packing
study). PMW has received consulting fees from
Boston Scientific, Cordis, UK Medical, and
Microvention.
1 Coats TJ. Subarachnoid haemorrhage: lumbar puncture
for every negative scan. BMJ 2006;333:396-7. (19 August.)
Strange things happen when
we never qualify the frequency
Editor
—
Aronson illustrated that the defini-
tion of frequency qualifiers cannot be taken
for granted.1However, we often omit them
completely as we condense complex
research findings into terse one liners. This
can dramatically distort our perception of
risk.
For example, nobody would disagree
that non-steroidal anti-inflammatory drugs
(NSAIDs) are an important cause of
avoidable iatrogenic mortality in elderly
patients, largely through ulceration and
perforation of the upper gastrointestinal
tract. These ideas profoundly influence pre-
scribing: doctors may avoid their use
altogether or co-prescribe prophylactic
measures.
Reputable studies show that, for NSAID
users over the age of 75, the annual risks for
serious gastrointestinal bleed and death are
1 in 110 and 1 in 650, respectively, and that
there is one episode of ulcer bleeding in eld-
erly people for every 2823 NSAID prescrip-
tions.23At least 83 patients need misopros-
tol prophylaxis to prevent one NSAID-
related gastrointestinal bleed, although a
subsequent systematic review was unable to
calculate any figure from the available
evidence.45
Using Aronson’s table of what frequency
qualifiers presently mean to people, we
would have to say that NSAIDs “never”
cause the problems described above, and the
most effective prophylactic measure against
these risks “never” works. Well I never.
John C Chambers Macmillan consultant
Katharine House Hospice, Adderbur y, Oxfordshire
OX17 3NL
dr.ch@mbers.info
Competing interests: None declared.
1 Aronson J. Sometimes, never. BMJ 2006;333:445. (26
August.)
2 Blower AL, Brooks A, Fenn GC, Hill A, Pearce MY, Morant
S, et al. Emergency admissions for upper gastrointestinal
disease and their relation to NSAID use. Alimentary
Pharmacol Therapeutics 1997;11:283-91.
3 Hawkey CJ, Cullen DJ, Greenwood DC, Wilson JV, Logan
RF.Prescribing of nonsteroidal anti-inflammator y drugs in
general practice: determinants and consequences. Alimen-
tary Pharmacol Therapeutics 1997;11:293-8.
4 Anon. GI complications and NSAIDs. Bandolier
1996;3(3):1-3. www.jr2.ox.ac.uk/bandolier/band25/b25-
1.html (accessed 29 Aug 2006).
5 Hooper L, Brown TJ, Elliott R, Payne K, Roberts C,
Symmons D. The effectiveness of five strategies for the
prevention of gastrointestinal toxicity induced by non-
steroidal anti-inflammatory drugs: systematic review. BMJ
2004;329:948.
Letters
550 BMJ VOLUME 333 9 SEPTEMBER 2006 bmj.com
Early intervention in acute
renal failure
Assessing fluid status is important
Editor
—
Bennett-Jones suggests that doc-
tors take a pragmatic and prompt approach
to intravenous fluid replacement, based on
the patient’s blood pressure, capillary refill
time, and venous filling.1Assessment of fluid
status needs to be much broader and incor-
porate a full history of any fluid gains and
losses from the patient, relatives, nurses,
fluid balance charts, prescription charts,
anaesthetic records, and daily weights. The
patient should be assessed for symptoms of
hypovolaemia, which can include postural
dizziness, thirst, dry mouth, reduced urine
output, feeling cold, shivering, shortness of
breath, and altered mental state.
Furthermore, in examining the patient,
of central importance are blood pressure, a
postural fall in blood pressure, tachycardia
(or rarely bradycardia with severe hypo-
volaemia) and postural changes in pulse
rate, whereas capillary refill time is not of
proved diagnostic value in adults.2Other
signs that should be sought are jugular
venous pressure, pallor, peripheral per-
fusion, the dryness of mucous membranes,
and the presence of pulmonary and periph-
eral oedema. If doubt about volume status
remains, central venous pressure monitor-
ing should be considered.
This careful assessment of fluid status is
crucial before the instruction to give intrave-
nous fluids, not loop diuretics to avoid
patients developing dangerous pulmonary
oedema, particularly since in some studies
fluid loading in intensive care has been asso-
ciated with a higher incidence of acute renal
failure.3
Jonathan M Gleadle university lecturer in nephrology
Oxford Kidney Unit, Oxford OX3 7LJ
jgleadle@well.ox.ac.uk
Competing interests: None declared.
1 Bennett-Jones DN. Early intervention in acute renal
failure. BMJ 2006;333:406-7. (26 August.)
2 McGee S, Abernethy WB, Simel DL. Is this patient
hypovolemic? JAMA 1999;281:1022–9.
3 Van Biesen W, Yegenaga I, Vanholder R, Verbeke F, Hoste
E, Colardyn F, et al. Relationship between fluid status and
its management on acute renal failure (ARF) in intensive
care unit (ICU) patients with sepsis: a prospective analysis.
J Nephrol 2005 18:54-60.
Evidence of inadequate intravenous fluid
treatment in UK hospitals
Editor
—
Bennett-Jones emphasises the
importance of prompt administration of
intravenous fluids for early intervention in
acute renal failure.1Determining the appro-
priate rate of fluid administration must
include an estimate of the degree of
intravascular volume depletion at the start of
treatment, with most aggressive volume
expansion targeted at patients with the
greatest deficits. To determine whether this
simple principle is followed in practice, we
audited intravenous fluid prescriptions for
114 consecutive acute surgical admissions to
three UK centres (one teaching hospital and
two district general hospitals).
A raised ratio of blood urea to creatinine
is commonly used as a quantitative refer-
ence standard for the diagnosis of hypo-
volaemia,2and similar rises may be seen in
patients with reduced effective intravascular
volume secondary to sepsis.3We therefore
compared the initial rate of intravenous
fluid administration for each patient with
their urea:creatinine ratio on admission. We
excluded from the analysis patients with
chronic renal failure or upper gastrointesti-
nal haemorrhage, or who were taking drugs
known to affect this ratio.
Across all admissions, the volume of
fluid prescribed over the first hour of
treatment ranged from 83 ml to 1250 ml.
The degree of correlation between rate of
administration and urea:creatinine ratio was
low, with a correlation coefficient for the
complete data set of only 0.23 (95%
confidence interval: 0.05 to 0.40). This
indicates that just 5.3% of the variation in
rate of fluid administration can be explained
by an association with urea:creatinine ratio
(and hence degree of intravascular volume
depletion).
The most likely explanation for this
finding is a failure by the admitting doctors
to appropriately diagnose and treat hypo-
volaemia. In UK hospitals, fluid prescription
is typically left to the most junior members
of medical and surgical teams, among whom
inadequate knowledge is common.4Training
and practice clearly need improving, and
courses such as ALERT (acute life-
threatening events
—
recognition and treat-
ment) may be a good start.5
Nicholas J Matheson senior house officer in medicine
nickmatheson@yahoo.com
St Thomas’Hospital, London SE1 7EH
Sarosh R Irani locum registrar in medicine
John Radcliffe Hospital, Oxford OX3 9DU
Anushka Irani clinical fellow in rheumatology
Great Western Hospital, Swindon SN3 6BB
We thank Nicola Alder, medical statistician at the
Centre for Statistics in Medicine, Oxford, for her
help with data analysis.
Competing interests: None declared.
1 Bennett-Jones DN. Early intervention in acute renal
failure. BMJ 2006;333:406-7. (26 August.)
2 McGee S, Abernethy WB, Simel DL. Is this patient
hypovolaemic? JAMA 1999;281:1022-9.
3 Robinson BE, Weber H. Dehydration despite drinking:
beyond the BUN/creatinine ratio. J Am Med Dir Assoc
2004;5(2 suppl):S68-71.
4 Lobo DN, Dube MG, Neal KR, Simpson J, Rowlands BJ,
Allison SP.Problems with solutions: drowning in the brine
of an inadequate knowledge base. Clin Nutrition
2001;20:125-30.
5 Smith GB, Osgood VM, Crane S, ALERT Course
Development Group. ALERT
—
a multiprofessional train-
ing course in the care of the acutely ill adult patient. Resus-
citation 2002;52:281-6.
Exempting mental health units
from smoke-free laws
Nicotine can have beneficial effects
Editor
—
Campion et al argue that psychia-
tric units should not be exempt from smok-
ing bans, but they do not discuss several
issues.1
Nicotine can have beneficial effects on
mood, anxiety, and cognition, and it amelio-
rates some of the side effects of psychotropic
drugs. Acute nicotine withdrawal can exac-
erbate psychiatric symptoms and cause
diagnostic difficulty. Cigarette smoke also
induces the metabolism of many different
psychotropic drugs.2Therefore, enforcing
acute smoking cessation in mentally unwell
patients may cause serious problems, includ-
ing making the patient feel worse, clouding
the clinical picture, worsening the side
effects of prescribed drugs, and precipitating
drug toxicity. When the patient starts smok-
ing again after discharge, the risk of relapse
is increased (secondary to re-stimulation of
the hepatic microsomal enzyme system and
associated reduction in plasma concentra-
tions of prescribed drugs).
To enforce a smoking ban on patients
who are free to leave hospital and who stay
of their own volition may be considered
acceptable. However, to enforce this on
patients who are detained against their will
under mental health legislation seems
unreasonable, especially in Scotland where
the principles of least restrictive alternative
and reciprocity are recognised.3
Even in studies where motivated patients
use smoking cessation aids in the absence of
acute mental illness, only the minority
remain abstinent in the medium to long
term. Is there any clear evidence that enforc-
ing a blanket smoking ban on acutely unwell
psychiatric patients will result in longer term
benefit for them in the real world?
I would love to live and work in a smoke-
free environment. However, I remain uncon-
vinced that we are treating patients as we
ourselves would wish to be treated if we ban
them from smoking against their will when
mentally ill. It is one thing to help smokers
give up when they are well enough to make
an informed choice for themselves. It is quite
another to enforce a smoking ban on acutely
unwell patients.
T Everett Julyan specialist registrar in psychiatry
Crosshouse Hospital, Ayrshire KA2 0BE
everett.julyan@nhs.net
Competing interests: None declared.
1 Campion J, McNeill A, Checinski K. Exempting mental
health units from smoke-free laws. BMJ 2006;333:407-8.
(26 August.)
2 Taylor D, Paton C, Kerwin R. The Maudsley 2005-2006 pre-
scribing guidelines. 8th ed. London: Taylor and Francis,
2006.
3 The Scottish Parliament. Mental Health (Care and
Treatment) (Scotland) Act 2003. www.opsi.gov.uk/
legislation/scotland/acts2003/20030013.htm (accessed
31 Aug 2006).
Issue should no longer be ignored
Editor
—
Campion et al wrote on no longer
exempting mental health units from smoke-
free laws.1By allowing smoking in psychia-
tric units the government will only increase
stigma towards psychiatric patients when
the Royal College of Psychiatrists is trying
hard to reduce it.
Admission of smokers with mental
illness to smoke-free psychiatric units may
lead to behavioural deterioration, but some
evidence from the United States refutes this
argument. In 1987 the Board of Trustees of
Southwest Washington Hospitals instituted
Letters
551BMJ VOLUME 333 9 SEPTEMBER 2006 bmj.com
a smoking ban in all of its facilities, including
general psychiatry units. The changes were
introduced successfully with minimal impact
on the successful function of the psychiatric
service.2The implementation of a smoking
ban, establishing a smoke-free psychiatric
service and abolishing tobacco products,
created minor management difficulties on a
locked psychiatric unit.3
The effects of prohibiting cigarette
smoking on the behaviour of patients on a
25 bed psychiatric inpatient unit were
assessed immediately after implementation
of a smoking ban and two years later. No
major behavioural disruptions were
observed after the ban. The number of calls
for security assistance, physical assaults,
instances of leather restraints and of
seclusions, and discharges against medical
advice did not increase significantly immedi-
ately after the restriction on smoking or two
years later.4
Signs and symptoms of nicotine with-
drawal and alterations in psychopathology
were evaluated among acutely ill psychiatric
patients admitted to a hospital with a smok-
ing ban.5Despite subjects’ reports of feeling
distressed and of experiencing nicotine
withdrawal symptoms, abrupt cessation of
smoking did not significantly affect either
the severity or the improvement of psycho-
pathological symptoms during admission.
The authors report no compelling reasons
to reverse the smoking ban.
With the growing concern for the harm-
ful effects of cigarette smoking and passive
smoking and the evidence above, exemp-
tions for mental health units from smoke-
free laws can no longer be ignored.
Faouzi Dib Alam specialist registrar
Royal Preston Hospital, Preston PR2 9HT
docftalam@aol.com
Competing interests: None declared.
1 Campion J, McNeill A, Checinski K. Exempting mental
health units from smoke-free laws. BMJ 2006;333:407-8.
(26 August.)
2 Thorward SR, Birnbaum. Effects of a smoking ban on a
general hospital psychiatric unit. Gen Hosp Psychiatry
1989;11:63-7.
3 Ryabik BM, Lippmann SB, Mount R. Implementation of a
smoking ban on a locked psychiatric unit. Gen Hosp
Psychiatry 1994;16:200-4.
4 Velasco J, Eells TD, Anderson R, Head M, Ryabik B, Mount
R, et al. A two yearfollow up on the effects of smoking ban
in an inpatient psychiatric service. Psychiatr Serv
1996;47:869-71.
5 Smith CM, Pristach CA, Cartagena M. Obligatory
cessation of smoking by psychiatric inpatients. Psychiatric
Services 1999;50:91-4.
HIV cannot be tackled in
isolation
Editor
—
The news extra by Clark with
much of the evidence on delivering health
care for HIV in less developed countries
highlights a key issue in all such countries.1
To raise one part of the country (one service,
one sector such as education or health) to a
much higher level is impossible without
raising the whole economy.
Health professionals, even if well paid,
are unlikely to want to work in areas where
their children cannot get a good education
or where they cannot rely on an energy sup-
ply at work or at home. Road safety alone
poses a major threat in many less developed
countries, as well as crime and civil unrest.
Faced with these additional “costs” of
working in such countries, many will choose
to go elsewhere, if only to more comfortable
cities in the same part of the world. The
developed world needs to come to terms
with its failure to do enough to raise general
standards of living, education, and public
health and safety in less developed coun-
tries. Until that happens, pouring large
amounts of money into single “vertical” dis-
ease programmes will not transform the
health prospects of populations of less
developed countries.
Peter A West senior associate
York Health Economics Consortium, University of
York, York YO10 5NH
paw11@york.ac.uk
Competing interests: York Health Economics
Consortium is a contract research company
owned by the University of York. It has a range of
public and private sector clients but is not active
in relevant research fields on HIV in the third
world. PAW has worked extensively on the cost
effectiveness of HIV prevention but is not doing
so currently. Past research has been funded by
the Overseas Development Institute and Depart-
ment for International Development and non-
governmental agencies in HIV prevention.
1 Clark J. HIV programmes in poor countries lack health
workers [News extra]. http://bmj.bmjjournals.com/cgi/
content/full/333/7565/412-d (accessed 31 Aug 2006.)
Three Bs, please
Don’t despise excellence
Editor
—
Choosing medical students is
more difficult than it might seem at first
glance.1Lowering entrance requirements
for medical school is not the answer
—
medical school and subsequent medical
practice require intellect and application.
Equally, favouring mature students with a
first degree or qualification in time may lead
to a situation where a pre-med qualification
becomes an advantage and may discrimi-
nate against school leavers.
Perhaps the fairest way to level the play-
ing field between state and private schools is
to introduce a standardised national qualifi-
cation for entry to medical school entry, in
addition to A level results. In that way, prob-
lem solving skills, knowledge, and emotional
intelligence could all be assessed, without
fear of bias towards one group.
In the meantime, to make excuses for
the failure of the state school system to
achieve good A level grades, by suggesting
that the private schools are “puffed up,” is
not helpful. Rather, ask why state schools
achieve such low results, even with recent
huge increases in funding? Private schools
can be academically selective, tend to have
smaller classes, be better disciplined, and
have more motivated pupils and parents
—
but why attack a system that appears to be
doing the job better?
It may be socially rewarding for Spence
to identify with what he sees as the
underdog, but with such an important issue
as this, perhaps it is time for all of us to put
aside our outdated social prejudices and just
try to get the best result.
Robin Ireland consultant psychiatrist
Brookhaven Psychiatric Unit, Bromsgrove,
Worcestershire B61 0BB
RIreland@doctors.org.uk
Competing interests: RI is privately (and
competitively) educated.
1 Spence D. Three Bs, please. BMJ 2006;333:453. (26
August.)
More academic excellence, please
Editor
—
Spence’s article highlights a grow-
ing divide in the perception of who the doc-
tors of tomorrow should be.1On the one
hand, we hear the cry for more students with
lower grades who have “the gift of the gab,
blarney, patter, or a silver tongue” (although
whether there is any evidence base on which
to support an association between these
attributes and low grades seems dubious).
At the same time, medicine needs
students of an academic bent more than
ever. We need doctors who are at home in
the world of primary research, who aspire to
further the limits of our knowledge, who
have a scientific approach to their profes-
sion, and the ability to accept uncertainty
and reassure.
When As make up 24.1% of all A levels
now awarded, surely to lower the require-
ments even further would remove an impor-
tant and useful hurdle for admissions
panels: are you motivated enough to achieve
the necessary grades?
Interviews are the appropriate point at
which to assess a student’s personality rather
than when surveying his or her grades. To
label all those students who want to go into
medicine and were willing to work to
achieve the necessary grades “neurotics” or
“no-social-skills types” is too absurd to be
offensive. Should Spence wish to verify the
accuracy of that statement, he has an open
invitation to dinner with the students in my
house (who have notaBgrade between us).
David McKean medical student
St John’s College, Oxford OX1 3JP
david.mckean@sjc.ox.ac.uk
Competing interests: More than three Bs at A
level.
1 Spence D. Three Bs, please. BMJ 2006;333:453. (26
August.)
We select the letters for these pages from the rapid
responses posted on bmj.com favouring those
received within five days of publication of the article
to which they refer.
Letters are thus an early selection of rapid responses
on a particular topic. Readers should consult the
website for the full list of responses and any authors'
re
p
lies
,
which usuall
y
arrive after our selection.
Letters
552 BMJ VOLUME 333 9 SEPTEMBER 2006 bmj.com
