Bias in meta-analysis detected by a simple,graphical test
Asymmetry detected in funnel plot was probably due to true heterogeneity Editor —Egger et al report that they “found bias in 38%of meta-analys published in four leading journals.”1This is misleading,at least insofar as our meta-analysis of inpa-tient geriatric consultations is concerned.2
Firstly,the bias obrved in our meta-analysis was not a retrospective detection of bias,as one might infer from Egger et al’s statements.We knew that there was evidence of heterogeneity for the pooled effect estimates of geriatric consultation pro-grammes and reported this finding.2Sec-ondly,the asymmetry detected in the funnel plot of the meta-analysis of inpatient consul-tation programmes was probably due not to bias (distortion of true effect)but to true het-erogeneity (true difference of effects between trials).We took the prence of het-erogeneity as an opportunity to examine whether we could identify the programme elements that might have resulted in the obrved effect differences between geriatric consultation programmes.Using a multi-variate logistic regression approach,we found that both geriatric asssment pro-grammes in which the consultant controlled the implementation of the recommenda-tions and tho that included long term fol-low up resulted in better outcomes than did programmes in which this was not the ca.
Thus,the meta-analytical methods of testing heterogeneity or drawing funnel plots should not be considered absolute cri-teria for parating good from bad meta-analys.Meta-analys reporting effect estimates that may contain bias should con-tinue to be published in leading medical journals,as long as the possibility of hetero-geneity is stated and potential underlying reasons for heterogeneity are addresd.This is especially true for meta-analys of complex interventions.Although they are methodologically difficult to deal with,varia-tions in effect estimates give us the opportunity to dintangle the black box of complex interventions,such as of geriatric asssment,and identify what the necessary ingredients of the programmes are.3
A third issue concerns the “mega-trial”to which our meta-analysis was being compared.4This trial was different from any of the trials considered in our meta-analysis.Among other things,it was bad in a health maintenance organisation system that had incorporated considerable geriatric exper-ti into its usual care for older people.Another important factor was that it involved four hospital sites,each with differ-ent characteristics,populations,and survival rates.If Egger et al had taken the same pains as we did in recovering unpublished data from primary trials,they would have found that the mega-trial they ud in questioning our meta-analysis was a multicentre trial with unreported variability in intervention components and outcomes across study sites.Analysts must consider rigor
ously any methodological issues unique to each trial,particularly when considering complex interventions.
Andreas E Stuck Chief
Department of Geriatrics and Rehabilitation,Zieglerspital,Berne,Switzerland
Laurence Z Rubenstein Professor of geriatric medicine
Education and Clinical Center,UCLA-Sepulveda VA Medical Center,Los Angeles,CA,USA Darryl Wieland Professor
Department of Medicine,Division of Geriatrics,University of South Carolina School of Medicine,Columbia,SC,USA
1Egger M,Smith GD,Schneider M,Minder C.Bias in meta-analysis detected by a simple graphical test.BMJ 1997;315:629-34.(13September.)
2Stuck AE,Siu AL,Wieland GD,Adams J,Rubenstein LZ.Comprehensive geriatric asssment:a meta-analysis of controlled trials.Lancet 1993;342:1032-6.
3Beck JC,Stuck AK.Preventing disability.Beyond the black box.JAMA 1996;276:1756-7.
4Reuben DB,Borok GM,W olde GT,Eshoff DH,Fishman LK,Ambrosini VE,et al.A randomid clinical trial of compre-hensive geriatric asssment consultation for hospitalid HMO patients.N Engl J Med 1995;332:1345-50.
Experts’views are still needed
Editor —Egger et al’s regression analysis of funnel plot asymmetry is an interesting exerci in descriptive statistics:most fasci-nating is their distribution of biadness in meta-analys.1The funnel plot test that they derive,however,rests on the assumption that it is the smaller trials that are the culprits.What if the larger trials are tho that were stopped judiciously at the right moment or underwent some data-analytic “massage”?As noted in the accompanying editorial,the predictive power of the test was validated retrospectively on eight specific instances and became positive only when its test boundaries were changed to a 10%value.2More experience with the test ems necessary.
If we accept the test,or any similar test of heterogeneity on meta-analys,what should we conclude from it?The main mes-sage from it is that there might be a problem becau the funnel plot is asym
metrical —which we also e on the plot.The real ques-tions to which we would like an answer are:what is the cau of the asymmetry and,more importantly,which trials should we believe?The cau of the asymmetry can be anything,from publication bias,“willingness to plea”during data collection,data massage in the analysis,unclear rules for stopping the trial,or downright fraud (as indicated by Egger et al);it can also be a mix of all the things.Alternatively,the source of heterogeneity might be a true difference in underlying populations.Most difficult to live with is the overall conclusion of the test that the literature is biad.If the test is posi-tive,should we dismiss all randomid trials on the subject?This means that we discard one trial by one group of investigators becau of the results of another trial by a completely unrelated group.We might try to u quality criteria,but a recent meta-analysis on homoeopathy teaches us that this will not suffice.3
In the end there is no escape from a return to “the expert,”who tells us which trial to believe,not only on the basis of methodology but also on the basis of insights in pathophysiology,pharmacology,and perhaps type of publication (supple-Advice to authors
W e receive more letters than we can publish:we can currently accept only about one third.W e prefer short letters that relate to articles
published within the past four weeks.W e also publish some “out of the blue”letters,which usually relate to matters of public policy.
When deciding which letters to publish we favour originality,asrtions supported by data or by citation,and a clear pro style.Letters should have fewer than 400words (plea give a word count)and no more than five references (including one to the BMJ article to which they relate);references should be in the V ancouver style.W e welcome pictures.
Letters,whether typed or nt by email,should give each author ’s current appointment and full address.Letters nt by email should give a telephone and fax number when possible.W e encourage you to declare any conflict of interest.Plea nd a stamped addresd
envelope if you would like to know whether your letter has been accepted or rejected.
W e may post some letters submitted to us on the world wide web before we decide on
publication in the paper version.W e will assume that correspondents connt to this unless they specifically say no.
Letters will be edited and may be shortened.
Letters
on 21 September 2008
< Downloaded from
ments,special interest or “throw away”jour-nals,etc).All that we can ask from the expert is a careful explanation of what arguments he or she ud in accepting or dismissing the evidence from certain trials.
Jan P Vandenbroucke Professor
Department of Clinical Epidemiology,Leiden
University Hospital,2300RC Leiden,Netherlands
1Egger M,Smith GD,Schneider M,Minder C.Bias in meta-analysis detected by a simple,graphical test.BMJ 1997;315:629-34.(13September.)
2Naylor CD.Meta-analysis and the meta-epidemiology of clinical rearch.BMJ 1997;315:617-9.(13September.)3Vandenbroucke JP .Homeopathy trials:going nowhere.Lancet 1997;350:824.
Graphical test is itlf biad
Editor —Although the concept is uful,the method propod by Egger et al to detect bias in meta-analys is itlf biad 1:it overesti-mates the occurrence and extent of publication bias.This is easily shown by simu-lating data for a meta-analysis of a hypo-thetical intervention that is effective (and therefore has a negative regression coefficient by Egger et al’s method)and is free of publication bias (and hence should have an intercept of zero in the regression analysis).
In our simulations,each study was of a treated group and a control group,both of equal size.For each simulated meta-analysis,studies ranging from 100per group to 1000per group,in increments of 100,were generated.The obrved number of events in each group was generated from a binomial distribution.
Here is one example in which the true event rate is 40%in the control group and 10%in the treatment group.When the true population values (which would not be known in practice)are ud to estimate pre-cision,the regression coefficient is −1.7942(an estimated log odds ratio equivalent to the expected value of 0.1667)and the inter-cept (0.0380,P =0.1)is clo to the expected value of zero,reflecting the lack of publication bias.However,the regression coefficient estimated when the pre
cision is bad on the obrved values,as would occur using Egger et al’s method,is −1.7169.More importantly,the intercept is −0.4492and significant (P <0.0001),incor-rectly suggesting that there has been publication bias.In general,our other simu-lations suggest that the bias in the estimated intercept is greater the more effective the intervention actually is and the smaller the sample size of the studies.
This problem has veral caus.Firstly,the estimates of precision are subject to ran-dom error due to sampling variability.This regression-dilution bias caus the regression slope to “tilt”around the mean of the predictor and respon variables so that its coefficient is clor to zero;this in turn leads to the intercept becoming negative.2Secondly,the estimated standardid log odds ratio is correlated with the estimated precision.Thirdly,the precision estimated by the method that we assume Egger et al ud 3is a biad estimate of the true precision,with the degree of bias increasing as sample size decreas.4
Clearly,until the caus of the problems we have outlined are better elucidated and solutions developed,one cannot rely on the method propod by Egger et al to detect publication bias.
Les Irwig Professor of epidemiology
Petra Macaskill Statistical rearch officer Geoffrey Berry Professor in epidemiology and biostatistics
Department of Public Health and Community Medicine,A27,University of Sydney,NSW 2006,Australia
Paul Glasziou Associate professor
Department of Social and Preventive Medicine,
University of Queensland,Medical School,Herston,QLD 4006,Australia
1Egger M,Davey Smith G,Schneider M,Minder C.Bias in meta-analysis detected by a simple,graphical test.BMJ 1997;315:629-34.(13September.)
2Draper NR,Smith H.Applied regression analysis.2nd ed.New Y ork:Wiley,1981:122-5.
3Galbraith RF.A note on graphical prentation of estimated odds ratios from veral clinical trials.Stat Med 1988;7:889-94.
4Agresti A.Categorical data analysis.New Y ork:Wiley,1990:54.
T est had 10%fal positive rate
Editor —With examples of results from meta-analys conflicting with tho from subquent large trials there is increasing need to distinguish the good from the not so good meta-analys.T o this end,Egger et al have developed a test for detecting bias in meta-analys bad on funnel plot asymme-try.1This test predicted discordance in meta-analys.But,as with any significance test,there is also the possibility of fally iden-tifying bias when none existed.Since signifi-cance was defined by P <0.1,the fal positive rate of this test would be 10%.For instance,the quoted 13%(5/38)of the systematic reviews in the Cochrane Databa showing bias may be attributed to chance alone.
Defining significance to be P <0.1enabled the test to predict discordant meta-analys —the conventional P <0.05pro-duced significant bias in only one of the four discordant meta-analys —but resulted in a 10%fal positive rate.Some may consider this rate of fal positive results to be unacceptably high.Be that as it may,the findings showed the continuing need for care in the interpretation of results of significance tests.The comments,however,should not detract from the importance of looking for bias in meta-analys and the potential benefits this test may bring to screening for such bias.
Valerie Seagroatt University rearch lecturer
Unit of Healthcare Epidemiology,Institute of Health Sciences,Oxford University,Oxford OX37LF Irene Stratton University rearch lecturer Diabetes Rearch Laboratories,Radcliffe Infirmary,Oxford OX26HE
1Egger M,Davey Smith G,Schneider M,Minder C.Bias in meta-analysis detected by a simple,graphical test.BMJ 1997;315:629-34.(13September.)
Authors’reply
Editor —Bias in meta-analysis is often reflected in asymmetrical funnel plots.As we discusd in our paper,both bias and true heterogeneity in underlying effects can lead to asymmetry.Complex interventions such as geriatric consultation rvices may be imple-
mented less thoroughly in larger studies,and this would explain the more positive results in smaller trials.Results of meta-analysis will then depend on how many,or how few,small or large studies are included.A thorough attempt should always be made to identify heterogeneity,and the analysis by Stuck et al is a good example of this.1W e maintain that in the situations the combined estimate is
likely to be biad and should not feature prominently in published reports.Stuck et al suggest that we should have considered differences in outcomes across centres in the health maintenance organisation trial.Post hoc analys of effects by study centres,how-ever,are likely to mislead,as recently shown for the blocker heart attack trial.2
Vandenbroucke could have benefited from a formal analysis of funnel plot asymmetry on at least two occasions.After visual asssment of a funnel plot he suggested that publication bias may explain the association found between passive smoking and lung cancer.3However,we found no evidence of asymmetry (P =0.80).Converly,when he discusd a recent meta-analysis of homoeopathy,4significant funnel plot asymmetry (P <0.001)would have lent support to his asrtion that bias had produced a body of fal positive evidence (fig).5
Letters
on 21 September 2008
< Downloaded from
Irwig et al claim that our method will overestimate the occurrence of bias.They simulated hypothetical trials of a treatment that reduced event rates from40%to10% (relative risk0.25)with sample sizes ranging from200to2000.Their example is not typical of the small effects usually examined in meta-analys.More importantly,when performing10000simulations bad on the same assumptions we found that on average 4.99%of tests were significant at the5%level and9.63%were significant at the10%level. Therefore,contrary to Irwig et al’s conten-tion,regression dilution bias did not produce fal positive results above what was expected by chance,and the P value they quote for the intercept(P<0.0001),presum-ably bad on a large number of simulations, is misleading.
Seagroatt and Stratton are concerned about the specificity of our test.Considering the many possible bias,we think that the low nsitivity is of greater concern.When meta-analys are bad on a few small trials no test will be able to detect or exclude bias reliably.No statistical solution exists in this situation,and the results should be treated with great caution.
Matthias Egger Reader in social medicine and epidemiology
George Davey Smith Professor of clinical epidemiology
Department of Social Medicine,University of Bristol,Bristol BS82PR
Christoph Minder Head,medical statistics unit Department of Social and Preventive Medicine, University of Berne,Switzerland
1Stuck AK,Siu AL,Wieland GD,Adams J,Rubenstein LZ. Comprehensive geriatric asssment:a meta-analysis of controlled trials.Lancet1993;342:1032-6.
2Davey Smith G,Egger M.Incommunicable knowledge? Interpreting and applying the results of clinical trials and meta-analys.J Clin Epidemiol(in press).
世界人口第一大国
3Vandenbroucke JP.Passive smoking and lung cancer:a publication bias?BMJ1988;296:391-2.
4Linde K,Clausius N,Ramirez G,Melchart D,Eitel F, Hedges LV,et al.Are the clinical effects of homoeopathy placebo effects?A meta-analysis of placebo-controlled trials.Lancet1997;350:834-43.
5Vandenbroucke JP.Homeopathy trials:going nowhere. Lancet1997;350:824.
Prospectively identified trials could be ud for comparison with meta-analys Editor—Egger et al’s paper about bias in meta-analysis outlines the value of compar-ing the results of a meta-analysis of small randomid trials with tho of a subquent large definitive trial.1Unfortunately,in many areas of clinical practice such as stroke reha-bilitation,large trials are difficult to carry out and unlikely to be available.2
One possible solution in this circum-stance is to compare the results of meta-analysis with tho of prospectively identified trials that could not have been subject to publication bias.This was possible with the recent publication of a systematic review by the Stroke Unit Trialists’Collaboration.3The funnel plot for veral small trials can be compared with the summary result of either six trials which were identified before they were fully published or two trials(in Perth and Nottingham)which were recruited to the systematic review project before data analysis had started.The figure shows the funnel plot results for individual trials and the summary
results for the two groups of prospectively
identified trials.
In this ca the results of meta-analysis
em to be compatible with tho of the
prospectively identified trials.With the
increasing move towards prospective regis-
tration of trials,this approach may allow
some asssment of bias in meta-analys
where no large definitive trial is available.
Peter Langhorne Senior lecturer
On behalf of the Stroke Unit Trialists’Collaboration
Academic Section of Geriatric Medicine,Royal
Infirmary,Glasgow G40SF
1Egger M,Davey Smith G,Schneider M,Minder C.Bias in
meta-analysis detected by a simple,graphical test.BMJ
1997;315:629-34.(13September.)
2Gladman J,Barer D,Langhorne P.Specialist rehabilitation
after stroke.BMJ1996;312:1623-4.
3Stroke Unit Trialists’Collaboration.A collaborative
systematic review of the randomid trials of organid
inpatient(stroke unit)care after stroke.BMJ1997;
314:1151-9.
Increa in studies of publication bias
coincided with increasing u of
meta-analysis
Editor—Egger et al suggest a method for
testing the possible existence of publication
bias,bad on the assumption that larger
trials are more likely to be published,
irrespective of their results.1Stern and
Simes,however,suggest that large sample
size is not sufficient,becau of the delay in
the publication of larger studies with
negative results.2A recent letter showed that
trials published at an early stage were more
likely to be positive.3
T o test the association between the year
of publication and treatment effect we iden-
tified38meta-analys published in BMJ or
JAMA during1992-6which provided sum-
mary data from individual studies.For each
meta-analysis we tested the association
between the year of publication and the
treatment effect of the individual studies,
using rank correlation analysis.We also
tested the correlation between the sample
size and the treatment effect.We ignored the
sign of the correlation coefficient becau it
is often difficult to decide which group was
the control when competing interventions
were compared.Using0.10as a level of sig-
nificance,we found that four meta-analys
showed a significant correlation between the
year of publication and the treatment effect
while10showed a significant correlation
between the sample size and the treatment
effect.In25meta-analys the correlation
coefficient between the sample size and the
treatment effect was greater than that
between the year of publication and the
treatment effect.Therefore,both the delay to
publication and the small sample size may
be associated with the negative results but
small sample size ems to be more
important as a risk of publication bias.
Publication bias jeopardis the validity
of meta-analysis as well as any other
attempts to u published literature.A
systematic approach is crucial to identify all
published studies,particularly in low circula-
tion or non-English journals and in the grey
literature,and to exclude duplicate publica-
tions of positive results.4We agree with
Naylor that“meta-analysis is an important
contribution to rearch and practice but it’s
not a panacea.”5In fact,it was meta-analysis
and systematic review that highlighted the
problem of publication bias.By arching
Medline,we found that the number of pub-
lished studies(empirical,methodological,or
editorial)of publication bias was71during
1993to June1997,41during1987-92,three
during1981-6,and zero during1966-80.
成长作文The increa in the number of articles co-
incides with an increasing u of meta-
analysis.It is naive to believe that publication
bias did not exist or was less important a
decade ago,when medical literature review
was dominated by conventional non-
systematic methods.
Fujian Song*Senior rearch fellow
Simon Gilbody*MRC fellow in health rvices
rearch
NHS Centre for Reviews and Dismination,
University of Y ork,Y ork YO15DD
*The authors are undertaking a review of publi-
cation bias in systematic reviews funded by the NHS
Health T echnology Asssment programme.
1Egger M,Davey Smith G,Schneider M,Minder C.Bias in
meta-analysis detected by a simple,graphical test.BMJ
1997;315:629-34.(13September.)
2Stern JM,Simes RJ.Publication bias:evidence of delayed
publication in a cohort study of clinical rearch projects.
BMJ1997;315:640-5.(13September.)
3Rothwell PM,Robertson G.Meta-analysis of randomid
controlled trials.Lancet1997;350:1181-2.
4Tramer M,Reynolds DJM,Moore RA,McQuay HJ.Impact
of covert duplicate publication on meta-analysis:a ca
study.BMJ1997;315:635-40.(13September.)
5Naylor CD.Meta-analysis and the meta-epidemiology of
clinical rearch.BMJ1997;315:617-9.(13September.)
British renal registry is fully
electronic
Editor—In his editorial on clinical data-
bas Black did not mention the UK Renal
Registry,1although it may be the most inno-
vative and ambitious registry in the United
Kingdom.
The registry was established by the
Renal Association in collaboration with the
British Transplant Society and the British
Association of Paediatric Nephrology,and it
received priming support from the Depart-
ment of Health.It has been t up to collect
quarterly data on patients treated for end
Odds ratio
P
r
e
c
i
s
i
o
n
Funnel plot results:odds ratio for combined adver
outcomes of death and needing institutional care
versus precision of trial or group of trials
RCT=randomid controlled trial
Letters
on 21 September 2008
<
Downloaded from
stage renal failure and has been carefully developed with the potential to collect data on patients with pre-end stage failure.
This registry is the only national or international renal registry to u fully elec-tronic data extraction and transmission.Unlike the intensive care registry,which col-lects a single patient episode,the renal regis-try will collect quential quarterly data on patients and can track patients as they move between treatments and centres.Data are collected by software links to existing clinical computer systems in renal units.
Most registries collect paper returns and transfer data to their computer systems.This slows retrieval and analysis —for example,the renal registries in the United States,Aus-tralia,and Italy are at least two years behind in analysing and reporting on the collected data.The UK Renal Registry has produced its first preliminary analysis of data this year,showing its ability to analy and report on patient activity within six months.
The registry aims to assist renal units with both comparative audit and audit against established British standards t jointly by the Royal College of Physicians and the Renal Association.It will help health care commissioners by providing agreed purchar datats.The data collected should prove an invaluable resource for planning,audit,and rearch in renal care.爱情电影排名前十
David Anll Clinical coordinator T erry Feest Chairman
UK Renal Registry,Southmead Hospital,Bristol BS105NB
John Wallis,President
Renal Association,Triangle Hou,London SW184HX
1Black N.Developing high quality clinical databas.BMJ 1997;315:381-2.(16August.)
Electronic record linkage to create diabetes registers
Impressive results can be obtained without record linkage
Editor —Morris et al have shown a gold standard method of compiling a community bad diabetes register using record linkage of multiple data sources.1They conclude that this method was “more nsitive than general practice registers in ascertaining cas of known diabetes.”We question the subliminal message in their paper —that with a diabetes register created from data from hospital clinics and general practices alone,effective diabetes care could not be delivered and outcomes monitored as outlined in the targets of the St Vincent declaration.2
五毒手
Using general practice registers and the hospital clinic register in the Borders region of Scotland,we have identified 2067live patients in the area from a total population of 106000(point prevalence 1.95%).We believe that we have achieved this quality of data collection by means of anonymid feedback of the prevalence of diabetes and other indicators to each practice annually.Between 1995and 1996this resulted in an
increa in the number of known patients with diabetes from 1825to 2067;ven of 24practices in the area have a prevalence of diabetes of over 2%,and in one practice the prevalence is 3.5%.The recording of retinopathy screening improved from 58%in 1995to 86%in 1996.For tho patients attending only general practice diabetic clin-ics the mean haemoglobin A 1c concentra-tion fell from 8.92%in 1995to 6.92%in 1996.When the data were analyd further 408ts of paired data were identified,and in the patients the haemo-globin A 1c fell from 7.25%in 1995to 7.00%in 1996(P <0.0001).
Our experience suggests that active par-ticipation in and ownership of a diabetes register between hospital clinics and general practices can achieve equally impressive results.We would encourage district diabe-tologists and general practitioner colleagues to continue to refine their data collection on the patients and not to be discouraged by the lack of electronic record linkage or sophisticated
monitoring facilities.
Peter J Leslie Consultant physician Sheena McDonald Audit facilitator
Ian A McDonald Director,public health medicine On behalf of the Borders diabetes register
Borders General Hospital,Melro,Roxburghshire TD69BS
1Morris AD,Boyle DIR,MacAlpine R,Emslie-Smith A,Jung RT,Newton RW ,et al for the DARTS/MEMO Collabora-tion.The diabetes audit and rearch in T ayside Scotland (DARTS)study:electronic record linkage to create a diabetes register.BMJ 1997;315:524-8.(30August.)
2World Health Organisation (Europe),International Diabetes Foundation (Europe).Diabetes care and rearch in Europe:the St Vincent Declaration.Diabetic Med 1990;7:360.
Non-insulin dependent diabetes is being misd
Editor —Morris et al highlight one of the fundamental difficulties associated with meet-ing the targets of the St Vincent declaration:identifying all diabetic patients.1It is hard to e how the targets can be met if the baline diabetic population is uncertain.
An audit in Scotland of patients who had had a leg amputated,carried out by the Scottish Physiotherapy Amputee Rearch Group,found that 30%of amputations were in patients known to be diabetic.2This figure was low compared with that in other studies.3Therefore the rearch group,in collaboration with the Scottish Vascular Audit Group,conducted a three month pro-spective study of the diagnosis of diabetes in 146patients prenting for lower limb amputation in Scotland.4The study found that over half of the “non-diabetic”patients tested (21/36)had fasting blood gluco concentrations above 5.5mmol/l.The posi-tive predictive value of the fasting plasma gluco test is incread in this high risk group of patients,and an oral gluco toler-ance test would probably confirm the diagnosis of diabetes in most cas.
Cas of non-insulin dependent diabetes are clearly being misd even in a group of elderly patients with vascular problems of sufficient verity to warrant amputation.Selective screening of high risk patients is one
solution to the problem of reducing the level of undiagnod diabetes,and the Scottish Vascular Audit Group intends to extend its screening programme to include all vascular patients.Whether earlier diagnosis of non-insulin dependent diabetes is of clinical benefit is still open to debate,5and maybe a system such as that described by Morris et al could be ud to study the effect of early diag-
nosis on clinical outcome.
Shaun Treweek Coordinator,Scottish Physiotherapy Amputee Rearch Group
Elizabeth Condie Chairman,Scottish Physiotherapy Amputee Rearch Group
National Centre for Training and Education in
Prosthetics and Orthotics,University of Strathclyde,Glasgow G40LS
Douglas Gilmour Consultant vascular surgeon Glasgow Royal Infirmary,Glasgow G40LS
1Morris AD,Boyle DIR,MacAlpine R,Emslie-Smith A,Jung RT,Newton RW ,et al for the DARTS/MEMO Collabora-tion.The diabetes audit and rearch in T ayside Scotland (DARTS)study:electronic record linkage to create a diabetes register.BMJ 1997;315:524-8.(30August.)
2Condie ME,Jones D,Scott H,Treweek SP .A one-year national survey of patients having a lower limb amputation.Physiotherapy 1996;32:14-20.
3Larsson J,Apelqvist J.T owards less amputations in diabetic patients:incidence,caus,cost,treatment and prevention —a review.Acta Orth Scand 1995;66:181-91.4Treweek SP,Condie ME,Gilmour DG.Screening for diabetes in lower limb amputees:a pilot study.J R Coll Surg Edinb (in press).
5Harris Ml,Modan M.Screening for NIDDM:why is there no national program?Diabetes Care 1994;17:440-4.
Registers constructed from primary care databas have advantages
Editor —Morris et al report having joined tho areas in T ayside that have developed diabetes registers and their u of electronic record linkage.1It is now clear from the literature that a comprehensive datat can be collected by a variety of means,either centrally led or built up from primary care.
The suggestion in Morris et al’s paper that general practice diabetes registers are not comprehensive is not supported either by their own regional data or by compari-sons with other areas.The nsitivity of gen-eral practice registers in their study was 0.91compared with 0.96for electronic linkage.This would em comparable,given the cost and effort entailed in producing a regi
ster by means of electronic linkage.
The prevalence of diabetes is increasing both through improved recognition and through incread morbidity.Comparison of point prevalence in T ayside in 1996with point prevalence from previous studies is therefore invalid.Our own experience over 13years of maintaining a district register has been of a steady increa in the prevalence of diabetes over time.Morris et al lectively cited the prevalence in North Tyneside in 1991(1.18%)while omitting the prevalence in 1994(1.8%)quoted in the same paper.2This has since rin to 2.2%in 1997,a figure that compares favourably with the prevalence in T ayside in 1996(1.94%).Figures from South Glamorgan support a similar ri in the prevalence of diabetes over time bad on a district register gener-ated by general practices.3
Other factors must also be taken into account.Although Morris et al claim to have shown “how clinical information can be
Letters
on 21 September 2008
< Downloaded from
harnesd electronically and exploited for the benefit of patients,”they have failed to state what benefit patients with diabetes in T ayside have derived from their register.Since a vari-ety of methods are equally effective in data collection,perhaps the choice of method should reflect the effect that the method itlf has on the commitment of tho involved, their feelings of ownership,and its ufulness. Diabetic registers constructed from primary care databas are not constrained by the problem of confidentiality associated with electronic linkage and are therefore free to fulfil the purpo for which they exist.They are therefore ud extensively for patient recall,the gathering of clinical data,screen-ing,audit,and rearch.Any presumed gain in nsitivity from electronic record linkage cannot compete with this overwhelming factor.
David L Whitford General practitioner
381West Farm Avenue,Longbenton,Newcastle upon Tyne NE128UT
谭咏麟父亲
Susan H Roberts Consultant diabetologist
North Tyneside General Hospital,North Shields, Tyne and Wear NE298NH
1Morris AD,Boyle DIR,MacAlpine R,Emslie-Smith A,Jung RT,Newton RW,et al for the DARTS/MEMO
Collabora-tion.The diabetes audit and rearch in T ayside Scotland (DARTS)study:electronic record linkage to create a diabetic register.BMJ1997;315:524-8.(30August.)
空调功能
2Whifford DL,Southern AJ,Braid E,Roberts SH.Compre-hensive diabetes care in North Tyneside.Diabetic Med 1995;12:691-5.
3Butler C,Smithers M,Stott N,Peters J.Audit-enhanced, district-wide primary care for people with diabetes mellitus.Eur J Gen Pract1997;3:23-7.
Under half of nior hou officers in Anglia in1997were United Kingdom graduates Editor—The nior hou officer grade in Britain has en the effects of the new deal on junior doctors’hours and conditions, recent rapid expansion,1problems in recruitment,and an imbalance of people and places.2T o find out the prent make up of the grade,the Anglian postgraduate dean’s databa of3July1997was analyd.
Of682trainees in post,469held nation-ality or residence status in countries in the European Economic Area(EEA).3A total of 314were United Kingdom graduates;155 others had EEA status and had qualified in other European countries(131)or el-where(24)—their medical schools had been mainly in Germany(57)and Spain(35). There were213nior hou officers without EEA status(referr
ed to here as over-as doctors);among the,the largest number by far(93)had graduated in India. The ratio of United Kingdom graduates to other EEA graduates to overas doctors in the region ranged from70:13:17at the teaching hospital to18:40:42at one district general hospital.The average ratio at the nine district hospitals in the deanery was 41:25:34.Differences between specialties were en,with United Kingdom doctors being in the majority in medicine(54:20:26) and in a minority in obstetrics and gynaecology(31:34:34).
Time since qualification showed large
differences between United Kingdom doc-
tors(average3.9years),other EEA doctors
(5.5years),and overas doctors(7.6years).
Many nior hou officers had qualified
more than10years previously(12(4%)
United Kingdom doctors,13(8%)other
EEA doctors,and43(20%)overas
doctors).Thirty doctors in the grade had
qualified more than15years previously,and
some up to28years previously.
The figures are from only one
deanery,but there must be doubt about
whether recruiting more than half of nior
hou officers from countries other than the
United Kingdom is sustainable.Is it appro-
priate for so many doctors still to be in the
nior hou officer grade—the general
professional training grade—more than10
years after qualification,and should a major
revision of the grade be undertaken?
John Biggs Postgraduate dean
University of Cambridge Clinical School,
Addenbrooke’s Hospital,Cambridge CB22SP
1Government Statistical Services.Statistical bulletin.Lon-中国年的来历
don:Department of Health,1997.
2Leman P,Little F,Duby A,Williams DJ.“Clearing hou”is
needed to match available junior doctors to unfilled SHO
posts.BMJ1997;315:1016-7.(18October.)
3Department of Health.A guide to specialist registrar training.
London:DoH,1996:51.
UK encourages unrealistic
expectations among overas
applicants for training
Editor—The current situation regarding
overas graduates and training posts in the
United Kingdom is unsatisfactory.I recently
went through147applications for four n-
ior hou officer posts in medicine at the
district general hospital where I work.Of
the,four were from British nationals(two
of whom were graduates of British universi-
ties).Most applicants were graduates from
India(77),with the others being from
African countries(16),Germany(9),Greece
(8),the Middle East(6),Myanmar(3),
Pakistan(3),and Sri Lanka(3).
Most of the overas graduates had not
worked in the United Kingdom;of the few
who had done,most had worked in accident
and emergency,care of the elderly,or
psychiatry departments.The doctors had,
however,stated in their curriculum vitae that
they intended to pursue training in general
medicine and various medical specialties.
Clearly,their choice of jobs had been
dictated not by their training needs or inten-
tions but by the rvice needs of the NHS.I
have since found that the wide gulf between
supply and demand for training posts is not
restricted to medicine.
Far more overas graduates are eking
higher professional training in the United
Kingdom than can be trained satisfactorily.
It thus ems inappropriate,if not callous,to
continue to permit large numbers of candi-
dates to sit the Professional and Linguistic
Asssment Board(PLAB)test;and even
more so to start holding examinations in the
candidates’home countries,encouraging
more to take them.If the United Kingdom is
sincere about its commitment to postgradu-
ate medical training it must institute
measures to reduce the number of overas
graduates sitting the PLAB test and coming
to the United Kingdom via overas doctors
training schemes.Institutions responsible
for examination and training should,in their
communications with overas doctors,state
clearly the dearth of training posts in certain
specialties.
The more knowing among the overas
doctors here,particularly the sizeable num-
bers with higher qualifications in medicine
and surgery who now work as general prac-
titioners,believe that the United Kingdom
has always ud overas graduates and the
PLAB test to recruit doctors for the less
sought after specialties.1Inaction about the
current unsatisfactory situation will only
rve to confirm this view.
Of equal cau for concern must be the
paucity of home graduates eking training
posts.2An oversupply of overas graduates
and an undersupply of local ones implies
that something is fundamentally wrong with
manpower planning in the NHS.The
sooner the issue is addresd riously the
better for all concerned.
M K Sridhar Consultant physician
Mid-Staffordshire NHS Trust Hospitals,Stafford
ST163SA
1Varman SS.Training for overas doctors.BMJ1993;
306:1545-6.
2Fletcher EWL.Home medical students account for less
than half the full registrants Britain requires.BMJ
1997;314:1278.
Adding methionine to every
paracetamol tablet
Paracetamol overdo is so rare in India
that adding methionine would be wrong
Editor—Krenzelok is perhaps right about
N-acetylcysteine and methionine not being
universally available in developing nations.1
But we do not agree with his“ethical”
suggestion that methionine should be added
to paracetamol tablets in developing coun-
tries and its u encouraged.Paracetamol
overdo is rare in India.Adding methio-
nine would be medically inappropriate and
the additional financial burden unjustified.
We studied the records of68adults and
children admitted to three major hospitals
with poisoning or drug overdos over six
months in1996in Pune.Pune is a city in
western India of about4million people.We
did not come across a single ca of
paracetamol overdo.We also interviewed
27physicians and paediatricians working in
acute care in five major hospitals.None had
en a ca of paracetamol overdo in the
previous year.One physician recalled eing
two cas,and another recalled eing six
over12years.According to the paediatri-
手绘风景cians,kerone poisoning was the common-
est poisoning in children.Organophospho-
rus poisoning due to a bed bug killer was the
commonest in adults.The data suggest
that paracetamol overdo is rare in Pune.
Letters
on 21 September 2008
<
Downloaded from
