World Bank must do more to develop safe and sustainabletransportation systems

Editor—In examinations of the WorldBank’s role in international health its influ-ence on the development of transportationsystems must not be overlooked.1 Each yearabout one million people die in road trafficaccidents, with perhaps 10 million perma-nently disabled.2 And for most of the worldthe epidemic of road deaths and injuries isjust beginning. It is estimated that by 2020road traffic accidents will be the thirdleading cause of disability adjusted life yearslost worldwide and the second leadingcause in the demographically developingcountries.3

The extent of the carnage will dependheavily on the type of transportationinfrastructure the bank promotes. Theimportance of bank policy on transporta-tion is underscored by the fact that 13% oflending is for transportation compared with11% for health, nutrition, and population.1

Close to 80% of the world’s cars areowned and produced by the 15% of theworld’s population who live in NorthAmerica, western Europe, and Japan. Thesame countries hold the greatest votingpower at the bank, which may not bode well

for the development of safe and sustainabletransportation systems. For example, Chinawith its millions of bicycles has one of themost equitable and sustainable transporta-tion systems on the planet.4 With Europeanand North American car markets reachingsaturation point, however, car manufactur-ers are looking east. The epidemic of roaddeaths in China has yet to begin, but alreadyan estimated 29 000 children are killed onthe roads each year.2

The midwife to this new epidemic maywell turn out to be the World Bank itself, byconflating motorisation and developmentand failing to count the true cost of cartravel. The neglect of road trauma by theWorld Health Organisation, and the fact thatglobal funding for research into road safetyis lower than that for almost any other causeof human misery, will do nothing to help.Ian Roberts DirectorChild Health Monitoring Unit, Department ofEpidemiology and Public Health, Institute of ChildHealth, London WC1N 1EHIan.Roberts@ich.ucl.ac.uk

1 Abbasi K. The World Bank and world health: Changingsides. BMJ 1999;318:865-9. (27 March.)

2 Murray CJL, Lopez AD. Global health statistics: acompendium of incidence, prevalence and mortality estimates forover 200 conditions. Harvard School of Public Health,Boston: Harvard University Press, 1996.

3 Murray CJL, Lopez AD. Alternative projections ofmortality and disability by cause 1990-2020: global burdenof disease study. Lancet 1997;349:1498-504.

4 Roberts I. Letter from Chengdu: China takes to the roads.BMJ 1995;310:1311-3.

The paradoxes of geneticallymodified foods

Protection of the public health shouldunderpin all decisions

Editor—In his editorial on geneticallymodified foods Dixon eschews certainimportant issues.1

Firstly, he does not point out that,although scientists claim this technology willhelp to feed the world’s burgeoning popula-tion, food production is not the problem.There are enough natural resources for usall to be fed: inequity of food distributionresults in the starvation of millions in poorernations.

Secondly, he dismisses the campaignerswho draw parallels between bovine spongi-form encephalopathy and genetically modi-fied foods. The connection between the two

in the public’s mind has little to do withgenetic manipulation and everything to dowith a lack of faith in policy makers. Not solong ago the government reassured thepublic that beef was safe. We now know thatthere was a risk of bovine spongiformencephalopathy, albeit very small. Thepublic is right to be cynical and mistrustful.

We must ask why we need these foodswhen there are no obvious benefits to thepopulation from their immediate introduc-tion. The driving force behind geneticallymodified foods is neither need nor demandbut certain multinational corporations.Thus, it is important to ensure that a desireto protect the public health—not the profitsof multinational organisations—underpinsall decisions on the introduction of geneti-cally modified foods.Emma Plugge Senior registrar in public healthmedicineBuckinghamshire Health Authority, Verney House,Aylesbury HP19 3ETemma.plugge@earthling.net

1 Dixon B. The paradoxes of genetically modified foods. BMJ1999;318:547-8. (27 February.)

Summary of electronic responses

We received 12 rapid responses to Dixon’seditorial1 on our website.2 Only onedefended genetic modification of food. “Forthousands of years farmers have used cross-breeding to genetically engineer crops,mixing genetic material on a large scale.But now, when genetic mixing is per-formed on a tiny scale, it suddenly becomesunacceptable” (S Root).

In general, there was a sense of fear. “It isthe question of crossing species’ boundariesand that, as yet, we have no answer regardingthe possible long term effects” (S Shrop-shire). Resistance to herbicides and pestsmight have unwarranted effects on theecosystem and possibly also directly onhealth. “It cannot be assumed that cropresistance to pests is merely due to ‘goodgenes.’ These genes may well be responsiblefor chemicals produced by the plant that inthemselves are responsible for the resist-ance” (J R Murray).

The subtitle to Dixon’s editorial says thata climate of mistrust is obscuring the manydifferent facets of genetic modification, andaccording to N Raithatha this is right: thereseems to be scientific evidence that the pub-lic trust neither politicians nor scientistsworking for the government.

The higher motive for producing geneti-cally modified food was also questioned.“Biotechnology companies … want to use

Advice to authorsWe prefer to receive all responses electronically,sent either directly to our website or to theeditorial office as email or on a disk. Processingyour letter will be delayed unless it arrives in anelectronic form.

We are now posting all direct submissions toour website within 72 hours of receipt and ourintention is to post all other electronicsubmissions there as well. All responses will beeligible for publication in the paper journal.

Responses should be under 400 words andrelate to articles published in the precedingmonth. They should include <5 references, in theVancouver style, including one to the BMJ articleto which they relate. We welcome illustrations.

Please supply each author’s currentappointment and full address, and a phone orfax number or email address for thecorresponding author. We ask authors to declareany competing interest. Please send a stampedaddressed envelope if you would like to knowwhether your letter has been accepted or rejected.

Letters will be edited and may be shortened.

www.bmj.comletters@bmj.com

Letters

Website: www.bmj.comEmail: letters@bmj.com

1694 BMJ VOLUME 318 19 JUNE 1999 www.bmj.com

this technology primarily to make money”(A Dowd).

1 Dixon B. The paradoxes of genetically modified foods. BMJ1999;318:547-8. (27 February.)

2 Electronic responses. Genetically modified foods. eBMJ1999;318 (www.bmj.com/cgi/content/full/318/7183/547#responses).

Preventing osteoporosis, falls,and fractures among elderlypeople

Few exercise programmes studied haveprevented falls

Editor—Kannus promotes physical activityfor the prevention of injurious falls amongelderly people.1 He acknowledges the uncer-tainty that surrounds the effectiveness ofspecific exercise programmes tested asinterventions in randomised controlledtrials. This is in contrast to the epidemiologi-cal evidence from longitudinal cohort orcase-control studies. We agree with him thatregular physical activity outside formal exer-cise programmes is likely to be beneficial toboth younger and older people.

In developing evidence based guidelinesfor the prevention of falls in older people wefound good evidence that exercise pro-grammes for unselected older people livingin the community do not prevent falls,2 withthe possible exception of balance training (taichi).3 Two trials found that selected olderpeople (those aged over 804 or with mild defi-cits in strength and balance5) benefit fromindividually tailored exercise programmesadministered by qualified professionals. Ourguidelines recommend that the implementa-tion of exercise programmes for unselectedolder people should not be a priority.

By contrast, multifaceted interventionprogrammes, including the identificationand treatment of postural hypotension,review of drug treatment, modification ofthe home environment, and possibly exer-cise training, do reduce the incidence offalls. Multiple risk assessment and modifica-tion have also proved successful in thecontext of an accident and emergencydepartment. The main research challenge isto develop these interventions further andtest them in pragmatic implementationtrials rather than to search for the optimalexercise activity.Gene Feder Senior lecturerg.s.feder@mds.qmw.ac.uk

Yvonne Carter ProfessorSheila Donovan FacilitatorDepartment of General Practice and Primary Care,St Bartholomew’s and the Royal London School ofMedicine and Dentistry, Queen Mary and WestfieldCollege, London E1 4NS

Colin Cryer StatisticianSouth East Institute of Public Health, King’sCollege London, Tunbridge Wells, Kent TN3 0XT

1 Kannus P. Preventing osteoporosis, falls, and fracturesamong elderly people. BMJ 1999;318:205-6. (23 March.)

2 Feder G, Cryer C, Donovan S for the development group.Guidelines for the prevention of falls in older people. London:Queen Mary and Westfield College, 1998.

3 Wolf SL, Barnhart HX, Kutner NG, McNeely E, Coogler C,Xu T, et al. Reducing frailty and falls in older persons: an

investigation of tai chi and computerized balance training.J Am Geriatr Soc 1996;44:489-97.

4 Campbell AJ, Robertson MC, Gardner MM, Norton RN,Tilyard MW, Buchner DM. Randomised controlled trial ofa general practice programme of home based exercise toprevent falls in elderly women. BMJ 1997;315:1065-9.

5 Buchner DM, Cress ME, de Lateur BJ, Esselman PC,Margherita AJ, Price R, et al. The effect of strength andendurance training on gait, balance, fall risk, and healthservices use in community-living older adults. J Gerontol1997;52A:M218-24.

Getting younger and older peoplemoving may seem sensible, but evidenceis lacking

Editor—Kannus’s editorial promoting life-long physical activity as essential in prevent-ing osteoporosis, falls, and fractures amongelderly people is not evidence based.1

Several findings need to be considered.Firstly, no randomised trials provide

evidence that regular exercise reducesfractures.

Secondly, vigorous exercise duringgrowth increases bone mineral density by10-20%2 and moderate exercise duringgrowth also increases bone mineral density.3

But these benefits may be lost after activity isstopped.4 Although residual benefits in bonemineral density have been reported inretired athletes aged under 50, falls and frac-tures are common in elderly people (thoseover 65). Whether elderly people who wereathletes or exercised moderately beforeretiring from sport have high bone mineraldensity is unknown.

Thirdly, most exercise interventionstudies in adults report no change in bonemineral density or increases of 1-3% (smallchanges of questionable biological signifi-cance given that the baseline bone mineraldensity is already low). One study suggeststhat exercise actually decreases bonemineral density.5

Fourthly, the lower prevalence of past orcurrent physical activity in patients with hipfracture than in controls is hypothesisgenerating, not hypothesis testing.

It may seem sensible to get “younger andolder people moving,” but evidence islacking. How many fewer fractures wouldoccur if lifelong exercise was widely taken upby younger and older people? Is the hypoth-esis one sided? Could lifelong exerciseincrease the risk of falls and fractures?E Seeman Associate professor of medicineAustin and Repatriation Medical Centre, Universityof Melbourne, Melbourne, Australiaego@austin.unimelb.edu.au

1 Kannus P. Preventing osteoporosis, falls, and fracturesamong elderly people. BMJ 1999;318:205-6. (23 March.)

2 Bass S, Pearce G, Hendrich E, Delmas P, Bradney M,Harding A, et al. Exercise before puberty may conferresidual benefits in bone density in adulthood: studies inactive prepubertal and retired female gymnasts. J BoneMiner Res 1998;13:500-7.

3 Bradney M, Pearce G, Naughton G, Starke KS, Ehsani AA,Slatopolsky E, et al. Differing effects of moderate exerciseon bone mass, size and volumetric density in pre-pubertalboys. J Bone Miner Res 1998;13:1814-21.

4 Dalsky GP, Starke KS, Ehsani AA, Slatopolsky E, Lee WC,Birge SJ Jr. Weight-bearing exercise training and lumbarbone mineral content in postmenopausal women. AnnIntern Med 1988;108:824-8.

5 Rockwell JC, Sorensen AM, Baker S, Leahey D, Stock JL,Michaels J, et al. Weight training decreases vertebral bonedensity in premenopausal women: A prospective study.J Clin Endocrinol Metab 1990;71:988-93.

Author’s reply

Editor—Feder et al emphasise that exercisein unselected groups of older people doesnot prevent falls while that in selectedgroups does. I do not know of any compara-tive study indicating which specific groups ofolder adults benefit from an exerciseprogramme, but the randomised trials thathave shown a benefit have included clearlydifferent age groups, from a mean of 651 toone of 84.2 Thus age does not seem to be astrong predictor of success or failure of theprogramme. I personally believe that theprogramme’s quality and the motivation ofits leaders are crucial.

Feder et al point out that multifacetedintervention programmes can reduce falls inolder adults. I agree. Tinetti et al and Closeet al have provided strong evidence forthis,3 4 Tinetti et al’s study including balanceand strengthening exercises.3 Unfortunately,neither study could separate the independ-ent role of each of the modified factors,although such analyses are important. Thusthese multifactorial interventions do notexclude exercise as one way of reducing therisk of falling.

Seeman is worried about the level ofevidence concerning exercise in the preven-tion of osteoporosis and related fractures.As in any medical condition, few actionshave been verified by randomised trials.This does not mean that we should stop try-ing to obtain better evidence, but at acertain point we have to summarise currentknowledge and give recommendations. Iagree that no randomised trial has provedthat regular exercise reduces the risk offracture; probably such a large and longtrial will never be conducted successfully.This does not mean, however, that exerciseis ineffective in preventing fractures relatedto age.

Many well controlled studies of physicalactivity and areal bone mineral density indi-cate that exercise during growing years ismore beneficial than exercise in adulthood.What is now needed is long term follow upof those who once obtained clear bone gainby exercise in early life. The studies showingsmall increases in areal bone mineral densityobtained by exercise in adulthood (1-3%)may have underestimated the true effect ofmechanical loading on bone strength. Arecent experimental intervention showedthat mechanical loading can improve bonestrength by reshaping the bone structurewithout increasing its areal bone mineraldensity.5 If this finding is repeated thecurrent pessimistic attitudes towards thepossibilities of improving adults’ bonestrength with exercise must be re-evaluated.

The final answer to whether lifelongphysical activity decreases the risk of fallsand fractures will probably never beavailable, but the evidence summarised inmy editorial speaks strongly for it.Pekka Kannus Chief physicianAccident and Trauma Research Centre, UKKInstitute, PO Box 30, FIN-33501 Tampere, FinlandKlpeka@uta.fi

Letters

1695BMJ VOLUME 318 19 JUNE 1999 www.bmj.com

1 McMurdo MET, Mole PA, Paterson CR. Controlled trial ofweight bearing exercise in older women in relation to bonedensity and falls. BMJ 1997;314:569.

2 Campbell AJ, Robertson MC, Gardner MM, Norton RN,Tilyard MW, Buchner DM. Randomised controlled trial ofa general practice programme of home based exercise toprevent falls in elderly women. BMJ 1997;315:1065-9.

3 Tinetti ME, Baker DI, McAvay G, Claus EB, Garrett P,Gottschalk M, et al. A multifactiroal intervention to reducethe risk of falling among elderly people living in the com-munity. N Engl J Med 1994;331:821-7.

4 Close J, Ellis M, Hooper R, Glucksman E, Jackson S, SwiftC. Prevention of falls in the elderly trial (PROFET): a ran-domised controlled trial. Lancet 1999;353:93-7.

5 Järvinen TLN, Kannus P, Sievänen H, Jolma P, HeinonenA, Järvinen M. Randomized controlled study of effects ofsudden impact loading on rat femur. J Bone Miner Res1998;13:1475-82.

*** A longer version of this letter is available on the BMJ ’swebsite www.bmj.com

Current census categories arenot a good match for identityEditor—Pfeffer discusses the complexitiesof theories of race, ethnicity, and culture.1

Surveys in the United Kingdom (UK) rely onthe census question that first appeared inthe 1991 census (incorporating colour andcountry of origin2) to define ethnic groups.The Office of Management and Budget’sclassification is dominant in the UnitedStates.3 Does the menu of terms given topeople included in such classifications offera good choice? Using data from the southTyneside heart study,4 we comparedrespondents’ identification of their ethnicityusing the census question, a description inan open question, country of birth, andcountry of family origin.

We recruited participants using thesnowball sampling technique.4 Communityworkers provided the names and addressesof people aged 16-74 of South Asian originwho had been resident in South Tynesidefor at least a year. (The term South Asian isused to refer to those individuals whoseancestral origin lies in the Indiansubcontinent—here, India, Pakistan, and

Bangladesh—and includes those born in theUK and others migrating to the UK via athird country (for example, Kenya).)

Participants (n = 334) were interviewedby a trained interviewer using a structuredquestionnaire in the preferred language,usually in the participant’s home. Therespondents first chose one of the categoriesfrom the census question (white, blackCaribbean, black African, black other,Indian, Pakistani, Bangladeshi, Chinese, andother), then provided a description of theirethnicity. They were also asked where theyand their mother and father were born.

Nineteen self descriptions were given.The most striking observation was the rarityof the term Asian and the absence of theterm South Asian—both commonly used inthe UK to describe people originating fromthe Indian subcontinent. These labels do notcapture ethnic identity. From the census cat-egories, 39% respondents chose Indian, 38%Bangladeshi, 13% Pakistani, 6% other, and 1black other (table 1). Only 81 (62%) of thosewho were Indian on the census questiondescribed themselves as Indian when giventhe open choice.

There was less variation among thosewho chose Bangladeshi from the census cat-egories. Altogether 74% of 27 respondentsborn in Pakistan described themselves asPakistani, whereas only 56% of 93 born inIndia described themselves as Indian. Of 68respondents born in the UK, 38% describedthemselves as Indian, 13% as British/English/Anglo Indian, and 10% as Sikh(table w1 on www.bmj.com). Self definedethnic origin and country of family originwere highly related when the country offamily origin was Bangladesh but less sowhen it was India or Pakistan (table w2 onwww.bmj.com).

Using census categories is insufficient tocapture self identification. If we had not

asked the census question first there mighthave been even less agreement betweenmodes of self identification. Too few catego-ries are offered to reflect the true hetero-geneity of ethnic groups. Similar issues alsoapply to other labels—for example, “black”and “white.”5 Our analysis emphasises theneed for fresh thinking if identity and selfidentification are to be the basis of ethnicgrouping.Judith Rankin Senior research associateDepartment of Epidemiology and Public Health,School of Health Sciences, Medical School,University of Newcastle, Newcastle upon TyneNE2 4HHJ.M.Rankin@newcastle.ac.uk

Raj Bhopal Bruce and John Usher professor of publichealthDepartment of Public Health Sciences, Universityof Edinburgh Medical School, Edinburgh EH8 9AGrbhopal@srv1.med.ed.ac.uk

1 Pfeffer N. Theories of race, ethnicity, and culture. BMJ1998;317:1381-4. (14 November.)

2 Nazroo JY. The health of Britain’s ethnic minorities: findingsfrom a national survey. London: Policy Studies Institute,1997.

3 Bennett T. “Racial” and ethnic classification: two steps for-ward and one step back? Public Health Rep 1997;112:477-80.

4 Rankin J, Bhopal R, Wallace B. Factors influencing heart dis-ease and diabetes in South Asians: the south Tyneside heart study.University of Newcastle upon Tyne: Department of Epide-miology and Public Health, 1997.

5 Bhopal R, Donaldson L. White, European, Western,Caucasian, or what? Inappropriate labeling in research onrace, ethnicity and health. Am J Public Health 1998;88:1303-7.

*** Additional tables are available on the BMJ ’s websitewww.bmj.com

Importance of healtheconomics must be recognisedwhen trials are designedEditor—Barbert and Thompson’s study iswelcome if it leads trial designers torecognise health economics as more than a(non-essential) afterthought to their trial.1

The points that the authors make wouldbenefit from a more complete understand-ing of the nature of economic data and thecontext in which economic analyses areundertaken.2

Clinical outcomes in trials tend to beuni-dimensional and unambiguous (such assurvival or response to treatment), whereaseconomic data are essentially multidimen-sional. Health care embraces the use of amultitude of resources, each measured indifferent units and attracting its own distinctpricing regime. The economic aspect of atypical clinical trial will require informationto be collected for 20-30 such items, eachsubject to a different statistical distribution.Theoretically, sample size should be calcu-lated for each of these and the trial’s recruit-ment target determined as that of thelargest. But this requires knowledge of theunderlying distribution of each variable—information that is rarely available—and sothere is no reliable basis for sizing a trialother than from estimates of sample sizesfor the clinical variables.

Assigning prices to resources consumedis difficult in the United Kingdom, where thecost of the same procedure can differ by

Respondents’ descriptions of their ethnic origin by 1991 census category (15 missing values). Figuresare numbers (percentages)

Self description

Ethnicity according to 1991 census categories

Indian Pakistani Bangladeshi Other

Indian 81 (62) 2 (5) — —

British/English/Anglo Indian 12 (9) — — 2 (11)

Indian Christian 1 (1) — — —

Kashmiri Indian 1 (1) — — —

Born in India but lived in Pakistan 1 (1) — — —

Pakistani — 29 (67) 1 (1) 1 (5)

Sikh 12 (9) — — 5 (26)

British Sikh 2 (2) — — 1 (5)

Indian Sikh 2 (2) — — 1 (5)

Bangladeshi — 1 (2) 118 (94) —

Bengali — — 5 (4) —

British Bengali — — — 1 (5)

Muslim 1 (1) 2 (5) 1 (1) 1 (5)

Kashmiri Muslim 2 (2) — — 1 (5)

British Muslim — 1 (2) — —

British 6 (5) 5 (12) 1 (1) —

British/English Asian 4 (3) 3 (7) — 5 (26)

Asian 4 (3) — — 1 (5)

Black or Asian 1 (1) — — 1 (5)

Total 130 (41) 43 (13) 126 (39) 20 (6)

Letters

1696 BMJ VOLUME 318 19 JUNE 1999 www.bmj.com

fivefold or 10-fold between hospitals—asmuch from different accounting schemes asfrom local variations in true cost. Applyinglocal prices to resources used in differenthospitals can disguise the resource changeswe are trying to detect, but applying somestandard or average price renders the resultsunrepresentative of any location.

The problems inherent in health eco-nomic analyses are much more fundamentalthan those addressed by Barber andThompson, which seem largely irrelevant atthis stage in the development of thediscipline.3 Sensitivity analysis is generallyrecognised to be the most effective way toaddress variability and uncertainty in aneconomic study, but even this must be usedselectively, otherwise the wide range of feasi-ble results generated renders the exerciseuninterpretable. Perhaps the most practicaladvice is to concentrate on changes in theuse of physical resources between compet-ing arms of a trial. Local decision makersmay then apply their own locally relevantcost weights to obtain meaningful cost peroutcome ratios.

Health economics remains in its infancy.There is a long way to go before the statisti-cal fine tuning suggested by Barber andThompson outweighs more pressing priori-ties, not the least of which is that theresource requirements of trials should begiven more than cursory attention at designtime.Alan Haycox Senior lecturer in health economicsA.R.Haycox@liverpool.ac.uk

Adrian Bagust Senior research fellowTom Walley DirectorPrescribing Research Group, Department ofPharmacology and Therapeutics, University ofLiverpool

Competing interests: None declared.

1 Barber JA, Thompson SG. Analysis and interpretation ofcost data in randomised controlled trials: review ofpublished studies. BMJ 1998;317:1195-1200. (31 October.)

2 Fayers PM, Hand DJ. Generalisation from phase III clinicaltrials: survival, quality of life and health economics. Lancet1997;350:1025-7.

3 Haycox A, Drummond M, Wally T. Pharmacoeconomics:integrating economic evaluation into clinical trials. Br JClin Pharmacol 1997;43:559-62.

Audit of use of ACE inhibitorsand monitoring in generalpractice

Guidelines on monitoring, on their own,are not sufficient

Editor—Kalra et al’s paper describing anaudit of monitoring renal function ingeneral practice suggests that practice isoften less than optimal, although associatedwith some uncertainty.1 Unfortunately, theauthors do not give confidence intervalsaround their estimates of effect—which weprovide here. The fact that only 34% (95%confidence interval 28% to 39%) of practi-tioners checked renal function after initiat-ing angiotensin converting enzyme inhibi-tors and 15% (11% to 19%) never checked itis an important finding. We disagree,however, with the authors’ conclusion that

producing guidelines on monitoring, on itsown, will lead to an improvement in practice.

The authors state that renal monitoringhas been neglected in recent guidelines andcite the North of England evidence basedguidelines on the use of angiotensinconverting enzyme inhibitors in primarycare (to which we contributed) as anexample of this.2 They recommend that“renal function should be checked beforeand 7-10 days after treatment is started in allpatients and thereafter regularly (for exam-ple, annually) only in those with risk factors.”Rather than neglect the issue of monitoring,the North of England Guidelines Groupprioritised it. The Agency for Health CarePolicy and Research also prioritised it in itsguidelines on heart failure,3 on which thissection of the North of England guidelinesdrew explicitly.

The North of England guidelines rec-ommend that “before initiation of angio-tensin converting enzyme inhibition . . .[patients] should have their blood pressure,renal function, and serum potassiummeasured. These measurements should berepeated one week after initiation oftreatment and again one week after eachsignificant increase in dosage.”2 The guide-lines go on to suggest monitoring of serumcreatinine concentration at least annually inall patients and describe specific criteria forpatients who develop renal insufficiency.Thus they actually propose monitoringstandards more stringent than those sug-gested by Kalra et al, although it is acknowl-edged that no basis for recommending onemonitoring period over another could befound. The evidence base is limited,although a recent small trial suggests thatcomplications are rare.4

Merely publishing guidelines on moni-toring seems insufficient if even those citingthe guidelines do not recall the recommen-dations. More active strategies, perhapsthrough computerised reminders built intoprescribing systems, are required; on thebasis of Kalra et al’s paper they should bedeveloped with some urgency.Nick Freemantle Senior research fellowJames Mason Senior research fellowMedicines Evaluation Group, Centre for HealthEconomics, University of York, York YO10 5DD

Martin Eccles Professor of clinical effectivenessCentre for Health Services Research, University ofNewcastle upon Tyne, Newcastle upon TyneNE2 4AA

1 Kalra PA, Kumwenda M, MacDowall P, Roland MO. Ques-tionnaire study and audit of use of angiotensin convertingenzyme inhibitor and monitoring in general practice: theneed for guidelines to prevent renal failure. BMJ1999;318:234-7. (23 January.)

2 Eccles M, Freemantle N, Mason JM for the North of Eng-land ACE-inhibitor Guideline Group. North of Englandevidence based development project: guideline for angio-tensin converting enzyme inhibitors in the primary caremanagement of adults with symptomatic heart failure.BMJ 1998;316:1369-75.

3 US Department of Health and Human Services, PublicHealth Service, Agency for Health Care Policy andResearch. Heart failure: evaluation and care of patients withleft-ventricular systolic dysfunction. Rockville, MD: USDepartment of Health and Human Services, 1994. (Clini-cal practice guideline 11.)

4 Lough M, Cleland J, Langan J, Cowley A, Wade A.Initiating angiotensin converting reuptake inhibitors inmild to moderate heart failure in general practice:randomised, placebo controlled trial. BMJ 1998;317:1352-3.

Authors’ reply

Editor—Although awareness of the possi-bility of serious uraemic complications ofangiotensin converting enzyme inhibitorsdates back over a decade,1 we found thatsuch complications are still frequentlyencountered in certain parts of the UnitedKingdom and are concerned by this.

Although non-evidence based recom-mendations for monitoring renal functionare available,2 we are worried that one keyarea of monitoring is often neglected, bothin the literature and by doctors supervisingthe care of patients receiving angiotensinconverting enzyme inhibitors. This concernspatients who become highly dependent ontheir renin-angiotensin-aldosterone axis tomaintain glomerular filtration because of anintercurrent illness or event (for example, anincrease in diuretic dose) accompanied byrenal hypoperfusion.

The most vulnerable patients are elderlypatients with heart failure (and consequentbaseline renal hypoperfusion) receivingdiuretics who then have a common illnesssuch as pneumonia, gastroenteritis, or evenflu.3 These are the patients most likely todevelop uraemia related to their angio-tensin converting enzyme inhibitor andindeed are the group whom we see in therenal service. No current guidelines orrecommendations emphasise the need tomonitor these patients carefully at the timeof their intercurrent illness; their renal dys-function would not, therefore, be identifiedby the application of existing monitoringprotocols.

As well as wishing to see this deficiencyrectified by the development of an appropri-ate evidence based guideline we would echoFreemantle et al’s call for more activestrategies that would consolidate doctors’awareness of the need to monitor renalfunction more judiciously in their most vul-nerable patients. Angiotensin convertingenzyme inhibitors provide excellent benefitsto many patients; more attention to moni-toring renal function, with a consequentincrease in doctors’ confidence, might para-doxically lead to a desired increase in the useof these agents.P A Kalra Consultant nephrologistDepartment of Renal Medicine, Hope Hospital,Salford M6 8HDphil@ren.srht.swest.nhs.uk

P MacDowall Specialist registrar in renal medicineRoyal Preston Hospital, Preston PR2 9HT

M Kumwenda Associate specialist in nephrologyDepartment of Renal Medicine, Glan ClwydHospital, Rhyl LL18 5UJ

M O Roland Director of research and developmentNational Primary Care Research and DevelopmentCentre, University of Manchester, ManchesterM13 6PL

1 Hricik DE, Browning PJ, Kopelman R, Goorno WE,Madias ME, Dzau VJ. Captopril-induced functional renalinsufficiency in patients with bilateral renal-artery stenosisin a solitary kidney. N Engl J Med 1983;308:373-6.

2 Eccles M, Freemantle N, Mason JM for the North of Eng-land ACE-inhibitor Guideline Group. North of Englandevidence based development project: guideline for angio-tensin converting enzyme inhibitors in the primary caremanagement of adults with symptomatic heart failure.BMJ 1998;316:1369-75.

3 Textor SC. Renal failure relating to angiotensin-convertingenzyme inhibitors. Semin Nephrol 1997;17:67-76.

Letters

1697BMJ VOLUME 318 19 JUNE 1999 www.bmj.com

Advice given to patients withfractures

Drug treatments that reduce fracture rateare underused after vertebral fractures

Editor—We recently undertook a costanalysis of osteoporosis, which showed thatthe disease is probably more costly than ear-lier estimates quoted by Pal suggested.1 Weestimated that the total annual cost offractures in a United Kingdom population isaround £940 million, with hip fracturescosting roughly £12 000 each.2 More impor-tantly, using data from the general practiceresearch database, like Pal we found littleevidence that patients who had sustained afracture were being offered treatment.3 Thetable (taken from our paper) summarisesour results.

In terms of evidenced based medicinethe only population of patients that hasbeen widely studied for secondary preven-tion is those who have sustained a vertebralfracture. Most, if not all, widely availabletreatments for the prevention of fractureshave been shown to be effective atpreventing new vertebral fractures inpatients who have had one previously.Despite this, over three fifths of patients witha vertebral fracture diagnosed in primarycare were not prescribed any drugs thatreduce the fracture rate to prevent new ver-tebral fractures. Such patients are also athigh risk of non-vertebral fractures.4 Thepriority in preventing fractures would seemto be to offer drug treatments that reducethe fracture rate to patients with existingvertebral fractures.David J Torgerson Senior research fellowCentre for Health Economics, University of York,York YO10 5DDdjt6@york.ac.uk

Paul Dolan Senior lecturer in economicsSheffield Health Economics Group andDepartment of Economics, University of Sheffield,Sheffield S1 4DA

1 Pal B. Questionnaire survey of advice given to patients withfractures. BMJ 1999;418:500-1. (20 February.)

2 Dolan P, Torgerson DJ. The cost of treating osteoporoticfractures in the United Kingdom female population.Osteoporosis Int 1999;8:611-7.

3 Torgerson DJ, Dolan P. Prescribing by general practitionersafter an osteoporotic fracture. Ann Rheum Dis 1998;57:378-9.

4 Burger H, van Daele PLA, Algra D, Hofman A, GrobbeeDE, Schütte HE, et al. Vertebral deformities as predictorsof non-vertebral fractures. BMJ 1994;309:991-2.

Fragility fractures establish diagnosis ofosteoporosis

Editor—Pal’s short article highlights thelack of appropriate advice received by manypatients with osteoporotic fracture, particu-larly those with hip fracture.1 Althoughcomprehensive recommendations for thecare of patients with hip fracture exist,1 2 itseems that they are not always implementedand opportunities for secondary preventionare missed. Measures that are effective inprotecting against hip fracture includecalcium and vitamin D supplementation andthe use of hip protectors,3 4 neither of whichis routinely advised. In addition, manypatients leave hospital without assessment ofrisk factors for falling, some of which, suchas poor vision and environmental hazards,may be remediable.

The editorial accompanying Pal’s articleadvocates dual energy x ray absorptiometryin older patients with osteoporotic fracture,both to identify underlying osteoporosis andto provide a baseline for monitoringtreatment.4 Although bone densitometrymay sometimes be of value in confirming orrefuting osteoporosis in individuals with afracture, it is often unnecessary: the presenceof a fragility fracture establishes a diagnosisof osteoporosis and is an independent riskfactor for subsequent fracture.

A strong case can be made for treatingpatients with a fragility fracture, regardlessof whether bone densitometry is available;this is particularly so for elderly patients withhip fracture, most of whom will satisfy thedensitometric criteria for osteoporosis.Further justification for this approach isprovided by the increasing evidence thatsafe and relatively short term interventionsin elderly people can produce significantreductions in the fracture rate.3 4

The rationale for monitoring treatmentcan also be challenged. Response rates toantiresorptive treatment are generallyreported as being greater than 90%. Againstthis background it is difficult to understandwhy monitoring should be regarded asessential, particularly since the time requiredaccurately to detect non-response may bethree years or more.

Finally, in view of the high prevalence ofvitamin D deficiency in the elderly popula-tion, routine vitamin D supplementationwould surely be a more cost effectiveapproach to reducing hip fracture thanmeasurement of serum 25-hydroxyvitaminD concentrations.5

Juliet Compston University lecturerDepartment of Medicine, Addenbrooke’s Hospital,Cambridge CB2 2QQjec1001@cam.ac.uk

1 Pal R. Questionnaire survey of advice given to patientswith fractures. BMJ 1991;318:500-1. (20 February.)

2 Royal College of Physicians. Fractured neck of femur: pre-vention and management. Summary and recommenda-tions of the report. J R Coll Phys Lond 1995;23:8-12.

3 Audit Commission. United they stand: co-ordinating carefor elderly patients with hip fractures. London: HMSO,1995.

4 Doube A. Managing osteoporosis in older people withfractures. BMJ 1999;318:477-8. (20 February.)

5 Compston JE. Vitamin deficiency: time for action. BMJ1998;317:1466-7. (28 November.)

Strategy to reformulate waitinglists

New Zealand has some suggestions forNHS priority system for elective surgery

Editor—I have some suggestions that theBMA might like to consider when it designsthe NHS priority system for electivesurgery.1 My four suggestions are: use a scor-ing system instead of “banding” patients;don’t use the words “severity score”—insteaduse “priority score”; include consumers inthe groups for developing the priorities; andset up a booking system when you introducepriority assessment.

I recommend that patients are scoredinstead of banded because clinicians will becompelled to use the assessment tool to geta score (the alternative is to determine a cat-egory for the patient, based on the currentclinical decision making). If you have toration (determine to which level you canprovide publicly funded operations) it ismuch easier to work with a scoring system asopposed to, for example, 25% of D category.

We have found that the assessment tool(and score) should reflect ability to benefitfrom surgery as well as “need” factors. If youuse only severity as an indicator then allpatients for day surgery will wait for a longtime or miss out on their surgery. This, infact, is a group who benefit greatly for eachhealth dollar/pound spent on them, bothdirectly and indirectly (days off work, qualityof life, etc). To determine your priorityassessment tool, please include consumers.Consumers are good at prioritising benefitand need and are essential to getting thecommunity to “buy in” to the scheme andfor transparency to priority setting.

Giving people an idea of when they willreceive their treatment means changing theway that waiting lists are managed. The mainimpact in our system has been projectingout our surgical booking lists to six months.I don’t think that you can introduce a prior-ity assessment tool without looking at capac-ity and booking issues. If you start prioritis-ing patients and give people certainty youneed to be able to say, “Yes, we can carry outsurgery on all of these people and the newpatients expected each month within thistime frame.” If you are not able to do thisthen you have to determine the level towhich you will be providing elective surgery.Janine Cochrane Project leader for booking systemsHealthCare Otago, Private Bag 1921, Dunedin,New ZealandJanineC@HealthOtago.co.nz

1 Fricker J. BMA proposes strategy to reformulate waitinglists. BMJ 1999;318:78. (9 January.)

Italy’s public health system is changingfrom waiting times to priority

Editor—Fricker reports that the BMA pro-poses a strategy to reformulate waiting listsin the United Kingdom.1 In the public healthsystem, issues such as priority setting andappropriateness ratings in the referral ofpatients by general practitioners to special-ists are usually faced in a hard (top down)

Number of patients receiving at least oneprescription for drug that reduces fracture rate(HRT, calcium, vitamin D, bisphosphonate) inyears before and after hip, wrist, or vertebralfracture (n=100)

Type of fracture

Hip Wrist Vertebral

Year before fracture(1994)

2 7 10

Year after fracture (1996) 4 5 39

HRT=Hormone replacement therapy.

Letters

1698 BMJ VOLUME 318 19 JUNE 1999 www.bmj.com

way. Negative or positive lists exist or proto-cols are set up, commonly by insurance orgovernment bodies, that have to be used byall professionals concerned.

In some areas of the Italian public healthsector we are now experimenting with deal-ing with this matter in a soft (bottom up)way. General practitioners and specialists arecalled together to set up criteria for a simplepriority model by specialty; each higherpriority level contains more causes (andother clinical signs or facts) for referral,related to the consequences that delay ofnon-emergency care (diagnostic proceduresin particular) would have.

Each priority level has its own space onthe specialist timetable, and bookings aregiven by a booking centre, where staff havereceived full instruction on the proceduresafter common planned criteria in identify-ing the right level (and, consequently, theright time for the consultation) have beenmet. Specialists are asked to report ifpatients they checked were not appropri-ately booked, and monitoring of thesereports gives the chance for remodellingand strengthening the model.

First results are positive, and theappropriateness of the referral system isimproving quickly. A composite evaluationby specialists of 570 cases from the quotedsurgeries was performed; in 67 of thesecases the waiting time was considered to bedelayed, in 53 cases it was early, and in 446cases it was appropriate. (In the remainingfour cases prescription was considered to beinappropriate.) We are now working withgeneral practitioners to identify moreclinical data able to define clinical categorieswith different levels of priority and tosimplify the booking procedure; the generalpractitioner will define his or her prescrip-tion as priority A (or B or C or D).

We will be able to identify major resultsas soon as the model starts in other areas ofthe Italian public health sector. We think thatboth an improvement in accessibility tohealthcare services and a more effectiveallocation of the resources will be importantconsequences of this model.Giuliano Mariotti Medical directorHealth Authority ULSS 2, 32032 Feltre, Italymariotti@tqs.it

Rosanna Sommadossi Trained nurseBooking Service, Health Authority APSS, 38100Trento, Italy

Tommaso Langiano Medical directorPaediatric Hospital “Bambin Gesy,” 00153 Rome,Italy

Roberto Raggi Public health consultant47037 Rimini, Italy

1 Fricker J. BMA proposes strategy to reformulate waitinglists. BMJ 1999;318:78. (9 January.)

Real-time priority scoringsystem must be used forprioritisation on waiting listsEditor—We would like to contribute to thedebate in the BMJ about the role of prioriti-sation scoring systems1–3 and respond to the

BMA’s recent discussion paper on thissubject (which briefly mentions our work fora project initiated by the Welsh Office).4

Weale dismisses the possibility of provid-ing an infinite variety of treatments to aninfinite number of patients from finiteresources.5 To channel the limited resourceseffectively, we have argued that the introduc-tion of a clinical management tool incorpo-rating a system of prioritisation based on thepatient’s degree of need is necessary.3 Work-ers at various centres throughout the worldhave investigated this problem without aclear consensus emerging.1 2 4 Prioritisationvariables in common use are social handi-cap, morbidity, and disease.

Our elective algorithm incorporatesthese principles and introduces time in aproportionate and cumulative manner, add-ing the patient’s derived priority score to anaccumulating waiting list score each week.The higher the need is scored the faster theprogression through the list. Current guar-antees in the patient’s charter have beenincorporated into the algorithm. An alterna-tive algorithm incorporating the rate ofprogress of the condition, the ability of treat-ment to influence outcome, and the patient’slevel of pain and distress is used in patientswith life threatening problems.

A computerised prioritisation systemthat changes with time allows moreadvanced analyses of the list, aiding effectivemanagement and resource allocation. Thecreation of a booking system is facilitated byprediction of admission dates. Potentialbreaches of the patient’s charter can be pre-dicted and patients diverted to other centres,or demand met with targeted resources.

The priority score for elective cases(termed the patient’s initial quotient) isentered at the bottom of the waiting list. As itgrows with time it is redefined as thepatient’s eligibility quotient, until the patientreaches the top of the list, at which point it isredefined as the patient’s exit quotient. Thesummed total of patients’ eligibility quo-tients, the average eligibility quotient, themean exit quotient, and various otherderivatives are amenable to analysis.

In our opinion the traditional method ofanalysing waiting lists, with its emphasis onmaximum wait and patient numbers, isinadequate. Alternative concepts are neededwhen prioritising and managing waiting listsand resources. We believe that understand-ing waiting list dynamics will be possibleonly when a real-time priority scoringsystem is used.B Davis Consultant otorhinolaryngologistS R Johnson Consultant trauma and orthopaedicsurgeonWest Wales General Hospital, CarmarthenSE31 2AF

1 Kee F, McDonald P, Kirwan JR, Patterson CC, Love AHG.Urgency and priority in cardiac surgery: a clinicaljudgment analysis. BMJ 1998;316:925-9.

2 Hadorn DC, Holmes AC. The New Zealand priority crite-ria project. Part 1 Overview. BMJ 1997;314:131-4.

3 Davis B, Johnson S. Rationing health care. BMJ 1998;316:1092-3.

4 BMA’s Health Policy and Economic Research Unit.Waiting list prioritisation scoring systems: a discussion paper.London: BMA, 1998.

5 Weale A. Rationing health care. BMJ 1998;316:410.

Eligibility criteria improvechildren’s access tolong term ventilationat homeEditor—Jardine et al comment that fundingand home carers are common obstacles todischarge home for children on long termventilation.1 Liverpool has a scheme whichlargely overcomes these obstacles.

Liverpool Health Authority and the localauthority have agreed on eligibility criteriafor joint funding for children with complexneeds, which has removed the delay previ-ously caused by funding negotiations. Theeligibility criteria are based on a discussiondocument produced in 19942 which aims todefine childhood disability by producing afunctional grade of severity. Impairment offunction is assessed in 10 categories thatcover mobility, sensory impairment, physicalhealth, learning, and personal care. We haveadded an 11th category, vulnerability, toinclude technology dependent children suchas those on long term ventilation. Childrenwho meet the jointly agreed criteria are eligi-ble for joint funding split equally betweensocial services and the health authority for ahome care package, or three ways with theeducation department if the packageincludes school support.

The joint funding scheme has been inexistence for 18 months. We have had twonew children on long term ventilationduring this time. It took 6 weeks after initialreferral for the health authority to confirmjoint funding with social services, with afurther 4 weeks for an additional edu-cational agreement to funding for one child.

Home care is provided by lay carers sup-ported by a qualified paediatric nurse. Thecarers are employed by the Royal LiverpoolChildren’s (NHS) Trust, which is responsiblefor recruitment, training, and supervision.Training is patient specific and is deliveredboth in hospital and at home by the supportnurse, assisted by high dependency unit andspecial school nursing staff. The main delayin the process is recruitment and training ofcarers, with a minimum training period of 6weeks.

We currently have three children in Liv-erpool on long term ventilation at home,supported by 10 carers. The annual cost is£108 000, which covers the care packagesbut not the capital cost of equipment. Wehave estimated, based on experience overthe past 6 years, that there will be two newchildren a year requiring long term ventila-tion at home.Sian Snelling Consultant community paediatricianAlder Hey Children’s Hospital, Liverpool L12 2AP

1 Jardine E, O’Toole M, Paton JY, Wallace C. Current statusof long term ventilation of children in the UnitedKingdom: questionnaire survey. BMJ 1999;318:295-9.(30 January.)

2 British Association for Community Child Health. Disabilityin childhood; towards nationally useful definitions. London:BACCH, 1994.

Letters

1699BMJ VOLUME 318 19 JUNE 1999 www.bmj.com

Maybe the time has come forthe primacy of the patient inthe NHSEditor—We wholeheartedly agree withHampton’s statement that cooperation andunderstanding between primary and sec-ondary care are crucial.1 Unfortunately, hisarticle has done little to promote these. Hestates that there is “no evidence that anNHS based on the primacy of primary carewill function any better” and then attacksprimary care with equally little evidence insupport of his uninformed, dogmaticstatements.

Firstly, the increase in admissions formedical emergencies does not prove thatgeneral practitioners are increasingly seeingthe importance of secondary care. It does,however, correspond with the evidence thatpatients are consulting their general practi-tioners more often,2 that the population isolder and better educated, and that generalpractitioners are more aware of the techni-cal support that can be provided only inlarger centres.

Secondly, there is no evidence that anemphasis on primary care will inevitablyslow medical advance. Hampton points outrightly that secondary care is responsible formost evidence based practice; this is surely awonderful argument for putting moreresources into general practice research,since 70-90% of patient contact occurs inprimary care.

Thirdly, primary care rightly emphasisescommon conditions. When did rare and dif-ficult problems become the “important”illnesses? Surely this is not the view ofpatients, who are more likely to have reallyimportant illnesses such as ischaemic heartdisease, depression, and back pain. Preven-tion has been proved to reduce mortality(for example, the Scandinavian simvastatinsurvival study3) and is also rightly empha-sised.

Finally, the statement that the “treatmentof important illnesses should be the first pri-ority” in medical education will be surprisingto those who have read the General MedicalCouncil’s recommendations for such educa-tion.4 A new generation of doctors is beingtrained to undertake and value preventionof disease, the importance of commonillnesses, and, most importantly, cooperationand mutual respect across primary andsecondary care.

One classic example of cooperation isthe management of diabetes mellitus. Moni-toring of diabetes can be done equally effec-tively in primary and secondary care,5 butthe answer is not competition but coopera-tion, with professionals acting as district-wide teams. The primary factor, and thedeciding factor, must be the patient. Has thetime come for the patient led NHS and theprimacy of the patient?Hugh Alberti Lecturer in primary health careGrey Towers Court, Middlesbrough TS7 0NJHugh@lone.demon.co.uk

George Alberti PresidentRoyal College of Physicians of London, LondonNW1 4LE

1 Hampton JR. The primacy of primary care. BMJ1998;317:1724-5. (19-26 December.)

2 Office of Health Economics. 10th compendium of healthstatistics. London: OHE, 1997.

3 Scandinavian Simvastatin Survival Study Group. Ran-domised trial of cholesterol lowering in 4444 patients withcoronary heart disease. Lancet 1994;344:1383-9.

4 General Medical Council. Tomorrow’s doctors: recommenda-tions on undergraduate medical education. London: GMC,1993.

5 Griffin S. Diabetes care in general practice: meta-analysisof randomised controlled trials. BMJ 1998;317:390-6.

Edinburgh college’s consensusstatements are not purely forUKEditor—As organisers of the Royal Collegeof Physicians of Edinburgh’s consensus con-ference in medical management of stroke,1

we would like to comment on some pointsthat Bogousslavsky raises in his editorial.2

The Edinburgh consensus statementhad the “flavour of UK practice” because itwas primarily directed at medical practice inthe UK. Accordingly, the consensus panelwas selected from within the UK, although itconsidered evidence that was internationaland collated through reviews (from, forexample, the Cochrane Collaboration),which were circulated as backgroundpapers. Furthermore, the conference hadinternational speakers, was advertised inter-nationally, and welcomed international del-egates. Finally, the consensus statement wasoffered to relevant international journals.

An account of the development andmethodology of the Edinburgh consensusconferences will be published later this yearin the college proceedings. The otherconsensus statements and guidelines thatBogousslavsky refers to should likewise beexamined for their methodological develop-ment, the representativeness of theirauthors, and their intended audience.

In keeping with the traditions andmission statement of this college, its consen-sus conferences are primarily for the UK butoffered to the world for debate. AsBogousslavsky notes, their consensus state-ments are simple but not simplistic; this isdeliberate in order to provide guidance forall people involved in stroke management—both those who are and those who are notexperts. As healthcare systems around theworld vary so widely, it seems unlikely that asingle consensus statement would be appli-cable to all countries.

We acknowledge that for antiplatelettreatment both lower and higher doses ofaspirin than that recommended in the state-ment (75-300 mg/day) are given. However,the only evidence for efficacy in acute strokeis for doses within this range, and such dosesalso have proved efficacy in secondaryprevention. Evidence suggests that bothmodified release dipyridamole and clopi-dogrel are acceptable alternatives in patientswith stroke or transient ischaemic attackwho are intolerant of aspirin,3 4 and dipyri-damole is cheaper.

We acknowledge that thrombolytic treat-ment in acute stroke is controversial becauseits risk-benefit ratio is uncertain. TheEdinburgh consensus panel thereforethought that further randomised trials wererequired before it could be recommended inroutine medical management. BMJ readerswill doubtless wish to review the evidencebefore deciding whether to be guided byconsensus panels or by “fulminating leadingexperts on both sides of the Atlantic.”Martin Dennis Reader in stroke medicineMargaret Farquhar College consensus conferenceorganiserPeter Langhorne Senior lecturer in geriatric medicineGordon Lowe College assessorCharles Warlow Professor of neurologyRoyal College of Physicians of Edinburgh,Edinburgh EH2 1LQ

1 Royal College of Physicians of Edinburgh. Consensus con-ference on medical management of stroke: consensusstatement. Proc R Coll Physicians Edinb 1998;28:367-9.

2 Bogousslavsky J. Consensus in stroke management? BMJ1999;318:140-1. (16 January.)

3 Diener HC, Cunha L, Forbes C, Sirenius J, Smets P,Lowenthal A. Dipyridamole and acetyl salicylic acid in thesecondary prevention of stroke. European Stroke Preven-tion Study (2) Group. J Neurol Sci 1996;143:1-13.

4 Caprie Steering Committee. A randomised, blinded, trialof clopidogrel versus aspirin in patients at risk ofischaemic events (CAPRIE). Lancet 1996;348:1329-39.

South Africa must admit thatHIV/AIDS is its greatest enemyEditor—Schuklenk is wrong in stating thatthe price that Glaxo-Wellcome demands forzidovudine is too high for South Africa totreat pregnant women infected with HIV.1 Intruth, Glaxo-Wellcome has set a standard forindustry cooperation with developing coun-tries by making zidovudine available toSouth Africa at a cost three quarters belowthat in Western countries for the preventionof perinatal transmission of HIV.

The company is correct to expect thatthe government should be prepared to pur-chase the drug at a minimal rate. The argu-ment that South Africa cannot afford thisreduced cost of zidovudine is nonsensical ifone considers that the drug would be usedonly for relatively short periods toward theend of pregnancy and that the governmenthas just invested many millions of dollars onthe purchase of fighter aircraft for its armedforces.

The real problem in South Africa is thegovernment’s continuing failure to admitthat HIV/AIDS is its greatest enemy and todevote necessary resources to limiting thespread of HIV.Mark A Wainberg President, International AIDSSocietyMcGill AIDS Centre, Montreal, Canadamdwa@musica.mcgill.ca

1 Schuklenk U. South African government’s response toAIDS crisis is sound. BMJ 1999;318:1143. (24 April.)

Letters

1700 BMJ VOLUME 318 19 JUNE 1999 www.bmj.com