Patients’ denial of disease may pose difficulty for achievinginformed consent

Editor—The 1998 Data Protection Act andthe General Medical Council’s guidance onconfidentiality have resulted in uncertaintyabout whether patient consent is necessaryto fulfil statutory requirements for fair andlawful processing of personal data such asthe identifiable health information collectedby cancer registries.1 2 A regulation isexpected under the Health and Social CareAct 2001 for England and Wales, while aconsultation process is ongoing in Scotlandand Northern Ireland, where health is adevolved matter.

If we wish to retain population basedsurveillance of cancer, the options includeanonymisation of data, informed consent,statutory regulation under the 2001 act, orprimary legislation to make cancer statuto-rily registrable, as for many infectiousdiseases. Anonymisation or pseudo-anonymisation would render the datainvalid, since identifiable data are required toensure quality control, investigate cancerclusters, and perform survival analysis bylinking cancer and death records. The costand difficulty of administering informedconsent and the unquantifiable bias incancer information due to the incomplete-ness of obtaining consent have beenhighlighted.3

Denial of the disease among a subgroupof patients would compound the threat tothe utility of health surveillance data from arequirement for patient consent. Denial is arecognised way of coping with bad newssuch as a cancer diagnosis and doctors may“collude” in denial to help the patient cope.4

Denial would prevent doctors from askingpermission to record the diagnosis in a can-cer registry.

An audit by retrospective case notereview of all patients diagnosed with cancerof the lung, breast, colon, rectum, orovary in Northern Ireland in 1996

(n = 2222) shows that non-discussion ofdiagnosis was part of active patient manage-ment in 14% of lung cancer patients, 9% ofcolorectal, 4% of breast cancer patients, and7% of women with ovarian cancer (table).These patients tended to be older thanaverage.

The actual figures for non-discussioncould be higher, since in 28-55% of notesthere was no record of such discussion. Aquestionnaire survey of 500 cancer patientsdiagnosed more recently (1998-9) in onepart of Northern Ireland showed that evensix months after diagnosis, 4% of patientsstill denied that they had cancer (M JThompson, personal communication).

Requiring informed consent for cancerregistration and patient denial of thediagnosis would cause further loss of data ofsome 4-14%, depending on the cancer.A T Gavin directorD Fitzpatrick biostatisticianR J Middleton data managerNorthern Ireland Cancer Registry, Department ofEpidemiology and Public Health, Queen’sUniversity of Belfast, Belfast BT12 6BJ

M P Coleman headCancer and Public Health Unit, London School ofHygiene and Tropical Medicine, LondonWC1E 7HT

Competing interests: The Northern Ireland CancerRegistry acknowledges funding from the Depart-ment of Health, Social Services and Public Safety,Northern Ireland.

1 Data Protection Act 1998. London: HMSO, 1998. (Clause29.)

2 General Medical Council. Confidentiality: protecting and pro-viding information. London: GMC, 2000. Available atwww.gmc-uk.org/standards

3 Brewster DH, Coleman MP, Forman D, Roche M. Cancerinformation under threat: the case for legislation. AnnOncol 2001;12:145-7.

4 Dunkel-Schetter C, Feinstein LG, Taylor SE, Falke RL. Pat-terns of coping with cancer. Health Psychol 1992;11:79-87.

Global Alliance on Vaccinesand Immunizations

Save the Children UK had concernsabout alliance that went further thanreport did

Editor—We would like to dissociate our-selves from the views reported in Fleck’snews item about the global initiative thatseeks to save millions of children’s lives byimmunisation.1 We were involved in thedesign and analysis of a study that was facili-tated, funded, and published jointly withSave the Children UK. The report looked atfour countries’ experience with the applica-tion process to the Global Alliance onVaccines and Immunizations (GAVI) andtheir perceptions about funding for systemssupport. Save the Children UK had severalconcerns about the alliance that went muchfurther than the report but unfortunatelygave the impression that these stemmedfrom the report.

Specifically, we wish to draw attention tofour points.

Firstly, the study did not criticise the glo-bal alliance “for including managers frompharmaceutical companies on its governingboard”; it made no comment on “a potentialconflict of interest,” or on the “the risk ofcommercial, product-oriented pressure.”

Secondly, the reference to the initiativehaving failed to “ensure that additionalresources were provided to countries” is anerror. An innovative feature of the globalalliance is that it did provide additionalresources.

Thirdly, the study made no suggestionthat such schemes could create “markets forcostly new vaccines while doing little totackle the biggest killer diseases.” Diseasesthat are prevented by vaccination are amongthe big killers in poor countries.

Fourthly, the report did not say that“Ghana was given only 10 days to decidewhether to accept a new hi tech vaccine forhepatitis B without any evidence that thiswas actually needed.” All four countries wel-comed the opportunity to introduce hepati-tis B vaccine. Some respondents in Ghanaexpressed concern about the introduction ofHaemophilus influenzae type B vaccine.

After the initial press release in Genevaon 15 January 2002, Save the Children UKissued the following statement: “The reportis based on country-specific research con-ducted by the London School of Hygieneand Tropical Medicine and facilitated,funded and published jointly with Save theChildren UK. . . . The media documents andstatements of Save the Children UK

Discussion of diagnosis of cancer with patient by type of cancer. Values are percentages unless statedotherwise

Lung (n=670) Colorectal (n=737) Breast (n=678) Ovary (n=137)

Diagnosis:

Discussed 58 52 41 52

Not discussed 14 9 4 7

Not recorded 28 39 55 41

Average age of patient (years):

Cancer discussed 67 66 58 59

Cancer not discussed 72 74 69 72

Letters

Website: bmj.comEmail: letters@bmj.com

974 BMJ VOLUME 324 20 APRIL 2002 bmj.com

therefore represent the considered views ofthat organisation alone and not those of theLondon School of Hygiene and TropicalMedicine. Save the Children UK regrets anyconfusion it may have inadvertently createdbetween its critique of GAVI and the analysisproduced by the London School of Hygieneand Tropical Medicine.”Gill Walt professor of international health policygill.walt@lshtm.ac.uk

Ruairi Brugha senior lecturer in public healthMary Starling independent consultantLondon School of Hygiene and Tropical Medicine,Department of Public Health and Policy, HealthPolicy Unit, London WC1E 7HT

1 Fleck F. Children’s charity criticises global immunisationinitiative. BMJ 2002;324:129. (19 January.)

Vaccine promotion is circumventingmarket forces

Editor—Save the Children UK has notedthe paradox of having vaccine manufactur-ers recommend the use of vaccines in devel-oping countries—the paradox of the sellerdictating demand.

In a market economy buyers evaluatethe cost against benefit. If buyers refuse tobuy then prices come down. The case of vac-cine manufacturers is unique—they havebroken loose from these market imperatives,using a subtle form of blackmail. They haveconvinced organisations such as the WorldHealth Organization that new vaccineresearch can be expected only if goodreturns are ensured. The onus of showingcost benefit has shifted from the manufac-turer to the consumer—organisations suchas the WHO. Two things that support thecharity’s findings are, firstly, exaggeratingbenefits and, secondly, promoting vaccinesthat are not needed.

As an example of exaggerating benefitsMiller (of the WHO’s children’s vaccineinitiative) and Kane suggest that a quarter ofcarriers of hepatitis B in India die at age 45.1

For this they rely on the incidence ofhepatocellular carcinoma in Taiwanesemales (495 cases per 100 000 carrier years).2

This ignores the statement in the report thatthis incidence is three or four times that inwomen. The real incidence in Taiwan worksout at around 320. In Montreal no cases ofhepatocellular carcinoma occurred,although 17 would have been expected.3

Selectively using figures for Taiwanese malesto project the dangers of hepatitis B on tothe world’s population grossly exaggeratesthose dangers.

Then there is the promotion of vaccinesthat are not needed. Nossal recommendedgiving four doses of Haemophilus influenzaetype b vaccine to each year’s birth cohort sothat the price comes down.4 India is home toa quarter of that birth cohort. We haveshown the existence of natural immunity toH influenzae type b in Indian infants.5 Thevaccine is not routinely used here at present,and thus Nossal is suggesting that we use thevaccine so that its price can come down inthe West.

Doctors think that because of vaccines’societal benefits vaccine manufacturers must

be considered differently from other phar-maceutical manufacturers. That, though, isto prejudge the issue. Not all vaccines willhave the same societal benefits. Vaccinemanufacturers cannot be defence and juryin arbitrating the issue. Organisations likethe WHO must avoid the blandishments ofvaccine manufacturers if they are to retaintheir credibility.R K Ojha senior house officerJacob Abraham senior house officerMeenakshi Khosla senior house officerJacob M Puliyel consultant paediatricianPuliyel@vsnl.comDepartment of Paediatrics, St Stephen’s Hospital,Tis Hazari, Delhi 110054, India

1 Miller MA, Kane M. Routine hepatitis B immunisation inIndia: cost-effectiveness assessment. Indian J Pediatr2000;67:299-300.

2 Beasley RP. Hepatitis B virus. The major aetiology of hepa-tocellular carcinoma. Cancer 1988;61:1942-56.

3 Villeneuve J-P, Desrochers M, Infante-Rivard C, Willems B,Raymond G, Bourcier M, et al. A follow up study of asymp-tomatic hepatitis B surface antigen-positive carriers inMontreal. Gastroenterology 1994;106:1000-5.

4 Nossal G. Living up to the legacy. Nature Med1998;5:475-6.

5 Puliyel JM, Agarwal KS, Abass FA. Natural immunity tohaemophilus b in infancy in Indian children Vaccine2001;19:4592-4.

Fifth revision of Declaration ofHelsinki

Clause 29 forbids trials from usingplacebos when effective treatment exists

Editor—I was delighted by the new clause29 in the revised Declaration of Helsinki,which forbids clinical trials comparing drugsagainst no treatment when an effective treat-ment exists. I was dismayed that the WorldMedical Association was retreating from thisposition.1 From reading Hirsch and Guess’spiece in the article about the latest revisionof the Declaration of Helsinki, I amconcerned that opposition from the phar-maceutical industry may lead to furtherrevision.2

I spent two years on a multicentreresearch ethics committee, which saw manyproposals for trials of active versus no treat-ment. In some of these cases an effectivetreatment already existed. I think it wrong toask people who have come seeking treat-ment to do without it for the sake ofresearch. People come to doctors for help,not to act as experimental subjects either forscientific curiosity or for drug regulation.

For example, a trial was proposed of adrug versus placebo in symptomatic benignprostatic hyperplasia. Men in the trial wereto take the placebo for more than a year. Wewere told that this drug was “equivalent” toanother established treatment. What wouldhappen to these men if they were not in thetrial? Surely they would be treated.

The patient information sheet statedthat the drug was available in many othercountries in x mg twice daily form, that anew 2x mg formulation was being studiedand might show the same effectiveness asthe original formulation, and that its effectscompared with the effects of dummycapsules had to be tested. This struck me as

a complete non sequitur. What is necessaryis to compare 2x mg with x mg twice daily.There was no need to test 2x mg against notreatment at all.

Another proposal was for a trial of adrug for Paget’s disease of the bone. Theapplicants had already shown that the drugwas better than placebo with smaller dosesin a larger trial. This was a six month trialwith a bone biopsy. Should we really be ask-ing patients to do this?

I can understand that clause 29 ascurrently stated might be interpreted as pre-venting trials in environments where thebest current clinical methods are notavailable because of their cost. It could beamended to permit such trials withoutallowing patients to be lured into foregoingproved effective treatment where it existsand is accessible to them.3

J M Bland professor of medical statisticsDepartment of Public Health Sciences, St George’sHospital Medical School, London SW17 0REmbland@sghms.ac.uk

1 Ferriman A. World Medical Association clarifies rules onplacebo controlled trials. BMJ 2001;323:825.

2 Tollman SM, Bastian H, Doll R, Hirsch LJ, Guess HA. Whatare the effects of the fifth revision of the Declaration ofHelsinki? BMJ 2001;323:1417-23. (15 December.)

3 Bland JM. WMA should not retreat on use of placebos.BMJ 2002;324:240. (26 January.)

Ethics and money are not goodbedfellows

Editor—Clause 30 of the latest revision ofthe Declaration of Helsinki says: “At theconclusion of the study, every patiententered into the study should be assured ofaccess to the best proven prophylactic, diag-nostic and therapeutic methods identified bythe study.”1

Recently, my unit has participated in atrial comparing continuous subcutaneousinsulin infusions (insulin pump therapy)with multiple daily injections in type 1diabetes. At the end of the trial more than80% of the participants chose to continuehaving the continuous infusions, which theyperceived were of more benefit to them.

At present these infusions are not avail-able to NHS patients, but locally more than50 patients have used their own money tobuy this treatment while remaining NHSpatients. In addition to the cost of purchas-ing a pump (£2000), however, there are alsorunning costs of roughly £1000 each year.Although the National Institute for ClinicalExcellence intends to examine these con-tinuous infusions, any decision is likely to bedelayed. Furthermore, seemingly no newmoney is available to pay for noveltreatments and something else will have togo.

The question is who should pay forthe ongoing costs—secondary care, whichinitiated the treatment; the patients; orprimary care? If primary care, should it beinvolved at the outset of clinical trials interms of being willing to continue support-ing new treatments as suggested by theabove declaration?

Data from randomised controlled clini-cal trials may indicate that one treatment is

Letters

975BMJ VOLUME 324 20 APRIL 2002 bmj.com

better than another, but there is the extradimension of patients being able to choosetheir treatment after participating in a clini-cal trial. Who decides, and should patientshave a choice?David Kerr consultant physicianBournemouth Diabetes and Endocrine Centre,Bournemouth BH7 7DLdavid.kerr@rbch-tr.swest.nhs.uk

1 Tollman SM, Bastian H, Doll R, Hirsch LJ, Guess HA. Whatare the effects of the fifth revision of the Declaration ofHelsinki? BMJ 2001;323:1417-23. (15 December.)

Bipolar affective disorder

Suicide statistics were misleading

Editor—Morriss et al say that the lifetimeprevalence of suicide attempts in peoplewith bipolar affective disorder is 50% andthat this figure could be reduced bytreatment.1 I have two comments.

Firstly, the lifetime prevalence of com-pleted suicides among bipolar patients is ofan order of magnitude less than 50% and isnot significantly different from that ofpeople suffering from schizophrenia, bor-derline personality disorder, alcoholdependency, or affective disorders otherthan bipolar illness.2 Secondly, there is noevidence that treatment prevents suicide inpatients with any mental disorder.3 4

Morriss et al are not the first todramatise the need for funding one oranother project in health service delivery byquoting misleading statistics on suicide. Thetemptation to do so should, however, havebeen resisted.Thomas J P Verberne clinical neuropsychologist(retired)14 Crampton Crescent, Rosanna, Victoria 3084,Australiaverberne@melbpc.org.au

1 Morriss R, Marshall M, Harris A. BMJ 2002;324:61-62Bipolar affective disorder left out in the cold: too late forthe national service framework but local initiatives may bepossible. (12 January.)

2 Inskip HM, Harris C, Barraclough, B. Lifetime risk ofsuicide for affective disorder, alcoholism and schizophre-nia. Br J Psychiatry 1998;172:35-7.

3 Baldessarini RJ, Jamison KR. Summary and conclusions:Effects of medical interventions on suicidal behavior. J ClinPsychiatry 1999;60(suppl 2):117-22.

4 Khan A, Warner HA, Brown WA. Symptom reduction andsuicide risk in patients treated with placebo in antidepres-sant clinical trials. Arch Gen Psychiatry 2000;57:311-7.

Author’s reply

Editor—We carefully worded our paper todiscuss the effects of management onsuicide attempts in bipolar affective disorder,not suicide, precisely because there is littleevidence that treatment prevents suicide inpatients with any mental disorder. We didnot make any claims at all about suicide inbipolar affective disorder.

There is evidence that treatment reducessuicide attempts in bipolar affective disorder.Retrospective and prospective follow up in5233 patient years of observation suggestedthat maintenance treatment with lithium wasassociated with a highly significant 6.4- to 7.5-fold reduction in suicide attempts in patientswith bipolar affective disorder.1 Nevertheless,

there is some evidence that suicide in patientswith bipolar affective disorder is associatedwith suboptimal treatment.

A psychological autopsy study of all 31suicide victims with bipolar affective disor-der in Finland in one year found that 12(39%) patients had expressed their suicidalintent to healthcare personnel in theprevious three months; only 10 (32%)patients were prescribed lithium, and onlythree of 24 patients with major depression atthe time of suicide had received adequatedoses of antidepressants.2 These data sug-gest that some suicides in patients withbipolar affective disorder may be prevent-able with active management of the suiciderisk, adequate treatment of the depressedphase of bipolar disorder with antidepres-sants, and maintenance treatment withlithium.Richard Morriss professor of psychiatryUniversity of Liverpool, Liverpool L69 3BXrmorris@liverpool.ac.uk

1 Tondo L, Baldessarini RJ, Hennen J, Floris G, Silvetti F,Tohen M. Lithium treatment and risk of suicidal behaviorin bipolar disorder patients. J Clin Psychiatry1998;59:405-14.

2 Isometsa ET, Henriksson MM, Aro HM, Lonnqvist JK. Sui-cide in bipolar disorder in Finland. Am J Psychiatry1994;151:1020-4.

First episode psychosis

Patients must be asked about suicidalideation and substance misuse

Editor—Lester’s article on first episode psy-chosis is welcome,1 given that general practi-tioners have a low index of suspicion for thepresence of psychosis and little confidencein diagnosing it.2 But the article omitted twoessential considerations—namely, the needto ask about hallucinations and about therisk of self harm and of harm to others;these may be closely related when commandhallucinations instruct the person to commitself harm or harm others.

Exploration of suicidal ideation (whichmay be independent of depression) is essen-tial. Suicide is the chief cause of prematuredeath among people with schizophrenia,with 4-13% of such people committingsuicide and 25-50% making a suicideattempt.

The risks may be further compoundedby misuse of alcohol and drugs, precipitat-ing impulsive self harm or harm to others.Identification of comorbid substance misusemay be crucial in managing this group,given the increased attendant risks and thestrong association between first episodeschizophrenia and substance misuse.3 Sub-stance misuse is likely to be related tosuicidal ideation and warrants early atten-tion.4 5

Regrettably, little systematic study hasbeen done of people with first episodepsychosis, but the development of a positivetrusting relationship with a single mentalhealth key worker as early as possible isadvisable. Probably the best results comewhen the person with schizophrenia and thekey worker search for common ground. The

key worker should avoid a prematureconfrontation about the explanatory modelof the illness and deliver treatment in asflexible a manner as possible.Andrew Sandor lecturer in social and communitypsychiatryRoyal Free and University College, Medical School,London N19 5NFa.sandor@ucl.ac.uk

1 Lester H. 10-minute consultation. First episode psychosis.BMJ 2001;323:1408. (15 December.)

2 Lincoln C, McGorry P. Who cares? Pathways to psychiatriccare for young people experiencing a first episode of psy-chosis. Psychiatr Serv 1995;46:1166-71.

3 Hambrecht M, Hafner H. Substance and onset ofschizophrenia. Biol Psychiatry 1996;40:1155-63.

4 Spencer E, Birchwood M, McGovern D. Management offirst episode psychosis. Advances in Psychiatric Treatment2001;7:133-42.

5 Kamali M, Kelly L, Gervin M, Browne S, Larkin C,O’Callaghan E.. The prevalence of comorbid substancemisuse and its influence on suicidal ideation amongin-patients with schizophrenia. Acta Psychiatr Scand2000;101:452-6.

GPs must assess patients’ risk behaviour

Editor—Lester highlights the difficulty thatdoctors in primary care have in detectingpsychosis in a 10 minute consultation.1 Shegives good advice on the breadth of theassessment and the action to be taken. Butself harm in psychosis is not mentioned. Thegovernment is aware of the risks of self harmand the risks to others posed by people withmental illness. The assessment of risk behav-iour has a prominent place in the care pro-gramme approach in mental health prac-tice. It is therefore equally important forgeneral practitioners to assess their patients’level of risk as part of the 10 minuteconsultation.

Mortality from unnatural causes ishigher in people with severe mental illnessessuch as psychosis, mood disorders, and per-sonality disorders. People with psychosis areat risk to themselves and others because oftheir potential to act in response tocommand hallucinations or delusionalbeliefs or because of feelings of guilt overpast actions. Harris refers to reports of therisk of suicide in schizophrenia being great-est at the inception of the illness—preciselythe time when the patient may present toprimary care.2 Therefore, inquiry aboutthoughts of self harm or harm to othersmust be an essential component of even themost pressured assessment of a person’smental state in primary care.Ken Courtenay specialist registrar, psychiatryDepartment of Psychiatry, St George’s HospitalMedical School, London SW17 0REkcourten@sghms.ac.uk

1 Lester H. 10-minute consultation: First episode psychosis.BMJ 2001;323:1408. (15 December.)

2 Harris EC, Barraclough B. Excess mortality of mental dis-order. Br J Psychiatry 1998;11:53.

Preimplantation geneticdiagnosis for donor babiescarries some harmEditor—Are the dangers in using preim-plantation genetic diagnosis to create donorbabies as minimal as Boyle and Savulescuclaim?1 The concerns about “commodifica-

Letters

976 BMJ VOLUME 324 20 APRIL 2002 bmj.com

tion” of donor babies are real. The fact thatKant’s dictum was “never use people solelyas a means” is not a reassurance, becausethese donor babies are solely means to anend (the successful treatment of a sicksibling).

To rebuff accusations of instrumental-ism, parents have claimed that they wantedanother child anyway. In the past, additionalchildren gave the hope of providing asuitable donor for an older sibling, but thiswas only a chance. Today, preimplantationgenetic diagnosis guarantees a tissue com-patible donor, and this certainty impacts onthe ethical quality of this decision.

Parents also risk psychological andphysical dangers to the donor baby. Indeed,the Royal College of Obstetricians andGynaecologists has expressed concernabout the potential risks to both mother andchild.2

Because the volume of cord blood issmall it is only effective in treating youngchildren; an older child will need to be givena bone marrow transplant. Molly Nash hasFanconi’s anaemia.1 People with this fre-quently have other organ abnormalities(particularly renal) and are more susceptibleto cancers, especially leukaemias. In thefuture, Molly may ask Adam, her brother, forbone marrow to treat her leukaemia, or akidney for her renal failure.

It is not known if, in choosing a certaintissue type, parents may be choosing to havea child who will later go on to have more ill-nesses and die earlier than a non-compatible child would.

In law, parents may consent to treatmenton behalf of their sick children as proxydecision makers. Not only is it unethical forparents to consent to unnecessary tests ortreatment that are not in the interests of thechild but, as the law stands, there can be nolegal justification for any non-therapeuticinvasion of the bodily integrity of anincapacitated person.

Parents will naturally be distraught at theprospect of the death of a much loved childand will do almost anything to prevent thatdeath. Does this really allow for free andinformed consent when they choose toundertake preimplantation genetic diagno-sis or in subsequent decisions involving thedonor baby?Sue Turner PhD studentInstitute for the Study of Genetics, Biorisks andSociety, University of Nottingham, NottinghamNG7 2RDlqxsct@nottingham.ac.uk

1 Boyle RJ, Savulescu J. Ethics of using preimplantationgenetic diagnosis to select a stem cell donor for an existingperson. BMJ 2001;323:1240-3. (24 November.)

2 Kmietowicz Z. Doctors object to companies offering tostore cord blood. BMJ 2001;323:1203. (24 November.)

Rehabilitation services remainimportant in multiple sclerosisEditor—Taylor’s enthusiasm for the pro-posed “sale or return” arrangement forinterferon beta seems ill informed.1 Giventhe variability of multiple sclerosis over time

in individual patients, how is the success oftreatment to be judged? Using historicalcontrols, such as the Canadian cohort,requires the availability of detailed datato match controls to the current treatmentcriteria.2

In making these choices, quality adjustedlife years and similar utility measures are allwe have, flawed as they are. A better debate isover the cut-off level. According to theNational Institute for Clinical Excellence,the upper threshold for interventionsjudged to offer acceptable NHS value isabout £30 000 per quality adjusted life year.The institute has provided no justificationfor this figure, which is high when comparedwith that of many treatments already in use.The economic output of two residents of theUnited Kingdom, or the disposable income(over the poverty line) of many more, wouldbe required to deliver each added year oflife. This figure could be sustained for a fewrarer diseases but would still infringe the“equal resources for equal need” rule of dis-tributive justice.

The pharmaceuticals market is not afree market as it lacks important characteris-tics: availability of consumer information(on appropriate outcomes), competitivepricing, and a mechanism equivalent to con-sumer choice (choices on treatments andpayment are made by third parties). TheNational Institute for Clinical Excellenceshould offer such a mechanism but is weakand further enfeebled by this risk sharingproposal. The Drug and Therapeutics Bulletinperforms more robust independent assess-ments, often contrary to those by theinstitute.3 4 The recent examples of riluzoleand acetylcholinesterase inhibitors, withtheir meagre benefits, indicate that the newdrugs for neurological diseases are just tooexpensive.

Drug companies control informationand are now more secretive than govern-ments. They put the best gloss on theirresults, such as reporting relative reductionrather than absolute benefit. Researchshould be removed from their controlthrough independent third parties choosingresearchers because of the quality of theirresearch, not the results. All research shouldbe openly registered, with an estimated dateof publication to guarantee openness fur-ther. Until that happens all NHS and univer-sity staff should be required to register theirresearch projects as part of governance.

Patients with multiple sclerosis doindeed need treatments that work. But theseare high quality rehabilitation services(though they are not glamorous or profit-able), which should be available to all,including patients who might use the newtreatments as well—after all, they will notabolish disability.Richard Richards consultant public health physicianNorth Nottinghamshire Health Authority,Rainworth, Mansfield NG21 0ERRichard.richards@nnotts-ha.nhs.uk

1 Taylor D. Funding medicines for people with multiple scle-rosis. BMJ 2001;323:1379-80. (15 December.)

2 Weinshenker BG, Bass B, Rice GP, Noseworthy J,Carriere W, Baskerville J, et al. The natural history of

multiple sclerosis: a geographically based study. I. Clinicalcourse and disability. Brain 1989;112:133-46.

3 Is sibutramine more than a slim hope? Drug Ther Bull2001;39:89-91.

4 Response to Drug and Therapeutics Bulletin–sibutramine.Available at: www.nice.org.uk/article.asp?a = 25608

Postoperativechemoradiotherapy helps ingastric adenocarcinomaEditor—Bowles and Benjamin’s clinicalreview of cancer of the stomach andpancreas omits any consideration of post-operative chemoradiotherapy for patientswith completely resected adenocarcinomaof the stomach.1 The Southwest OncologyGroup has shown significant improvementin overall survival in patients who receivedpostoperative chemoradiotherapy.2 3 Theproportion of patients surviving three yearsafter surgery and chemoradiotherapy was50%, compared with 41% in patients whohad surgery alone.

This magnitude of survival benefit isequivalent to gains reported with preopera-tive chemotherapy in oesophageal cancer4:investigators reported survival at two yearsas being 43% for patients who hadchemotherapy and surgery, compared with34% for those who had surgery alone. Theyconcluded that neoadjuvant chemotherapyfollowed by surgery should become stand-ard treatment in trials in this group ofpatients.

Presumably neoadjuvant chemotherapyshould also be considered as standard forpatients with oesophageal cancer, for whomsuch trials are unavailable. In order to havethe opportunity for equivalent survivalgains, patients with completely resected gas-tric adenocarcinoma should have a clinicaloncology assessment for consideration ofpostoperative chemoradiotherapy.Kenneth S Wilson consultant in medical oncologyBritish Columbia Cancer Agency, Vancouver IslandCentre, Victoria, British Columbia, CanadaV8R 6V5KWilson@bccancer.bc.ca

1 Bowles MJ, Benjamin IS. ABC of the upper gastrointestinaltract: Cancer of the stomach and pancreas. BMJ2001;323:1413-6. (15 December.)

2 Macdonald JS, Smalley S, Benedetti J, Estes N, Haller DG,Ajani JA, et al. Postoperative combined radiation andchemotherapy improves disease-free survival and overallsurvival in resected adenocarcinoma of the stomach andG.E. junction. Results of intergroup study INT-0116(SWOG 9008). Proc Am Soc Clin Oncol 2000;plenaryabstract 1.

3 Macdonald JS, Smalley SR, Benedetti J, Hundahl SA, EstesNC, Stemmermann GN, et al. Chemoradiotherapy aftersurgery compared with surgery alone for adenocarcinomaof the stomach or gastroesophageal junction. N Engl J Med2001;345:725-30.

4 Clark PI. Neoadjuvant chemotherapy in oesophageal can-cer. Eur J Cancer 2001;37 (suppl 6):140.

Endoscopy is useful duringpercutaneous tracheostomyEditor—Percutaneous tracheostomy is nowwell established in intensive care units in theUnited Kingdom.1 Routine use of endos-copy during this procedure is less well estab-lished, and Susanto’s comment that mostpercutaneous tracheostomies are done

Letters

977BMJ VOLUME 324 20 APRIL 2002 bmj.com

under endoscopic guidance requires com-ment.2 A survey of intensive care units in theUnited Kingdom found that less than onethird of units routinely used fibreopticendoscopy during the procedure.1

We reviewed seven papers published in2000 or 2001 that either compared percuta-neous tracheostomy with surgical tracheos-tomy or compared different methods ofpercutaneous tracheostomy. In only two ofthese studies was endoscopy routinely used.

The use of endoscopy may decrease asexperience of the technique increases. Inone prospective study of 100 cases endos-copy was used for the first 14 cases and thenabandoned as a routine procedure.3

We use continuous fibreoptic endoscopywith a video display for all percutaneous tra-cheostomies. This allows assessment of theairway before the procedure and detectionof complications during the procedure. Howcan early, initially subclinical, complicationsbe detected if they are not looked for? Com-plications may occur because of compo-nents of the procedure that are common tomany other techniques, such as intratrachealplacement of the needle and guidewire, orbecause of the method of dilating the tract.Continuous endoscopic inspection mayallow a distinction to be made between thesecomplications.

Future studies of percutaneous trache-ostomy should include endoscopy as aroutine component of the procedure.Paul Jefferson consultant in anaesthetics andintensive carep.jefferson@dgri.scot.nhs.uk

David R Ball consultant in anaesthetics and intensivecareDumfries and Galloway Royal Infirmary, DumfriesDG1 4AP

1 Cooper RM. Use and safety of percutaneous tracheostomyin intensive care. Report of a postal survey of ICU practice.Anaesthesia 1998;53:1209-12.

2 Susanto I. Comparing percutaneous tracheostomy withopen surgical tracheostomy. BMJ 2002;324:3-4. (5January.)

3 Velmahos GC, Gomez H, Boicey CM, Demetriades D. Bed-side percutaneous tracheostomy: prospective evaluation ofa modification of the current technique in 100 patients.World J Surg 2000;24:1109-15.

Income inequality andpopulation health

Better measures of social differentiationand hierarchy are needed

Editor—Mackenbach’s editorial reads likean obituary for the hypothesis that incomeinequality is related to population health.1

But a substantial body of evidence of such arelation has accumulated over the past 20years, not only in the United States but alsoin Brazil, Russia, Taiwan, and England.Attempts to explain away this relation arerarely relevant to more than one of themany contexts in which it occurs.

The fact that health is more closelyrelated to income in developed societiesthan to differences in income between themsuggests effects of relative income or socialstatus.2 But if income distribution has itsmain effect through differences in social sta-

tus, and individual income (or education) isa proxy for individual social status, control-ling income distribution for individualincome makes little sense.3 It is a differencewithout a distinction, and more status equal-ity is likely to improve average health.

As with individual income, the assump-tion that the median income of small areasmeasures material consumption rather thansocial relativities is unwarranted. Differencesin median income between small areas arecomponents of the income inequality of thelarger areas. Choosing smaller areas con-verts variance, which would be incomeinequality in larger areas, into what is naivelytaken to be absolute income. Hence associa-tions between income inequality and healthtend to be strongest in larger areas andweakest in smaller areas, while the oppositeis true of associations between medianincome and health.

Poor social affiliations and low statuscarry high population attributable risks.More unequal societies not only suffer morerelative deprivation but tend to have lowerrates of trust and of community involvementand—as over 40 studies show—more vio-lence. More unequal societies will be moredifferentiated by social rank into relations ofdominance and subordination and less ableto enjoy more egalitarian and inclusive rela-tions consistent with higher social capitaland less class and racial prejudice.4 The linkswith violence show that inequality haspsychosocial effects, and, given the powerfulassociation between violence and othercauses of death, critics should tread withcaution.

Income may work better as a proxy fordifference in social status in the UnitedStates than elsewhere. European differencesin status ranking may be smaller and needmore subtle markers. To test hypotheses thatare fundamentally about differences insocial status we need better measures ofsocial differentiation and the importance ofhierarchy in a society. The social dominanceorientation scale might be a good place tostart.5

Richard Wilkinson professor of social epidemiologyDivision of Public Health Sciences, University ofNottingham Medical School, NottinghamNG7 2UHRichard.Wilkinson@Nottingham.ac.uk

1 Mackenbach JP. Income inequality and population health.BMJ 2002;324:1-2. (5 January.)

2 Wilkinson RG. Health inequalities: relative or absolutematerial standards? BMJ 1997;314:591-5.

3 Marmot M, Wilkinson RG. Psychosocial and materialpathways in the relation between income and health: aresponse to Lynch et al. BMJ 2001;322:1233-6.

4 Wilkinson RG. Mind the gap: hierarchies, health and humanevolution. London: Weidenfeld & Nicolson, 2000.

5 Sidanius J, Pratto F. Social dominance: an intergroup theory ofsocial hierarchy and oppression. Cambridge: Cambridge Uni-versity Press, 1999.

Hierarchy and health are related

Editor—Mackenbach’s editorial and theaccompanying four studies suggest thatindividual income rather than income distri-bution is the important determinant ofpopulation health.1

The studies presented do not deny arelation between hierarchy and health, but

the results suggest that income distributionmay not always be a good measure of thathierarchy. Non-income aspects of socialrankings operating in specific cultures mayoverpower single economic measures suchas income distribution. In addition, as theJapanese study suggests, high levels ofinequality may be required before incomeeffects are shown. Given that a perfect meas-ure of inequality or hierarchy does notcurrently exist, however, the lack of such ameasure does not refute the possibility of animportant relation between hierarchy andhealth.

In Japan the poorest region (Okinawa)seems to have the lowest mortality,2 which isat odds with Mackenbach’s individual incomehypothesis. The Japanese study of Shibuya etal looked at self rated health, but this measurehas been validated as related to mortalitymeasures by only one study in Japan,compared with 13 studies of American popu-lations.3 Only 0.8% of Shibuya et al’s samplerated their health as poor, in contrast to stud-ies in other populations with higher rates,such as 4.8% in an American study.4

The Copenhagen study confirms previ-ous evidence. Denmark is the only rich coun-try that has worse health, as measured by lifeexpectancy, than the United States (theworld’s richest and most powerful country).Unlike the United States government, theDanish government has an official website(www.sum.dk/health/sider/print.htm) thatcalls attention to its relatively poor healthstatus compared with that of other Europeancountries and to its not having had thesame secular time improvements as othercountries.

This site also shows Denmark’s poor lifeexpectancy ranking, an unusual admissionby a country. A higher prevalence ofsmoking may not be a satisfactory explana-tion: Japan is the world’s healthiest countrywhen ranked by life expectancy but has thehighest male smoking rates of any countryin the Organisation for Economic Coopera-tion and Development.

If individual income, rather than itsdistribution, is so important then why doesKerala, one of India’s poorest states, have alife expectancy approaching that of theUnited States?5 The key piece of evidencerequired to help convince the wealthy wouldbe that the health of the richer is betterwhere hierarchy is decreased and everyoneshares more in the fruits of society.Stephen Bezruchka affiliate associate professorDepartment of Health Services, School of PublicHealth and Community Medicine, University ofWashington, Box 357660, Seattle, WA 98195-3576,USAsabez@u.washington.edu

1 Mackenbach JP. Income inequality and population health.BMJ 2002;324:1-2. (5 January.)

2 Cockerham WC, Hattori H, Yamori Y. The social gradientin life expectancy: the contrary case of Okinawa in Japan.Soc Sci Med 2000;51:115-22.

3 Idler EL, Benyamini Y. Self-rated health and mortality: areview of twenty-seven community studies. J Health SocBehav 1997;38:21-37.

4 Kawachi I, Kennedy BP, Glass R. Social capital andself-rated health: a contextual analysis. Am J Public Health1999;89:1187-93.

5 Sen A. Development as freedom. New York: Knopf, 2000.

Letters

978 BMJ VOLUME 324 20 APRIL 2002 bmj.com

If only lottery winnings wererandomly assignedEditor—The causal effect of income onhealth has important policy implications,and yet its identification is dogged by theproblem of income not being randomlyassigned. Researchers continue to ignorethis, including Rodgers in his editorial,1 butthe idea that the problem can be overcomeby exploiting lottery winnings as a “naturalexperiment”2 is superficial because winnersare not randomly selected from the popula-tion since playing is not a random event.

Although playing is popular, it is farfrom universal. The probability of winning isdirectly proportional to the number of tick-ets bought, so winners, on average, will bedisproportionately heavy players.

Survey evidence suggests that playersand non-players differ systematically interms of their observable characteristicssuch as age, education, and sex.3 Indeed,expenditure survey data show that, com-pared with non-players, players insure less,smoke more, work less, have more modestpension provision, are more likely to live insocial housing, have larger credit card debts,and save less. Controlling for all of theseobservable differences through matching ormultivariate methods may well be possible,but considerable unobservable heterogen-eity that would be difficult to control forwould probably remain.

Data suggest that lottery players havehigher rates of “time preference” thannon-players do. In games in the UnitedStates most states offer jackpot winners thechoice between $x a year for the next 20years or $10x in cash today; most winnerschoose the latter even though the presentvalue of the former far exceeds it at typicalinterest rates. This suggests that the timepreference among players is high—higherthan would be suggested by typical interestrates. This raises the problem that winnersare more likely to be impatient people with ahigh time preference who have invested lessthan others in their health (by diet, exercise,not smoking, etc) throughout their lives.

It is important to know not only whethermoney matters but also when (if ever) it mat-ters. Does having more money when we areyoung—perhaps money given by ourparents—make for better health in the longrun than having more money when we areold (and perhaps already sick)? Havinglottery winnings randomly distributedwould have been a useful way of answeringimportant policy questions—if only winningwere random.Ian Walker professorDepartment of Economics, University of Warwick,Coventry CV4 7ALi.walker@warwick.ac.uk

1 Rodgers, A. Income, health, and the National Lottery. BMJ2001;323:1438-9. (22-29 December.)

2 Osler M, Prescott E, Grønbæk M, Christensen U, Due P,Engholm G. Income inequality, individual income, andmortality in Danish adults: analysis of pooled data fromtwo cohort studies. BMJ 2002;324:13-6. (5 January.)

3 Farrell L, Walker I. The welfare effects of lotto: evidencefrom the UK. J Public Economics 1999;72:99-120.

Surgeons should beware ofplans to print individualperformanceEditor—Before the NHS proceeds muchfurther with the idea of publishing details ofthe performance of individual surgeons,1

note ought to be taken of experience in theUnited States, the country with most experi-ence in the public release of such data. Themain consequence will be a reluctance totreat riskier patients.

Since 1992 Pennsylvania has publisheda consumer guide to coronary artery bypasssurgery; this lists annual risk adjustedmortality for all hospitals and surgeons pro-viding such surgery in the state. Schneiderand Epstein conducted a survey of cardiolo-gists and cardiac surgeons to seek theirviews on its usefulness and limitations.2 Alto-gether, 59% of cardiologists reportedincreased difficulty in finding surgeonswilling to perform coronary artery bypasssurgery in severely ill patients who requiredit and 63% of the cardiac surgeons reportedthat they were less willing to operate on suchpatients.

Similarly, one study suggested that themovement of severely ill patients to an adja-cent state had been a measurable effect ofNew York State’s public reporting of data oncoronary artery bypass surgery.3 Whereasthis movement can occur in a consumer, pri-vately funded, system, it is unclear wherethese patients would move to if faced withsimilar problems in the United Kingdom.Access to health care for seriously ill patientswould probably be decreased, ultimatelyleading to a decrease in NHS costs—abyproduct that would be welcomed by thegovernment.

A further complicating factor is thatoperative mortality is always attributed tothe surgeon. Thus the subtle but importantinfluences of (in heart cases) cardiologicalmanagement and referral, anaesthetic care,and intensive care resources are ignored. Arecent editorial highlighted the importanceof anaesthetic management in influencingsurgical wound healing, which is arguablythe commonest cause of postoperative mor-bidity, long hospital stays, and increasedcosts.4 These factors are some of those usedto compare surgical outcome and yet maybe outside surgical control. In other words,surgeons may be criticised for their postop-erative morbidity when compared with theircolleagues, but it may all be the fault of theanaesthetist.Nicholas Pace consultant anaesthetistWestern Infirmary, Glasgow G11 6NTnp@blanefield.com

1 Vass A. Performance of individual surgeons to bepublished. BMJ 2002;324:189. (26 January.)

2 Schneider EC, Epstein AM. Influence of cardiac surgeryperformance reports on referral practices and access tocare. N Engl J Med 1996;335:251-6.

3 Omoigui NA, Miller DP, Brown KJ, Annan K, Cosgrove D3rd, Lytle B, et al. Outmigration for coronary bypasssurgery in an era of public dissemination of clinicaloutcomes. Circulation 1996;93:27-33.

4 Buggy D. Can anaesthetic management influence surgicalwound healing? Lancet 2000;356:355-7.

Open letter to the NationalBlood ServiceEditor—Another invitation to give bloodhas arrived, bringing with it a now familiarwash of guilt. I have not given since last July,and I am not sure whether I shall ever do itagain. My sister, whose rare blood groupmakes her contribution particularly valuable,will not go either, and for the same reasons.

We both had blood taken by someonewho was not competent to do the job. Weboth suffered pain and extensive bruising.Giving blood is not pleasant but it shouldnot leave the arm discoloured and sore fromelbow to wrist for over a week. Ourexperience the previous time had been bad,but the second time, it was unacceptable.

I left a message informing the local serv-ice in Caernarfon, and their public relationswoman called me back. She promised thatshe would make sure that the nurse respon-sible would ring me. I heard nothing. I rangthe central London office and spoke toanother woman who seemed to take mycomplaint seriously, although, she said, therewas nothing she could do. She promised tofax the medical director without delay.Again, I heard nothing. Nobody, it seemed,was willing to accept responsibility.

Although inadequate skill of a particularphlebotomist is, presumably, a local issue, Ibelieve that my experience highlights awider problem with the blood service. Giventhat donors are needed, should not the serv-ice do its best to minimise the unpleasant-ness of the experience?

I have been giving blood for 30 years.Little has changed. There is no welcome,only a series of queues; staff do notintroduce themselves, nobody takes care ofnovice donors, nobody takes any notice if adonor says her arm hurts (a reliable sign, Ihave discovered, that the needle has notbeen efficiently inserted). Donors are nottreated as people—even less, as volunteerswho should be encouraged.

But few people will put up with this sortof treatment nowadays. If donors can bepersuaded to come, they are not likely toreturn. I have never been invited to givefeedback on my experiences. It seems thatthe blood service does not want to knowwhat happens to its donors: it did not followup my calls, nor question why my sister didnot return. I would feel reassured if I knewthe local problem was being taken seriouslyand the phlebotomist had received furthertraining, but I have no reason to believe anyaction has been taken. No wonder the NHSis short of blood.Arabella Melville independent consultant for the NHSBryn Derwen, 1 Osmond Terrace, Porthmadog,Gwynedd LL49 9AN

Correspondence submitted electronicallyis available on our website

Letters

979BMJ VOLUME 324 20 APRIL 2002 bmj.com