191
professions of restricted admission, extensive training, and punishment mistakes still happen. Our studies of errors involving health
drug names indicate that these mistakes may be related more to the organisation of memory and the processing of information than to laziness or negligence. A practitioner’s memory appears to be organised so that typical or familiar drug names are centrally located and easily accessed; then familiar objects may be recognised without using conscious attentionThus a less familiar drug name (’Losec’ when it came onto the market) is likely to be recognised as the more familiar ’Lasix’, as happened in the case reported by Dr Faber and colleagues (May 25, p 1287). We reviewed reports of wrong drug dispensing errors by pharmacists and concluded that the mistakes were unlikely to be due to chance and could be caused by the organisation of memory. Reports published between 1977 and 1987 totalled 52. We designated drug products reported to be in the top 200 drugs prescribed each year as familiar and other drugs as unfamiliar. We then assigned each reported error to one of four classes: a familiar drug substituted for a familiar drug (n = 11), a familiar drug substituted for an unfamiliar one (n = 19), an unfamiliar substituted for a familiar drug (n =1 ), and an unfamiliar drug substituted for an unfamiliar drug (n=21). The distribution was significantly different from that predicted by chance (=364, p<0’001). Similar results were obtained when we studied drug name recognition in the laboratory. Ten pharmacists, with an average of 13 years of experience, were shown a drug name on the screen of a microcomputer so that the name was barely perceptible. When asked
indicate which of two names, one familiar and one was shown they were more likely to substitute the familiar for the unfamiliar. If errors are indeed secondary to information processing failure, drug names could be tested before product release to prevent errors such as Lasix for Losec, or drug administration procedures could be altered to decrease the risk of recognition error. This source of iatrogenic disease may be preventable by considering recognition errors as information processing failures not moral failures.
rate in the Minnesota group (in contrast with the outpatients) that made analysis difficult. The researchers pointed out that inpatient treatment has repeatedly been shown to be no better than outpatient treatment; according to Miller and Hester it is, if anything, less effective, although much more profitable for the private sector. The apparent superiority of the Minnesota unit in the Finnish trial may merely reflect the fact that its staff were new and perhaps more enthusiastic than those working in the traditional unit. Only 14% of the Minnesota patients abstained for 12 months. About 20% of both Minnesota and traditional patients were reported as controlled drinkers but controlled drinking is anathema to the Minnesota model and thus cannot be regarded as a specific effect of Minnesota treatment. By contrast Azrin et al3 reported very high levels of abstinence when using a combination of outpatient behavioural psychotherapy and supervised disulfiram. It is ironic that Salaspuro and Keso urge the adoption of laboratory markers for evaluation. Their advice is sound but their original study showed no significant difference in gammaglutamyltransferase (GGT) activity between the two groups,44 whereas a UK randomised controlled trial of outpatient-supervised disulfiram found a significant advantage attributable to disulfiram in terms of GGT reduction as well as in several measures of alcohol
consumption.5 The Stapleford Centre, 25A Eccleston Street, London SW1W 9NP, UK
Health Sciences Center, University of Arizona, Tucson, Arizona 85721, USA
Forgive and remember: managing medical failure. Chicago: University of
E, Mervis CB. Family resemblances studies categories. Cogn Psychol 1975; 7: 573-605.
2 Rosch
m
the internal structure of
SIR,—While the details ot their case are very well described by Ur Faber and colleagues their conclusion is not the most appropriate one. The real lesson from this tragedy is that the mistake would not have occurred if the doctors caring for this patient had written legibly (perhaps even printed) "omeprazole" rather than "Losec". There is no justification for implying that it would have been better to have prescribed an alternative brand name. Department of Medicine, Division of Digestive Diseases and Nutrition, University of South Carolina School of Medicine, Columbia, South Carolina 29203, USA
COLIN W. HOWDEN
Alcoholism treatment according to the Minnesota model SIR,-Professor Salaspuro and Dr Keso (June 22, p 1550) imply treatment
may be useful because
to a Minnesota unit abstained for the next 12 months did patients than inpatient randomised to a traditional inpatient unit. Although their investigation of this important and contentious issue is impartial, their findings hardly constitute support for the Minnesota model as the first-choice treatment. Another randomised study! showed no difference in outcome between outpatient treatment, Minnesota inpatient treatment, or no treatment, but there was a very high non-acceptance and drop-out
substantially
more
1986; 41: 794-805. N, Sisson R, Meyers R, Godley M Alcoholism
treatment by disulfiram and community reinforcement therapy. J Behav Ther Exp Psychiatry 1982; 13: 105-12. 4. Keso L, Salaspuro M. Inpatient treatment of employed alcoholics: a randomised clinical trial of Hazelden-type and traditional treatment. Alcohol Clin Exp Res 1990; 14: 584-89. 5 Chick J, Gough K, Falkowski W, et al. Disulfiram treatment of alcoholism. Br J
3. Azrin
Psychiatry (in press).
hormone treatment
MARION SLACK
Chicago Press, 1979.
that "Minnesota model"
2.
Early electroretinogram alterations in Creutzfeldt-Jakob disease after growth
College of Pharmacy,
1. Bosk CL
S, Parmander M, Allebeck P. A trial that failed and the reasons why: comparing the Minnesota model with outpatient treatment and non-treatment for alcohol disorders. Scand J Soc Med 1990; 18: 221-24. Miller W, Hester R. Inpatient alcoholism treatments: who benefits? Am Psychologist
1. Andreasson
to
unfamiliar,
COLIN BREWER
patients randomly assigned
SIR,-Dr Billette de Villemeur and colleagues (April 6, p 864) describe Creutzfeldt-Jakob disease (CJD) in two children (aged 10 and 11) treated with pituitary human growth hormone (hGH). These were the first two French cases to be reported. We have done repeated neurophysiological investigations in these children, as well as in a third patient (aged 20) with a very similar clinical course of CJD following hGH treatment. In these three consecutive cases, electroretinograms (ERG) showed early and profound anomalies. Pathological changes were obvious at the first recording one month after clinical onset--slight increases in "a" and "b" wave peak times, morphological changes, and a large reduction of "b" wave amplitude (a/b amplitudes ratio 0-70, normal range 027-041). The ERG was completely absent a few months later, whereas the optic fundus remained normal. Foster et aP described a retinal involvement in scrapie-infected rodents and emphasised that retinopathogenicity can vary according to both the strain of scrapie and the host genotype. Curtis et aF found that ERG changes, detectable in some mice 118 days after intracerebral injection of scrapie, coincided with the earliest morphological evidence of photoreceptor damage. Our findings show that the retina is affected early, at the same time as the first neurological signs, in ataxic CJD after hGH treatment in childhood. Early ERG recordings might also be useful for diagnostic assessment of CJD, especially in its ataxic forms, and even of other encephalopathies due to prions in man and animals. Laboratory of Paediatric Neurophysiology, Hôpital Trousseau, 75571 Paris, France 1. Foster
JD, Fraser H, Bruce ME. Retinopathy Neuropathol Appl Neurobiol 1986; 12: 185-96
2. Curtis
FRANCIS RENAULT PASCAL RICHARD in mice with
experimental scrapie.
R, Fraser H. Foster JD, Scott JR The correlation of electroretinographic and histopathological findings in the eyes of mice infected with the 79A strain of scrapie. Neuropathol Appl Neurobiol 1989; 15: 75-89.