1085 in your leader indicates that it is at least as potent as metiamide in pharmacological study. However, clinical experience is limited and, although controlled clinical studies are under way, cimetidine cannot at present be made generally available other than in exceptional circumstances where the patient may be suffering from severe and life-threatening complicated peptic-ulcer disease. Patients with the Zollinger-Ellison syndrome will be able to transfer to treatment with cimetidine under strictly controlled circumstances. to
Letters
to
the Editor
REVERSAL OF METIAMIDE-INDUCED
AGRANULOCYTOSIS DURING TREATMENT WITH CIMETIDINE leader SiR,-Your (Oct. 25, p. 802) stated that double-blind clinical trials had shown metiamide, a histamine H2-receptor antagonist, to have significantly and beneficially affected pain and heating of duodenal ulceration and to have controlled gastric secretion in the Zollinger-Ellison syndrome. Metiamide has also been shown to be very effective in the management of acute hxmorrhage from the upper-gastrointestinal tract.’ As indicated by you, these promising early reports were marred by reversible agranulocytosis in three patients. Since that article appeared four more patients have developed agranulocytosis during metiamide treatment for severe and complicated pepticulcer disease (including the Zollinger-Ellison syndrome and generalised systemic mastocytosis). Three of these patients recovered once metiamide was withdrawn, but one has died. One of the former patients is particularly interesting. A man, aged 45 years had a 7-year history of duodenal ulceration for which he had a vagotomy and pyloroplasty (Mr B. Williams) on July 29, 1975. On Aug. 4 he had a massive haemorrhage from a postbulbar duodenal ulcer which was under-sewn at operation. He had copious gastric aspirates postoperatively and a further bleed from an anastomotic ulcer for which he had a partial gastrectomy on Aug. 20. There was a carcinoid-like malignant tumour under the duodenal mucosa in the excised specimen. Serum-gastrin was measured on September 2, 3, and 4 and was found to be 1800, 229, and 336 pg/ml respectively (Dr A. Evans). Postoperatively the patient continued to have copious aspirates of between 2 and 4 I/day and further bleeding. Metiamide 100 mg was infused intravenously on Sept. 5 and within 2 hours the aspirates had ceased. He was treated with metiamide by mouth, his general condition rapidly improved, and he was discharged home to continue taking metiamide 200 mg twice daily and 400 mg at night. At the time of discharge his Hb was 12.4 g/dl, and total white blood-cells were
10 000/mmB On Oct. 30 he developed
a
staphylococcal infection of his abdominal
wound, his nose, and an eye. His total white blood-cells were 2600/mm3,
neutrophils were identified in the peripheral film. Hb was 11-6g/dl. On his admission to hospital on Nov. 3 his total white blood-cells were 700/mm3 with 15% neutrophils. The platelet-count was normal. Metiamide was withdrawn. He had a melxna stool and, his hasmogtobm fell to 8.6 g/dl. Cimetidine was infused intravenously, 1 g in each 24h. Bleeding stopped. On Nov. 6 his total white blood-cells were 3400/mm3 with 47% neutrophils. He maintained his recovery and his white-cell count continued to improve. Treatment with cimetidine 200 mg four times a day was continued by mouth and he was discharged home. He has had no dyspepsia and on Nov. 20 his total white bloodcells were 11 000/mm’ with 60% neutrophils. A bone-marrow aspirate had been obtained on Nov. 11 by which time he was already established on treatment with cimetidine. This showed evidence of recovering agranulocytosis (Dr A. Green).
no
This
case
is both
important and interesting
because it
demonstrates that in a patient with metiamide-induced agranulocytosis the bone-marrow depression was reversed during treament with cimetidine, another H2-receptor antagonist. These data are additional evidence that it is the thiourea moiety of metiamide rather than the H2-receptor blockade itself which is responsible for the bone-marrow depression seen in association with metiamide reatment. Previous evidence has been the absence of hxmatological changes in 66 dogs treated with cimetidine in doses up to 504 mg/kg/day for 3 to 9 months in chronic toxicity studies. This contrasts with agranulocytosis in 5 of 68 dogs treated with doses of up to 244 of metiamide for 1 to 12 months.22 In view of the risk of agranulocytosis from metiamide those few clinicians currently treating patients with metiamide have been advised to withdraw the drug from their patients at the earliest opportunity. The experience with cimetidine referred
mg/kg/day
1. MacDonald, A. S., Steele, B. J., Bottomley, M. G. Lancet (in the press) 2. Brimblecombe, R. W., Duncan, W. A. M., Parsons, M E S Afr med. 1974, 48, 2253.
J
Research Institute, Smith Kline & French Laboratories Limited, Welwyn Garden City, Hertfordshire.
Queen Alexandra Hospital, Cosham, Portsmouth.
W. L. BUR LAND P. C. SHARPE
D.G. COLIN- JONES
P. R. G. TURNBULL P. BOWSKILL
METIAMIDE IN DUODENAL ULCER
SIR,-Iwrite in connection with Mr Earlam’s
comments
(Nov. 15, p. 973) on the multicentre trial of metiamide (Oct. 25, p. 779) to point out that the presence of and change in the symptom of pain cannot be used as a primary index in assessing the course of duodenal ulceration during therapeutic trials. It is true that the main concern of patients with duodenal ulcer is pain, but all the complications of duodenal ulcer can occur in the absence of pain. Conversely, it has often been shown that pain can be dramatically relieved and even cured by therapeutic manipulations of placebo type, without any change in the appearance of the ulcer and the incidence of complications. Therapeutic efficacy in the treatment of duodenal ulcer must, therefore, be assessed endoscopically by monitoring the disappearance, caused by healing, of the duodenal ulceration. Ninewells
Hospital,
Ninewells, Dundee DD2 1UB.
K. G. WORMSLEY
NORMALISATION: A POSSIBLE GENERAL SOLUTION TO THE UNITS PROBLEM SiR,—The extensive and often entertaining correspondence on pascals and newtons and moles and the like in your columns and elsewhere has been concerned entirely with the pros and cons (mostly cons) of the change to SI. No contributor seems to have considered the possibility of a more fundamental change in the way of recording biological values that would liberate us once and for all from dependence on arbitrary and evanescent "absolute" units, and free us in consequence to a very large extent from the burden of memorising normal values and their ranges. We must accept that "absolute" units are always liable to change. The perfectly logical unitary system may be a will-o’the-wisp, but there will always be those who pursue it, dragging us simple pragmatists with them; just as microbiologists insist on changing the names of familiar microorganisms without reference to medical convenience. Somewhere already there must be a committee (very probably several rival committees) drafting Version II of the SI and by the year 2000 the pascal may be as old-fashioned as the scruple. If people like me claim crossly that they have done their bit to winkle medicine out of minims and Fahrenheit into a metric system that is genuinely more convenient to use than AngloSaxon measures, and see no great benefit to ourselves in going further into SI, we must recognise that a good part of the complaint is mere Blimpish conservatism, a mere assertion of a vested interest in keeping things as they are. Yet there is more to it than Blimpery. It is a very considerable labour to learn to remember the normal values of the large number of variables with which one is expected to be acquainted, and to alter them involves not only more labour but a very considerable risk of error. To those who produce the figures, it is a mere matter of altering the calibration of their dials and the parameters in
1086
,
their equations, and to the specialist users change is only a brief episode of re-learning: but to the generality of the profession, who meet most of the figures only occasionally, it will remain a problem for a long time. One must also pity students in the transition phase-not to mention all those who write or edit papers and textbooks. We could avoid all this trouble by recording all variables that behave in anything like a normal statistical fashion (and that is the great majority) by using the process that statisticians call normalisation. For this one counts the normal mean value as 0, and records all other values in terms of the number of standard deviations by which they exceed or fall short of the mean. Thus if the normal mean serum concentration of substance X was 20 mg/100 ml, and its standard deviation 3 mg/100 ml (giving a probable "normal" range of 14-26 mg/100 ml) one would call the mean 0, and the range -2 to +2, and for instance 17 mg/100 ml would become -1 and 33 mg would become +4.3. And the normalised figures would remain exactly the same when the original calculations were done in millimoles, or for that matter in carats per Imperial gallon or molecules per firkin. And one would know that there was at least the strongest of possibilities, without burdening one’s memory with the figures, that the -1 value was within the normal range and the +4.3 well above it, and that exactly the same figures would serve for nearly every substance. There would be exceptions, but remembering them is a far better use for the memory than clinging to all the raw figures. To those who complain that they do not know enough statistics to follow this, I would suggest first that they should read Bradford Hill,’ who makes all this business of means and standard deviations appear just as simple as it really is, and, second, that one does not need to understand statistics to read a normalised value. If the value lies between -2 and +2, it is probably normal: outside that range it is increasingly probably abnormal, until a value below -3 or above +3has only one chance in a thousand of being normal. That is not the whole truth, of course, but then normal ranges as ordinarily expressed in absolute units are far from being the whole truth either. A little practice in their use would make it seem very easy. For those who are still frightened, it would no doubt be usual practice to produce both types of figures during a transition period. There are snags, of course, but the obvious ones are not as great as might seem. Some extra calculations are involved in producing the results, but since most results involve calculations already, usually on a computer or desk calculator, an extra step is of little significance. This is in fact one of the ways in which computers could help by predigesting the information they supply, instead of so often being used simply to produce large amounts of unprocessed data. Things like blood-pressure need only a new scale to be pasted on beside the old. The arguments about precise values of the "normal" mean are mostly of little importance: small variations make little difference to the normalised value. The method would make it easy to allow for variations in value due to differences between laboratories and between populations: it would of course be necessary to signal in some way the fact that the normalised result was adjusted to the population concerned and not based on the textbook "normal". There will be trouble with skewed and irregular distributions like blood-urea: some of them could be handled by using a logarithmic scale or some such device, but in any case it would be at least as easy to remember a +4 value for urea is not as alarming as a -4 would be than it is to remember the absolute figures. There are admittedly many situations where absolute values are required, mostly for calculations : but even if the absolute value was not supplied, provided that one knew the mean and s.D., the finding of the absolute value would merely be a simple first stage of such calculations. Adoption of the system by one hospital would cause difficulties when their students or staff moved elsewhere, and had to start learning the absolute values: but this applies to any change of units or system. 1. Bradford
Hill, A., Principles of Medical Statistics. London, 1971
The real obstacles are human. A surprising number of biochemists seem to have an inbuilt resistance to any device which appears to commit them to interpreting the results ("I’ll give you the figures, it’s your job to say what they mean"), and normalisation is a kind of inbuilt interpretation. Physicians take a great pride in their knowledge of masses of numerical data, and have something of a vested interest in ensuring that the game is not made easier. But most resistance, I believe, derives from our pursuit of this problem in one direction for so long that it is now hard to reconcile people to the idea (a piece of lateral thinking perhaps) that a totally new direction might give a better answer. There are of course many numerical values used in medicine to which the concept could not be applied. However attractive the idea of prescribing in terms of proportions of a "normal" dose (as one does in effect in asking for tablets of aspirin or bottles of blood), most of the numerical values of pharmacology and toxicology will probably remain inaccessible to normalisation. But that is no reason why we should not normalise those values that can be so treated. Western
Infirmary,
BERNARD LENNOX
Glasgow G11 6NT.
SI UNITS
SIR,-Professor Alberti (Oct. 25, p. 820) and Dr Bold and Dr Wilding (Nov. 1, p. 870) make the most modest claim for the benefits of change to Systeme International in clinical medicine. This is certainly not a "major non-event", and there is contention about prior consultation, the scope of which remains obscure. There is no evidence, however, that the protagonists of change now have carte blanche. Clinicians, who hold ultimate responsibility for patient care, have had no say in the proposals. It is unreasonable to suggest that we were fully informed by the Journal of Clinical Pathology and to expect us to debate the issue therein. The first notification received by clinicians of what was to happen was in July, 1975, when the D.H.S.S. instructions (HSC[ 1 s] 140) were received at district level with the mandatory demand that implementation must be made by Dec. 1. These instructions were vague and indefinite and did not specify the precise changes. In July, 1975, the Royal College of Physicians of London expressed its concern in Comitia and requested a delay of one year to allow discussion. This was refused on the ground that it was too late! We are told that a minority of centres have painlessly changed to SI, but this is no justification for its use in the periphery, where the major burden of clinical work is carried out. The present changes are being forced through with unnecessary haste and pressure; and they are the forerunners of the imposition of other scientific measurements which also have disturbing implications in practice. We have decided in the Brighton Health District, by an overwhelming majority, not to make any change, at present, towards SI. We have also been joined by other areas who share our concern. We feel that Systeme International should not be imposed against our will and judgment and that the proposed changes should be carefully studied to see how they can be best applied in clinical practice. If clinicians in sufficient numbers agree and take similar action, it may not be too late to influence events. Brighton General Hospital, Elm Grove, Brighton BN2 3EW.
A. N. G. CLARK
J. SHELDON
DENTAL CARIES AND SUGAR INTAKE
SIR,-We feel compelled to reply to the letter by Dr Walker and Dr Cleaton-Jones (Oct. 18, p. 765), in view of the bias with which their evidence was presented. We do not dispute the detailed facts they chose to present, including possible
Letters
to
the Editor
REVERSAL OF METIAMIDE-INDUCED
AGRANULOCYTOSIS DURING TREATMENT WITH CIMETIDINE leader SiR,-Your (Oct. 25, p. 802) stated that double-blind clinical trials had shown metiamide, a histamine H2-receptor antagonist, to have significantly and beneficially affected pain and heating of duodenal ulceration and to have controlled gastric secretion in the Zollinger-Ellison syndrome. Metiamide has also been shown to be very effective in the management of acute hxmorrhage from the upper-gastrointestinal tract.’ As indicated by you, these promising early reports were marred by reversible agranulocytosis in three patients. Since that article appeared four more patients have developed agranulocytosis during metiamide treatment for severe and complicated pepticulcer disease (including the Zollinger-Ellison syndrome and generalised systemic mastocytosis). Three of these patients recovered once metiamide was withdrawn, but one has died. One of the former patients is particularly interesting. A man, aged 45 years had a 7-year history of duodenal ulceration for which he had a vagotomy and pyloroplasty (Mr B. Williams) on July 29, 1975. On Aug. 4 he had a massive haemorrhage from a postbulbar duodenal ulcer which was under-sewn at operation. He had copious gastric aspirates postoperatively and a further bleed from an anastomotic ulcer for which he had a partial gastrectomy on Aug. 20. There was a carcinoid-like malignant tumour under the duodenal mucosa in the excised specimen. Serum-gastrin was measured on September 2, 3, and 4 and was found to be 1800, 229, and 336 pg/ml respectively (Dr A. Evans). Postoperatively the patient continued to have copious aspirates of between 2 and 4 I/day and further bleeding. Metiamide 100 mg was infused intravenously on Sept. 5 and within 2 hours the aspirates had ceased. He was treated with metiamide by mouth, his general condition rapidly improved, and he was discharged home to continue taking metiamide 200 mg twice daily and 400 mg at night. At the time of discharge his Hb was 12.4 g/dl, and total white blood-cells were
10 000/mmB On Oct. 30 he developed
a
staphylococcal infection of his abdominal
wound, his nose, and an eye. His total white blood-cells were 2600/mm3,
neutrophils were identified in the peripheral film. Hb was 11-6g/dl. On his admission to hospital on Nov. 3 his total white blood-cells were 700/mm3 with 15% neutrophils. The platelet-count was normal. Metiamide was withdrawn. He had a melxna stool and, his hasmogtobm fell to 8.6 g/dl. Cimetidine was infused intravenously, 1 g in each 24h. Bleeding stopped. On Nov. 6 his total white blood-cells were 3400/mm3 with 47% neutrophils. He maintained his recovery and his white-cell count continued to improve. Treatment with cimetidine 200 mg four times a day was continued by mouth and he was discharged home. He has had no dyspepsia and on Nov. 20 his total white bloodcells were 11 000/mm’ with 60% neutrophils. A bone-marrow aspirate had been obtained on Nov. 11 by which time he was already established on treatment with cimetidine. This showed evidence of recovering agranulocytosis (Dr A. Green).
no
This
case
is both
important and interesting
because it
demonstrates that in a patient with metiamide-induced agranulocytosis the bone-marrow depression was reversed during treament with cimetidine, another H2-receptor antagonist. These data are additional evidence that it is the thiourea moiety of metiamide rather than the H2-receptor blockade itself which is responsible for the bone-marrow depression seen in association with metiamide reatment. Previous evidence has been the absence of hxmatological changes in 66 dogs treated with cimetidine in doses up to 504 mg/kg/day for 3 to 9 months in chronic toxicity studies. This contrasts with agranulocytosis in 5 of 68 dogs treated with doses of up to 244 of metiamide for 1 to 12 months.22 In view of the risk of agranulocytosis from metiamide those few clinicians currently treating patients with metiamide have been advised to withdraw the drug from their patients at the earliest opportunity. The experience with cimetidine referred
mg/kg/day
1. MacDonald, A. S., Steele, B. J., Bottomley, M. G. Lancet (in the press) 2. Brimblecombe, R. W., Duncan, W. A. M., Parsons, M E S Afr med. 1974, 48, 2253.
J
Research Institute, Smith Kline & French Laboratories Limited, Welwyn Garden City, Hertfordshire.
Queen Alexandra Hospital, Cosham, Portsmouth.
W. L. BUR LAND P. C. SHARPE
D.G. COLIN- JONES
P. R. G. TURNBULL P. BOWSKILL
METIAMIDE IN DUODENAL ULCER
SIR,-Iwrite in connection with Mr Earlam’s
comments
(Nov. 15, p. 973) on the multicentre trial of metiamide (Oct. 25, p. 779) to point out that the presence of and change in the symptom of pain cannot be used as a primary index in assessing the course of duodenal ulceration during therapeutic trials. It is true that the main concern of patients with duodenal ulcer is pain, but all the complications of duodenal ulcer can occur in the absence of pain. Conversely, it has often been shown that pain can be dramatically relieved and even cured by therapeutic manipulations of placebo type, without any change in the appearance of the ulcer and the incidence of complications. Therapeutic efficacy in the treatment of duodenal ulcer must, therefore, be assessed endoscopically by monitoring the disappearance, caused by healing, of the duodenal ulceration. Ninewells
Hospital,
Ninewells, Dundee DD2 1UB.
K. G. WORMSLEY
NORMALISATION: A POSSIBLE GENERAL SOLUTION TO THE UNITS PROBLEM SiR,—The extensive and often entertaining correspondence on pascals and newtons and moles and the like in your columns and elsewhere has been concerned entirely with the pros and cons (mostly cons) of the change to SI. No contributor seems to have considered the possibility of a more fundamental change in the way of recording biological values that would liberate us once and for all from dependence on arbitrary and evanescent "absolute" units, and free us in consequence to a very large extent from the burden of memorising normal values and their ranges. We must accept that "absolute" units are always liable to change. The perfectly logical unitary system may be a will-o’the-wisp, but there will always be those who pursue it, dragging us simple pragmatists with them; just as microbiologists insist on changing the names of familiar microorganisms without reference to medical convenience. Somewhere already there must be a committee (very probably several rival committees) drafting Version II of the SI and by the year 2000 the pascal may be as old-fashioned as the scruple. If people like me claim crossly that they have done their bit to winkle medicine out of minims and Fahrenheit into a metric system that is genuinely more convenient to use than AngloSaxon measures, and see no great benefit to ourselves in going further into SI, we must recognise that a good part of the complaint is mere Blimpish conservatism, a mere assertion of a vested interest in keeping things as they are. Yet there is more to it than Blimpery. It is a very considerable labour to learn to remember the normal values of the large number of variables with which one is expected to be acquainted, and to alter them involves not only more labour but a very considerable risk of error. To those who produce the figures, it is a mere matter of altering the calibration of their dials and the parameters in
1086
,
their equations, and to the specialist users change is only a brief episode of re-learning: but to the generality of the profession, who meet most of the figures only occasionally, it will remain a problem for a long time. One must also pity students in the transition phase-not to mention all those who write or edit papers and textbooks. We could avoid all this trouble by recording all variables that behave in anything like a normal statistical fashion (and that is the great majority) by using the process that statisticians call normalisation. For this one counts the normal mean value as 0, and records all other values in terms of the number of standard deviations by which they exceed or fall short of the mean. Thus if the normal mean serum concentration of substance X was 20 mg/100 ml, and its standard deviation 3 mg/100 ml (giving a probable "normal" range of 14-26 mg/100 ml) one would call the mean 0, and the range -2 to +2, and for instance 17 mg/100 ml would become -1 and 33 mg would become +4.3. And the normalised figures would remain exactly the same when the original calculations were done in millimoles, or for that matter in carats per Imperial gallon or molecules per firkin. And one would know that there was at least the strongest of possibilities, without burdening one’s memory with the figures, that the -1 value was within the normal range and the +4.3 well above it, and that exactly the same figures would serve for nearly every substance. There would be exceptions, but remembering them is a far better use for the memory than clinging to all the raw figures. To those who complain that they do not know enough statistics to follow this, I would suggest first that they should read Bradford Hill,’ who makes all this business of means and standard deviations appear just as simple as it really is, and, second, that one does not need to understand statistics to read a normalised value. If the value lies between -2 and +2, it is probably normal: outside that range it is increasingly probably abnormal, until a value below -3 or above +3has only one chance in a thousand of being normal. That is not the whole truth, of course, but then normal ranges as ordinarily expressed in absolute units are far from being the whole truth either. A little practice in their use would make it seem very easy. For those who are still frightened, it would no doubt be usual practice to produce both types of figures during a transition period. There are snags, of course, but the obvious ones are not as great as might seem. Some extra calculations are involved in producing the results, but since most results involve calculations already, usually on a computer or desk calculator, an extra step is of little significance. This is in fact one of the ways in which computers could help by predigesting the information they supply, instead of so often being used simply to produce large amounts of unprocessed data. Things like blood-pressure need only a new scale to be pasted on beside the old. The arguments about precise values of the "normal" mean are mostly of little importance: small variations make little difference to the normalised value. The method would make it easy to allow for variations in value due to differences between laboratories and between populations: it would of course be necessary to signal in some way the fact that the normalised result was adjusted to the population concerned and not based on the textbook "normal". There will be trouble with skewed and irregular distributions like blood-urea: some of them could be handled by using a logarithmic scale or some such device, but in any case it would be at least as easy to remember a +4 value for urea is not as alarming as a -4 would be than it is to remember the absolute figures. There are admittedly many situations where absolute values are required, mostly for calculations : but even if the absolute value was not supplied, provided that one knew the mean and s.D., the finding of the absolute value would merely be a simple first stage of such calculations. Adoption of the system by one hospital would cause difficulties when their students or staff moved elsewhere, and had to start learning the absolute values: but this applies to any change of units or system. 1. Bradford
Hill, A., Principles of Medical Statistics. London, 1971
The real obstacles are human. A surprising number of biochemists seem to have an inbuilt resistance to any device which appears to commit them to interpreting the results ("I’ll give you the figures, it’s your job to say what they mean"), and normalisation is a kind of inbuilt interpretation. Physicians take a great pride in their knowledge of masses of numerical data, and have something of a vested interest in ensuring that the game is not made easier. But most resistance, I believe, derives from our pursuit of this problem in one direction for so long that it is now hard to reconcile people to the idea (a piece of lateral thinking perhaps) that a totally new direction might give a better answer. There are of course many numerical values used in medicine to which the concept could not be applied. However attractive the idea of prescribing in terms of proportions of a "normal" dose (as one does in effect in asking for tablets of aspirin or bottles of blood), most of the numerical values of pharmacology and toxicology will probably remain inaccessible to normalisation. But that is no reason why we should not normalise those values that can be so treated. Western
Infirmary,
BERNARD LENNOX
Glasgow G11 6NT.
SI UNITS
SIR,-Professor Alberti (Oct. 25, p. 820) and Dr Bold and Dr Wilding (Nov. 1, p. 870) make the most modest claim for the benefits of change to Systeme International in clinical medicine. This is certainly not a "major non-event", and there is contention about prior consultation, the scope of which remains obscure. There is no evidence, however, that the protagonists of change now have carte blanche. Clinicians, who hold ultimate responsibility for patient care, have had no say in the proposals. It is unreasonable to suggest that we were fully informed by the Journal of Clinical Pathology and to expect us to debate the issue therein. The first notification received by clinicians of what was to happen was in July, 1975, when the D.H.S.S. instructions (HSC[ 1 s] 140) were received at district level with the mandatory demand that implementation must be made by Dec. 1. These instructions were vague and indefinite and did not specify the precise changes. In July, 1975, the Royal College of Physicians of London expressed its concern in Comitia and requested a delay of one year to allow discussion. This was refused on the ground that it was too late! We are told that a minority of centres have painlessly changed to SI, but this is no justification for its use in the periphery, where the major burden of clinical work is carried out. The present changes are being forced through with unnecessary haste and pressure; and they are the forerunners of the imposition of other scientific measurements which also have disturbing implications in practice. We have decided in the Brighton Health District, by an overwhelming majority, not to make any change, at present, towards SI. We have also been joined by other areas who share our concern. We feel that Systeme International should not be imposed against our will and judgment and that the proposed changes should be carefully studied to see how they can be best applied in clinical practice. If clinicians in sufficient numbers agree and take similar action, it may not be too late to influence events. Brighton General Hospital, Elm Grove, Brighton BN2 3EW.
A. N. G. CLARK
J. SHELDON
DENTAL CARIES AND SUGAR INTAKE
SIR,-We feel compelled to reply to the letter by Dr Walker and Dr Cleaton-Jones (Oct. 18, p. 765), in view of the bias with which their evidence was presented. We do not dispute the detailed facts they chose to present, including possible
