- Email: [email protected]
HOME TREATMENT FOR HÆMOPHILIA
230
loped this yet. No practical way has been suggested of using the hospital’s expertise after it has been closed. Representation to the authorities- of these and many other points has been fruitless. Our only hope now is to try to mobilise professional and public opinion in order to our own new A.H.A., or the R.H.A., or the C.H.C. many c.H.c.s have to be consulted about the closure
influence
(how
of a regional unit ?), or the Secretary of State. My colleagues and I would therefore be grateful for support of any kind from anybody who cares about the incurably mentally disabled and who would not like to see them deprived even further of their fair share of attention and resources. St. Wulstan’s
Hospital,
Malvern,
ROGER MORGAN,
Worcestershire.
Director of Rehabilitation.
YERSINIA ENTEROCOLITICA INFECTION AND THYROID DISORDERS JIR,--1 he incidence ot sarcoidosis
is
high
in
Scan-
dinavians,! as are infections by Yersinia enterocolitica (Y.E.). Dr Lidman and others (Dec. 14, p. 1449) found that Y.E. antigens cross-react with thyroid epithelial cells, but, using higher titres, Dr Bech and her colleagues (Oct. 19, p. 951) did not find a prozone phenomenon. A common factor was apparent in the three Y.E.-positive cases reported by Dr Bonsdorff and Dr Friman (Dec. 28, p. 1565). Their first case had de Quervain’s thyroiditis, and this condition is sometimes caused by sarcoidosis.2,3 Sarcoid infiltration may have also caused the thyroiditis in their second example, since the third case had sarcoid thyroiditis. Sarcoid granulomas formed in the thyroid gland produce hyperthyroidism,4 Hashimoto’s thyroiditis, or myxredema,5 depending on the activity of the immune response. Sarcoidosis sometimes also produces false-positive L.E. tests 6 and non-specific elevation of brucella antibodies.7 Possibly, the negative tuberculin test found in 70% of cases, -results from skin hyperimmunity rather than immune deficiency.8 Sarcoid infiltration is still diagnosed less often than it occurs. Wood Cottage,
Chilworth, Surrey.
G. A. MACGREGOR.
HOME TREATMENT FOR HÆMOPHILIA SIR,-As the father of a schoolboy with severe factor-ix deficiency, I should like to say that the home treatment programme for patients with haemophilia, as described by Le Quesne et al.,9 has revolutionised our family life. All the upheaval, the delays, and the pain associated with having to go to hospital each time a bleed occurred put a great strain on the whole family. Now, a 20-minute treatment, administered by our son in the comfort of his home, is usually all that is required. As a result he is free to attend school regularly, to enjoy an active social life, and to travel at will, safe in the knowledge that he has his own treatment with him. The correspondence in your columns, initiated by Biggs,lO about the short supply of factor-viii concentrate for the of hasmophilic patients, has made how fortunate we are that sufficient factor-ix treatment
1. 2. 3. 4.
5. 6. 7. 8. 9.
10.
us
realise
concentrate
Scadding, J. G. Sarcoidosis. London, 1967. Birchall, G. Br. J. clin. Pract. 1966, 20, 586. MacGregor, G. A. Br. med. J. 1969, ii, 385. MacGregor, G. A. Proc. R. Soc. Med. 1974, 67, 221. Karlish, A. J., MacGregor, G. A. Lancet, 1970, ii, 330. Matz, R., Zucker, S. Metabolism, 1967, 16, 522. Ellman, P., Hanson, A. Br. J. Tuberc. 1958, 52, 218. MacGregor, G. A. Br. med. J. 1971, ii, 471. Le Quesne, B., Britten, M. I., Maragaki, C., Dormandy, K. M. Lancet, 1974, ii, 507. Biggs, R. ibid. 1974, i, 1339.
is produced in the United Kingdom to allow our son and other patients to have home treatment. It must be terrible for the parents of a haEmophilic boy on treatment with freeze-dried concentrate to realise that they are privileged, while others go without, and also to have to face the possibility that they may have to give up home treatment because the State finds it too expensive. Those in authority should realise that, whether viewed in terms of human suffering or Treasury finance, it can never be worth denying this sort of treatment. 64 Knoll
Drive,
London N14.
L. B. LEVEN.
UNDERSTANDING AND PRESENTING VARIATION
SiR,-I have just looked at the latest issue of The Lancet (Jan. 4) for examples of variability in biological measurement for use in teaching undergraduate medical students. I am a regular reader of articles in my field in your journal, but this may be the first occasion I have consistently gone
through each
paper. It was disappointing to see the way in which data on variation was presented. I will use two examples, chosen accidentally without knowledge of
authors
or subject material. First, from Smith et all can be taken " serum-growthhormone concentrations were higher in group A at 2 hours (17-6±7-4 v. 5-2±2-3 ng. per ml.)". Is, for example, the value 7-4 the standard deviation, or the standard error of the mean, or half the length of some unspecified confidence interval, or twice the standard deviation, or twice the
standard error of the mean, or what ? I suspect it is the standard deviation, in which case it would be better to say something like " the mean was 17-6 ng. per ml., with a standard deviation of 7-4 ng. per ml.", followed by whatever arguments are to be based on it. If I am wrong, and the value 7-4 is the standard error of the mean, then this implies that the standard deviation must be 7-4x6=18-1ng. per ml. This figure is larger than the mean itself, and suggests that statistics based on the normal distribution should not be used for interpretation of this data, without first transforming to a more appropriate scale to remove excessive skewness. Secondly, a comment on the symbol :::1::, which is widespread in the medical literature. It is used in all the tables in your issue I examined where the standard error of the mean is presented. The implication to me is that, to take an example from table i of Ghosh et a1.2, the standard error of the mean plasma-triglyceride concentration after acute myocardial infarction in a placebo group of patients one month after start of treatment was ±26-87 mg. per 100 ml. This is just not so-the standard error of the mean can only be positive, and so here it is 26-87 mg. per 100 ml. The ability to use the standard error of the mean to form so-called confidence limits, for an interval in which the true value of the mean may lie with some stated probability, is where the shorthand ± may be helpful. In this case, approximate 95% confidence limits for the true mean plasma-triglyceride concentration are 180-23 ±1-96x26-87 -i.e., an interval covering 127-56 to 232-90 mg. per 100 ml. The approximation is due to the use of the multiplying factor 1-96 (taken from normal tables); more correctly for small samples a percentage point taken from t tables should be used.3 The fault for these common inadequacies and misunderstandings must be shared between teachers, users, and publishers. It would be helpful to all, if the wide use of the Smith, R., Fuller, D. J., Wedge, J. H., Williamson, D. H., Alberti, K. G. G. M. Lancet, Jan. 4, 1975, p. 1. 2. Ghosh, P., Cochrane, A. M. G., de Bono, D. ibid. p. 9. 3. Armitage, P. Statistical Methods in Medical Research; p. 110. 1.
Oxford and Edinburgh, 1971.
loped this yet. No practical way has been suggested of using the hospital’s expertise after it has been closed. Representation to the authorities- of these and many other points has been fruitless. Our only hope now is to try to mobilise professional and public opinion in order to our own new A.H.A., or the R.H.A., or the C.H.C. many c.H.c.s have to be consulted about the closure
influence
(how
of a regional unit ?), or the Secretary of State. My colleagues and I would therefore be grateful for support of any kind from anybody who cares about the incurably mentally disabled and who would not like to see them deprived even further of their fair share of attention and resources. St. Wulstan’s
Hospital,
Malvern,
ROGER MORGAN,
Worcestershire.
Director of Rehabilitation.
YERSINIA ENTEROCOLITICA INFECTION AND THYROID DISORDERS JIR,--1 he incidence ot sarcoidosis
is
high
in
Scan-
dinavians,! as are infections by Yersinia enterocolitica (Y.E.). Dr Lidman and others (Dec. 14, p. 1449) found that Y.E. antigens cross-react with thyroid epithelial cells, but, using higher titres, Dr Bech and her colleagues (Oct. 19, p. 951) did not find a prozone phenomenon. A common factor was apparent in the three Y.E.-positive cases reported by Dr Bonsdorff and Dr Friman (Dec. 28, p. 1565). Their first case had de Quervain’s thyroiditis, and this condition is sometimes caused by sarcoidosis.2,3 Sarcoid infiltration may have also caused the thyroiditis in their second example, since the third case had sarcoid thyroiditis. Sarcoid granulomas formed in the thyroid gland produce hyperthyroidism,4 Hashimoto’s thyroiditis, or myxredema,5 depending on the activity of the immune response. Sarcoidosis sometimes also produces false-positive L.E. tests 6 and non-specific elevation of brucella antibodies.7 Possibly, the negative tuberculin test found in 70% of cases, -results from skin hyperimmunity rather than immune deficiency.8 Sarcoid infiltration is still diagnosed less often than it occurs. Wood Cottage,
Chilworth, Surrey.
G. A. MACGREGOR.
HOME TREATMENT FOR HÆMOPHILIA SIR,-As the father of a schoolboy with severe factor-ix deficiency, I should like to say that the home treatment programme for patients with haemophilia, as described by Le Quesne et al.,9 has revolutionised our family life. All the upheaval, the delays, and the pain associated with having to go to hospital each time a bleed occurred put a great strain on the whole family. Now, a 20-minute treatment, administered by our son in the comfort of his home, is usually all that is required. As a result he is free to attend school regularly, to enjoy an active social life, and to travel at will, safe in the knowledge that he has his own treatment with him. The correspondence in your columns, initiated by Biggs,lO about the short supply of factor-viii concentrate for the of hasmophilic patients, has made how fortunate we are that sufficient factor-ix treatment
1. 2. 3. 4.
5. 6. 7. 8. 9.
10.
us
realise
concentrate
Scadding, J. G. Sarcoidosis. London, 1967. Birchall, G. Br. J. clin. Pract. 1966, 20, 586. MacGregor, G. A. Br. med. J. 1969, ii, 385. MacGregor, G. A. Proc. R. Soc. Med. 1974, 67, 221. Karlish, A. J., MacGregor, G. A. Lancet, 1970, ii, 330. Matz, R., Zucker, S. Metabolism, 1967, 16, 522. Ellman, P., Hanson, A. Br. J. Tuberc. 1958, 52, 218. MacGregor, G. A. Br. med. J. 1971, ii, 471. Le Quesne, B., Britten, M. I., Maragaki, C., Dormandy, K. M. Lancet, 1974, ii, 507. Biggs, R. ibid. 1974, i, 1339.
is produced in the United Kingdom to allow our son and other patients to have home treatment. It must be terrible for the parents of a haEmophilic boy on treatment with freeze-dried concentrate to realise that they are privileged, while others go without, and also to have to face the possibility that they may have to give up home treatment because the State finds it too expensive. Those in authority should realise that, whether viewed in terms of human suffering or Treasury finance, it can never be worth denying this sort of treatment. 64 Knoll
Drive,
London N14.
L. B. LEVEN.
UNDERSTANDING AND PRESENTING VARIATION
SiR,-I have just looked at the latest issue of The Lancet (Jan. 4) for examples of variability in biological measurement for use in teaching undergraduate medical students. I am a regular reader of articles in my field in your journal, but this may be the first occasion I have consistently gone
through each
paper. It was disappointing to see the way in which data on variation was presented. I will use two examples, chosen accidentally without knowledge of
authors
or subject material. First, from Smith et all can be taken " serum-growthhormone concentrations were higher in group A at 2 hours (17-6±7-4 v. 5-2±2-3 ng. per ml.)". Is, for example, the value 7-4 the standard deviation, or the standard error of the mean, or half the length of some unspecified confidence interval, or twice the standard deviation, or twice the
standard error of the mean, or what ? I suspect it is the standard deviation, in which case it would be better to say something like " the mean was 17-6 ng. per ml., with a standard deviation of 7-4 ng. per ml.", followed by whatever arguments are to be based on it. If I am wrong, and the value 7-4 is the standard error of the mean, then this implies that the standard deviation must be 7-4x6=18-1ng. per ml. This figure is larger than the mean itself, and suggests that statistics based on the normal distribution should not be used for interpretation of this data, without first transforming to a more appropriate scale to remove excessive skewness. Secondly, a comment on the symbol :::1::, which is widespread in the medical literature. It is used in all the tables in your issue I examined where the standard error of the mean is presented. The implication to me is that, to take an example from table i of Ghosh et a1.2, the standard error of the mean plasma-triglyceride concentration after acute myocardial infarction in a placebo group of patients one month after start of treatment was ±26-87 mg. per 100 ml. This is just not so-the standard error of the mean can only be positive, and so here it is 26-87 mg. per 100 ml. The ability to use the standard error of the mean to form so-called confidence limits, for an interval in which the true value of the mean may lie with some stated probability, is where the shorthand ± may be helpful. In this case, approximate 95% confidence limits for the true mean plasma-triglyceride concentration are 180-23 ±1-96x26-87 -i.e., an interval covering 127-56 to 232-90 mg. per 100 ml. The approximation is due to the use of the multiplying factor 1-96 (taken from normal tables); more correctly for small samples a percentage point taken from t tables should be used.3 The fault for these common inadequacies and misunderstandings must be shared between teachers, users, and publishers. It would be helpful to all, if the wide use of the Smith, R., Fuller, D. J., Wedge, J. H., Williamson, D. H., Alberti, K. G. G. M. Lancet, Jan. 4, 1975, p. 1. 2. Ghosh, P., Cochrane, A. M. G., de Bono, D. ibid. p. 9. 3. Armitage, P. Statistical Methods in Medical Research; p. 110. 1.
Oxford and Edinburgh, 1971.