184
is
diagnostic of thrombocythaemic erythromelalgia. In untreated thrombocythaemic erythromelalgia frequently progresses to painful ischaemic acrocyanosis and frank gangrene. Low-dose aspirin can improve the peripheral ischaemic circulation,3 but Raynaud’s-like symptoms may continue because of the persistence of acrocyanotic microvascular ischaemia. The incurable variant-burning distress in the absence of any detectable underlying disorder-is rare. I have designated this entity "idiopathic or primary erythermalgiaand postulated six criteria for the condition: (1) attacks of local red vasodilatation with increased local heat and burning pain constitute the basic disturbances; (2) bilateral distribution; (3) production and aggravation by exercise and heat; (4) relief always provided by cold, rest, and elevation of the affected extremities; (5) no primary or associated disease; and (6) no response to treatment. In contrast with the asymmetric or unilateral localisation of thrombocythaemic erythromelalgia in toes and fingers occurring in adulthood or old age 2,3 primary erythermalgia arises in childhood or puberty as bilateral, more or less symmetrical, burning distress in the feet, ankles, and lower legs.4 There is relative sparing of the toes and no progression to peripheral ischaemia and gangrene. Secondary erythermalgia, without platelet involvement, has been cases,
described in association with inflammatory and circulatory disturbances such as gout, systemic lupus erythematosus, rheumatoid arthritis, cryoglobulinaemia, and endarteritis obliterans, for example, and also with diabetes, neurological conditions, and treatment with vasoactive drugs. These secondary forms of erythermalgia are not identical to the specific condition of thrombocythaemic erythromelalgia, because prompt and lasting relief of pain by aspirin is lacking. Department of Haematology, Dijkzigt Univeristy Hospital, Erasmus University, 3015 Rotterdam, Netherlands
JAN J. MICHIELS
1. Michiels JJ, Ten Kate FWJ,
Vuzevski VD, Abels J Histopathology of erythomelalgia thrombocythaemia. Histopathology 1984; 8: 669-78. 2. Michiels JJ, Abels J, Steketee J, Van Vliet HHDM, Vuzevski VD. Erythromelalgia caused by platelet mediated artenolar inflammation and thrombosis m thrombocythemia. Ann Intern Med 1985; 102: 466-71. 3. Michiels JJ, Van Joost Th. Erythromelalgia and thrombocythemia: a causal relation. J Am Acad Dermatol 1990; 22: 107-11. 4. Michiels JJ, Van Joost Th, Vuzevski VD. Idiopathic erythermalgia: a congenital disorder. J Am Acad Dermatol 1989; 21: 1128-30. m
Graves’ ophthalmopathy and smoking SIR,-In a case-control study the purpose of the comparison (control) group is to identify factors that occur more (or less) often in the cases and which may increase or reduce the risk of the condition under investigation.1 The control group, difficult though it may be to identify, must therefore consist of an unbiased sample of the base population from which the cases arise. They must not be biased with respect to the exposure of interest. Unfortunately Dr Shine and colleagues (May 26, p 1261) did not clearly state the hypothesis that they intended to test, and one consequence was the selection of three small control groups. The validity of each is debatable and the rationale for amalgamating them for comparison with the case group is not discussed. The control groups may have been seriously biased with respect to smoking because the information was derived from questionnaires, to which there was a poor response. Thus the individuals who contributed smoking information were self-selected responders and thus unlikely to have been representative of the groups from which they were derived.2 In one group the response rate was only 32% and the reported prevalence of current smoking was only 13 %. This group almost certainly provided a biased underestimate of smoking, and the apparently significant association of Graves’ disease with smoking may have arisen simply as a consequence of control group selection bias. We agree with your May 12 editorial that criticism of epidemiological studies should not be based on methodological pedantry. However, the fundamental problems relating to the control groups used by Shine et al mean that their findings must be
viewed with considerable circumspection. The apparent association between smoking and Graves’ ophthalmopathy does merit further examination but this would be best achieved in a study large enough to test a single hypothesis and with attention to complete data collection. The effects of potential confounders such as age and sex might also be taken into account. Department of Public Health Medicine, Leicestershire Health Authority, Leicester LE1 6TP, UK
R. SHUKLA
Department of Community Health, University of Leicester
J. J. KURINCZUK
NE, Day NE. Statistical methods in cancer research: vol I-the analysis of case-control studies. Lyon: International Agency for Research on Cancer, 1980. 2. Seltzer CC, Bosse R, Garvey AJ. Mail response by smoking status. Am J Epidemiol 1974; 100: 453-57 1. Breslow
***This letter has been shown to Dr Shine and Dr Weetman, whose reply follows.-ED. L. SIR,- We thank Dr Shukla and Dr Kurinczuk for their comments. Our hypotheses were that there is an association between Graves’ ophthalmopathy and smoking, and that within the group of patients with Graves’ ophthalmopathy there is an association between the degree of ophthahnopathy and amount of tobacco consumption. Our control groups were all of a similar size to the patient group, with much the same responses in three of the four groups. Although self-selection may lead to bias, the direction and magnitude of the bias might be expected to be similar in all groups. It is often only possible to detect bias after the event, by appropriate statistical analysis. We detected no significant differences between the three control groups either with respect to prevalence (chi-squared 47, 4 DF excluding the Graves’ ophthalmopathy group), or to pack-year history (chi-squared 8-4, 4 DF excluding the Graves’ ophthalmopathy group). We are thus happy with the homogeneity of the control groups, and that the first hypothesis is most probably correct. This hypothesis is further supported by our retrospective analysis of data from case records.’ The second hypothesis is lent support by the association of smoking history with degree of ophthalmopathy (our fig 2). Here, there is no need for a specific control group, since we are making a within-group comparison. The low prevalence of smoking seen in both the Moorfields and Cambridge control groups, might be expected from the high overall ages of our subjects. Our survey was conducted three years after the last general survey reported by the Office of Population Censuses and Surveys. If the trend of the previous fifteen years has continued, we would expect a smaller proportion of the general population to be smokers than in that series. There may, in addition, be regional differences, for which our comparison groups should have controlled. Institute of Ophthalmology, University of London, London EC1V 9AT, UK
BRIAN SHINE TONY WEETMAN
1. Shme B, Edwards
OM, Weetman AP. Graves’ ophthalmopathy and smoking Acta Endocrinologica 1989 121 (suppl 2): 182-84 2. Office of Population Censuses and Surveys Cigarette smoking 1972 to 1986. OPCS Monitor. London: Government Statistical Service, 1988; Feb 9: 1-10.
Schizophrenia and encephalitis lethargica SIR,-My spirits were lifted by Dr Boyle’s letter (April 7, p 853) on prevalence of the sequelae of encephalitis lethargica in the population studied by Kraepelin and Bleuler in their identification of dementia praecox and schizophrenia. During my early apprenticeship in 1949, when I was charged with the care of 400 or 500 chronic patients, there was in every ward a sprinkling of cases labelled "encephalitis lethargica, effects of’. These were distinguishable to me because of parkinsonism, choreoathetosis, or night calls to deal with oculogyric crises. Although I had read one or two good books before starting my apprenticeship, I was not able to distinguish these patients from those labelled "chronic paranoid schizophrenia", except by observing their disorders of movement. I the
is
diagnostic of thrombocythaemic erythromelalgia. In untreated thrombocythaemic erythromelalgia frequently progresses to painful ischaemic acrocyanosis and frank gangrene. Low-dose aspirin can improve the peripheral ischaemic circulation,3 but Raynaud’s-like symptoms may continue because of the persistence of acrocyanotic microvascular ischaemia. The incurable variant-burning distress in the absence of any detectable underlying disorder-is rare. I have designated this entity "idiopathic or primary erythermalgiaand postulated six criteria for the condition: (1) attacks of local red vasodilatation with increased local heat and burning pain constitute the basic disturbances; (2) bilateral distribution; (3) production and aggravation by exercise and heat; (4) relief always provided by cold, rest, and elevation of the affected extremities; (5) no primary or associated disease; and (6) no response to treatment. In contrast with the asymmetric or unilateral localisation of thrombocythaemic erythromelalgia in toes and fingers occurring in adulthood or old age 2,3 primary erythermalgia arises in childhood or puberty as bilateral, more or less symmetrical, burning distress in the feet, ankles, and lower legs.4 There is relative sparing of the toes and no progression to peripheral ischaemia and gangrene. Secondary erythermalgia, without platelet involvement, has been cases,
described in association with inflammatory and circulatory disturbances such as gout, systemic lupus erythematosus, rheumatoid arthritis, cryoglobulinaemia, and endarteritis obliterans, for example, and also with diabetes, neurological conditions, and treatment with vasoactive drugs. These secondary forms of erythermalgia are not identical to the specific condition of thrombocythaemic erythromelalgia, because prompt and lasting relief of pain by aspirin is lacking. Department of Haematology, Dijkzigt Univeristy Hospital, Erasmus University, 3015 Rotterdam, Netherlands
JAN J. MICHIELS
1. Michiels JJ, Ten Kate FWJ,
Vuzevski VD, Abels J Histopathology of erythomelalgia thrombocythaemia. Histopathology 1984; 8: 669-78. 2. Michiels JJ, Abels J, Steketee J, Van Vliet HHDM, Vuzevski VD. Erythromelalgia caused by platelet mediated artenolar inflammation and thrombosis m thrombocythemia. Ann Intern Med 1985; 102: 466-71. 3. Michiels JJ, Van Joost Th. Erythromelalgia and thrombocythemia: a causal relation. J Am Acad Dermatol 1990; 22: 107-11. 4. Michiels JJ, Van Joost Th, Vuzevski VD. Idiopathic erythermalgia: a congenital disorder. J Am Acad Dermatol 1989; 21: 1128-30. m
Graves’ ophthalmopathy and smoking SIR,-In a case-control study the purpose of the comparison (control) group is to identify factors that occur more (or less) often in the cases and which may increase or reduce the risk of the condition under investigation.1 The control group, difficult though it may be to identify, must therefore consist of an unbiased sample of the base population from which the cases arise. They must not be biased with respect to the exposure of interest. Unfortunately Dr Shine and colleagues (May 26, p 1261) did not clearly state the hypothesis that they intended to test, and one consequence was the selection of three small control groups. The validity of each is debatable and the rationale for amalgamating them for comparison with the case group is not discussed. The control groups may have been seriously biased with respect to smoking because the information was derived from questionnaires, to which there was a poor response. Thus the individuals who contributed smoking information were self-selected responders and thus unlikely to have been representative of the groups from which they were derived.2 In one group the response rate was only 32% and the reported prevalence of current smoking was only 13 %. This group almost certainly provided a biased underestimate of smoking, and the apparently significant association of Graves’ disease with smoking may have arisen simply as a consequence of control group selection bias. We agree with your May 12 editorial that criticism of epidemiological studies should not be based on methodological pedantry. However, the fundamental problems relating to the control groups used by Shine et al mean that their findings must be
viewed with considerable circumspection. The apparent association between smoking and Graves’ ophthalmopathy does merit further examination but this would be best achieved in a study large enough to test a single hypothesis and with attention to complete data collection. The effects of potential confounders such as age and sex might also be taken into account. Department of Public Health Medicine, Leicestershire Health Authority, Leicester LE1 6TP, UK
R. SHUKLA
Department of Community Health, University of Leicester
J. J. KURINCZUK
NE, Day NE. Statistical methods in cancer research: vol I-the analysis of case-control studies. Lyon: International Agency for Research on Cancer, 1980. 2. Seltzer CC, Bosse R, Garvey AJ. Mail response by smoking status. Am J Epidemiol 1974; 100: 453-57 1. Breslow
***This letter has been shown to Dr Shine and Dr Weetman, whose reply follows.-ED. L. SIR,- We thank Dr Shukla and Dr Kurinczuk for their comments. Our hypotheses were that there is an association between Graves’ ophthalmopathy and smoking, and that within the group of patients with Graves’ ophthalmopathy there is an association between the degree of ophthahnopathy and amount of tobacco consumption. Our control groups were all of a similar size to the patient group, with much the same responses in three of the four groups. Although self-selection may lead to bias, the direction and magnitude of the bias might be expected to be similar in all groups. It is often only possible to detect bias after the event, by appropriate statistical analysis. We detected no significant differences between the three control groups either with respect to prevalence (chi-squared 47, 4 DF excluding the Graves’ ophthalmopathy group), or to pack-year history (chi-squared 8-4, 4 DF excluding the Graves’ ophthalmopathy group). We are thus happy with the homogeneity of the control groups, and that the first hypothesis is most probably correct. This hypothesis is further supported by our retrospective analysis of data from case records.’ The second hypothesis is lent support by the association of smoking history with degree of ophthalmopathy (our fig 2). Here, there is no need for a specific control group, since we are making a within-group comparison. The low prevalence of smoking seen in both the Moorfields and Cambridge control groups, might be expected from the high overall ages of our subjects. Our survey was conducted three years after the last general survey reported by the Office of Population Censuses and Surveys. If the trend of the previous fifteen years has continued, we would expect a smaller proportion of the general population to be smokers than in that series. There may, in addition, be regional differences, for which our comparison groups should have controlled. Institute of Ophthalmology, University of London, London EC1V 9AT, UK
BRIAN SHINE TONY WEETMAN
1. Shme B, Edwards
OM, Weetman AP. Graves’ ophthalmopathy and smoking Acta Endocrinologica 1989 121 (suppl 2): 182-84 2. Office of Population Censuses and Surveys Cigarette smoking 1972 to 1986. OPCS Monitor. London: Government Statistical Service, 1988; Feb 9: 1-10.
Schizophrenia and encephalitis lethargica SIR,-My spirits were lifted by Dr Boyle’s letter (April 7, p 853) on prevalence of the sequelae of encephalitis lethargica in the population studied by Kraepelin and Bleuler in their identification of dementia praecox and schizophrenia. During my early apprenticeship in 1949, when I was charged with the care of 400 or 500 chronic patients, there was in every ward a sprinkling of cases labelled "encephalitis lethargica, effects of’. These were distinguishable to me because of parkinsonism, choreoathetosis, or night calls to deal with oculogyric crises. Although I had read one or two good books before starting my apprenticeship, I was not able to distinguish these patients from those labelled "chronic paranoid schizophrenia", except by observing their disorders of movement. I the
