data show that the most striking improvement was a reduction in deaths from cardiogenic shock. This may well have occurred because of poor matching of the placebo and treatment groups. The treatment group were younger and had a lower prevalence of previous ischaemic heart disease than the placebo group. The idea that intravenous magnesium could be beneficial by causing coronary vasodilatation is interesting, but to our knowledge there are no trials in vivo comparing the use of magnesium with intravenous nitrates.

In the United Kingdom, about half the patients with Graves' disease who are given antithyroid drugs are still in remission one year after treatment is stopped. The most widely held view is that such remission rates are due only to the biochemical effects of the drugs, the disease either spontaneously remitting or abating when the immune system is no longer subject to the stimulatory effects of excessive thyroid hormone. We review here the accumulating evidence against both of these alternatives. In contrast, there is now a large body of work which shows that thyrotrophin receptor antibody levels, central to the aetiology of Graves' hyperthyroidism, fall during antithyroid treatment and that remission may be related to this fall in a fashion which is dependent on the dose and duration of treatment. This immunosuppressive effect is supported by experimental data and on the basis of these results we propose that antithyroid drugs may modify the natural history of Graves' disease and contribute to the remission which occurs in a proportion of treated patients.

To investigate the distribution of thyroid-stimulating antibody (TSAb) activity between IgG subclasses, sera from 11 patients with Graves disease (including the National Institute of Biological Standards and Control (NIBSC) Research Standard, long acting thyroid stimulator-B) were fractionated by chromatography on affinity columns of monoclonal IgG subclass antibodies or protein A to deplete all but a single subclass. The resulting fractions were 98% or more pure for a single subclass. In all 11 patients, TSAb activity appeared to be confined to the IgG, fraction as determined by cAMP production on addition of the fractions to the FRTL-5 rat thyroid cell line. In all of eight specimens from seven patients so tested, the whole serum activity was recovered in the IgG, fraction, after adjusting for the recovery of the isotype from the column. TSAb activity in one serum comprised both lambda and kappa light chains but was IgG1 restricted. This IgG subclass restriction was not found when the same fractions were tested for thyroglobulin, microsomal/thyroid peroxidase, or tetanus toxoid antibody activity. Together with previous results showing marked restriction of both light chain usage and isoelectric point of TSAb, these results support the idea that Graves' disease may be the result of an oligo-or possibly monoclonal response at the B cell level. (J. Clin. Invest. 1990. 86:723-727.)

Up to half of the children in our study had Cl difficile isolated from the stool with no associated faecal toxin; this has been noted before. The absence of any clinical differences between those who do and those who do not have the organism (and whether toxigenic or non-toxigenic in vitro) and a similar incidence of isolation of the organism among controls support the notion that the presence of Cl difficile alone and without demonstrable faecal toxin is not primarily pathogenic.The lesser incidence of the organism in our control group despite a threefold increase in antibiotic usage is of interest and confirms the findings of the Swedish group who found that Cl difficile occurred numerically more often in children who had not been treated by antibiotics than in those who had been given such drugs.8 This may be related to the more frequent use of the antianaerobic antibiotic erythromycin in the controls. Alternatively children who have gastroenteritis may be more prone to Cl difficile associated diarrhoea and colitis. Thus other microorganisms are found in three quarters of patients (Ellis et al, to be published); the incidence of toxin in our gastroenteritis group was twice that of the controls; and parallel experience from adult cases suggests that disturbance in bowel function due to causes other than antibiotics may also predispose to Cl difficile diar-The role of Cl difficile cytotoxin in children with gastroenteritis is not totally clear; many recover spontaneously, whereas the clinical condition of others demands treatment, to which they apparently respond. For children who are not that ill guidelines for treatment are lacking. Sigmoidoscopic and histological appearances of the rectal mucosa may be normal (especially if the disease is proximal) or non-specific; and toxin titres often bear little relation to severity of the disease. In view of the disastrous consequences of failure to treat some cases4 it seems prudent to treat all children who have symptoms and toxin in the stool with at least five days of oral vancomycin.We are most grateful to D M Jones, director of the Public Health Laboratory, Withington, for his generous contribution of the use of the laboratory facilities in this study. We also thank Sisters M Coyle, N Fosbrook, and S Thompson and the nursing staff of the paediatric wards at Monsall Hospital for their efficient and untiring contribution in making the study run smoothly. Mr Andrzej Makowski, of the department of statistics, North West Regional Health Authority, gave expert help with the statistical data. Serial samples of peripheral blood mononuclear cells from 42 patients with hyperthyroid Graves' disease were collected at monthly intervals before, during, and for 12 References

A study was conducted investigating the possibility that the immunosuppressive action of methimazole (the active metabolite of the antithyroid drug carbimazole) might be due to an effect on the production of oxygen radicals by monocytes. Techniques comprised measurement of luminol dependent chemoluminescence in monocytes and a spectrophotometric assay for production of hydrogen peroxide.The results showed definite inhibition of formation of oxygen radicals by resting and stimulated monocytes, which may explain the immunosuppressive action of the drug in Graves' disease. The findings also suggest that the formation of oxygen radicals and the initiation of the immune response may be closely related.