Background. This study describes a decision‐theoretic approach to nutrient assessment based on Bayesian methods, which can be used to give accurate estimates of optimum intakes. Analysis of risk is an incomplete technique for dealing with nutrients and other substances that, by definition, have an associated benefit.Results. This paper shows that the risk analysis methods being developed by the Codex Commission on Nutrition and European Food Safety Authority, among others, are inappropriate for assessing safe nutrient intake levels. Decision theoretic methods incorporate benefits associated with these essential nutrients, as well as potential risk. Conclusion. These methods allow for missing or incomplete data, which conventional risk analysis does not.

Regression models with the variance function depending on unknown parameters appear in a number of practical problems (variogram fitting and mixed effect models are popular examples). We found that estimation of parameters entering both response and variance functions can be combined in a rather simple way. The proposed estimator belongs to the class of iterated estimators and is numerically very close to the fixed point method, which takes the form of the reweighted least squares method in our setting. The proposed estimators lead to design problems with additive information matrices and therefore can be treated within traditional convex design theory.

1. There has been an enormous increase in the production and use of synthetic chemicals and in exposure to pollutants this century, which has accelerated in the last 50 years. Five million different chemicals are now recognized and relatively few of the chemicals in use have been adequately tested for toxicity to man or the environment. Of the 3000 or so in large-scale use, the 'minimal' toxicity data required by the Organization for Economic Co-operation and Development (OECD) for a preliminary assessment of health hazards to humans are not publicly available for 75%. There has been little study of exposure to combinations of different chemicals, either simultaneously or in sequence. 2. Establishing the chronic effects of any exposure is very difficult, even in experimental animals. In humans who are regularly exposed to chemical cocktails that vary from day to day and year to year, it becomes almost impossible unless the outcome is a very rare and well-documented condition. However, it is equally impossible to establish that such exposures are without risk. 3. Long-term ill effects have been reported from chronic exposure to heavy metals and by a minority of individuals after the exposure of defined populations to synthetic or pollutant chemicals-chemical accidents, pesticide exposure, 'sick' buildings, implants, etc. Most of these individuals report that their late symptoms are exacerbated by ambient exposures to volatile organic chemicals (VOCs) emitted by everyday products (for instance, perfumes, washing powders and cigarette smoke), which were previously tolerated without trouble and indeed often enjoyed. Similar chronic conditions occurring in other individuals, after single toxic or sub-toxic exposures, or prolonged low-dose exposure, tend to be overlooked. This condition is referred to as multiple chemical sensitivity (MCS). 4. The genuine nature of MCS has been recognized by officially commissioned reports from independent scientists in the USA and the UK, who have concluded that it is a valid diagnosis and a sometimes disabling condition, although all have stressed the need for further research. 5. Because of their biological roles, it has been suggested that pesticides (designed to interfere with enzymes) and hormone mimics may have a particular place in inducing MCS. However, the evidence also implicates VOCs and other compounds such as diesel particulates and the nitrogen oxides. 6. A large study in Bristol has found associations of air freshener and aerosol use with increased prevalence of certain symptoms in both mothers and babies. 7. Patients with MCS are often treated as if their symptoms are psychologically caused. There is no good evidence for this. On the other hand, double-blind challenges have provided positive evidence for the provocation of symptoms by certain chemicals (varying in different patients) at levels too low to be recognized by the patient; challenge has been reported to cause changes in brain activity on SPECT scans. 8. These patients are usually found, on investigation, to be suffering also from the effects of hidden' allergy (intolerance) to a number of different foods and food additives. The symptom reactions they experience are often dissociated in time from the ingestion of the trigger food in such a way that the connection has not previously been made.

Editor—Beral et al have reported an increased relative risk of death from cervical cancer among women using oral contraception.1 Their study controlled for smoking but not for the other principal risk factors for cervical cancer, which include infection with human papillomavirus of a high risk type, sexual behaviour, and inadequate cervical screening. There is therefore a considerable risk of confounding, which could account for the results. The authors do not give patterns of use of contraception by never users and non-users of oral contraceptives; it would be interesting to know what proportion of this comparison group was using barrier methods. Such women are at reduced risk of cervical cancer,2 and therefore a comparison that includes a large proportion of women using barrier methods will seem, artificially, to increase any risk in pill users. It should also be noted that durations of use of less than 10 years are not associated with a significant increase in risk in this study. This apparent duration of use effect may well be related to the risk of acquiring an infection with an oncogenic human papillomavirus.

(1996). Effective Nutritional Medicine: The Application of Nutrition to Major Health Problems: British Society for Allergy and Environmental Medicine with the British Society for Nutritional Medicine. Journal of Nutritional & Environmental Medicine: Vol. 6, No. 2, pp. 191-232.

ED1TOR,-We agree with David Pearson that psychological factors can influence health,'2 but we take exception to his statement that "orthodox medicine has already looked fairly at the claims of clinical ecology." This is simply not true. The Royal College of Physicians' report was highly selective in its literature citations, which served the clear bias of its authors.2 This led us to publish an editorial addressing the potential pitfalls of consensus statements and their influence on medical progress and the quality of patient care.' We subsequently published a detailed critique of the Royal College of Physicians' report, highlighting specifically the numerous errors it contained.4 We acknowledge that the results of well conducted clinical trials are invaluable in assisting doctors in their decision making, but we take issue with the idea that "unproved" treatments should not be used "except possibly as part of a controlled clinical trial." This idea has been promulgated in recent years but is directly contrary to the Declarations of Geneva and Helsinki, which emphasise that a physician's ethical responsibility is to act at all times in his or her patient's best interest. We cannot support an approach that places scientific investigation as senior to acting, in one's judgment, in the best interest of a patient; to do so is to violate the most fundamental ethical principle underlying the practice of medicine. This is the level at which the discussion of innovative treatmnents should be taking place, rather than at the level of the tactics used by Pearson and others. STEPHEN DAVIES

ABSTRACTSingle factor changes from a standard canning process indicated that significant reductions in splitting resulted from higher soak Ca concentrations, higher soak temperatures, higher brine Ca concentrations, and shorter cooking times. Splitting in canned kidney beans was markedly reduced by soaking beans before cooking at temperatures of 66–71°C in solutions containing 150–350 ppm CaCl2. Treatments giving lower gain in weight during soaking led to less splitting. Lower splitting was associated with lower drained weight and firmer cooked beans.