authors' reply We note the comments of Drs Bernsen and van der Wouden. With regard to their first point, the graphs clearly show the changing incidence of new episodes of asthma and acute bronchitis. We isolated and modelled the trend by other methods which gave similar results. However, we acknowledge that the sudden rise in asthma during 1991 (though not evident for acute bronchitis) was not allowed for in our model, and we cannot explain this feature. It may be that the 1989 and 1990 values were, for some reason, unduly low as is supported by the graphs of annual new episode rates over the last 20 years displayed by the LAIA (fact sheet 2000/1 included as an insert in the same edition of Thorax) using the same data base. Whatever view is taken, the new episode rates for both asthma and bronchitis peaked in 1993/4, and rates at the end of the decade were marginally lower than at the beginning.

We referred in the text of the paper to a peaking of total respiratory infections in 1994, but we can only speculate as to the cause. Clearly, there is a strong relationship between acute respiratory infections and asthma attacks.

For asthma, the distinction between new episodes and ongoing consultations is difficult. However, for acute bronchitis there are comparatively few ongoing consultations and our findings are not obscured by this difficulty. While we cannot rule out the possibility of changes in the behaviour of doctors with regard to the allocation of episode type, we know that, individually, doctors are consistent in their recording behaviour.

We can confirm that nationally the number of hospital admissions per annum with a diagnosis of asthma peaked in 1993 for adults and in 1990 for children (Department of Health reports from finished consultant episodes), although caution is needed when interpreting these data since persons treated in accident and emergency departments(especially children) are not necessarily admitted.

The role of nurse facilitators and the relationship between asthma attacks and prevalence has been addressed in the paper. We agree with Drs Bernsen and van der Wouden that the explanation of the findings remains elusive.

I was particularly interested to read the paper by Fleming and colleagues2-1 in which they described evidence for a peaking of the UK asthma epidemic in about 1993 and a decline in incidence thereafter. In speculating about possible causes they exclude improvements in pollution, exposure to allergens, or diet at the relevant time.

In 1994 my colleagues and I proposed that dietary change—specifically, the observed reductions in average intake of fresh fruit, vegetables and fish—had been responsible for making the populations of advanced countries more susceptible to allergy,2-2 and that this was the explanation for the increases in the prevalence of asthma observed worldwide. Since then we have published three studies, all different but all showing evidence of a 3–7-fold increase in risk of wheezy illness in relation to the lowest intakes of foods containing antioxidant vitamins.2-3-2-5 A poor diet does indeed appear to be an important risk factor for asthma.

In the final sentence of our original paper we stated “. . . if the dietary hypothesis is correct, the favourable trend in eating habits between 1985 and 1991 may already be having a beneficial effect”. The trend we referred to was a clear increase in intake of the three foods referred to above, as recorded in the annual national household food consumption and expenditure surveys. We had in mind a decrease in the prevalence of asthma in 10–12 year olds from about the mid 1990s.

The paper by Fleming and colleagues seems to give some further indirect support to our hypothesis. My colleagues and I are currently investigating the influence of maternal diet during pregnancy on allergy in the child, including in these studies fatty acids as well as antioxidants. We believe that a dietary hypothesis for the aetiology of asthma is worthy of very serious scientific investigation, not least because it points to an obvious and simple public heath strategy for prevention.

authors' reply We note Professor Seaton's comments in relation to the potential benefits of an improved diet. We find it difficult to reconcile the very diffuse changes in the incidence of episodes of asthma and of acute bronchitis in all age groups peaking at roughly the same time with a diet based hypothesis. It is unlikely that dietary deterioration in the 1980s and improvement in the 1990s would have occurred simultaneously in all age groups and all regions of the country. The seasonal pattern of asthma attacks3-1 shows a relationship to pollens (a well recognised allergenic factor), but an even stronger relationship to viral respiratory infections, where the links with allergy are much weaker.

The paper by Fleming et al 4-1 reporting the decline in the incidence of asthma episodes raises some interesting questions. While they discuss possible reasons for the decline, we suggest the authors have underestimated the impact of trained asthma nurses.

There has been a threefold rise in whole time equivalent practice nurse numbers since 1988 following the new GP contract and the introduction of payment for asthma chronic disease management (CDM) clinics in the early 1990s. While we agree that historically the nurse's role did not include disease diagnosis, the level of asthma care they provide has increased.4-2 Nurse responsibility for CDM ranges from supporting the GP to the diagnosis and management of asthma; specialist training is recognised as important.

The 1992–4 peak in the mean weekly episodes, both by quarter and by region, may reflect increased nurse involvement at that time. Completion of a recognised asthma course has been shown to be associated with favourable patterns of structure, process, and clinical outcomes in general practice,4-3 as well as reductions in asthma symptoms and numbers of acute attacks where a specialist asthma nurse was employed.4-4 Evidence of improved management in hospitals4-5 by trained asthma nurses has also been reported. Recognised training, as well as the British Thoracic Society's asthma guidelines, have facilitated a more structured management approach and nurses now diagnose and treat new episodes of asthma and suggest appropriate treatment, leading to the development of asthma management protocols and their implementation in primary and secondary care.4-6

We propose that the decrease in the incidence of asthma episodes reported by Fleming et al may reflect better management of asthma by GPs as well as increased and improved asthma management by nurses. Although the introduction of specialist nurses would not have had a significant effect on numbers of consultations fornew episodes of asthma, improved management is likely to have substantially reduced numbers of asthmaexacerbations. We therefore suggest that the impact of nurses' specialist training on reducing episodes of asthma is greater than is currently recognised.

authors' reply The introduction of trained nurse based management programmes for asthma has undoubtedly been good for asthmatic patients. The contribution made to the changing incidence of new episodes of asthma is difficult to estimate but, as the writers recognise, it is likely to be small. We suspect the main factors associated with the decline are linked with those associated with the increase in the 1980s and with those factors associated with the decline in acute bronchitis—a condition much more frequent than asthma and not generally associated with nurse based care.