Data on seasonality of deaths due to asthma were omitted from paper but editorial's author did not know

Editor—Readers of Ann J Woolcock's editorial about our paper may have wondered at her comments on the seasonality of deaths due to asthma1 although data on this were not presented in our paper.2 This was because we had been asked to remove these results to shorten the paper and she had not been told. However, we wish to present them here. The dataset was the same as that referred to in our paper. Briefly, all deaths due to asthma (International Classification of Diseases code 493) in England and Wales from 1 January 1983 to 31 December 1995 were included. Deaths were divided into calendar months, an adjustment being made for the different lengths of the months. Age specific rates were calculated using the mid-period population (1988) and the age groups 0-4, 5-14, 15-44, 45-64, 65-74, 75-84, and 85. Analysis was by Poisson regression in STATA 5 (StataCorp, 1997). Seasonality was tested by including both sine and cosine terms with an annual period in the regression. Each age group, except for those who died aged under 5, showed significant seasonality. For those aged 45–64 the significance was marginal (2=8.4, df=2, P=0.015), but for all others the P value was <0.01. The phase of the seasonality was different for each age group (1). Deaths in the younger age groups peaked in the summer and deaths in the older age groups peaked in the winter. Deaths of people aged 5–14 peaked in August, and in those aged 15–44 the peak was slightly earlier, in July, but the rate in August was also high. Deaths in people aged 45–64 showed an intermediate pattern, with excess mortality from November to March and a separate peak in August. In those aged 65–74 there was a winter excess lasting from November to April but no …