Search

With great interest we read the thought provoking contribution by Fleming et al on general practice consultation patterns for asthma.1 We would like to offer some alternative viewpoints to those posed by the authors. First of all, we wonder whether the choice of statistical techniques obscures the view of what happened to respiratory morbidity. Given the sudden rise in asthma episodes at the end of 1991 and the subsequent fall after 1994, a step function would be more appropriate. Besides, the amplitude within the years studied appears to increase during the period 1991–4. This can be modelled by adding an interaction term season*time to the model. Furthermore, given the nature of the data, Poisson regression would be preferable to linear regression techniques. Secondly, it is unclear whether the trend is specific for asthma and bronchitis or is relevant for all respiratory morbidity or, even broader, applies to all morbidity. In the discussion paragraph the authors point out that the broader category of respiratory infections shows the same trend, but they fail to explore the possible implications of this finding. How do their data compare with hospital data over the years? How about general practice consultation patterns? Thirdly, the distinction between new episodes and repeat consultations may be a difficult one, especially for chronic diseases like asthma. Subtle changes in registration routines may have affected the outcomes of this study. Apart from practice nurses, we wonder whether the introduction of asthma facilitators2 may be possible causes for the trends that are shown. The authors suggest that the observed trends are due to fluctuations in prevalence. There is no evidence for this. The rise in consultations in the years 1991–4 could be due to a temporary increase in complaints in roughly the same number of prevalent asthma patients. We invite …