Geographical variation of Legionnaires' disease: a critique and guide to future research.

This review considers the value of the observation that Legionnaires' disease varies geographically. Estimates of disease incidence, derived from case registers and from studies measuring the proportion of cases of pneumonia which are Legionnaires' disease, and of the prevalence of the population with antibody, show that there is geographical variation in disease frequency. Much, but not all, of this variation is artefact due to differences in definitions, diagnostic methods, surveillance systems and data presentation. Some of the variation is attributed to publication bias, e.g. in 10 small studies (< 100 patients) 13.2% of pneumonia patients had Legionnaires' disease but in five large studies (> or = 500 patients) the figure was 3.6%. Research to explain variations has been neglected but a few studies have provided important insights into disease transmission. Future studies should: be based on agreed disease definitions and data collection and analysis methods; analyse subgroups separately; and collect data to develop explanations for geographical variation.