Risk of selected eye diseases in people admitted to hospital for hypertension or diabetes mellitus: Record linkage studies

摘要

We read with great interest the article by Goldacre et al. The authors conducted a retrospective matched cohort study to measure the association between hyperten-sion or diabetes mellitus and various eye diseases (eg, cataracts and glaucoma). The authors report the rate ratios, the calculation of which is detailed in the manuscript, comparing the rate of cataracts, for example, among individuals with and without hypertension. However, when interpreting their results they use the word 'risk'. Use of this word has major implications on the interpretation of their results as the definition of risk is different from rate. Rate is the speed at which the disease develops in a population, ranging from zero to infinity. Risk is the instantaneous probability that an individual will develop the disease, ranging from zero to one.2 Rates can be used as an indicator of risk only when the follow-up is short and the rate of disease is relatively low (<10%) and constant over time. However, rate ratios based on outcomes that occur frequently overestimate the risk ratio when the the rate ratio is more than 1 and underestimate the risk ratio when the rate ratio is less than I. For example, in this study, the rate of cataract development, particularly among older adults, is not a rare condition; the rate will be markedly discrepant from the risk. In the future, the authors are strongly urged to use good epidemiological practice and interpret a rate ratio as a rate ratio, not as a risk ratio and vice versa.