Classification criteria: methodological considerations and results from a 12 year following study in the general population.

In a 12 year followup study on knee osteoarthritis (OA) in the general population, several methodological aspects of the development and choice of classification criteria were studied. It cannot be assumed that the American College of Rheumatology (ACR) criteria sets developed in a clinical setting with comparable sensitivity and specificity will give comparable results in the general population. Moreover, validity needs to be studied in the general population to show whether there is consistency of research findings in the study of the prevalence and the relationship with putative risk factors. We found that the clinical criteria sets, with or without laboratory criteria in a traditional format, do not give results consistent with the other sets of ACR criteria. Kellgren and Lawrence's criteria were shown to predict the presence of future OA according to the other sets of ACR criteria.