Reproducibility of a questionnaire on risk factors for osteoporosis in a multicentre prevalence survey: the European Vertebral Osteoporosis Study.

Background: The European Vertebral Osteoporosis Study Group (EVOS) developed a questionnaire, back translated into 14 different European languages, for use in a multinational epidemiological study of vertebral osteoporosis. We investigated the reproducibility of this questionnaire in four of the participating study centres.

Methods: In all 151 men and women, aged 50-85 years, from Lubeck (Germany), Malmo (Sweden), Warsaw (Poland) and Oviedo (Northern Spain), were retested with the questionnaire on two occasions using a different observer within a 28-day period.

Results: Questions relating to personal or medical history were more reproducible than questions concerning subjective symptoms or aspects of lifestyle. The level of agreement for the non-ordinal categorical variables, as estimated by kappa, varied from 0.38 to 1.00 across the four centres. Agreement for the multicategory ordinal, mainly lifestyle, questions was in general poorer though improved when a weighted analysis was performed. For continuous data the 95% limits of agreement were narrow, and there was no evidence of bias between interviewers. There were no important differences in reproducibility across the four centres for either categorical or continuous data.

Conclusion: The study indicates that the questionnaire may produce useful and comparable information concerning risk factors for osteoporosis across different countries and in different languages. It also highlights that questionnaire instruments designed for use in multinational population-based studies may provide data of comparable quality across a range of settings.