We invited 1604 randomly selected women, all 75 years of age, to participate in a study on the risk factors for fracture. The women were divided into three groups consisting of 1044 (65%) who attended the complete study, 308 (19%) respondents to the study questionnaire only and 252 (16%) who did not respond. The occurrence of the life-time fracture was ascertained from radiological records in all groups and by questionnaires from the attendees and respondents.

According to the radiological records, fewer of the questionnaire respondents (88 of 308, 28.6%) and non-respondents (68 of 252, 27%) had sustained at least one fracture when compared with the attendees (435 of 1044, 41.7%; chi-squared test, p < 0.001). According to the questionnaire, fewer of the respondents (96 of 308, 31.1%) had sustained at least one previous fracture when compared with the attendees (457 of 1044, 43.7%; chi-squared test, p < 0.001).

Any study concerning the risk of fracture may attract those with experience of a fracture which explains the higher previous life-time incidence among the attendees. This factor may cause bias in epidemiological studies.