Osteoporosis is a major public health problem that is responsible for a considerable morbidity, mortality and health care costs. Evaluation of fracture risk is essential to select patients who will benefit most of interventional strategies. Most prospective cohort studies, with fracture outcomes, rely on participant self-report as the main or only source of information on fracture incidence. Systematic validation by screening of medical files is time-consuming, costly and almost impossible to accomplish in large scale multicenter epidemiological studies. We found in a well characterized prospective population-based cohort of 3560 postmenopausal, volunteer women, aged 6085 years, included in the Fracture Risk Brussels Epidemiological Enquiry (FRISBEE) cohort that the global rate of unconfirmed self-reported fractures (false positives) was 14.4% (BES 2018 and ECTS 2019, Abstracts). Over a median follow-up period of 8.9 years, the global percentages of validated fractures were 77.1% for the four classical major osteoporotic fractures (MOFs, ie. hip, vertebra, shoulder/upper arm, wrist), 75.4% for other major and 53.6% for minor fractures. The percentages of confirmed fractures varied by fracture site, with the hip having the highest proportion of confirmed fractures (90.3%, n=65), followed by fractures of the pelvic bone (88.5%, n=46), wrist and shoulder/proximal humerus (80.7%, n=129 and 80.8%, n=84, respectively). Participant self-report could also lead to bias in the classification of fracture status if a significant proportion of fractures are not reported. This other cause of misclassification has been little studied so far. We thus assessed the proportion of non-reported fractures (false negatives) in order to evaluate the possible impact of this phenomenon in epidemiologic cohort studies and in models of fracture risk prediction. Participants are followed by yearly phone calls for the occurrence of incident fragility fractures. We had access to the medical records of 67.9% of our study participants. After a thorough verification of these medical records, we found a total of 209 unreported fractures. The false negative rate for all fractures was 21.1%, including 21.7% for the 4 MOFs, 13.1% for other major fractures and 25.8% for minor fractures. The percentages of fractures showed to be false negatives varied by fracture site: 1.4% (n=1/73) at the hip, 46.3% at the spine (n=99/214), 5.3% at the shoulder/proximal humerus (n=5/94) and 7.1% at the wrist (n=11/154). We analyzed participants baseline characteristics that could have influenced the rate of false negatives fractures. In a multivariate analysis, older subjects (OR 0.6; 95% CI, 0.92.4; P=0.007) and subjects with a lower education level (OR 1.6; 95%CI, 1.12.3; P=0.008) were more likely to underreport a fracture event. Besides false positive reports, underreporting of a substantial proportion of fracture events is another major cause of misclassification of fracture events. Both types of inadequate reports will influence any model of fracture risk prediction and decrease statistical power when estimating the associations between candidate risk factors and incident fractures.
21 Oct 2019
Belgian Endocrine Society