Citation

Hundrup YA, Høidrup S, Obel EB, Rasmussen NK. Scand. J. Public Health 2004; 32(2): 136-143.

Affiliation

National Institute of Public Health, Copenhagen, Denmark. yah@niph.dk

Copyright

(Copyright © 2004, Associations of Public Health in the Nordic Countries Regions, Publisher SAGE Publishing)

DOI

unavailable

PMID

15255503

Abstract

BACKGROUND: The authors compared self-reported non-spine fractures obtained from a cohort of Danish female nurses with fracture diagnoses registered in the Danish National Hospital Register (DNHR). METHOD: The self-reported fracture history was obtained from a questionnaire and was related to fracture information registered with the DNHR by means of the unique person identification code of Danish citizens. A total of 166 self-reported hip fractures, 391 self-reported wrist fractures, and 121 self-reported upper arm fractures were available for the comparison. The self-reported fractures were initially compared with the anatomic specific fracture diagnoses registered in the DNHR. Second, the comparison also included fracture diagnoses of adjacent skeletal sites (unspecific fracture diagnoses). RESULTS: The positive predictive value of a positive report of hip fracture was 89%. Inclusion of unspecific registered hip fractures increased the positive predictive value to 94%. The same figures for wrist fractures were 75% and 84%, respectively, and for upper arm fractures 54% and 83%, respectively. The predictive value of a negative report of hip fracture was 99.5%. The fracture year was correctly reported in 76% of the hip fracture cases, 81% of the wrist fracture cases, and 82% of the upper arm fracture cases. Predictors of false-positive report of fractures were young age ( < 60 years), report of indoor falls in the previous year, and use of hormone replacement therapy (HRT). CONCLUSION: The authors conclude that self-report of hip, wrist, or upper arm fractures among Danish nurses is relatively accurate but varies by the site of fracture. False positive reports of fracture introduce only modest bias fracture risk estimates and tend to dilute the association between exposures and fracture.

Language: en