1 Park Nicollet Clinic and HealthPartners Institute, Minneapolis, MN, United States of America; Division of Health Policy and Management, University of Minnesota, Minneapolis, MN, United States of America. Electronic address: scho0600@umn.edu.
2 George & Fay Yee Centre for Healthcare Innovation, University of Manitoba, Winnipeg, Canada.
3 Department of Radiology, University of Manitoba, Winnipeg, Canada.
4 Nutrition Health Innovation Research Institute, Edith Cowan University, Perth, Australia; Centre for AI & ML, School of Science, Edith Cowan University, Perth, Australia; Computer Science and Software Engineering, University of Western Australia, Perth, Australia.
5 Centre for AI & ML, School of Science, Edith Cowan University, Perth, Australia.
6 Department of Computer Science, Concordia University, Montreal, Canada.
7 Nutrition Health Innovation Research Institute, Edith Cowan University, Perth, Australia; Medical School, University of Western Australia, Perth, Australia; Centre for Kidney Research, School of Public Health, The University of Sydney, Sydney, Australia.
8 Department of Radiology, University of Manitoba, Winnipeg, Canada; Department of Medicine, University of Manitoba, Winnipeg, Canada.

Background: It is unknown if prevalent vertebral fracture (PVFx) captured on bone density vertebral fracture assessment (VFA) images predicts incident CVD events.

Methods: 11,760 individuals (mean [SD] age 75.7 [6.8] years, 94 % female) had VFA contemporaneously with bone densitometry in Manitoba, Canada, between February 2010 and December 2017, of whom 1919 (16.3 %) had =1 PVFx. This cohort was followed over a mean (SD) 3.8 (2.3) years for Major Adverse Cardiovascular Events (MACE, composite of hospitalization for myocardial infarction, non-hemorrhagic stroke, or all-cause mortality) and other CVD events (hospitalization for coronary artery disease, congestive heart failure, peripheral vascular disease, or coronary bypass/stenting). Proportional hazards models were used to estimate hazard ratios (HR) for incident MACE and other CVD events in those with compared to those without PVFx.

Results: Adjusted for age and sex, those with PVFx had HR for incident MACE of 1.34 (95 % C·I. 1.19, 1.51), hospitalization for myocardial infarction (HR 1.35, 95 % C.I. 1.02, 1.79), all-cause mortality (HR 1.36, 95 % C.I, 1.19, 1.56), and other CVD events (HR 1.40, 95 % C.I. 1.21, 1.61). These associations were only slightly attenuated with further adjustment for prior CVD disease and additional CVD risk factors.

Conclusion: Prevalent vertebral fracture identified on VFA images in routine clinical practice is robustly associated with incident MACE, independent of other risk factors including AAC which can be simultaneously ascertained on the same images. VFA may have utility for prediction of fractures and CVD outcomes via ascertainment of both prevalent vertebral fracture and AAC.