Abstract

Background: Vertebral fractures in elderly patients contribute to huge economic and social burden and seriously low quality of life. This study aims to identify the associated risk factors for vertebral fractures in the non-institutionalized old American population.

Methods: A total of 31034 people from National Health and Nutrition Examination Survey (NHANES) during 2005-2010were included. Therein, 108 individuals with exact history of vertebral fractureswere enrolled froma series of 4268 people (aged between 65 and 85 years old). The related demographic and laboratory examination data were collected. Further, univariate and multivariate analysis were applied to identify the related risk factors of vertebral fractures in the elderly. A nomogram risk prediction model was constructed and internally validated through the boot strapping method.

Results: Univariate analysis indicated these factors, age (χ² = 5.75, p = 0.017), smoking (χ² = 4.46, p = 0.035), and combined with osteoporosis (χ² = 41.26, p < 0.001) were statistically different between the vertebral fracture group and no-vertebral fracture group. In multivariate analysis, the parameters of smoking (OR = 1.620, 95% CI: 1.080-2.440, p = 0.021), combined with osteoporosis (OR = 3.370, 95% CI: 2.220-5.110, p < 0.001), and the serum K (OR = 0.360, 95% CI: 0.230-0.560, p < 0.001) were screened as the independent risk factors for vertebral fractures in the elderly. The nomogram demonstrated a favorable level of discrimination with the area under the curve was 0.71. The calibration curves exhibited satisfactory agreement between the actual observation and nomogram prediction.

Conclusion: The nomogram (including osteoporosis, smoking, and serum K) was capable of predicting vertebral fracture and has certain auxiliary value in clinical applications.