A predictive nomogram for selective screening of asymptomatic vertebral fractures: The Vietnam Osteoporosis Study

Abstract

Objectives

Vertebral fractures are associated with disability and mortality, but most vertebral fractures are asymptomatic. The present study aimed to determine the incidence of and develop a predictive nomogram for asymptomatic vertebral fractures in Vietnamese adults.

Methods

This cohort study as a part of the Vietnam Osteoporosis Study involved 168 men and 287 women aged 50 years and older without a clinically diagnosed vertebral fracture. Their spine x-rays were taken at the recruitment and subsequent 2-year visit. Vertebral fractures were ascertained using the Genant's semi-quantitative method. We employed the Bayesian Model Averaging method to search for the optimal model for predicting asymptomatic vertebral fractures. A predictive nomogram was also developed to facilitate risk prediction.

Results

During a median of 2.38 years of follow-up, 13 men and 16 women developed an asymptomatic vertebral fracture, yielding the overall incidence rate of 28 fractures per 1000 person-years, or 33 fractures/1000 person-years in men and 24 fractures/1000 person-years in women, respectively. Most asymptomatic vertebral fractures were moderate, almost 1.5 times more common than mild fractures. The optimal model for predicting incident asymptomatic vertebral fractures included age, male sex and lower femoral neck T-score. The area under the receiver's operating characteristic curve was 0.91, with 95% CI ranging from 0.86 to 0.96.

Conclusions

Asymptomatic vertebral fractures were relatively common among adults in Vietnam. A simple model with sex, age and femoral neck T-score is helpful for selective screening of asymptomatic vertebral fractures in Vietnamese individuals.