Session Information
Session Type: Abstract Submissions (ACR)
Background/Purpose: Fracture Risk Assessment Tool (FRAX) is a useful clinical tool for assessment of fracture risk and a major development in evaluation of osteoporosis. Its external validity is not established. This study was done to assess how well FRAX correlates with BMD in postmenopausal females and males of White and African-American ethnicity and calculate its test characteristics in a cohort of patients undergoing BMD.
Methods: Retrospective chart review of 1026 postmenopausal women and 245 men who underwent DEXA scan performed at UMC from 2005-2009. Data regarding demographics, BMD, FRAX and other relevant variables were abstracted, tabulated and analyzed using Statistical software STATA . WHO classification of bone density as well as standard FRAX stratification was used. Scatter plot and correlation coefficient was used to establish the association, concordance of FRAX-defined outcome and BMD-defined outcome using sensitivity, specificity, and the kappa statistic were calculated. BMD-defined osteoporosis status was treated as the gold standard.
Results: In males, FRAX demonstrates a low agreement with the BMD with Kappa=0.496, Low Sensitivity and Negative Predictive Value (NPV). Agreement was low/poor (K=0.167) in African-Americans (AA). In postmenopausal females, FRAX demonstrates a moderate agreement with the BMD with Kappa=0.647, adequate Sensitivity and NPV. It had good agreement (K= 0.847) in whites and low/poor agreement (K=0.313) in African-Americans.
It had adequate/good test characteristics in whites but poor sensitivity and NPV in AA males and females. Table 1 and 2 describe these results in detail. These highlight the lack of correlation with a gold standard bone mass measurement and raise doubts on the validity of using FRAX in white males as well as in AA males and PM females.
Conclusion: Use of FRAX to analyze fracture risk in US males and AA postmenopausal females may not lead to valid results. FRAX analysis did not have good agreement with BMD in these populations in this analysis. The lack of agreement was significantly worse in African-Americans as compared to Whites. These results suggest the need to do further population studies in larger cohorts of these subset populations to assess its external validity.
Table 1: Characteristics of the Study Population.
|
|
Overall |
Whites |
African-Americans |
|
MALES (N= 112) |
|
|
|
|
Sensitivity |
54.55 |
78.13 |
21.74 |
|
Specificity |
94.74 |
93.18 |
100 |
|
Positive Predictive Value (PPV) |
90.91 |
89.29 |
100 |
|
Negative Predictive Value (NPV) |
68.35 |
85.42 |
41.94 |
|
Agreement (Kappa) |
0.496 (CI=0.335-0.658) |
0.725 (CI=0.567-0.884) |
0.167 (CI=0-0.439) |
|
PM-FEMALES (N= 469) |
|
|
|
|
Sensitivity |
67.75 |
89.61 |
30.76 |
|
Specificity |
97.37 |
95.40 |
100 |
|
Positive Predictive Value (PPV) |
96.50 |
95.80 |
100 |
|
Negative Predictive Value (NPV) |
73.84 |
88.80 |
41.94 |
|
Agreement (Kappa) |
0.647 (CI=0.576-0.713) |
0.8467 (CI=0.785-0.908) |
0.313 (CI=0.175-0.451) |
Table 2: Agreement between FRAX and BMD/ Test Characteristics.
Disclosure:
V. Majithia,
None;
K. Aujla,
None.
« Back to 2013 ACR/ARHP Annual Meeting
ACR Meeting Abstracts - https://acrabstracts.org/abstract/analysis-of-fracture-risk-assessment-score-frax-correlation-with-bone-mineral-density-bmd-in-males-and-post-menopausal-females/