Calcaneal quantitative ultrasound and phalangeal radiographic absorptiometry alone or in combination in a triage approach for assessment of osteoporosis: a study of older women with a high prevalence of falls

Abstract

Background: The objective of this study was to investigate if application of United Kingdom National Osteoporosis Society (UK-NOS) triage approach, using calcaneal quantitative ultrasound (QUS), phalangeal radiographic absorptiometry (RA), or both methods in combination, for identification of women with osteoporosis, would reduce the percentage of women who need further assessment with Dual Energy X-ray Absorptiometry (DXA) among older women with a high prevalence of falls.

Methods: We assessed 286 women with DXA of hip and spine (Hologic Discovery) of whom 221 were assessed with calcaneal QUS (Achilles Lunar), 245 were assessed with phalangeal RA (Aleris Metriscan), and 202 were assessed with all three methods. Receiver operator characteristics (ROC) curve for QUS, RA, and both methods in combination predicting osteoporosis defined by central DXA were performed. We identified cutoffs at different sensitivity and specificity values and applied the triage approach recommended by UK-NOS. The percentage of women who would not need further examination with DXA was calculated.

Results: Median age was 80 years (interquartile range [IQR]) [75-85], range 65-98. 66.8% reported at least one fall within the last 12 months. Prevalence of osteoporosis was 44.4%. Area under the ROC-curve (AUC) (95% confidence interval (CI)) was 0.808 (0.748-0.867) for QUS, 0.800 (0.738-0.863) for RA, and 0.848 (0.796-0.900) for RA and QUS in combination. At 90% certainty levels, UK-NOS triage approach would reduce the percentage of women who need further assessment with DXA by 60% for QUS, and 43% for RA. The false negative and false positive rates ranged from 4% to 5% for QUS and RA respectively. For the combined approach using 90% certainty level the proportion of DXAs saved was 22%, the false negative rate was 0% and false positive rate was 0.5%. Using 85% certainty level for the combined approach the proportion of DXAs saved increased to 41%, but false negative and false positive values remained low (0.5%, and 0.5% respectively).

Conclusions: In a two-step, triage approach calcaneal QUS and phalangeal RA perform well, reducing the number of women who would need assessment with central DXA. Combining RA and QUS reduces misclassifications whilst still reducing the need for DXAs.

Figure 1

Receiver operator characteristics (ROC) curves. ROC-curves for calcaneal QUS (BUA), phalangeal RA (RA T-score), and a combination of BUA and RA T-score (combined) for discrimination between osteoporotic and non-osteoporotic individuals. Hosmer and Lemeshow’s goodness-of-fit test: P = 0.61. Abbreviations: QUS = Quantitative Ultrasound, RA = Radiographic Absorptiometry, BUA = Broadband Ultrasound Attenuation.

Figure 2

The distribution of the results of calcaneal QUS (A) and phalangeal RA (B). Plots showing the distribution of the results of QUS of the calcaneus (A) and phalangeal RA (B). Horizontal lines represent the upper and lower triage thresholds at a 90% certainty level. A: 90% sensitivity threshold: BUA=105.88, 90% specificity threshold: BUA=86.63. B: 90% sensitivity threshold: T-score= -0.65, 90% specificity threshold T-score=-2.95. Abbreviations: QUS = Quantitative Ultrasound, RA = Radiographic Absorptiometry, BUA = Broadband Ultrasound Attenuation.