Objectives To analyse if using risk indices and PIXI scans jointly would reduce requirement for central DEXA to diagnose postmenopausal OP.
Methods A stratified sample of postmenopausal women was selected: a random sample of 305 postmenopausal women from Primary care, and 200 consecutive postmenopausal women referred for central DXA measurement from tertiary care. Inclusion criteria: Caucasian female, age ≥50 yrs, and full menopause (amenorrhea ≥12 mo.). Exclusion criteria: previous diagnose of OP, previous treatment with OP drugs (calcium and/or vitamin D and/or estrogens for menopausal symptoms treatment were allowed) or steroids and other drugs related with low BMD, and institutionalized persons OR Steinbrocker’s functional grade 4. Informed consent was obtained. Four risk indices were calculated: SCORE, ORAI, OSIRIS, and OST. All participants underwent two different BMD measurements: a non-dominant heel BMD (PIXI Lunar, Software #50699, GE Corp.), and a central DXA of the hip and lumbar spine (Lunar Prodigy Advance, Software ENCORE 2006, PA+300274, GE Corporation). OP definition according to the WHO was used. Statistical analyses: The diagnostic utility was measured by ROC curves. We calculated the sensitivity and specificity for risk indices, PIXI, and all possible combinations. Logistic regression was performed to build a risk model with the presence or absence of osteoporosis at the central DEXA as dependent variable. The thresholds was established in 2 ways: 1) a cutoff point where the sensitivity and specificity are maximized, and 2) with two cutoff points, where both it sensitivity and specificity reached 90%.
Results 505 Caucasian women with a mean (SD) 61 (8) yrs, were recruited. Median (p25-p75) scores for each risk index were: OST 1 (0-3), ORAI 10 (7-14), and SCORE 8 (6-11). The mean (SD) PIXI T-score of the calcaneus was -0.33 (1.14). The mean (SD) femoral neck T-score was -1.01 (1.05), total femur was -0.59 (1.19) and lumbar T-score was -1.18 (1.36). The prevalence of osteoporosis by central DXA was 20% (n=102), 19% (57) in primary care and 23% (45) in tertiary care. The combined algorithm PIXI + OST + SCORE was the greatest area under the curve obtained: 75% (95% CI, 71 to 79). Most favorable threshold for this algorithm stratified subjects into high, medium, and low risk according to 2 cutoff (-20 and -5, respectively), and the most favorable subsequent decision was referral for central DXA if the medium or high risk categories were reached. According to this algorithm (table) 11 (2.2%) false negatives were obtained, but 257 (52%) central DEXA were avoided and a 20 – 35% in costs could have been saved.
Conclusions A triage based on a combined algorithm composed of PIXI + OST + SCORE reduces reduce the requirements of central DEXA a 52%, and may savings cost.
Disclosure of Interest None Declared