Abstract
Background/Aims Screening DXA is recommended for women aged 65+ to identify individuals likely to benefit from treatments to prevent osteoporotic fracture. This strategy is most cost-saving in women 75+, but DXA use remains low among older women. We determined whether estimates of absolute fracture or fall risk impact DXA screening.
Methods Using administrative data, we performed a cross-sectional analysis of women aged 65+ with >=2 years of enrollment in Kaiser Permanente Colorado during 2005–2009. DXA was determined by procedure code or scan documentation prior to the end of observation. To isolate screening, we excluded persons with diagnoses of osteoporosis or fragility fractures prior to DXA or end of enrollment. Covariates were obtained from the most recent data in the year prior to DXA or, if no DXA, the year prior to enrollment end. Absolute 10-yr hip fracture risk was calculated via FRAX™, fall risk was estimated using a modified Falls Risk Assessment Tool (FRAT), and comorbidity was assessed via Quan score. Relative risks were estimated from Poisson regression.
Results Screening DXA prevalence was 43% (10,061/23,479). Unscreened women were older (77.0±7.3 yrs vs 72.3±5.7; mean±SD) and had higher comorbidity (4.2±3.4 vs 3.8±2.9), FRAX™ hip fracture risk (6.9±6.9 vs 4.2±4.5), and fall risk (1.0±0.9 vs 0.6±0.81) [all p<0.05]. Multivariate modeling demonstrated lower DXA likelihood for: each added year of age (0.945; 0.042–0.948); each FRAT point (0.765; 0.747–0.783); Blacks (RR: 0.821; 95%CI: 0.756–0.892) and Hispanics (0.892; 0.837–0.950) vs Caucasians; and BMI<=18.5 kg/m2 vs all other categories (0.763; 0.756–0.892). FRAX™ was not associated with DXA in adjusted analyses and each Quan point increased DXA likelihood (1.030; 1.025–1.036).
Discussion The only variable associated with screening DXA that is also known to predict hip fracture risk was Caucasian race. Older age, being underweight and increased fall risk lowered the chance of DXA, but all these factors are known to increase hip fracture risk. Multivariate modeling found greater comorbidity associated with screening, which may represent increased health care utilization and more preventive care opportunities. Further studies should determine whether patient preferences, provider practice patterns, or both underlie the neglect of screening DXA in women at high fracture risk.
