Pathology
Partner With Us
Advertisement
From the Journals

Assessing Osteoporosis Risk in Younger Postmenopausal Women

Commonly used tools may fall short in identifying at-risk women, study finds.

Conexiant

A new cross-sectional study evaluated the accuracy of three commonly used osteoporosis risk assessment tools in identifying osteoporosis among postmenopausal women aged 50 to 64 years.

In their recent study published in JAMA Network Open, researchers analyzed data from 6,067 participants in the Women’s Health Initiative Bone Density Substudy across clinical centers in Arizona, Pennsylvania, and Alabama. Participants, who had no prior osteoporosis therapy, underwent bone mineral density (BMD) measurements at the femoral neck, total hip, and lumbar spine. Osteoporosis was defined as a T score of –2.5 or lower at any site. Among the study participants, 14.1% (n = 857) had osteoporosis at one or more of the anatomical sites.

The tools—Osteoporosis Risk Assessment Instrument (ORAI), Osteoporosis Index of Risk (OSIRIS), and Osteoporosis Self-Assessment Tool (OST)—demonstrated fair to moderate discrimination, while area under the receiver operating characteristic curve (AUC) values ranged from 0.633 to 0.663, using published score thresholds:

  • ORAI: AUC = 0.663, sensitivity = 53.3%, specificity = 79.4%
  • OSIRIS: AUC = 0.633, sensitivity = 37.8%, specificity = 88.8%
  • OST: AUC = 0.654, sensitivity = 62.4%, specificity = 68.5%

"All three tools were better able to predict a BMD T score of –2.5 or lower at the femoral neck," compared with the other two sites, the researchers noted.

When applied to a theoretical cohort of 1,000 women, ORAI identified 75 true cases but resulted in 180 unnecessary BMD tests. OST, while more sensitive, led to 266 extraneous tests. Comparatively, OSIRIS identified 51 cases and led to 94 extraneous tests.

“Screening is essential to reducing the burden of osteoporosis and fractures, yet this study reveals a shortcoming in identifying at-risk women using common clinical risk factors,” wrote Henry W. Zheng, BS, of UCLA’s Medical and Imaging Informatics department, with colleagues.

The study population consisted of relatively healthy women with a higher socioeconomic status, which may have limited generalizability. The researchers suggested refining existing tools or exploring alternative methods, such as machine learning models, to enhance osteoporosis risk assessment in this demographic.

No conflicts of interest were reported.