Postmenopausal Bone Density Referral Decision Rules: Correlation with Clinical Fractures

Military Medicine, Dec 2004 by Wei, Gina S, Jackson, Jeffrey L

Objective: To test decision rules for bone mineral density (BMD) against fractures. Methods: We surveyed postmenopausal women in a military primary care clinic and tested three national clinical decision rules (Osteoporosis Risk Assessment Instrument; age, body size, no estrogen; weight) for correlation with fracture history. Outcome measures included relative risk (RR), area under the receiver operating characteristics curve (aROC), sensitivity, and specificity. Results: Patients were 69 years old on average, 53% were Caucasian, 38% were African American, and 15% had a history of fractures. Caucasian women (RR, 1.8; 95% confidence interval [CI], 1.1-3.1) and those older than 65 years (RR, 2.0; 95% CI, 1.2-3.5) had higher prevalence of fractures. The Osteoporosis Risk Assessment Instrument decision rule had the highest aROC (0.65; 95% CI, 0.57-0.73) and sensitivity (sensitivity, 0.83; specificity, 0.31). Age, Body Size, No Estrogen had the next highest aROC (0.63; 95% CI, 0.54-0.71) and sensitivity (sensitivity, 0.74; specificity, 0.46). Weight criterion was the most specific (aROC, 0.60; 95% CI, 0.52-0.68; sensitivity, 0.64; specificity, 0.56). Conclusions: Current postmenopausal bone density referral decision rules only modestly correlate with clinical fractures.

Introduction

Osteoporosis and its related fractures are major public health threats to postmenopausal women. In the United States, an estimated 250,000 hip fractures and 500,000 vertebral fractures occur annually, ' with direct medical expenditures approximating $10 to $15 billion per year.2 Among admissions to military hospitals, hip fracture was one of the top 10 orthopedic reasons for admission.3 This is a source for considerable expenditure of military health commands. Fractures also lead to significant morbidity and mortality. The 1-year mortality after hip fractures is 20% among those less than 70 years old, 30% for ages 70 to 79 years old, and 40% in those more than 80 years old;4-6 even among the survivors, only 50% regain their independent ability to walk and only 13% regain their ability to climb stairs.7

Presently, universal screening of all postmenopausal women for low bone mineral density (BMD) is not recommended, perhaps because of a lack of convincing cost-effectiveness data and the paucity of studies demonstrating improved outcomes with universal screening. Although several risk factors for osteoporosis have been identified,2 it remains debatable which postmenopausal women merit screening. In recent years, clinical decision rules have been developed to help clinicians reach evidence-based decisions. A clinical decision rule quantifies the individual components of a patient's history, physical examination, and/or basic laboratory results toward the diagnosis, prognosis, or likely response to treatment in that patient and is formally tested in a population of interest.8 To date, several clinical decision rules have been created to guide clinicians in selecting women for BMD testing. A number of these rules base their selection criteria on very simple and easily assessed clinical risk factors; most use a combination of weight, age, and estrogen exposure status to predict women at risk for low BMD (Table I). Specific examples of such rules include the Osteoporosis Risk Assessment Instrument (ORAI),9 age, body size, no estrogen (ABONE),10 and body weight

Methods

This cross-sectional study was conducted at the general internal medicine clinic of the Walter Reed Army Medical Center in Washington, DC. As a large U.S. tertiary medical center, the Center provides comprehensive medical service to active duty military members, retirees, and their civilian dependents. Such patients have been found to be similar to those seen in civilian clinical practices across the country.12,13 The study protocol was approved by the center's Institutional Review Board. Self-administered surveys were consecutively distributed from May through November 2001 to women at least 40 years old as they presented to the primary care clinic for routine medical care. Questionnaires were excluded from final analysis if respondents' answers indicated that they did not meet the criteria for postmenopausal status, defined as not having menstruated for at least 12 months. The survey assessed respondent demographics, risk factors for osteoporosis, and personal history of fractures (i.e., wrist, hip, and spine). No unique identifiers were obtained.

Three published clinical decision rules were assessed in this study: ORAI, ABONE, and body weight

Each decision rule was applied to the study cohort using individual responses from the survey questionnaire. Main outcome measures included relative risk (RR), area under the receiver operating characteristics (ROC) curve (aROC), sensitivity, and specificity of each of the three decision rules for identifying respondents with a personal history of fractures (wrist, hip, or spine). In our study, the aROC of a clinical decision rule measures the overall ability of that rule to identify women who do, versus those who do not, have a personal history of fractures. In general, a larger aROC (i.e., approaching a value of 1.0) indicates that a decision rule has better overall test characteristics.