Validation of the Osteoporosis Self-Assessment Tool in US Male Veterans

Transcription

1 Journal of Clinical Densitometry: Assessment & Management of Musculoskeletal Health, vol. 17, no. 1, 32e37, 2014 Published by Elsevier Inc. on behalf of The International Society for Clinical Densitometry /17:32e37/$ Section I: Fracture Risk Assessment Validation of the Osteoporosis Self-Assessment Tool in US Male Veterans J. Steuart Richards,*,1,2 Antonio A. Lazzari, 3 Denise A. Teves Qualler, 4,5 Sameer Desale, 6 Robert Howard, 7 and Gail S. Kerr 1,2,8 1 Veterans Affairs Medical Center, Washington, DC, USA; 2 Department of Medicine, Georgetown University, Washington, DC, USA; 3 Boston Division VA Health Care System, Boston University Medical School, Boston, MA, USA; 4 Zablocki VA Medical Center in Milwaukee, Milwaukee, WI, USA; 5 Medical College of Wisconsin, Milwaukee, WI, USA; 6 Medstar Health Research Institute, Hyattsville, MD, USA; 7 Biloxi VA Health Care System, Biloxi, MS, USA; and 8 Department of Medicine, Howard University, Washington, DC, USA Abstract The osteoporosis self-assessment tool (OST) is a screening instrument that uses age and weight as parameters to predict the risk of osteoporosis. This study was designed to evaluate OST in predicting osteoporosis in males. Male veterans aged 50 yr and older with no prior diagnosis of osteoporosis and no prior bone densitometry (dual-energy X-ray absorptiometry [DXA]) testing were eligible for the study. Sociodemographic information, medical history, and risk factors for osteoporosis were recorded. Anthropometric measurements were taken and DXA testing performed. The OST index for each subject was calculated and predictive values and receiver operating characteristic (ROC) curves were evaluated for OST and osteoporosis. Five hundred eighteen subjects underwent DXA, 92 (17.8%) had osteoporosis, 281 (54.2%) had low bone mass, and 145 (28.0%) had normal bone mineral density. The OST index ranged from 8 to 23 with a mean of 4 (standard deviation 4.3). An OST index of 6 or lower predicted osteoporosis with a sensitivity of 82.6%, specificity of 33.6%, and an area under the curve for the ROC curve of OST index performed better in non-hispanic whites and males O65 yr. OST predicts osteoporosis with moderate sensitivity and poor specificity in men. Key Words: Dual-energy X-ray absorptiometry; men; osteoporosis self-assessment tool; screening. Introduction Osteoporosis is diagnosed frequently in males, but osteoporotic fracture-related mortality in males is twice that of females, emphasizing the importance for early diagnosis in men with increased fracture risk (1). Secondary etiologies are responsible for 47% of osteoporosis in men but may not correlate with densitometric results (2). The densitometric diagnosis of osteoporosis is based on calculations of fracture risk established in postmenopausal Caucasian women, Received 11/28/12; Accepted 02/05/13. *Address correspondence to: J. Steuart Richards, MBBS, Washington DC VA Medical Center, 50 Irving Street, NW, Washington, DC john.richards1@va.gov therefore making the diagnosis in younger women, non- Caucasians, and men less precise (3). The International Society of Clinical Densitometry has included recommendations for obtaining densitometry in men, where adherence is low even in the presence of recognized risk factors (4). Various tools are available to evaluate the risk of osteoporosis and initiate screening in groups. One such tool, the osteoporosis self-assessment tool (OST), is a simplified instrument that is derived by including only weight and age and was demonstrated to predict osteoporosis in postmenopausal Asian women at the greatest risk and where bone densitometry (dual-energy X-ray absorptiometry [DXA]) was not readily available (5). OST was also found to be predictive of osteoporosis in both Caucasian women and men, but in small or retrospective studies (6e8). We conducted this study to 32

2 Validation of OST 33 evaluate the predictive value of OST in a population of male veteran patients with a high prevalence of comorbidity. Methods and Materials Male patients older than 50 yr attending primary care clinics at 4 participating VA Medical Centers were invited to participate. Exclusion criteria included a prior diagnosis or treatment for osteoporosis, prior DXA, and metabolic bone diseases including osteomalacia, renal osteodystrophy, or osteogenesis imperfecta. Inability to undergo a DXA at 2 of 3 sites (lumbar spine, hip, or forearm) or excessive weight above the table limit for the instrument were additional exclusion criteria. All participating subjects gave informed signed consent, approved by the local institutional review board. Data collected included sociodemographics, medical history, and a questionnaire of risk factors for osteoporosis (Appendix). A physical examination was performed for anthropometric measurements (height and weight). The OST index was calculated for each subject using the formula: OST index 5 ðweight ½kgŠ Age ½yrŠÞ 0:2 The OST index is reported as an integer by rounding down its value to the nearest whole number (5). Subjects had bone mineral density (BMD) measured at the hip (femoral neck and total hip or greater trochanter), lumbar spine (anterior-posterior L1eL4 or L2eL4), or distal forearm. Trained technicians performed DXA on either the Hologic (Hologic Inc., Bedford, MA) or the Lunar (GE Healthcare, Madison, WI) scanner, specific to each participating center. To adjust for systematic differences in BMD by DXA, values were standardized to the Hologic BMD using published equations (9). Site-specific T-scores were calculated. For the femoral neck and total hip, we used maleand race-specific reference data from the National Health and Nutrition Examination Survey III (10). Osteoporosis was defined using the lowest site-specific T-score (3,11). The threshold for establishing the diagnosis of osteoporosis was based on the World Health Organization (WHO) definition and consisted of a BMD value 2.5 standard deviation (SD) below the young adult mean (or a T-score 2.5). Low bone mass or osteopenia was defined as a BMD value at any site between! 1.0 and O 2.5. Demographic data were reported as mean and SDs. The associations of demographic data and risk factors for osteoporosis with a diagnosis of osteoporosis (DXA T-score 2.5) were calculated by the nonparametric Wilcoxon test. Logistic regression analyses for osteoporosis were performed; variables with p-values!0.1 were included in the model. The sensitivity, specificity, and predictive values of the OST index for osteoporosis were calculated and receiver operating characteristic (ROC) curves created. The ability of OST to predict osteoporosis in subgroups based on ethnicity (Caucasians vs African Americans) and age (65 vs O65 yr) were analyzed. The 10-yr probability of a hip or other major fracture was calculated using the WHO Fracture Risk Assessment Tool (FRAX Ò ), without BMD (12). The FRAX Ò was developed to aid physicians with the decision to initiate antiosteoporosis therapy. The FRAX Ò was compared with the OST index. All analyses were performed using SAS version 9.1 (SAS Inc., Cary, NC). Significance was set at an alpha of Results Five hundred twenty men were enrolled in the study from the 4 participating Veterans Affairs Medical Centers. The mean age of the cohort was 66 yr (SD 10.2) with a mean weight of 90.9 kg (SD 17.7), height of cm (SD 7.6), and body mass index of 30.4 kg/m 2 (SD 7.6). The ethnic distribution of the cohort was 374 (72.2%) Caucasians, 130 (25.1%) African Americans, and the remaining 14 (2.7%) comprised Hispanics, Asians, and other ethnic groups. As is typical of cohorts of elderly male veterans, comorbid diseases were common (Table 1) Risk factors for osteoporotic fractures were common in this cohort; the most frequent were smoking (ever) 73.6%, alcohol (O4 ounces/d) 50.3%, a history of weight loss (O10% of body weight) 41.4%, history of fracture 39.9% (Table 1). Approximately 45% of the subjects were designated by the investigator to have other risk factors for osteoporosis, which included other inflammatory diseases. The differences in comorbid diseases and risk factors for osteoporosis between Caucasians and African Americans and between patients 65 yr and O65 yr are shown in Table 1. Five hundred eighteen subjects underwent DXA and 92 (17.8%) had results that satisfied the diagnosis of osteoporosis; 281 (54.2%) had low bone mass (osteopenia) and 145 (28.0%) had normal BMD. The OST index ranged from 8 to 23 with a mean value of 4 (SD 4.3). An OST index of 6 or lower predicted osteoporosis with a sensitivity of 82.6% and specificity of 33.6%. The predictive values for osteoporosis using varying OST indices are shown in Table 2. Factors associated with a densitometric diagnosis of osteoporosis included other risk factors for osteoporosis ( p! ), renal disease ( p! ), weight ( p! ), and OST index 6 (p! ); all 3 variables remained statistically relevant in multivariate analysis ( p , p , and p , respectively). If OST was examined as a continuous variable in the multivariate analysis, a lower OST index had a p value of! The area under the curve for the ROC curve was 0.67 (Fig. 1). The predictability of OST was evaluated in different subgroups (Table 2). An OST index of 5 operated with the best combination of sensitivity (75.4%) and specificity (41.4%) for Caucasian men. An OST index of 6 in African American men predicted osteoporosis with a sensitivity of 70.0% and a specificity of 36.4%. In subjects aged 65 yr, an OST cutoff of 7 predicted osteoporosis with a sensitivity of 76.2% and a specificity of 39.5%, whereas in subjects O65 yr, an OST index 2 operated best, with a sensitivity of 80% and a specificity of 52.8%). Five hundred twelve

3 34 Richards et al. Table 1 Characteristics of the Study Population Variables Total a Caucasians African Americans p Value Age 65 yr Age O65 yr p Value Number / /518 d 270/ /520 d Age (mean) ! Weight (mean) Height (mean) ! BMI (mean) Hypertension 361/ /372 99/ / /248! Diabetes mellitus 157/ /372 47/ /269 86/ Coronary artery disease 155/ /369 32/ / /246! Pulmonary disease 70/517 55/372 12/ /269 34/ Renal disease 60/517 30/372 26/ /269 39/ Ever smoking 380/ /372 97/ /268 71/ Alcohol 260/ /373 65/ / / History of weight loss O10% body 205/ /355 33/126! /262 94/ weight History of fracture 206/ /373 40/ /267 83/ Family history of hip fracture 85/516 75/373 8/ /267 28/ Ever use of glucocorticoids 52/517 38/373 12/ /269 31/ Androgen deprivation therapy (prostate cancer) 20/518 12/374 7/ /269 14/ Abbr: BMI, body mass index. a Unless otherwise indicated, number is average or total count of the data provided for the total number indicated. subjects had data available to calculate the 10-yr probability of fracture or FRAX Ò ; 70 (13.7%) had a 10-yr probability of a major osteoporotic fracture 20%, 164 (32.0%) had a 10-yr probability of a hip fracture 3%, and 174 (34.0%) had either of the 10-yr fracture risk probabilities. The FRAX Ò predicted BMD 2.5 with a sensitivity of 40.7% vs 82.6% for an OST index 6(p! ) and a specificity of 74.1% vs 33.6% for an OST index 6 (p! ). For an OST 6, the FRAX predicted BMD 2.5 vs BMD O 2.5 with a sensitivity of 40.7% vs 82.6% and a specificity of 74.1% vs 33.6%. Discussion Bone densitometric studies, specifically DXA, for the diagnosis of osteoporosis are underused in men and not always available in various populations (4). Therefore, tools have been developed to optimize the yield of DXA in both men and women, particularly where access is limited. OST is one such instrument derived from a study of 860 postmenopausal Asian women (5). Multiple regression analyses of potential risk factors with item reduction were performed to determine which combination of factors identified women with osteoporosis. The final model included only age and weight and had a sensitivity of 91% and a specificity of 45% for osteoporosis. OST has since been validated in community-dwelling middle-aged and elderly Caucasian women (6,7). Our study was designed to evaluate the utility of OST in predicting osteoporosis (BMD 2.5) in male veterans attending primary care clinics. An OST index of 6 operated with optimal sensitivity (82.6%) but had low specificity (33.6%), significantly lower than that reported in a study of 181 men at the Richmond VA facility (8). In that study, patients were recruited from pulmonary and rheumatology outpatient clinics and included patients at high risk for osteoporosis. An OST index of 3 was used and found excellent sensitivity (93%) and good specificity (66%). The lower OST index cutoff may be explained by these study patients having greater risk for osteoporosis; 24% were on glucocorticoids and being exclusively from pulmonary and rheumatology outpatient clinics potentially with greater inflammatory disease. In contrast, our cohort consisted of mostly noninflammatory diseases and may have contributed to the weaker predictive value of OST. The optimal OST cutoff to predict an osteoporotic BMD result may vary in different ethnic groups. Although there was similar specificity for OST in Caucasians and African Americans (32.2% vs 36.4%) in our cohort, an OST index of 6 had greater sensitivity in Caucasians compared with African Americans (85.5% vs 70%). A similar trend, although not statistically significant, was also reported in the Richmond VA cohort: OST 3 had greater sensitivity for Caucasian men compared with African American men (95% vs 83%) but less specificity in Caucasians (60% vs 80%) (8). In contrast, using an OST index cutoff of 4 to predict

4 Validation of OST 35 Table 2 Sensitivity and Specificity for Various OST Values in Predicting Sensitivity and Specificity for Osteoporosis (T-score 2.5) OST index Sensitivity (%) Specificity (%) AUC Total population (n 5 520) Caucasians (n 5 373) African Americans (n 5 130) Age 65 yr (n 5 270) Age O 65 yr (n 5 250) Abbr: OST, osteoporosis self-assessment tool; AUC, area under the curve. femoral neck BMD of 2.5, superior sensitivity and specificity in African American men (93%, 76%, respectively) rather than Caucasian men (85%, 51%, respectively) was found in 639 older males in the New Jersey VA System (13). These findings support the need for different OST indices in different ethnic groups as was used for Caucasian and Asian men in another study (14). Age may also have an impact on the validity of OST indices. A multisite retrospective review of OST in men with rheumatoid arthritis performed in the Veterans Affairs Administration found OST to be more predictive for men Fig. 1. Receiver operating characteristic curve for OST index in comparison with dual-energy X-ray absorptiometry 2.5 for predicting osteoporosis. O60 yr (15). A retrospective study of the OST index in 2 large cohorts of men, the Rotterdam study of 2445 men older than 55 yr and the Baltimore Men s Osteoporosis Study of 503 men older than 65 yr, suggested utility of OST (16). An OST of 2 had sensitivities for femoral neck BMD 2.5 of 79% and 88% for the Rotterdam and Baltimore studies, respectively, and a specificity of 51% and 32%. OST indices varying from 2 to 4 are reported to be optimal for predicting osteoporosis in men; we found the optimal integer to be slightly higher at 6. Similar to our results, the study from Richmond reported greater specificity among younger subjects (8). A major strength of our study is the large number of prospectively enrolled subjects from 4 separate regions of the countrydsouth, Mid-Atlantic, Northeast, and the Midwest. We believe that the regional and ethnic diversity, with 28% of the cohort comprising ethnic minorities, broadens the applicability of our results. However, although ethnic diversity could have resulted in the less robust correlations that we found for OST in predicting osteoporosis, our cohort is more reflective of the community and general practice. But, similar to prior reports, we found the correlation between osteoporosis risk factors and BMD to be poor, where only the presence of renal disease, lower weight, and subjects who were assessed by the investigator to have other osteoporosis risk factors were more likely to have osteoporosis. Established risks for osteoporotic fracture including cigarette smoking, oral prednisone, alcohol, fractures, or family history of fractures were not associated with osteoporosis in our study. Practical use of OST in the clinical setting would allow the physician to determine, based on the score, if a patient required a DXA. Other investigators have stratified the OST index into risk groups (low, medium, and high risk) (8,13), implying further interpretation of the index is required and further limiting its applicability. The WHO Fracture Risk Assessment Tool (FRAX Ò ) was developed to give a 10-yr absolute fracture risk and so aid

5 36 Richards et al. in the decision process for the initiation of therapy for osteoporosis (12). The FRAX Ò assessment can be calculated with or without the BMD result, broadening its applicability to those regions of the world where DXA machines are scarce. The FRAX Ò thus differs from OST in that it does not help the clinician in deciding if the patient requires referral for a DXA. Although more specific compared with OST in our population, the FRAX Ò without BMD had sensitivity far lower than the OST index (40.7% vs 82.6%), limiting its utility as a screening tool for osteoporosis in this population. Our study had a number of limitations. Although the multisite nature of the study permits broader applicability of the results, it also highlights the differences in measurements of BMD and reporting of BMDs by different centers. DXA machines from different manufacturers were used to obtain BMD measurements (GE, Lunar and Hologic, Inc.), and the results from the 4 sites were not standardized by a common phantom. Hence, the reports for hip BMD differed among sites. Some sites reported total proximal hip, others sites reported femoral neck, greater trochanter, or the combination of those measurements. The concordance between these regions of interest in males has not been adequately studied. We converted BMDs from multiple sites to standardized values to reduce any possible bias related to the manufacturer of the DXA machine, a procedure well-described and used in prior studies (15,17). In conclusion, our study found that OST index correlates highly with the diagnosis of osteoporosis in men, but in our cohort, it was demonstrated to be only moderately predictive with good sensitivity and fair specificity. OST may have greater predictive value if different OST cutoff indices are validated for different ethnic and age groups. Further studies with a larger population and including more Asian subjects need to be performed. Acknowledgments The authors thank Phil Dussault, Sam Davis, and Deborah Cales for their help with the enrollment of patients for this study, and Jeffrey Huang for assisting with the organization of data. This project was supported by FosamxR Investigator Initiated Studies Program #31236 from Merck & Co, Inc. Statistical analysis for this project was funded in part with Federal funds (Grant # UL1RR031975) from the National Center for Research Resources (NCRR) and the National Center for Advancing Translational Sciences (NCATS), National Institutes of Health (NIH), through the Clinical and Translational Science Awards Program (CTSA), a trademark of DHHS, part of the Roadmap Initiative, Re-Engineering the Clinical Research Enterprise. References 1. Forsen L, Sogaard AJ, Meyer HE, et al Survival after hip fracture: short- and long-term excess mortality according to age and gender. Osteoporos Int 10:73e Pye SR, Adams KR, Halsey JP, et al Frequency and causes of osteoporosis in men. Rheumatology (Oxford) 42:811e Binkley NC, Schmeer P, Wasnich RD, Lenchik L What are the criteria by which a densitometric diagnosis of osteoporosis can be made in males and non-caucasians? J Clin Densitom 5(Suppl):S39eS Solomon DH, Katz JN, Jacobs JP, et al Management of glucocorticoid-induced osteoporosis in patients with rheumatoid arthritis: rates and predictors of care in an academic rheumatology practice. Arthritis Rheum 46:3136e Koh LK, Sedrine WB, Torralba TP, et al A simple tool to identify Asian women at increased risk of osteoporosis. Osteoporos Int 12:699e Gourlay ML, Miller WC, Richy F, et al Performance of osteoporosis risk assessment tools in postmenopausal women aged years. Osteoporos Int 16:921e Richy F, Gourlay M, Ross PD, et al Validation and comparative evaluation of the osteoporosis self-assessment tool (OST) in a Caucasian population from Belgium. QJM 97: 39e Adler RA, Tran MT, Petkov VI Performance of the Osteoporosis Self-assessment Screening Tool for osteoporosis in American men. Mayo Clin Proc 78:723e Genant HK, Grampp S, Gl}uer CC, et al Universal standardization for dual x-ray absorptiometry: patient and phantom cross-calibration results. J Bone Miner Res 9:1503e Looker AC, Wahner HW, Dunn WL, et al Updated data on proximal femur bone mineral levels of US adults. Osteoporos Int 8:468e Hamdy RC, Petak SM, Lenchick L Which central dualenergy x-ray absorptiometry skeletal sites and regions of interest should be used to determine the diagnosis of osteoporosis. J Clin Densitom 5(Suppl):S11eS Kanis JA, Johnell O, Oden A, et al FRAX and the assessment of fracture probability in men and women from the UK. Osteoporos Int 19:385e Zimering MB, Shinn JJ, Shah J, et al Validation of a novel risk estimation tool for predicting low bone density in Caucasian and African American men veterans. J Clin Densitom 10:289e Lynn HS, Woo J, Leung PC, et al An evaluation of osteoporosis screening tools for the osteoporotic fractures in men (MrOS) study. Osteoporos Int 19:1087e Richards JS, Peng J, Amdur RL, et al Dual x-ray absorptiometry and evaluation of the osteoporosis self assessment tool in men with rheumatoid arthritis. J Clin Densitom 12:434e Hochberg MC, Tracy JK, van der Klift M, Pols H Validation of a risk index to identify men with an increased likelihood of osteoporosis. Proc 24th Meeting of the American Society for Bone and Mineral Research, San Antonio, TX, p Mikuls TR, Saag KG, Curtis J, et al Prevalence of osteoporosis and osteopenia among African Americans with early rheumatoid arthritis: the impact of ethnic-specific normative data. J Natl Med Assoc 97:1155e1160.

6 Validation of OST 37