Clinical risk factors for fracture in postmenopausal Canadian women: A population-based prevalence study

Bone 40 (2007) 991 – 996 www.elsevier.com/locate/bone Clinical risk factors for fracture in postmenopausal Canadian women: A population-based prevalence study William D. Leslie a,⁎, William A. Anderson b , Colleen J. Metge a , Lori-Jean Manness c for the Maximizing Osteoporosis Management in Manitoba (MOMM) Steering Committee a University of Manitoba, Winnipeg, Canada b Manitoba Health, Winnipeg, Canada c Merck Frosst Canada, Winnipeg, Canada Received 1 May 2006; revised 24 October 2006; accepted 10 November 2006 Available online 19 December 2006 Abstract Clinical risk factor assessment can be used to enhance fracture risk estimation based upon bone densitometry alone. Population- and agespecific risk factor prevalence data are required for the construction of these risk models. Our objective was to derive population-based prevalence estimates of specific clinical risk factors for postmenopausal women resident in the Province of Manitoba, Canada. A random sample of 40,300 women age 50 or older identified from the provincial health plan was mailed a validated self-report risk factor survey. The response rate was 8747 (21.7%) with a final study population of 8027 women after exclusions. The individual prevalence for each clinical risk factor ranged from 5.8% for hyperthyroidism to 33.0% for a fall in the preceding 12 months. Most point prevalence estimates were similar to other large cohort studies, though the prevalences of inactivity and poor mobility were higher than expected while height at age 25 and the prevalence of any fracture after age 50 were lower than expected. Most of the respondents (86.9%) had at least one non-age clinical risk factor, 60.6% had two or more, and 33.5% had three or more. Age affected risk factor prevalence, and older age was associated with a higher rate of multiple risk factors. The availability of age-specific risk factor prevalence rates in this population may allow for more accurate fracture risk modeling. © 2006 Elsevier Inc. All rights reserved. Keywords: Osteoporosis; Fracture; Population-based; Epidemiology Introduction Fragility fractures are the hallmark of osteoporosis, a common condition characterized by compromised bone strength [1]. Hip fractures have the greatest short-term effect on morbidity and mortality, [2] but spine fractures are also associated with significant excess mortality [3,4]. The health policy implications are considerable, with the average cost of caring for a hip fracture patient in Canada $26,527 in the first year alone [5] and annual economic implications of hip fractures expected to quadruple by 2041 [5]. Accurate methods for assessing fracture risk should enhance the identification of those most likely to benefit ⁎ Corresponding author. Fax: +1 204 237 2007. E-mail address: bleslie@sbgh.mb.ca (W.D. Leslie). 8756-3282/$ - see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.bone.2006.11.008 from intervention. Bone density measurement has been widely used in the diagnosis of osteoporosis [6], but absolute fracture risk is increasingly viewed as the preferred method for expressing fracture risk and intervention thresholds [7]. A large volume of literature shows that clinical risk factors can be used for fracture prediction [7–10] and improve on fracture risk estimates based upon bone density measurement alone [11]. Mathematical methods have been proposed which combine the independent contributions of age, bone density and selected clinical risk factors in the fracture risk estimation [12]. The mathematical models for predicting fracture risk require estimates for the prevalence of the clinical risk factors and these may in turn vary geographically or across the age spectrum [13]. The need for assessment of risk factor prevalence in varied populations is underscored by the enormous global variation in fracture incidence [14,15]. The objective of this study was to cha- 992 W.D. Leslie et al. / Bone 40 (2007) 991–996 racterize the prevalence of selected clinical risk factors in postmenopausal women in the Province of Manitoba, Canada. Materials and methods Population Manitoba Health maintains computerized databases of the physician services and hospitalizations provided for all persons registered with the system [16,17]. Manitoba Health generated a random sample of registered women who were 50 years of age or older for the calendar year 2000. These individuals were then mailed an initial letter explaining the study and seeking their consent to participate. In order to preserve confidentiality, the study investigators were not involved in the initial sampling or mailing. The study protocol was approved by the University of Manitoba's Health Research Ethics Board and the Province of Manitoba's Health Information Privacy Committee. The population sample was stratified by area of residence as determined by postal code (urban versus non-urban). For rural residents, 1600 women 50 years of age and older were randomly selected from each of Manitoba's 11 rural regional health authorities (RHAs) with the exception of one isolated northern RHA (Churchill) where all women age 50 and over were selected (N = 85). For urban residents, 24,000 women aged 50 and over and who had a postal code designation within the city of Winnipeg were randomly selected. The combined urban and rural sample represented more than 40,000 women from all regions of the province. Women who were subsequently determined to be premenopausal or younger than 50 years of age were excluded from the study. Survey questionnaire Between April 2000 and September 2000 a brief questionnaire was mailed to each of the women in the study population. This included a short inventory of clinical risk factors for osteoporotic fracture adapted by the Manitoba Bone Density Program [18] from the Study of Osteoporotic Fracture (SOF) [8]: fracture after age 50, falls in previous 12 months, current smoking, activity level, overall health, height at age 25, family history of osteoporotic fracture, height loss, previous hyperthyroidism, mobility level, and current weight. Questions were coded using five ordinal categories or continuous measures whenever possible, and these were subsequently collapsed into two categories of risk using previously established cutoffs [8,19]. A combined clinical Risk Factor Score (RFS) was calculated based upon these items as well as age (weighting +1 for all risk factors except for height loss ≥ 5 cm and age 75 or older which were weighted +2) [18]. Medical risk factors, which are generally considered sufficient grounds for bone density testing and/or intervention even when they occur in isolation, were the focus of additional questions on systemic corticosteroid use, prolonged amenorrhea prior to age 45 (including surgical menopause or premature menopause) and a physician diagnosis of osteoporosis. Copies of the questionnaire are available upon request from the authors. It was necessary to validate the reliability of the self-reported subject information collected by the survey since the comparator studies from which we selected our clinical risk factors had used trained interviewers. A convenience sample of 63 women (50 urban and 13 rural) attending a clinical bone density testing site were recruited to complete the clinical risk factor inventory on two occasions, initially 3 weeks prior to the bone density test using the selfadministered questionnaire and later when the woman attended the bone density clinic using a single trained interviewer. There was excellent agreement in each of the survey items (mean kappa = 0.79, all P < 0.001). High agreement was also found between the total number of clinical risk factors (R = 0.90, P < 0.001) confirming that interviewer-assisted and patient self-report produced equivalent information. Statistical analysis Prevalence rates for clinical and medical risk factors were stratified by age and compared using the Chi-square test. Percentages were corrected for missing values. The degree of correlation between individual clinical risk factors was assessed using a multivariable factor analysis and the principal components extraction method on both continuous (raw data values) and transformed (dichotomous data coded to indicate the presence or absence of a given risk factor) datasets using parametric and non-parametric methods as appropriate. The Chi-square summary statistic was partitioned to determine the degree of linearity (P-for-trend) between the proportions of women with a given clinical risk factor and age category (50–54, 55–64, 65–74, and 75 or older) [20]. Analysis was performed with SPSS Version 12.0.1 (SPSS Inc., Chicago, IL). P < 0.05 was taken to indicate a statistically significant effect. Results Study population A total of 40,330 surveys were mailed out and 8747 (21.7%) responses were returned. Due to the anonymous nature of the mailing process, no information was available on the nonrespondents. Of the returned surveys, 46 could not be scanned due to damage, 628 women indicated that they were still premenopausal (spontaneous menses within the preceding 12 months), 43 asked to be removed from the study and 3 indicated they were not yet 50 years of age. These 720 women were excluded from further analysis. This left a final study population of 8027 women. Of these, the median age was 63 years (range 50–99) with 56.4% being 50–64 years of age and 43.6% being 65 or older. Clinical risk factor prevalence The overall and age-specific prevalences for individual clinical risk factors are summarized in Table 1. Most clinical risk factors showed an increasing prevalence with advancing age, though smoking was an exception and became less prevalent in older women. The correlation between pairs of clinical risk factors was found to be weak (R2 < 0.2), whether based upon the dichotomized or continuous data. The degree of dependency was assessed in multivariable models in which each risk factor was treated as the dependent variable with all other variables used as predictors. This showed a high degree of independence among the clinical risk factors (R2 ≤ 0.17 for all dichotomized variables and R2 ≤ 0.34 for all untransformed variables). All clinical risk factors displayed a significant linear relationship with age (Table 1). When age was excluded, 13.1% of the study population reported no clinical risk factors, a single risk factor was reported by 26.3%, 27.1% reported two, and 33.5% had three or more clinical risk factors (Table 2). If older age was included as a risk factor then the respective values were 8.4% (no risk factors), 21.8% (single risk factor), 24.8% (two risk factors) and 44.9% (three or more risk factors). The absence of clinical risk factors (excluding age) decreased from 13.7% in the 50–54 age group to 6.7% in the 75 and older age group with a significant negative correlation with age (P-for-trend < 0.001). Conversely, the presence of three or more non-age risk factors increased from 25.9% to 54.5% across the age spectrum, and again showed a strong positive correlation with age (P-fortrend < 0.001). W.D. Leslie et al. / Bone 40 (2007) 991–996 993 Table 1 Percentage of individual clinical risk factors for fracture in women ≥ 50 years of age and linear correlation with age Risk factor Age (years) 50–54 N = 1376 55–64 N = 3143 65–74 N = 2271 ≥75 N = 1219 Overall N = 8027 Fracture after age 50 Fall in preceding 12 months Current smoking Activity level (on feet less than 4 h/day) Fair or poor self-rated health status Height at age 25 over 168 cm Family history of hip fracture Family history of osteoporotic fracture (hip, wrist or spine) Height loss since age 25 ≥5 cm Previous hyperthyroidism Mobility level (inability to rise from a chair without using arms) Current weight below 58 kg 4.8 34.7 28.9 33.7 13.5 18.5 11.1 27.5 2.9 4.2 10.1 13.6 Risk categorization The RFS increased progressively with age (Fig. 1; R = 0.29, P < 0.001 when age was not included as a risk factor, R = 0.57, P < 0.001 with age included). Mean RFS was 1.86 (95% CI 1.79–1.93) for the 50–54 age group and increased linearly to 5.1 (95% CI 2.17–2.24) for women 75 years of age and older. There was a strong linear relationship between the proportions of women with a RFS of 3 or more and age (P-for-trend < 0.001) (Table 3). After age 50, more than 67% of women in this study were considered at increased risk for fracture based on RFS ≥ 3 or a medical risk factor. Physician-confirmed osteoporosis (by self-report) increased from 7.7% of women aged 50–54 to 29.9% of women aged 75 years and older, with an overall prevalence of 15.9%. Comparison with the Study of Osteoporotic Fractures The prevalence of clinical risk factors in the Manitoba population was compared with the reported rates from SOF (Table 4) [8,9,21,22]. Since SOF was limited to women age 65 and older, for comparison purposes the Manitoba data were restricted to women age 65 and older. Prevalence rates in these studies were similar (less than 5% difference) for 9 of the 13 11.4 31.8 23.7 30.3 12.9 18.1 14.0 30.4 4.6 4.9 12.0 13.5 21.4 30.2 16.0 28.7 15.6 17.4 17.5 32.8 15.5 7.1 18.2 16.1 32.6 33.1 7.2 40.0 25.9 14.8 18.6 35.0 36.4 7.9 35.3 26.4 16.3 33.0 20.2 31.9 15.7 17.5 15.2 31.3 12.3 5.8 17.0 16.2 P-fortrend <0.001 0.031 <0.001 0.011 <0.001 0.010 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 clinical risk factors investigated (falls, smoking, overall health status, history of maternal hip fracture, height loss ≥ 5 cm, previous hyperthyroidism, low current weight, taking anticonvulsants and/or oral steroid medications). There were some notable exceptions. Manitoba women reported fewer fractures after age 50 and a lower rate of tall peak height (height at age 25 over 168 cm), but higher frequencies of inactivity and poor mobility. Despite these differences, the average prevalence across all clinical risk factors listed in Table 4 was similar for the current study (17.7%) and SOF (15.7%). Discussion Quantitative models that use clinical risk factor information to estimate fracture risk should adjust for the prevalence of the risk factor in the population [19,23]. Clinical risk factor prevalence may not be the same in different countries and populations, and may also vary with age. Ettinger et al. [13] reported a significant relationship between age, the probability of thinness, smoking, and a sister or mother with hip fracture. Table 2 Percentage of zero, single, two and multiple (three or more) clinical risk factors for fracture in women ≥ 50 years of age Number of risk factors 0 Age (years) 50–54 N = 1376 (13.7) [13.7] 1 (31.1) [31.1] 2 (29.3) [29.3] 3 or more (25.9) [25.9] 55–64 N = 3143 65–74 N = 2271 ≥75 N = 1219 Overall N = 8027 (15.4) [15.4] (30.0) [30.1] (27.0) [26.9] (27.6) [27.6] (0.0) [12.9] (12.9) [24.7] (24.7) [27.3] (62.4) [35.1] (0.0) [6.7] (6.8) [14.4] (14.4) [24.4] (78.8) [54.5] (8.4) [13.1] (21.8) [26.3] (24.8) [27.1] (44.9) [33.5] Risk factor count calculated with () and without [] age as a risk factor. Fig. 1. Mean clinical risk factor score (RFS) in women ≥ 65 years of age. Solid line excludes age as a risk factor whereas broken line includes age (+1 for age 65–74, +2 for age 75 and older). Error bars are for 95% confidence intervals. 994 W.D. Leslie et al. / Bone 40 (2007) 991–996 Table 3 Percentage of women ≥ 50 years of age with a risk factor score (RFS) of 3 or more, medical risk factor(s), or a physician diagnosis (self-report) of osteoporosis and linear correlation with age Risk factors Age (years) P-for-trend 50–54 N = 1376 55–64 N = 3143 65–74 N = 2271 ≥75 N = 1219 Overall N = 8027 Clinical RFS 3 or more 26.6 28.5 63.8 90.8 47.7 <0.001 Medical Systemic corticosteroid therapy Prolonged amenorrhea prior to age 45 Premature menopause 45 or younger Physician diagnosis of osteoporosis Any medical risk factor 5.0 3.0 34.7 7.7 42.4 4.9 3.6 31.4 11.5 42.0 5.2 3.6 29.7 20.0 42.6 7.6 2.6 26.3 29.9 44.5 5.4 3.3 30.7 15.9 42.6 0.004 0.667 <0.001 <0.001 0.229 Any of the above 55.5 55.6 76.8 94.1 67.4 <0.001 Our results confirm the importance of considering age as a factor that interacts with clinical risk factors in these models. Despite some differences between our study cohort and an equivalent age group from SOF, it is reassuring to note that the overall prevalence of clinical risk factors in Manitoba women age 65 and older was very similar. The largest discrepancy was for the assessment of mobility (inability to rise from a chair without using arms 24.2% in our study versus 4.4% in SOF for women age 65 years and older). This may reflect methodologic differences, since our measure was based upon self-report whereas SOF directly tested whether the subject could rise up from a chair five times without using arms [8]. It is unclear why the prevalence of a previous fracture after age 50 years was lower in the Manitoba cohort than in SOF, but it is worth noting that hip, clinical spine and wrist fracture rates were individually and collectively similar (data not shown). Confirmation of similar prevalence patterns in populations outside of North America would simplify the development of global fracture risk models [7]. A meta-analysis on corticosteroid use and fracture risk found a history of prior corticosteroid Table 4 Percentage of clinical risk factors and steroid use in Manitoba women ≥ 65 years of age compared with results reported from the Study for Osteoporotic Fractures (SOF) [8,9,21,22] Current study N = 3490 Fracture after age 50 25.2 Falls in previous 12 months 31.2 Current smoking 13.0 Activity level (on feet less than 4 h/day) 32.6 Fair or poor self-rated health status 19.2 Height at age 25 over 168 cm 16.5 History of maternal hip fracture 11.1 Height loss since age 25 ≥5 cm 22.8 Previous hyperthyroidism 7.4 Mobility level (inability to rise from a chair 24.2 without using arms) Current weight below 58 kg 19.7 Currently take anticonvulsant drugs 0.9 Currently take oral steroid medications 6.0 SOF N = 9516 35.9 30.2 9.9 9.7 16.7 27.7 10.0 25.0 9.2 4.4 21.8 1.1 2.8 use of 1.9–9.0% for women from seven large cohort studies [24]. Although the definition of corticosteroid use differed between cohorts and from our assessment of current use, our finding that 5.4% of all respondents reported current corticosteroid use is within this range. A related meta-analysis of smoking found that prevalence decreased from 26.8% for a 50 year old woman to 6.2% for an 85 year old woman which is again similar to our study (50–54 year old stratum 28.9% versus age 75 years and older stratum 7.2%) [25]. Therefore, our findings appear to be consistent with global prevalence rates. The importance of research in clinical risk factors relates to their importance in osteoporotic fracture assessment complementary to bone density measurement [7]. SOF assessed risk factors for hip fracture in 9516 white women 65 years of age or older and found that although the individual effect of most factors was moderate, their combined impact was substantial and operated largely independent of bone density [8]. Women with multiple risk factors and low bone density were at especially high risk (1.1 per 1000 woman-years among women with two risk factors or less and normal bone density versus 27 per 1000 woman-years among those with five or more risk factors and bone density in the lowest third). Black et al., [9] working from the same cohort, derived a semi-quantitative FRACTURE Index based upon age and five clinical risk factors, and showed that even without bone density this provided excellent risk stratification for hip fractures (range from 0.6% in the lowest quintile to 8.2% in the highest quintile), non-vertebral fractures (10.5% in the lowest quintile to 26.1% in the highest quintile) and vertebral fractures (1.4% in the lowest quintile to 9.9% in the highest quintile). This suggests that it is not necessary to develop different risk assessment tools for each fracture type. The use of a single risk assessment tool for a variety of fracture types is certainly convenient for clinical use, and is consistent with the current effort to develop a globally applicable system for determining absolute fracture probability [7]. Others have proposed a more quantitative approach to fracture risk estimation. De Laet et al. [26] showed that including combinations of clinical risk factors improved the gradient of risk over that obtained with bone density measurement alone. They estimated W.D. Leslie et al. / Bone 40 (2007) 991–996 that only 21% of women might need bone density measurement for clinical decision making. Strategies that combine clinical risk indicators might enhance cost-effectiveness and be an efficient method for selecting women for further assessment with bone density measurement [27]. There are significant limitations to our study. Firstly, the response rate (21.7%) was relatively low and bias cannot be excluded. Secondly, our information was all self-reported though we have shown that the self-report and interviewerreported information were in close agreement. Self-reported fracture data, widely used in large epidemiologic studies, is reasonably accurate for hip, forearm/wrist and major nonvertebral osteoporotic fractures but is less reliable for clinical spine fractures [28–31]. SOF has previously reported that over 70% of incident radiographic spine fractures are not recognized clinically, and therefore self-report measures greatly underestimate vertebral fracture burden [32]. Although our risk factor survey was patterned after similar SOF reports, other clinical risk factors may be important. The proposed system from the WHO Collaborating Centre includes “alcohol intake >2 units daily” and “rheumatoid arthritis” (among other factors) [7], and these are not accurately assessed by self-report. Finally, our survey was limited to postmenopausal women and cannot be extrapolated to men. We did not ask about ethnicity, income, education or other socioeconomic determinants which could also be important. For example, the lower prevalence of height greater than 168 cm at age 25 could relate to the inclusion of non-White ethnic groups. Census data indicate that the vast majority of the Manitoba population reports White European ethnicity with 8% visible minorities (predominantly Asian) [33]. The most commonly stated ethnic origins (in descending order) are English, German, Scottish, Ukrainian, Irish, and French [34]. The public health implications of our findings are straightforward and compelling. The high frequency of multiple clinical risk factors in our study population is strongly associated with high fracture risk [8,9]. Approximately one-third of the population reported three or more risk factors (other than age), and this increased to over one-half after age 75. At the same time, the conventional medical model for intervention following fracture has been generally unsuccessful with very low rates of secondary prevention measures [35–38]. Expanded use of clinical risk factor screening is conceptually simple and inexpensive, and would serve two practical purposes: (i) improved identification of those at risk for fracture and (ii) targeting of health care resources such as clinical densitometry and disease-modifying therapies to those at highest risk. Acknowledgments We are indebted to Health Information Services at Manitoba Health for providing data. The results and conclusions are those of the authors, and no official endorsement by Manitoba Health is intended or should be inferred. The authors would like to acknowledge the important contributions of Dr. C.K. Yuen and Dr. B. Kvern to the success of this project, and also to thank Ms. Tanis Wilson for assistance in data collection and interviewing. 995 This study was funded through an unrestrcited grant from the Patient Health Management Department of Merck Frosst Canada Limited. References [1] NIH Consensus Development Panel. Osteoporosis prevention, diagnosis, and therapy. JAMA 2001;285:785–95. [2] Forsen L, Sogaard AJ, Meyer HE, Edna T, Kopjar B. Survival after hip fracture: short- and long-term excess mortality according to age and gender. Osteoporos Int 1999;10:73–8. [3] Center JR, Nguyen TV, Schneider D, Sambrook PN, Eisman JA. Mortality after all major types of osteoporotic fracture in men and women: an observational study. Lancet 1999;353:878–82. [4] Johnell O, Kanis JA, Oden A, Sernbo I, Redlund-Johnell I, Petterson C, et al. Mortality after osteoporotic fractures. Osteoporos Int 2004;15:38–42. [5] Wiktorowicz ME, Goeree R, Papaioannou A, Adachi JD, Papadimitropoulos E. Economic implications of hip fracture: health service use, institutional care and cost in Canada. Osteoporos Int 2001;12:271–8. [6] Brown JP, Josse RG. 2002 clinical practice guidelines for the diagnosis and management of osteoporosis in Canada. CMAJ 2002;167:S1–S34. [7] Kanis JA, Borgstrom F, De Laet C, Johansson H, Johnell O, Jonsson B, et al. Assessment of fracture risk. Osteoporos Int 2005;16:581–9. [8] Cummings SR, Nevitt MC, Browner WS, Stone K, Fox KM, Ensrud KE, et al. Risk factors for hip fracture in white women. Study of Osteoporotic Fractures Research Group. N Engl J Med 1995;332:767–73. [9] Black DM, Steinbuch M, Palermo L, Dargent-Molina P, Lindsay R, Hoseyni MS, et al. An assessment tool for predicting fracture risk in postmenopausal women. Osteoporos Int 2001;12:519–28. [10] Espallargues M, Sampietro-Colom L, Estrada MD, Sola M, del Rio L, Setoain J, et al. Identifying bone-mass-related risk factors for fracture to guide bone densitometry measurements: a systematic review of the literature. Osteoporos Int 2001;12:811–22. [11] Leslie WD, Metge C, Ward L. Contribution of clinical risk factors to bone density-based absolute fracture risk assessment in postmenopausal women. Osteoporos Int 2003;14:334–8. [12] De Laet, C, Oden A, Johansson H, Johnell O, Jonsson B, Kanis JA. The impact of the use of multiple risk indicators for fracture on case-finding strategies: a mathematical approach. In: 2004. [13] Ettinger B, Hillier TA, Pressman A, Che M, Hanley DA. Simple computer model for calculating and reporting 5-year osteoporotic fracture risk in postmenopausal women. J Women's Health (Larchmt) 2005;14:159–71. [14] Kanis JA, Johnell O, De Laet C, Jonsson B, Oden A, Ogelsby AK. International variations in hip fracture probabilities: implications for risk assessment. J Bone Miner Res 2002;17:1237–44. [15] Lunt M, Felsenberg D, Reeve J, Benevolenskaya L, Cannata J, Dequeker J, et al. Bone density variation and its effects on risk of vertebral deformity in men and women studied in thirteen European centers: the EVOS Study. J Bone Miner Res 1997;12:1883–94. [16] Roos NP, Shapiro E. Revisiting the Manitoba Centre for Health Policy and Evaluation and its population-based health information system. Med Care 1999;37:JS10–4. [17] Roos NP. Establishing a population data-based policy unit. Med Care 1999;37:JS15–26. [18] Leslie WD, Metge C. Establishing a regional bone density program: lessons from the Manitoba experience. J Clin Densitom 2003;6:275–82. [19] Leslie WD, Metge C, Salamon EA, Yuen CK. Bone mineral density testing in healthy postmenopausal women. The role of clinical risk factor assessment in determining fracture risk. J Clin Densitom 2002;5:117–30. [20] Selvin S. Epidemiologic analysis. New York: Oxford Univ. Press; 2001. [21] Baltzan MA, Suissa S, Bauer DC, Cummings SR. Hip fractures attributable to corticosteroid use. Study of Osteoporotic Fractures Group. Lancet 1999;353:1327. [22] SOF Coordinating Center. SOF Online-V1HTLOSS. URL: http://sof.ucsf. edu (last visited October 1, 2006). [23] Development Committee of the National Osteoporosis Foundation. Osteoporosis: review of the evidence for prevention, diagnosis and 996 [24] [25] [26] [27] [28] [29] [30] [31] W.D. Leslie et al. / Bone 40 (2007) 991–996 treatment and cost-effectiveness analysis. Introduction. Osteoporos Int 1998;8(Suppl 4):S7–S80. Kanis JA, Johansson H, Oden A, Johnell O, de Melton III LJ, Tenenhouse A, et al. A meta-analysis of prior corticosteroid use and fracture risk. J Bone Miner Res 2004;19:893–9. Kanis JA, Johnell O, Oden A, Johansson H, De Laet C, Eisman JA, et al. Smoking and fracture risk: a meta-analysis. Osteoporos Int 2005;16:155–62. De Laet C, Oden A, Johansson H, Johnell O, Jonsson B, Kanis JA. The impact of the use of multiple risk indicators for fracture on case-finding strategies: a mathematical approach. Osteoporos Int 2005;16:313–8. Johansson H, Oden A, Johnell O, Jonsson B, De Laet C, Oglesby A, et al. Optimization of BMD measurements to identify high risk groups for treatment—A test analysis. J Bone Miner Res 2004;19:906–13. Honkanen K, Honkanen R, Heikkinen L, Kroger H, Saarikoski S. Validity of self-reports of fractures in perimenopausal women. Am J Epidemiol 1999;150:511–6. Joakimsen RM, Fonnebo V, Sogaard AJ, Tollan A, Stormer J, Magnus JH. The Tromso study: registration of fractures, how good are self-reports, a computerized radiographic register and a discharge register? Osteoporos Int 2001;12:1001–5. Ivers RQ, Cumming RG, Mitchell P, Peduto AJ. The accuracy of selfreported fractures in older people. J Clin Epidemiol 2002;55:452–7. Chen Z, Kooperberg C, Pettinger MB, Bassford T, Cauley JA, LaCroix AZ, et al. Validity of self-report for fractures among a multiethnic cohort of [32] [33] [34] [35] [36] [37] [38] postmenopausal women: results from the Women's Health Initiative observational study and clinical trials. Menopause 2004;11:264–74. Nevitt MC, Ettinger B, Black DM, Stone K, Jamal SA, Ensrud K, et al. The association of radiographically detected vertebral fractures with back pain and function: a prospective study. Ann Intern Med 1998;128:793–800. Statistics Canada Visible minority population, by provinces and territories (2001 Census). http://www40.statcan.ca/l01/cst01/demo52b.htm (last visited September 24, 2006). 5 A.D. Statistics Canada. Population by selected ethnic origins, by province and territory (2001 Census-Manitoba). URL: http://www40.statcan.ca/l01/ cst01/demo26h.htm?sdi=manitoba%20european (last visited October 1, 2006). Juby AG, Geus-Wenceslau CM. Evaluation of osteoporosis treatment in seniors after hip fracture. Osteoporos Int 2002;13:205–10. Feldstein AC, Nichols GA, Elmer PJ, Smith DH, Aickin M, Herson M. Older women with fractures: patients falling through the cracks of guideline-recommended osteoporosis screening and treatment. J Bone Jt Surg, Am 2003;85-A:2294–302. Follin SL, Black JN, McDermott MT. Lack of diagnosis and treatment of osteoporosis in men and women after hip fracture. Pharmacotherapy 2003;23:190–8. Ettinger B, Black DM, Palermo L, Nevitt MC, Melnikoff S, Cummings SR. Kyphosis in older women and its relation to back pain, disability and osteopenia: the Study of Osteoporotic Fractures. Osteoporos Int 1994;4:55–60.