Bone 76 (2015) 1–4
Contents lists available at ScienceDirect
Bone journal homepage: www.elsevier.com/locate/bone
Original Full Length Article
FRAX (Aus) and falls risk: Association in men and women Kara L. Holloway a,⁎, Mark A. Kotowicz a,b,c, Stephen E. Lane a,c, Sharon L. Brennan a,b,d, Julie A. Pasco a,b,c a
School of Medicine, Deakin University, Geelong, Australia NorthWest Academic Centre, Department of Medicine, The University of Melbourne, Melbourne, Australia c Barwon Health, Geelong, Australia d Australian Institute of Musculoskeletal Sciences, The University of Melbourne, Melbourne, Australia b
a r t i c l e
i n f o
Article history: Received 20 October 2014 Revised 2 March 2015 Accepted 3 March 2015 Available online 11 March 2015 Edited by: Kristine Ensrud Keywords: FRAX score Falls risk Fracture prediction Elderly Falls Screening Test
a b s t r a c t Purpose: The WHO fracture risk prediction tool (FRAX®) utilises clinical risk factors to estimate the probability of fracture over a 10-year period. Although falls increase fracture risk, they have not been incorporated into FRAX. It is currently unclear if FRAX captures falls risk and whether addition of falls would improve fracture prediction. We aimed to investigate the association of falls risk and Australian-specific FRAX. Methods: Clinical risk factors were documented for 735 men and 602 women (age 40–90 yr) assessed at followup (2006–2010 and 2000–2003, respectively) of the Geelong Osteoporosis Study. FRAX scores with and without BMD were calculated. A falls risk score was determined at the time of BMD assessment and self-reported incident falls were documented from questionnaires returned one year later. Multivariable analyses were performed to determine: (i) cross-sectional association between FRAX scores and falls risk score (Elderly Falls Screening Test, EFST) and (ii) prospective relationship between FRAX and time to a fall. Results: There was an association between FRAX (hip with BMD) and EFST scores (β = 0.07, p b 0.001). After adjustment for sex and age, the relationship became non-significant (β = 0.00, p = 0.79). The risk of incident falls increased with increasing FRAX (hip with BMD) score (unadjusted HR 1.04, 95% CI 1.02, 1.07). After adjustment for age and sex, the relationship became non-significant (1.01, 95% CI 0.97, 1.05). Conclusions: There is a weak positive correlation between FRAX and falls risk score, that is likely explained by the inclusion of age and sex in the FRAX model. These data suggest that FRAX score may not be a robust surrogate for falls risk and that inclusion of falls in fracture risk assessment should be further explored. © 2015 Elsevier Inc. All rights reserved.
Introduction Fractures are a major health concern because of the public health costs associated with diagnosis (X-ray) and management that may include hospitalisation and rehabilitation [1]. Additionally, fracture may be associated with excess morbidity [2] and mortality [3] and increases the risk of subsequent fractures [4] that further increases the public health burden [1]. It is important to identify those at the highest risk of fracture to ensure that treatment is targeted to those who are most likely to benefit and to avoid expensive treatments in those at lower risk [5]. FRAX is a fracture risk assessment algorithm developed by the World Health Organization (WHO) and was designed for use in primary care to assess 10-year fracture probability in men and women aged 40 to 90 years who have not previously received pharmacological treatment to prevent fractures [5]. It combines clinical risk factors (age, body mass index, prior fracture, parental hip fracture, smoking, glucocorticoid ⁎ Corresponding author at: Epi-Centre for Healthy Ageing, IMPACT Strategic Research Centre, School of Medicine, Deakin University, PO Box 281, Geelong, VIC 3220, Australia. Fax: +61 342153491. E-mail address: [email protected] (K.L. Holloway).
http://dx.doi.org/10.1016/j.bone.2015.03.004 8756-3282/© 2015 Elsevier Inc. All rights reserved.
use, rheumatoid arthritis, secondary osteoporosis and daily alcohol intake) to estimate the 10 year probability of hip and major osteoporotic fracture in men and women [6]. Bone mineral density (BMD) can be incorporated into the FRAX calculation to improve the predictions, but it is not necessary for fracture risk calculations [5]. Additionally, FRAX has multiple versions with different predictions for 53 countries, allowing more accurate fracture predictions based on data from the country of interest. The FRAX tool is widely accepted for predicting fractures, but there have been suggestions as to how to improve it [5]. One important fracture risk factor that is not included is falls history; this was not included due to two reasons. First, uniform falls data were not available from all cohorts that contributed to FRAX and second, pharmaceutical intervention has not been shown to reduce fracture risk in those with a high falls risk [7]. For the latter, identification of those at high risk might only be useful if the risk can be reduced through some type of intervention. Falls strongly predict hip and other non-vertebral fracture independent of BMD and bone strength [5,8,9], and fallers are suggested to be at a higher fracture risk than FRAX may estimate [7]. Falls are common in the community-dwelling elderly occurring in 28–35% of 65 year olds and 42% of those over 75 years [10]. One method of measuring falls risk is the “Timed Up and Go Test” (TUG) which involves recording
2
K.L. Holloway et al. / Bone 76 (2015) 1–4
the time taken for an individual to rise from a chair, walk 3 m and return to sitting in the chair. The TUG test is mainly used for assessing the physical performance and functional mobility among elderly individuals, but it is also a good technique for determining individuals at a high risk of falls [11]. Falls risk assessment can be enhanced by incorporating more measurements into the assessment as in the Elderly Falls Screening Test (EFST) which includes prior falls, injurious falls, near falls, slow walking speed and gait [8]. There are recent data suggesting that the exclusion of falls risk from FRAX fracture risk assessments is not critical because the FRAX score is a surrogate for falls risk in elderly men [12]. Therefore, we hypothesised that falls risk is dependent on FRAX score. Thus, the aim of our study was to investigate the association of falls risk and Australian-specific FRAX 10-year probability scores from an Australian cohort study to determine if falls risk is predicted by FRAX score.
Prospective relationship between the FRAX score and the time to a fall We used Cox proportional hazards regression to estimate hazard ratios (HRs) for unit increases in each of the four FRAX scores. We presented the HRs as an increase in risk for a 1% increase in FRAX score. We chose 1% because “high risk” FRAX scores are considered to be ≥ 3% for hip and ≥20% for major osteoporotic fracture [14]. We determined that it would be more relevant to assess 1% increases in FRAX than for example, 5% or 10% because this would miss the cut-off for “high risk” hip fracture FRAX scores. Baseline hazards were stratified by 10 year age groups. We developed models that included age, sex and prior falls to allow adjustment for these variables [12]. Statistical analyses were performed using R version 3.0.2. Results Descriptive characteristics
Materials and methods Participants This current work incorporates data from the Geelong Osteoporosis Study (GOS) which includes residents form the Barwon Statistical Division (BSD), a region of south-eastern Australia [13]. This region has a number of factors that make it ideal for epidemiological studies including a large, stable population (n ~ 250,000) as well as multiple social, cultural and geographical settings. It is also representative of the Australian population. Further details of the study are provided elsewhere [13]. Participants were selected at random from Commonwealth Electoral Rolls. In Australia, voting is compulsory and therefore the electoral rolls are a comprehensive list of all adult Australians. Men were recruited from 2001 to 2006 (1540 recruited, 67% response). Women underwent baseline assessments from 1993 to 1997 (1494 recruited, 77% response). The data for this study come from the 5 and 6 year follow-up assessments for men (2006 to 2010) and the 6 and 7 year follow-up assessments for women (2000 to 2003). At the 5 year follow-up for men and 6 year follow-up for women, there were 735 and 602 individuals, aged 40 to 90 years, respectively. One year later, participants were sent questionnaires regarding falls and fractures that had occurred since their previous assessment; 473 men and 523 women completed these and returned them. A comparison of those who did and did not return the follow-up questionnaire revealed no differences in mean age, BMD, FRAX scores or prior fracture or falls (all p N 0.05, data not shown). The age of 5 year follow-up (for men) or 6 year follow-up (for women) was used for the cross-sectional analysis and as baseline age for the longitudinal analysis. Written, informed consent was obtained from all participants and this study was approved by the Barwon Health Human Research Ethics Committee. FRAX scores (hip and major osteoporotic fracture, with and without BMD; N = 4 scores) and Elderly Falls Screening Test (EFST) scores were calculated for all participants (see Appendix 1).
The descriptive characteristics of the participants are shown in Table 1. Those with a fall prior to baseline (5 year follow-up for men and 6 year follow-up for women) were older, lighter and shorter than those who did not fall prior to baseline. Prior fallers also had more prior fractures (45.3% vs 34.1%) and a higher proportion of rheumatoid arthritis (9.4% vs 3.9%). Individuals who fell during the 12 month followup were older, had a higher proportion of prior fractures (45.5% vs 35.1%) as well as prior falls (37% vs 20%). No other associations were found between the descriptive characteristics for prior and incident fall groups. Some differences in characteristics between men and women are noted. The median age for men and women was similar (men 62.2 IQR 52.2–74.3 years vs women 62.1 IQR 51.0, 73.0 years, p = 1.000), mean weight (83.9 ± 13.6 kg vs 70.1 ± 13.9 kg, p = 0.000), height (174.1 ± 6.9 cm vs 159.8 ± 6.3, p = 0.000), BMI (27.6 ± 4.0 kg/m2 vs 27.4 ± 5.3 kg/m2, p = 0.469), prior fracture (36.6% vs 36.9%, p = 0.916), history of parent hip fracture (6.1% vs 5.8%, p = 0.908) and current smoking (10.9 vs 10.5%, p = 0.805). Women had higher proportions of glucocorticoid use (4.2% vs 24.4%), rheumatoid arthritis (2.4 vs 8.5%) and secondary osteoporosis (9.8% vs 27.7%); men had a higher proportion of individuals with excessive alcohol intake (25.2% vs 2.7%) and a higher femoral neck BMD (0.965 ± 0.141 vs 0.898 ± 0.162 g/m2) (all p b 0.001). Cross-sectional analysis: EFST and FRAX There was a trend of increasing EFST scores with increasing FRAX scores, but the magnitude of the effect was small and unlikely to have clinical significance. After adjusting for the association of EFST scores on FRAX scores by age and sex, this relationship was not statistically significant (p N 0.2). This pattern was observed for all four FRAX scores investigated. The R-squared for the four interaction models were low; 0.196, 0.186, 0.194, and 0.186 for major osteoporotic fracture (with BMD), major osteoporotic fracture (without BMD), hip fracture (with BMD) and hip fracture (without BMD), respectively.
Statistical analyses
Longitudinal analysis: incident falls and FRAX
Cross-sectional association between the FRAX score and EFST The cross-sectional analysis aimed to investigate the association between a fracture risk score (FRAX) and a falls risk score (EFST). This was assessed using linear regression; an initial model was developed using the FRAX score as the independent variable and EFST as the dependent variable, adjusting for sex. We then fitted a further model adjusting for age (centred around age 60 years) and its interaction with sex, following the methods of Johansson et al. [12]. Beta coefficients were calculated, which represent the change in EFST for every 1% change in FRAX. QQ plots were used to assess normality of residuals, so that inference on regression coefficients was valid.
Among the 735 men at baseline, 132 (18%) reported a prior fall over the past 12 months. For the 602 women at baseline, 177 (29%) reported a prior fall over the past 12 months. During the subsequent 12 month period 78 out of 473 men (16%) and 110 out of 523 women (21%) reported falling. There were five men and 17 women who recorded two falls, and two men and five women who recorded three falls during the 12 month follow-up period. The unadjusted and adjusted hazard ratios for each of the four FRAX scores are presented in Table 2. In the unadjusted models, the risk of incident falls increased with increasing FRAX score. After adjustment for age and sex, the risk became non-significant. In the fully adjusted
K.L. Holloway et al. / Bone 76 (2015) 1–4
3
Table 1 Participant characteristics. Values expressed as median (interquartile range, IQR), mean ± SD or n (percentage). Participants were aged 40–90 years. Prior fall (baseline)a
Incident fall (over the 12 month FU)b
Risk factors
All (n = 1337)
Yes (n = 309)
No (n = 1028)
Yes (n = 189)
No (n = 807)
Age (years, median (IQR)) Weight (kg) Height (cm) BMI (kg/m2) Prior fracture Parent fractured hip Current smoking Glucocorticoid use Rheumatoid arthritis Secondary osteoporosis Excessive alcohol intake Femoral neck BMD (g/cm2) Prior falls
62.0 (51.7, 73.3) 77.7 ± 15.4 167.7 ± 9.7 27.6 ± 4.6 491 (36.7) 80 (6.0) 143 (10.7) 178 (13.3) 69 (5.2) 239 (17.9) 201 (15.0) 0.935 ± 0.155 309 (23.1)
68.0 (56.7, 77.5) c 75.9 ± 15.0 c 165.2 ± 9.8 c 27.8 ± 4.9 140 (45.3) c 20 (6.5) 29 (9.4) 49 (15.9) 29 (9.4) c 66 (21.4) 38 (12.3) 0.908 ± 0.164 c –
60.3 (50, 70)c 78.2 ± 15.4 c 168.4 ± 9.6 c 27.5 ± 4.5 351 (34.1) c 60 (5.8) 114 (11.1) 129 (12.5) 40 (3.9) c 173 (16.8) 163 (15.9) 0.943 ± 0.151 c –
66.0 (54.0, 76.5)d 76.4 ± 15.5 165.7 ± 9.7 27.8 ± 4.9 86 (45.5) d 12 (6.3) 12 (6.3) 32 (16.9) 15 (7.9) 41 (21.7) 26 (13.8) 0.914 ± 0.171 70 (37.0) d
60.8 (52, 72) d 76.4 ± 15.3 166.8 ± 9.5 27.4 ± 4.7 283 (35.1) d 51 (6.3) 86 (10.7) 111 (13.8) 44 (5.5) 151 (18.7) 99 (12.3) 0.939 ± 0.153 161 (20.0) d
a b c d
Baseline corresponds to the 5 year FU for men and 6 year FU for women from the Geelong Osteoporosis Study. Prior falls were counted if they occurred before this baseline visit. The 12 month FU occurred after the baseline visit and corresponds to the 6 year FU for men and 7 year FU for women. Significant difference (p b 0.05) between the prior falls and no prior falls groups. Significant difference (p b 0.05) between the incident falls and no incident falls groups.
models there were independent contributions from age, sex and prior falls. No interaction terms were identified. The Harrell's C values were 0.616, 0.613, 0.615, and 0.616 for major osteoporotic fracture (with BMD), major osteoporotic fracture (without BMD), hip fracture (with BMD) and hip fracture (without BMD), respectively. Discussion In this study, we report a clinically non-significant cross-sectional association between FRAX and EFST score. This relationship became statistically non-significant when adjusted for age and sex. Additionally, there was a weak association between FRAX and incident falls that became non-significant after adjusting for age and sex. A meeting involving a joint Task Force including members from the International Osteoporosis Foundation (IOF) and the International Society for Clinical Densitometry (ISCD) during 2010 resulted in the development of resource documents that present recommendations on how to improve FRAX and better inform clinicians who use the tool [15]. It was considered that exclusion of falls risk from FRAX could result in underestimation of fracture risk in individuals with recurrent falls [7]. However, the discussion also included reasons why falls risk had not been incorporated into FRAX to-date. Reasons included a lack of data regarding falls, lack of information about interaction of falls with other FRAX variables, lack of evidence to suggest that pharmaceutical intervention can reduce falls, heterogeneity in falls questions, variability in falls between populations and that falls risk is inherently
taken into account in the FRAX algorithm, though not as an input variable [7,15]. Including falls in FRAX would not help identify those who could benefit from pharmaceutical therapies, since there is no evidence that drugs can reduce high fracture risks associated with falling. There are options available to help reduce risk from falling (e.g. reduce psychotropic medications, exercise (to improve balance and stability), home modifications, management of fainting/dizziness [16]), though these are not the same types of therapies that would be effective for preventing fractures associated with low BMD and represents a challenge for the reduction of fracture risk. There is evidence from previous studies to include falls in fracture risk assessments [10] and also studies that suggest that inclusion of additional risk factors does not improve the predictive ability of FRAX or other models [17–19]. This seems paradoxical if FRAX does not predict falls, then it would be expected that the addition of falls would improve fracture prediction. The reason for this could be that both age and sex, which were predictors of falls risk in this study, are already included in FRAX and other fracture prediction tools [16]. Since FRAX did not predict falls well, it should not be used to determine falls risk. Other fracture risk tools have been developed that incorporate falls risk, including the Garvan Fracture Risk Calculator (FRC) [20] and the Geelong Osteoporosis Study FRISK Score [9] and may be more suitable for fracture prediction when an individual has a high falls risk. The challenges of prevention of fracture still remain, however. The major strength of this study is that participants were randomly selected from the general population. Additionally, we recruited men
Table 2 Estimated hazard ratios (95% CIs) of an absolute increase of one percentage point in FRAX risks for various FRAX risk calculations. FRAX score type Major osteoporotic fracture, with BMD Past falls Female FRAX score Major osteoporotic fracture, without BMD Past falls Female FRAX score Hip fracture, with BMD Past falls Female FRAX score Hip fracture, without BMD Past falls Female FRAX score
Unadjusted hazard ratio (95% CI)
Adjusted (sex and age) hazard ratio (95% CI)
Adjusted (sex, age and prior falls) hazard ratio (95% CI)
1.03 (1.02, 1.04)
1.39 (1.05, 1.83) 1.01 (0.98, 1.03)
1.64 (1.27, 2.13) 1.30 (0.98, 1.72) 1.01 (0.98, 1.03)
1.02 (1.01, 1.03)
1.45 (1.09, 1.93) 1.00 (0.98, 1.01)
1.63 (1.26, 2.12) 1.36 (1.02, 1.82) 1.00 (0.98, 1.01)
1.04 (1.02, 1.07)
1.41 (1.08, 1.83) 1.01 (0.97, 1.04)
1.64 (1.27, 2.13) 1.31 (1.00, 1.72) 1.01 (0.97, 1.05)
1.02 (1.01, 1.04)
1.45 (1.10, 1.89) 0.99 (0.97, 1.02)
1.63 (1.26, 2.11) 1.35 (1.02, 1.78) 0.99 (0.97, 1.02)
4
K.L. Holloway et al. / Bone 76 (2015) 1–4
and women over the entire age range covered by FRAX (40–90 years), whereas many previous studies have focussed on one sex within a narrower age range (usually above 50 years). We also used a detailed falls risk assessment [8] including five different criteria to assess falls risk, which would be more informative than a single question regarding falls in the past 12 months. This falls risk assessment has been reviewed in several meta-analyses [21–23] and has sufficient sensitivity and specificity for predicting fallers and non-fallers [24]. However, we recognised that our study has some limitations. The majority of the study participants were Caucasian (99%) and this limits the generalisibility of the findings to other populations. Finally, we relied on some self-reported fracture and falls data in order to calculate FRAX and EFST scores; however, self-reported fractures have previously been shown to be reliable [25–27]. Furthermore, we acknowledge limitations in relying on self-reported falls dates. Conclusions There is a weak positive correlation between FRAX and falls risk score, that is likely explained by the inclusion of age and sex in the FRAX model. These data suggest that FRAX score may not be a robust surrogate for falls risk and that inclusion of falls in fracture risk assessment should be further explored. Acknowledgments The study was funded by grants from the National Health and Medical Council (NHMRC 251638, 299831, 628582), The University of Melbourne Research Grant Scheme, American Society for Bone and Mineral Research (ASBMR), Perpetual Trustees, Amgen (Europe) GmBH and the Geelong Region Medical Research Foundation. Authors' roles: Study design: KLH, MAK, and JAP. Study conduct: JAP. Data analysis: SEL. Data interpretation: KLH, MAK, SEL, SLB, and JAP. Drafting manuscript: KLH. Revising manuscript content: KLH, MAK, SEL, SLB, and JAP. Approving final version of manuscript: KLH, MAK, SEL, SLB, and JAP. SEL takes responsibility for the integrity of the data analysis. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.bone.2015.03.004. References [1] Watts JJ, Abimanyi-Ochom J, Sanders KM. Osteoporosis costing all Australians: a new burden of disease analysis—2012 to 2022. Australia: osteoporosis Australia; 2013. [2] Pasco JA, Sanders KM, Hoekstra FM, Henry MJ, Nicholson GC, Kotowicz MA. The human cost of fracture. Osteoporos Int 2005;16:2046–52. [3] Otmar R, Kotowicz MA, Brennan SL, Bucki-Smith G, Korn S, Pasco JA. Personal and psychosocial impacts of clinical fracture in men. J Mens Health 2013;10:22–7.
[4] Pasco JA, Seeman E, Henry MJ, Merriman EN, Nicholson GC, Kotowicz MA. The population burden of fractures originates in women with osteopenia, not osteoporosis. Osteoporos Int 2006;17:1404–9. [5] Bauer D. FRAX, falls, and fracture prediction: predicting the future. Arch Intern Med 2011;171:1661–2. [6] World Health Organization Collaborating Centre for Metabolic Bone Diseases University of Sheffield UK. FRAX ® WHO fracture risk assessment tool; 2011. [7] Masud T, Binkley N, Boonen S, Hannan MT. Official Positions for FRAX® clinical regarding falls and frailty: can falls and frailty be used in FRAX®? From Joint Official Positions Development Conference of the International Society for Clinical Densitometry and International Osteoporosis Foundation on FRAX®. J Clin Densitom 2011;14:194–204. [8] Cwikel J, Fried A, Biderman A, Galinsky D. Validation of a fall-risk screening test, the Elderly Fall Screening Test (EFST), for community-dwelling elderly. Disabil Rehabil 1998;20:161–7. [9] Henry MJ, Pasco JA, Sanders KM, Nicholson GC, Kotowicz MA. Fracture risk (FRISK) score: Geelong Osteoporosis Study. Radiology 2006;241:190–6. [10] Edwards MH, Jameson K, Denison H, Harvey NC, Sayer AA, Dennison EM, et al. Clinical risk factors, bone density and fall history in the prediction of incident fracture among men and women. Bone 2013;52:541–7. [11] Zhu K, Devine A, Lewis J, Dhaliwal S, Prince R. “Timed up and go” test and bone mineral density measurement for fracture prediction. Arch Intern Med 2011;171:1655–61. [12] Johansson H, Odén A, Karlsson M, Rosengren B, Ljunggren O, McCloskey E, et al. FRAX® predicts future falls in elderly men. MrOs Sweden. J Bone Miner Res 2013; 28(Suppl. 1). [13] Pasco JA, Nicholson GC, Kotowicz MA. Cohort profile: Geelong Osteoporosis Study. Int J Epidemiol 2012;41:1565–75. [14] Unnanuntana A, Gladnick BP, Lane EDJM. The assessment of fracture risk. J Bone Joint Surg 2010;92:743–53. [15] Kanis JA, Hans D, Cooper C, Baim S, Bilezikian JP, Binkley N, et al. Interpretation and use of FRAX in clinical practice. Osteoporos Int 2011;22:2395–411. [16] Hill K, Schwarz J. Assessment and management of falls in older people. Intern Med J 2004;34:557–64. [17] Sambrook PN, Flahive J, Hooven FH, Boonen S, Chapurlat R, Lindsay R, et al. Predicting fractures in an international cohort using risk factor algorithms without BMD. J Bone Miner Res 2011;26:2770–7. [18] Melton III LG, Crowson CS, O'Fallon WM. Fracture incidence in Olmsted county, Minnesota: comparison of urban with rural rates and changes in urban rates over time. Osteoporos Int 1999;9:29–37. [19] Melton III J, Atkinson EJ, Achenbach SJ, Kanis JA, Therneau TM, Johansson H, et al. Potential extensions of the US FRAX algorithm. J Osteoporos 2012. http://dx.doi.org/10. 1155/2012/528790. [20] Institute Garvan. Bone Fracture Risk Calculator - Garvan Institute of Medical Research; 2008. [21] Kehinde J. Instruments for measuring fall risk in older adults living in long-term care facilities. J Gerontol Nurs 2009;35:46–55. [22] Perell KL, Nelson A, Goldman RL, Luther SL, Prieto-Lewis N, Rubenstein LZ. Fall risk assessment measures: an analytic review. J Gerontol A 2001;56:M761–6. [23] Scott V, Votova K, Scanlan A, Close J. Multifactorial and functional mobility assessment tools for fall risk among older adults in community, home-support, longterm and acute care settings. Age Ageing 2007;36:130–9. [24] Oliver D, Daly F, Martin F, McMurdo M. Risk factors and risk assessment tools for falls in hospital in-patients: a systematic review. Age Ageing 2004;33:122–30. [25] Chen Z, Kooperberg C, Pettinger M, Bassford T, Cauley J, LaCroix A, et al. Validity of self-report for fractures among a multiethnic cohort of postmenopausal women: results from the Women's Health Initiative observational study and clinical trials. J North Am Menopause Soc 2004;11:264–74. [26] Gehlbach S, Saag KG, Adachi JD, Hooven FH, Flahive J, Boonen S, et al. Previous fractures at multiple sites increase the risk for subsequent fractures: the global longitudinal study of osteoporosis in women. J Bone Miner Res 2012;27:645–53. [27] Nevitt MC, Cummings SR, Browner WS, Seeley DG, Cauley JA, Vogt TM, et al. The accuracy of self-report of fractures in elderly women: evidence from a prospective study. Am J Epidemiol 1992;135:490–9.
Contents lists available at ScienceDirect
Bone journal homepage: www.elsevier.com/locate/bone
Original Full Length Article
FRAX (Aus) and falls risk: Association in men and women Kara L. Holloway a,⁎, Mark A. Kotowicz a,b,c, Stephen E. Lane a,c, Sharon L. Brennan a,b,d, Julie A. Pasco a,b,c a
School of Medicine, Deakin University, Geelong, Australia NorthWest Academic Centre, Department of Medicine, The University of Melbourne, Melbourne, Australia c Barwon Health, Geelong, Australia d Australian Institute of Musculoskeletal Sciences, The University of Melbourne, Melbourne, Australia b
a r t i c l e
i n f o
Article history: Received 20 October 2014 Revised 2 March 2015 Accepted 3 March 2015 Available online 11 March 2015 Edited by: Kristine Ensrud Keywords: FRAX score Falls risk Fracture prediction Elderly Falls Screening Test
a b s t r a c t Purpose: The WHO fracture risk prediction tool (FRAX®) utilises clinical risk factors to estimate the probability of fracture over a 10-year period. Although falls increase fracture risk, they have not been incorporated into FRAX. It is currently unclear if FRAX captures falls risk and whether addition of falls would improve fracture prediction. We aimed to investigate the association of falls risk and Australian-specific FRAX. Methods: Clinical risk factors were documented for 735 men and 602 women (age 40–90 yr) assessed at followup (2006–2010 and 2000–2003, respectively) of the Geelong Osteoporosis Study. FRAX scores with and without BMD were calculated. A falls risk score was determined at the time of BMD assessment and self-reported incident falls were documented from questionnaires returned one year later. Multivariable analyses were performed to determine: (i) cross-sectional association between FRAX scores and falls risk score (Elderly Falls Screening Test, EFST) and (ii) prospective relationship between FRAX and time to a fall. Results: There was an association between FRAX (hip with BMD) and EFST scores (β = 0.07, p b 0.001). After adjustment for sex and age, the relationship became non-significant (β = 0.00, p = 0.79). The risk of incident falls increased with increasing FRAX (hip with BMD) score (unadjusted HR 1.04, 95% CI 1.02, 1.07). After adjustment for age and sex, the relationship became non-significant (1.01, 95% CI 0.97, 1.05). Conclusions: There is a weak positive correlation between FRAX and falls risk score, that is likely explained by the inclusion of age and sex in the FRAX model. These data suggest that FRAX score may not be a robust surrogate for falls risk and that inclusion of falls in fracture risk assessment should be further explored. © 2015 Elsevier Inc. All rights reserved.
Introduction Fractures are a major health concern because of the public health costs associated with diagnosis (X-ray) and management that may include hospitalisation and rehabilitation [1]. Additionally, fracture may be associated with excess morbidity [2] and mortality [3] and increases the risk of subsequent fractures [4] that further increases the public health burden [1]. It is important to identify those at the highest risk of fracture to ensure that treatment is targeted to those who are most likely to benefit and to avoid expensive treatments in those at lower risk [5]. FRAX is a fracture risk assessment algorithm developed by the World Health Organization (WHO) and was designed for use in primary care to assess 10-year fracture probability in men and women aged 40 to 90 years who have not previously received pharmacological treatment to prevent fractures [5]. It combines clinical risk factors (age, body mass index, prior fracture, parental hip fracture, smoking, glucocorticoid ⁎ Corresponding author at: Epi-Centre for Healthy Ageing, IMPACT Strategic Research Centre, School of Medicine, Deakin University, PO Box 281, Geelong, VIC 3220, Australia. Fax: +61 342153491. E-mail address: [email protected] (K.L. Holloway).
http://dx.doi.org/10.1016/j.bone.2015.03.004 8756-3282/© 2015 Elsevier Inc. All rights reserved.
use, rheumatoid arthritis, secondary osteoporosis and daily alcohol intake) to estimate the 10 year probability of hip and major osteoporotic fracture in men and women [6]. Bone mineral density (BMD) can be incorporated into the FRAX calculation to improve the predictions, but it is not necessary for fracture risk calculations [5]. Additionally, FRAX has multiple versions with different predictions for 53 countries, allowing more accurate fracture predictions based on data from the country of interest. The FRAX tool is widely accepted for predicting fractures, but there have been suggestions as to how to improve it [5]. One important fracture risk factor that is not included is falls history; this was not included due to two reasons. First, uniform falls data were not available from all cohorts that contributed to FRAX and second, pharmaceutical intervention has not been shown to reduce fracture risk in those with a high falls risk [7]. For the latter, identification of those at high risk might only be useful if the risk can be reduced through some type of intervention. Falls strongly predict hip and other non-vertebral fracture independent of BMD and bone strength [5,8,9], and fallers are suggested to be at a higher fracture risk than FRAX may estimate [7]. Falls are common in the community-dwelling elderly occurring in 28–35% of 65 year olds and 42% of those over 75 years [10]. One method of measuring falls risk is the “Timed Up and Go Test” (TUG) which involves recording
2
K.L. Holloway et al. / Bone 76 (2015) 1–4
the time taken for an individual to rise from a chair, walk 3 m and return to sitting in the chair. The TUG test is mainly used for assessing the physical performance and functional mobility among elderly individuals, but it is also a good technique for determining individuals at a high risk of falls [11]. Falls risk assessment can be enhanced by incorporating more measurements into the assessment as in the Elderly Falls Screening Test (EFST) which includes prior falls, injurious falls, near falls, slow walking speed and gait [8]. There are recent data suggesting that the exclusion of falls risk from FRAX fracture risk assessments is not critical because the FRAX score is a surrogate for falls risk in elderly men [12]. Therefore, we hypothesised that falls risk is dependent on FRAX score. Thus, the aim of our study was to investigate the association of falls risk and Australian-specific FRAX 10-year probability scores from an Australian cohort study to determine if falls risk is predicted by FRAX score.
Prospective relationship between the FRAX score and the time to a fall We used Cox proportional hazards regression to estimate hazard ratios (HRs) for unit increases in each of the four FRAX scores. We presented the HRs as an increase in risk for a 1% increase in FRAX score. We chose 1% because “high risk” FRAX scores are considered to be ≥ 3% for hip and ≥20% for major osteoporotic fracture [14]. We determined that it would be more relevant to assess 1% increases in FRAX than for example, 5% or 10% because this would miss the cut-off for “high risk” hip fracture FRAX scores. Baseline hazards were stratified by 10 year age groups. We developed models that included age, sex and prior falls to allow adjustment for these variables [12]. Statistical analyses were performed using R version 3.0.2. Results Descriptive characteristics
Materials and methods Participants This current work incorporates data from the Geelong Osteoporosis Study (GOS) which includes residents form the Barwon Statistical Division (BSD), a region of south-eastern Australia [13]. This region has a number of factors that make it ideal for epidemiological studies including a large, stable population (n ~ 250,000) as well as multiple social, cultural and geographical settings. It is also representative of the Australian population. Further details of the study are provided elsewhere [13]. Participants were selected at random from Commonwealth Electoral Rolls. In Australia, voting is compulsory and therefore the electoral rolls are a comprehensive list of all adult Australians. Men were recruited from 2001 to 2006 (1540 recruited, 67% response). Women underwent baseline assessments from 1993 to 1997 (1494 recruited, 77% response). The data for this study come from the 5 and 6 year follow-up assessments for men (2006 to 2010) and the 6 and 7 year follow-up assessments for women (2000 to 2003). At the 5 year follow-up for men and 6 year follow-up for women, there were 735 and 602 individuals, aged 40 to 90 years, respectively. One year later, participants were sent questionnaires regarding falls and fractures that had occurred since their previous assessment; 473 men and 523 women completed these and returned them. A comparison of those who did and did not return the follow-up questionnaire revealed no differences in mean age, BMD, FRAX scores or prior fracture or falls (all p N 0.05, data not shown). The age of 5 year follow-up (for men) or 6 year follow-up (for women) was used for the cross-sectional analysis and as baseline age for the longitudinal analysis. Written, informed consent was obtained from all participants and this study was approved by the Barwon Health Human Research Ethics Committee. FRAX scores (hip and major osteoporotic fracture, with and without BMD; N = 4 scores) and Elderly Falls Screening Test (EFST) scores were calculated for all participants (see Appendix 1).
The descriptive characteristics of the participants are shown in Table 1. Those with a fall prior to baseline (5 year follow-up for men and 6 year follow-up for women) were older, lighter and shorter than those who did not fall prior to baseline. Prior fallers also had more prior fractures (45.3% vs 34.1%) and a higher proportion of rheumatoid arthritis (9.4% vs 3.9%). Individuals who fell during the 12 month followup were older, had a higher proportion of prior fractures (45.5% vs 35.1%) as well as prior falls (37% vs 20%). No other associations were found between the descriptive characteristics for prior and incident fall groups. Some differences in characteristics between men and women are noted. The median age for men and women was similar (men 62.2 IQR 52.2–74.3 years vs women 62.1 IQR 51.0, 73.0 years, p = 1.000), mean weight (83.9 ± 13.6 kg vs 70.1 ± 13.9 kg, p = 0.000), height (174.1 ± 6.9 cm vs 159.8 ± 6.3, p = 0.000), BMI (27.6 ± 4.0 kg/m2 vs 27.4 ± 5.3 kg/m2, p = 0.469), prior fracture (36.6% vs 36.9%, p = 0.916), history of parent hip fracture (6.1% vs 5.8%, p = 0.908) and current smoking (10.9 vs 10.5%, p = 0.805). Women had higher proportions of glucocorticoid use (4.2% vs 24.4%), rheumatoid arthritis (2.4 vs 8.5%) and secondary osteoporosis (9.8% vs 27.7%); men had a higher proportion of individuals with excessive alcohol intake (25.2% vs 2.7%) and a higher femoral neck BMD (0.965 ± 0.141 vs 0.898 ± 0.162 g/m2) (all p b 0.001). Cross-sectional analysis: EFST and FRAX There was a trend of increasing EFST scores with increasing FRAX scores, but the magnitude of the effect was small and unlikely to have clinical significance. After adjusting for the association of EFST scores on FRAX scores by age and sex, this relationship was not statistically significant (p N 0.2). This pattern was observed for all four FRAX scores investigated. The R-squared for the four interaction models were low; 0.196, 0.186, 0.194, and 0.186 for major osteoporotic fracture (with BMD), major osteoporotic fracture (without BMD), hip fracture (with BMD) and hip fracture (without BMD), respectively.
Statistical analyses
Longitudinal analysis: incident falls and FRAX
Cross-sectional association between the FRAX score and EFST The cross-sectional analysis aimed to investigate the association between a fracture risk score (FRAX) and a falls risk score (EFST). This was assessed using linear regression; an initial model was developed using the FRAX score as the independent variable and EFST as the dependent variable, adjusting for sex. We then fitted a further model adjusting for age (centred around age 60 years) and its interaction with sex, following the methods of Johansson et al. [12]. Beta coefficients were calculated, which represent the change in EFST for every 1% change in FRAX. QQ plots were used to assess normality of residuals, so that inference on regression coefficients was valid.
Among the 735 men at baseline, 132 (18%) reported a prior fall over the past 12 months. For the 602 women at baseline, 177 (29%) reported a prior fall over the past 12 months. During the subsequent 12 month period 78 out of 473 men (16%) and 110 out of 523 women (21%) reported falling. There were five men and 17 women who recorded two falls, and two men and five women who recorded three falls during the 12 month follow-up period. The unadjusted and adjusted hazard ratios for each of the four FRAX scores are presented in Table 2. In the unadjusted models, the risk of incident falls increased with increasing FRAX score. After adjustment for age and sex, the risk became non-significant. In the fully adjusted
K.L. Holloway et al. / Bone 76 (2015) 1–4
3
Table 1 Participant characteristics. Values expressed as median (interquartile range, IQR), mean ± SD or n (percentage). Participants were aged 40–90 years. Prior fall (baseline)a
Incident fall (over the 12 month FU)b
Risk factors
All (n = 1337)
Yes (n = 309)
No (n = 1028)
Yes (n = 189)
No (n = 807)
Age (years, median (IQR)) Weight (kg) Height (cm) BMI (kg/m2) Prior fracture Parent fractured hip Current smoking Glucocorticoid use Rheumatoid arthritis Secondary osteoporosis Excessive alcohol intake Femoral neck BMD (g/cm2) Prior falls
62.0 (51.7, 73.3) 77.7 ± 15.4 167.7 ± 9.7 27.6 ± 4.6 491 (36.7) 80 (6.0) 143 (10.7) 178 (13.3) 69 (5.2) 239 (17.9) 201 (15.0) 0.935 ± 0.155 309 (23.1)
68.0 (56.7, 77.5) c 75.9 ± 15.0 c 165.2 ± 9.8 c 27.8 ± 4.9 140 (45.3) c 20 (6.5) 29 (9.4) 49 (15.9) 29 (9.4) c 66 (21.4) 38 (12.3) 0.908 ± 0.164 c –
60.3 (50, 70)c 78.2 ± 15.4 c 168.4 ± 9.6 c 27.5 ± 4.5 351 (34.1) c 60 (5.8) 114 (11.1) 129 (12.5) 40 (3.9) c 173 (16.8) 163 (15.9) 0.943 ± 0.151 c –
66.0 (54.0, 76.5)d 76.4 ± 15.5 165.7 ± 9.7 27.8 ± 4.9 86 (45.5) d 12 (6.3) 12 (6.3) 32 (16.9) 15 (7.9) 41 (21.7) 26 (13.8) 0.914 ± 0.171 70 (37.0) d
60.8 (52, 72) d 76.4 ± 15.3 166.8 ± 9.5 27.4 ± 4.7 283 (35.1) d 51 (6.3) 86 (10.7) 111 (13.8) 44 (5.5) 151 (18.7) 99 (12.3) 0.939 ± 0.153 161 (20.0) d
a b c d
Baseline corresponds to the 5 year FU for men and 6 year FU for women from the Geelong Osteoporosis Study. Prior falls were counted if they occurred before this baseline visit. The 12 month FU occurred after the baseline visit and corresponds to the 6 year FU for men and 7 year FU for women. Significant difference (p b 0.05) between the prior falls and no prior falls groups. Significant difference (p b 0.05) between the incident falls and no incident falls groups.
models there were independent contributions from age, sex and prior falls. No interaction terms were identified. The Harrell's C values were 0.616, 0.613, 0.615, and 0.616 for major osteoporotic fracture (with BMD), major osteoporotic fracture (without BMD), hip fracture (with BMD) and hip fracture (without BMD), respectively. Discussion In this study, we report a clinically non-significant cross-sectional association between FRAX and EFST score. This relationship became statistically non-significant when adjusted for age and sex. Additionally, there was a weak association between FRAX and incident falls that became non-significant after adjusting for age and sex. A meeting involving a joint Task Force including members from the International Osteoporosis Foundation (IOF) and the International Society for Clinical Densitometry (ISCD) during 2010 resulted in the development of resource documents that present recommendations on how to improve FRAX and better inform clinicians who use the tool [15]. It was considered that exclusion of falls risk from FRAX could result in underestimation of fracture risk in individuals with recurrent falls [7]. However, the discussion also included reasons why falls risk had not been incorporated into FRAX to-date. Reasons included a lack of data regarding falls, lack of information about interaction of falls with other FRAX variables, lack of evidence to suggest that pharmaceutical intervention can reduce falls, heterogeneity in falls questions, variability in falls between populations and that falls risk is inherently
taken into account in the FRAX algorithm, though not as an input variable [7,15]. Including falls in FRAX would not help identify those who could benefit from pharmaceutical therapies, since there is no evidence that drugs can reduce high fracture risks associated with falling. There are options available to help reduce risk from falling (e.g. reduce psychotropic medications, exercise (to improve balance and stability), home modifications, management of fainting/dizziness [16]), though these are not the same types of therapies that would be effective for preventing fractures associated with low BMD and represents a challenge for the reduction of fracture risk. There is evidence from previous studies to include falls in fracture risk assessments [10] and also studies that suggest that inclusion of additional risk factors does not improve the predictive ability of FRAX or other models [17–19]. This seems paradoxical if FRAX does not predict falls, then it would be expected that the addition of falls would improve fracture prediction. The reason for this could be that both age and sex, which were predictors of falls risk in this study, are already included in FRAX and other fracture prediction tools [16]. Since FRAX did not predict falls well, it should not be used to determine falls risk. Other fracture risk tools have been developed that incorporate falls risk, including the Garvan Fracture Risk Calculator (FRC) [20] and the Geelong Osteoporosis Study FRISK Score [9] and may be more suitable for fracture prediction when an individual has a high falls risk. The challenges of prevention of fracture still remain, however. The major strength of this study is that participants were randomly selected from the general population. Additionally, we recruited men
Table 2 Estimated hazard ratios (95% CIs) of an absolute increase of one percentage point in FRAX risks for various FRAX risk calculations. FRAX score type Major osteoporotic fracture, with BMD Past falls Female FRAX score Major osteoporotic fracture, without BMD Past falls Female FRAX score Hip fracture, with BMD Past falls Female FRAX score Hip fracture, without BMD Past falls Female FRAX score
Unadjusted hazard ratio (95% CI)
Adjusted (sex and age) hazard ratio (95% CI)
Adjusted (sex, age and prior falls) hazard ratio (95% CI)
1.03 (1.02, 1.04)
1.39 (1.05, 1.83) 1.01 (0.98, 1.03)
1.64 (1.27, 2.13) 1.30 (0.98, 1.72) 1.01 (0.98, 1.03)
1.02 (1.01, 1.03)
1.45 (1.09, 1.93) 1.00 (0.98, 1.01)
1.63 (1.26, 2.12) 1.36 (1.02, 1.82) 1.00 (0.98, 1.01)
1.04 (1.02, 1.07)
1.41 (1.08, 1.83) 1.01 (0.97, 1.04)
1.64 (1.27, 2.13) 1.31 (1.00, 1.72) 1.01 (0.97, 1.05)
1.02 (1.01, 1.04)
1.45 (1.10, 1.89) 0.99 (0.97, 1.02)
1.63 (1.26, 2.11) 1.35 (1.02, 1.78) 0.99 (0.97, 1.02)
4
K.L. Holloway et al. / Bone 76 (2015) 1–4
and women over the entire age range covered by FRAX (40–90 years), whereas many previous studies have focussed on one sex within a narrower age range (usually above 50 years). We also used a detailed falls risk assessment [8] including five different criteria to assess falls risk, which would be more informative than a single question regarding falls in the past 12 months. This falls risk assessment has been reviewed in several meta-analyses [21–23] and has sufficient sensitivity and specificity for predicting fallers and non-fallers [24]. However, we recognised that our study has some limitations. The majority of the study participants were Caucasian (99%) and this limits the generalisibility of the findings to other populations. Finally, we relied on some self-reported fracture and falls data in order to calculate FRAX and EFST scores; however, self-reported fractures have previously been shown to be reliable [25–27]. Furthermore, we acknowledge limitations in relying on self-reported falls dates. Conclusions There is a weak positive correlation between FRAX and falls risk score, that is likely explained by the inclusion of age and sex in the FRAX model. These data suggest that FRAX score may not be a robust surrogate for falls risk and that inclusion of falls in fracture risk assessment should be further explored. Acknowledgments The study was funded by grants from the National Health and Medical Council (NHMRC 251638, 299831, 628582), The University of Melbourne Research Grant Scheme, American Society for Bone and Mineral Research (ASBMR), Perpetual Trustees, Amgen (Europe) GmBH and the Geelong Region Medical Research Foundation. Authors' roles: Study design: KLH, MAK, and JAP. Study conduct: JAP. Data analysis: SEL. Data interpretation: KLH, MAK, SEL, SLB, and JAP. Drafting manuscript: KLH. Revising manuscript content: KLH, MAK, SEL, SLB, and JAP. Approving final version of manuscript: KLH, MAK, SEL, SLB, and JAP. SEL takes responsibility for the integrity of the data analysis. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.bone.2015.03.004. References [1] Watts JJ, Abimanyi-Ochom J, Sanders KM. Osteoporosis costing all Australians: a new burden of disease analysis—2012 to 2022. Australia: osteoporosis Australia; 2013. [2] Pasco JA, Sanders KM, Hoekstra FM, Henry MJ, Nicholson GC, Kotowicz MA. The human cost of fracture. Osteoporos Int 2005;16:2046–52. [3] Otmar R, Kotowicz MA, Brennan SL, Bucki-Smith G, Korn S, Pasco JA. Personal and psychosocial impacts of clinical fracture in men. J Mens Health 2013;10:22–7.
[4] Pasco JA, Seeman E, Henry MJ, Merriman EN, Nicholson GC, Kotowicz MA. The population burden of fractures originates in women with osteopenia, not osteoporosis. Osteoporos Int 2006;17:1404–9. [5] Bauer D. FRAX, falls, and fracture prediction: predicting the future. Arch Intern Med 2011;171:1661–2. [6] World Health Organization Collaborating Centre for Metabolic Bone Diseases University of Sheffield UK. FRAX ® WHO fracture risk assessment tool; 2011. [7] Masud T, Binkley N, Boonen S, Hannan MT. Official Positions for FRAX® clinical regarding falls and frailty: can falls and frailty be used in FRAX®? From Joint Official Positions Development Conference of the International Society for Clinical Densitometry and International Osteoporosis Foundation on FRAX®. J Clin Densitom 2011;14:194–204. [8] Cwikel J, Fried A, Biderman A, Galinsky D. Validation of a fall-risk screening test, the Elderly Fall Screening Test (EFST), for community-dwelling elderly. Disabil Rehabil 1998;20:161–7. [9] Henry MJ, Pasco JA, Sanders KM, Nicholson GC, Kotowicz MA. Fracture risk (FRISK) score: Geelong Osteoporosis Study. Radiology 2006;241:190–6. [10] Edwards MH, Jameson K, Denison H, Harvey NC, Sayer AA, Dennison EM, et al. Clinical risk factors, bone density and fall history in the prediction of incident fracture among men and women. Bone 2013;52:541–7. [11] Zhu K, Devine A, Lewis J, Dhaliwal S, Prince R. “Timed up and go” test and bone mineral density measurement for fracture prediction. Arch Intern Med 2011;171:1655–61. [12] Johansson H, Odén A, Karlsson M, Rosengren B, Ljunggren O, McCloskey E, et al. FRAX® predicts future falls in elderly men. MrOs Sweden. J Bone Miner Res 2013; 28(Suppl. 1). [13] Pasco JA, Nicholson GC, Kotowicz MA. Cohort profile: Geelong Osteoporosis Study. Int J Epidemiol 2012;41:1565–75. [14] Unnanuntana A, Gladnick BP, Lane EDJM. The assessment of fracture risk. J Bone Joint Surg 2010;92:743–53. [15] Kanis JA, Hans D, Cooper C, Baim S, Bilezikian JP, Binkley N, et al. Interpretation and use of FRAX in clinical practice. Osteoporos Int 2011;22:2395–411. [16] Hill K, Schwarz J. Assessment and management of falls in older people. Intern Med J 2004;34:557–64. [17] Sambrook PN, Flahive J, Hooven FH, Boonen S, Chapurlat R, Lindsay R, et al. Predicting fractures in an international cohort using risk factor algorithms without BMD. J Bone Miner Res 2011;26:2770–7. [18] Melton III LG, Crowson CS, O'Fallon WM. Fracture incidence in Olmsted county, Minnesota: comparison of urban with rural rates and changes in urban rates over time. Osteoporos Int 1999;9:29–37. [19] Melton III J, Atkinson EJ, Achenbach SJ, Kanis JA, Therneau TM, Johansson H, et al. Potential extensions of the US FRAX algorithm. J Osteoporos 2012. http://dx.doi.org/10. 1155/2012/528790. [20] Institute Garvan. Bone Fracture Risk Calculator - Garvan Institute of Medical Research; 2008. [21] Kehinde J. Instruments for measuring fall risk in older adults living in long-term care facilities. J Gerontol Nurs 2009;35:46–55. [22] Perell KL, Nelson A, Goldman RL, Luther SL, Prieto-Lewis N, Rubenstein LZ. Fall risk assessment measures: an analytic review. J Gerontol A 2001;56:M761–6. [23] Scott V, Votova K, Scanlan A, Close J. Multifactorial and functional mobility assessment tools for fall risk among older adults in community, home-support, longterm and acute care settings. Age Ageing 2007;36:130–9. [24] Oliver D, Daly F, Martin F, McMurdo M. Risk factors and risk assessment tools for falls in hospital in-patients: a systematic review. Age Ageing 2004;33:122–30. [25] Chen Z, Kooperberg C, Pettinger M, Bassford T, Cauley J, LaCroix A, et al. Validity of self-report for fractures among a multiethnic cohort of postmenopausal women: results from the Women's Health Initiative observational study and clinical trials. J North Am Menopause Soc 2004;11:264–74. [26] Gehlbach S, Saag KG, Adachi JD, Hooven FH, Flahive J, Boonen S, et al. Previous fractures at multiple sites increase the risk for subsequent fractures: the global longitudinal study of osteoporosis in women. J Bone Miner Res 2012;27:645–53. [27] Nevitt MC, Cummings SR, Browner WS, Seeley DG, Cauley JA, Vogt TM, et al. The accuracy of self-report of fractures in elderly women: evidence from a prospective study. Am J Epidemiol 1992;135:490–9.
