Access the full text.
Sign up today, get DeepDyve free for 14 days.
Loading next page...
/lp/springer-journals/ireland-dxa-frax-may-differ-significantly-and-substantially-to-web-UjHjrss2XJ
- Publisher
- Springer Journals
- Copyright
- Copyright © The Author(s) 2023
- ISSN
- 1862-3522
- eISSN
- 1862-3514
- DOI
- 10.1007/s11657-023-01232-y
- Publisher site
- See Article on Publisher Site
Abstract
Summary Appropriate use of FRAX reduces the number of people requiring DXA scans, while contemporaneously deter- mining those most at risk. We compared the results of FRAX with and without inclusion of BMD. It suggests clinicians to carefully consider the importance of BMD inclusion in fracture risk estimation or interpretation in individual patients. Purpose FRAX is a widely accepted tool to estimate the 10-year risk of hip and major osteoporotic fracture in adults. Prior calibration studies suggest this works similarly with or without the inclusion of bone mineral density (BMD). The purpose of the study is to compare within-subject differences between FRAX estimations derived using DXA and Web software with and without the inclusion of BMD. Method A convenience cohort was used for this cross-sectional study, consisting of 1254 men and women aged between 40 and 90 years who had a DXA scan and complete validated data available for analysis. FRAX 10-year estimations for hip and major osteoporotic fracture were calculated using DXA software (DXA-FRAX) and the Web tool (Web-FRAX), with and without BMD. Agreements between estimates within each individual subject were examined using Bland–Altman plots. We performed exploratory analyses of the characteristics of those with very discordant results. Results Overall median DXA-FRAX and Web-FRAX 10-year hip and major osteoporotic fracture risk estimations which include BMD are very similar: 2.9% vs. 2.8% and 11.0% vs. 11% respectively. However, both are signic fi antly lower than those obtained without BMD: 4.9% and 14% respectively, P < 0.001. Within-subject differences between hip fracture estimates with and without BMD were < 3% in 57% of cases, between 3 and 6% in 19% of cases, and > 6% in 24% of cases, while for major osteoporotic fractures such differences are < 10% in 82% of cases, between 10 and 20% in 15% of cases, and > 20% in 3% of cases. Conclusions Although there is excellent agreement between the Web-FRAX and DXA-FRAX tools when BMD is incorpo- rated, sometimes there are very large differences for individuals between results obtained with and without BMD. Clinicians should carefully consider the importance of BMD inclusion in FRAX estimations when assessing individual patients. Keywords Osteoporosis · Fracture Risk · Bone Mineral Density · FRAX * Lan Yang School of Management, Guangxi Minzu University, Nanning, lan.yang@universityofgalway.ie China Nuffield Department of Medicine, University of Oxford, Insight SFI Research Centre for Data Analytics, Data Oxford, UK Science Institute, University of Galway, IDA Business Park, Lower Dangan, Galway H91 AEX4, Ireland Department of Rheumatology, Galway University Hospitals, Galway, Ireland School of Engineering, College of Science and Engineering, University of Galway, Galway, Ireland Department of Industrial Engineering, Tsinghua University, Beijing, China School of Computer Science, College of Science and Engineering, University of Galway, Galway, Ireland School of Medicine, College of Medicine, Nursing and Health Sciences, University of Galway, Galway, Ireland Vol.:(0123456789) 1 3 43 Page 2 of 12 Archives of Osteoporosis (2023) 18:43 undertaking such exposures if alternative methods are Background available which can achieve the same objectives [31]. Clinical risk factors included in FRAX are prevalent in Osteoporosis and the associated fractures are a major Irish adults, including those who are hospitalized and those global health burden for patients, their social network, with and without fractures [17, 32–34]. FRAX probabili- and society [1–3]. Ireland has one of the greatest illness ties and potential intervention thresholds for Ireland were burdens, and the highest projected increases in osteo- derived using a limited data set of public hospital admis- porotic fractures in Europe over the coming decade [3, 4]. sions, population statistics, and several assumptions, though National publications suggest that available data reflect neither individual-level data nor DXA results were available the ‘tip of the iceberg,’ and the financial costs of managing [4]. Resulting software exists to calculate fracture risk esti- people with these fractures will double during this decade, mates either via the FRAX website (www. she ffield. ac. uk/ rising to €2billion by 2030 [5–7]. Recent programs estab- FRAX/), or on most modern DXA machines. Sometimes we lished national standards for the management and audit of note discordance between FRAX estimates in correspond- hip fracture care among adults aged ≥ 60 years of age and ence we receive and those derived from our DXA machines fracture liaison services [8, 9]. These reflect the current which include BMD. state of fragility fracture care, variation in osteoporosis The DXA-HIP cohort was established to examine and val- diagnosis, risk assessment and management, some pro- idate international DXA criteria and osteoporosis diagnostic gress, while also highlighting substantial needs including and prediction algorithms for Irish adults [32]. In order to increases in resources, data, policy, priority, and logistics understand the importance of BMD inclusion when calcu- [3, 5–10]. lating FRAX probabilities for our population, we compare Access to quality risk assessment, diagnosis, and treat- the agreement between Web-based and DXA-based FRAX ment of osteoporosis is heterogeneous and inadequate derivatives for Ireland. around the globe [1–3, 11–17]. Many algorithms are available to decide whom to test, how to assess fracture risk, when to intervene, and how to monitor the effects of Methods interventions [1–3, 12, 15, 16, 18–21]. Their performance varies considerably among different populations, with no Details of the entire DXA-HIP cohort have been described single method substantially superior to others [18–22]. [25, 32]. In brief, a convenience cohort was established The Osteoporosis Self-assessment Tool (OST) is one of for clinical research using DXA data from 3 centers which the simplest algorithms requiring only age and weight to include 4 GE-Lunar Prodigy DXA machines, using G.E. aid in the identification of people likely to have a DXA Encore software version 17. Femoral neck T-scores for diagnosis of osteoporosis [18–23]. Appropriate use of the men and women are generated using NHANES III ISCD- OST index could significantly reduce the number of people recommended calculations [35]. All scans are performed requiring a DXA screening test [21–25]. and reported by staff trained to ISCD standards and rec- Fracture risk may be estimated using various methods, ommendations. The staff have regular weekly meetings to each with strengths and limitations [2, 18–22]. Substantial discuss discrepancies, complex cases and audit procedures, efforts over many years supported the development of the performance, and reports. The collection and analysis of the FRAX tool such that it has become the dominant fracture data for the DXA-HIP project were approved by our Institu- risk assessment tool worldwide [26] and is the preferred tions Ethics Committee and in compliance with G.D.P.R. algorithm of global professional bodies in skeletal health regulations [25, 32, 36]. In this study, due to the inherent [2, 3, 26–29]. FRAX estimates the 10-year probability of thresholds in the FRAX tool [26], we include Caucasian hip fracture (HF) and certain major osteoporotic fractures subjects aged between 40 and 90 years of age with weight (MOF) in people aged 40 to 90 years, with country-spe- less than 125 kg. cific estimate options [26, 29]. Strengths of the FRAX Preliminary data for this study were supplied from 1 algorithm include the availability of an online calcula- clinical site between June and December 2021. Data were tion option, with or without DXA testing, and the ability collected and compiled when auditing our DXA-FRAX esti- to include additional risk factors such as glucocorticoids, mates at the time of scanning and reporting (G.E. Lunar secondary causes, and a parental history of hip fracture FRAX estimates for Ireland, version 3.8). Contemporane- [26, 29]. This is particularly attractive when access to ously, we derived Web-FRAX estimates for these same men DXA is limited [14, 20, 26, 29], such as in our center, and women from the FRAX website for Ireland (https:// where reducing unnecessary testing is important [25, 30]. www.she ffield.ac. uk/ FRAX/ , version 4, country code = 48), Irish legislation governing the justification of the risks with and without femoral neck BMD values in g/cm for GE associated with exposure to ionizing radiation prohibits 1 3 Archives of Osteoporosis (2023) 18:43 Page 3 of 12 43 Lunar. All data were subsequently rechecked on 2 further Table 1 Characteristics of study subjects occasions by 2 of the investigators, and FRAX estimates Female Male P value were recalculated, to ensure the information being used was Number of subjects 964 290 accurate, consistent, and complete. The data were merged, Mean age in years (SD) 68.5 (9.7) 69.2 (11.4) 0.38 anonymized, and stored for analysis. Mean weight in kg (SD) 69.1 (14.4) 83.3 (15.4) < 0.001 We chose to compare FRAX estimates between DXA- Mean height in m (SD) 1.6 (0.1) 1.7 (0.1) < 0.001 FRAX and Web-FRAX with and without the inclusion of Mean body mass index (kg/m ) (SD) 27.1 (5.4) 28.0 (4.8) < 0.001 femoral neck BMD for both men and women. In order to Prevalent fracture, N (%) 458 (47.5) 141 (48.6) 0.79 highlight the extent and magnitude of the difference between Parent fractured hip, N (%) 71 (7.4) 6 (2.1) < 0.001 various FRAX estimations, we show the proportion of men Current smokers, N (%) 91 (9.4) 33 (11.4) 0.39 and women whose difference between DXA-FRAX with Current glucocorticoid use, N (%) 132 (13.7) 116 (40.0) < 0.001 BMD and their corresponding Web-FRAX without BMD Rheumatoid arthritis, N (%) 106 (11.0) 32 (11.0) 1.00 HF which were: < 3%, between 3 and 6%, and > 6%, and Secondary osteoporosis, N (%) 344 (35.7) 103 (35.5) 1.00 MOF which were < 10%, between 10 and 20%, and > 20%. Alcohol ≥ 3 units/day, N (%) 5 (0.5) 10 (3.4) < 0.001 We used box and whisker plots and Bland–Altman plots Mean femoral neck T-score (SD) − 1.6 (0.8) − 1.0 (1.1) < 0.001 to assess the overall and within-person differences between Web-FRAX with and without BMD and DXA-FRAX and the patterns of bias. We used paired T-tests, Chi-squared tests, Fisher’s exact tests, and Wilcoxon Rank Sum tests to compare means and medians as appropriate. All analy- ses were planned ad hoc. All analyses were performed on BMD or DXA-FRAX scores for both men and women, and both HF and MOF, shown in Table 2, though for individu- Python 3.6. We performed sensitivity analyses by excluding those whose prior fracture site was unknown, and for those als they were sometimes lower, Figs. 3 and 4. Differences between DXA-FRAX and Web-FRAX with BMD were very with multiple prior fragility fractures. small and not statistically significant (P values HF: 0.914, MOF: 0.967) shown in Appendix Figs. 5 and 6, whereas the Results differences between DXA-FRAX and Web-FRAX without BMD as well as the differences between Web-FRAX with A total of 2090 records were collected during an audit of BMD and Web-FRAX without BMD are sometimes large *** and were statistically significant, P < 0.001 (Figs. 3 and vertebral fracture assessment (VFA) scans between 2019 and 2021 including patients’ demographic information such 4). The prevalence of hip fracture and major osteoporotic fracture for women is 3.4% for HF and 47.5% for MOF, as age, gender, weight, height, and BMI; risk factors such as previous fracture and femoral neck BMD; and results of while for men is 6.6% for HF and 48.6% for MOF. In contrast, we found substantial differences within indi- DXA-FRAX estimations. Subjects with missing or incom- plete information were excluded. Complete data on 1254 viduals when we compared the absolute difference in HF and MOF between their DXA-FRAX and their Web-FRAX scores adults aged between 40 and 90 years were available for this study, including 290 (23.1%) men and 964 (76.9%) women. without BMD, particularly among those with higher scores, as shown in the Bland–Altman plots in Figs. 3 and 4. How- A summary of patient details including the variables used in FRAX calculations is shown in Table 1, broken down ever, for those with low scores: HF < 5% and MOF < 10%, the absolute die ff rences were generally small. A similar pattern by gender. Women are significantly lighter and shorter than men, had lower BMD, and were less likely to take corti- was noted when we compared Web-FRAX scores with BMD to Web-FRAX scores without BMD, data not shown. The dif- costeroids or drink excessively, but more likely to have a parent who had a hip fracture. Almost half of the men and ferences observed between DXA-FRAX and the correspond- ing Web-FRAX with BMD estimates for HF and MOF were women had a previous MOF, while almost 36% have another disorder strongly associated with osteoporosis such as early generally very small, with limits of agreement of < 1% and maximum differences of < 2.4%, shown in Appendix Figs. 5 menopause, diabetes mellitus, or coeliac disease. Women had higher FRAX scores than men using all 3 and 6. Table 3 presents the breakdown of the proportion of individuals with various differences between DXA-FRAX calculation methods, for both HF and MOF, shown in Figs. 1 and 2. The majority of men and women had DXA-FRAX HF and Web-FRAX scores without BMD, HF: < 3%, 3 to 6%, and > 6%, and MOF: < 10%, 10–20%, and > 20% difference. scores below 5% and MOF less than 20%. A small number of female patients have very high scores (> 50%) for both HF These show a greater proportion of women have larger abso- lute differences > 10% than men for MOF. However, 43% and MOF. Overall, Web-FRAX scores without BMD were significantly higher, P < 0.001, than Web-FRAX scores with of the patients have an absolute difference in HF of > 3%. 1 3 43 Page 4 of 12 Archives of Osteoporosis (2023) 18:43 Fig. 1 Box and whisker plots comparing DXA-FRAX, Web-FRAX with BMD, and Web-FRAX without BMD for MOF by gender. Note: the * ** *** significance is reported for the following levels: ns: not significant; P < 0.05, P < 0.01, and P < 0.001 Fig. 2 Box and whisker plots comparing DXA-FRAX, Web-FRAX with BMD, and Web-FRAX without BMD for HF by gender. Note: the sig- * ** *** nificance is reported for the following levels: ns: not significant; P < 0.05, P < 0.01, and P < 0.001 1 3 Archives of Osteoporosis (2023) 18:43 Page 5 of 12 43 Table 2 Comparison of DXA- Gender Fracture risk DXA-FRAX Web-FRAX with BMD Web-FRAX without BMD FRAX and Web-FRAX scores with and without BMD for Female Median MOF (IQR) 12.1 (7.9–18.1) 12.0 (7.8–18.0) 16.0 (9.7–26.0) MOF and HF Median HF (IQR) 2.9 (1.0–6.0) 2.9 (1.1–6.0) 5.7 (2.2–12.0) Male Median MOF (IQR) 8.0 (5.3–11.1) 7.8 (5.4–11.0) 10.0 (6.5–14.0) Median HF (IQR) 2.8 (1.0–4.7) 2.6 (1.1–4.6) 3.8 (1.7–7.3) All Median MOF (IQR) 11.0 (7.0–16.2) 11.0 (7.0–16.0) 14.0 (8.5–23.0) Median HF (IQR) 2.9 (1.0–5.5) 2.8 (1.1– 5.5) 4.9 (2.0–10.0) Fig. 3 Bland–Altman plots comparing DXA-FRAX to Web-FRAX without BMD for HF for by gender Moreover, 28% of females and 14% of males have an absolute respectively, whose differences between their DXA-FRAX and difference in HF estimates of > 6%. The range of differences Web-FRAX scores were greater than or less than the limits for women is 0–39.2% for MOF and 0–45.3% for HF while of agreement derived from our Bland–Altman results (Figs. 3 for men is 0–17.0% for MOF and 0–14.5% for HF. and 4). Women with more extreme differences were older and Table 4 summarizes the characteristics of people whose lighter and had lower BMI and BMD; a greater prevalence of DXA-FRAX and corresponding Web-FRAX without BMD fractures, rheumatoid arthritis, and glucocorticoid use; and a scores differ by a small, moderate, or large amount. Frac- much higher or lower prevalence of secondary osteoporosis, tures, secondary osteoporosis, and rheumatoid arthritis were tobacco use, or a parent with a previous hip fracture. Men with more prevalent among those with larger differences, who are more extreme differences were similarly lighter and had a lower also older, lighter, and have lower BMI. In Tables 5 and 6, BMI, a greater prevalence of fractures and excessive alcohol we summarize the characteristics of those women and men, use, a lower prevalence of parents with a hip fracture, a higher 1 3 43 Page 6 of 12 Archives of Osteoporosis (2023) 18:43 Fig. 4 Bland–Altman plots comparing DXA-FRAX to Web-FRAX without BMD for MOF by gender Table 3 Proportion of Gender Hip fracture Major osteoporotic fracture individuals with various differences in HF and MOF |x|< 3% 3 ≤|x|< 6% |x|≥ 6% |x|< 10% 10% ≤|x|< 20% |x|> 20% between DXA-FRAX and Web- Female 524 (54%) 174 (18%) 266 (28%) 757 (78%) 170 (18%) 37 (4%) FRAX without BMD scores Male 190 (65%) 60 (21%) 40 (14%) 275 (95%) 15 (5%) 0 (0%) All 714 (57%) 234 (19%) 306 (24%) 1032 (82%) 185 (15%) 37 (3%) x indicates DXA-FRAX score minus Web-FRAX without BMD score or lower age, BMD, and prevalence of smoking, glucocorticoid estimations without BMD to estimations that included use, rheumatoid arthritis, and secondary osteoporosis. BMD, there were notable differences for some individuals or extreme cases which at times were quite large, up to 40% absolute difference for major osteoporotic fracture and 46% Discussion absolute difference for hip fracture, shown in Figs. 3 and 4. Such differences are more likely to be observed at extremes In this paper comparing different FRAX calculations in of weight, BMI, BMD, or prevalence of rheumatoid arthritis older Irish men and women, we found excellent agreement or secondary causes of osteoporosis, as well as where frac- between the Web version and the DXA version when femo- tures or glucocorticoid use is present. ral neck BMD was included. However, when we compared 1 3 Archives of Osteoporosis (2023) 18:43 Page 7 of 12 43 Table 4 Characteristics of individuals with various differences in HF and MOF between DXA-FRAX and Web-FRAX without BMD scores Hip fracture Major osteoporotic fracture |x|< 3% 3 ≤|x|< 6% |x|≥ 6% |x|< 10% 10% ≤|x|< 20% |x|> 20% Number of subjects 713 233 308 1031 186 37 Mean age in years (SD) 64.0 (9.0) 72.3 (7.4) 76.7 (7.9) 66.8 (9.7) 76.7 (6.9) 80.5 (4.9) Mean weight in kg (SD) 76.4 (16.2) 71.0 (13.2) 64.2 (13.1) 74.4 (15.8) 63.9 (12.7) 60.0 (10.0) Mean height in m (SD) 1.6 (0.1) 1.6 (0.1) 1.6 (0.1) 1.6 (0.1) 1.6 (0.1) 1.6 (0.1) Mean BMI in kg/m (SD) 28.4 (5.3) 26.9 (4.5) 25.0 (5.0) 27.8 (5.2) 25.3 (4.9) 23.9 (3.9) Prevalent fracture, N (%) 259 (36.3) 121 (51.9) 219 (71.1) 439 (42.6) 124 (66.7) 36 (97.3) Parent fractured hip, N (%) 44 (6.2) 14 (6.0) 19 (6.2) 60 (5.8) 11 (5.9) 6 (16.2) Current smokers, N (%) 67 (9.4) 24 (10.3) 33 (10.7) 102 (9.9) 15 (8.1) 7 (18.9) Current glucocorticoid use, N (%) 141 (19.8) 45 (19.3) 62 (20.1) 204 (19.8) 39 (21.0) 5 (13.5) Rheumatoid arthritis, N (%) 63 (8.8) 30 (12.9) 45 (14.6) 105 (10.2) 24 (12.9) 9 (24.3) Secondary osteoporosis, N (%) 201 (28.2) 98 (42.1) 148 (48.1) 331 (32.1) 93 (50.0) 23 (62.2) Alcohol ≥ 3 units/day, N (%) 3 (0.4) 7 (3.0) 5 (1.6) 12 (1.2) 2 (1.1) 1 (2.7) Mean femoral neck T-score (SD) − 1.3 (0.9) − 1.5 (0.9) − 1.6 (1.0) − 1.4 (0.9) − 1.6 (1.0) − 1.2 (0.8) x indicates DXA-FRAX score minus Web-FRAX without BMD score Table 5 Comparison of extreme Hip fracture Major osteoporotic f09rac- differences between DXA- ture FRAX and Web-FRAX for women x > 8.82% x < − 16.88% x > 9.71% x < − 9.25% Number of subjects 9 44 12 42 Mean age in years (SD) 70.3 (9.1) 79.8 (5.3) 69.4 (7.9) 80.2 (5.1) Mean weight in kg (SD) 60.5 (10.3) 55.4 (9.3) 70.5 (22.8) 59.8 (10.5) Mean height in m (SD) 1.6 (0.1) 1.6 (0.1) 1.6 (0.1) 1.6 (0.1) Mean BMI in kg/m (SD) 24.8 (4.7) 22.1 (3.1) 29.0 (9.8) 23.7 (4.0) Prevalent fracture, N (%) 8 (88.9) 36 (81.8) 10 (83.3) 39 (92.9) Parent fractured hip, N (%) 0 (0) 11 (25.0) 1 (8.3) 7 (16.7) Current smokers, N (%) 0 (0) 8 (18.2) 1 (8.3) 8 (19.0) Current glucocorticoid use, N (%) 2 (22.2) 14 (31.8) 3 (25.0) 8 (19.0) Rheumatoid arthritis, N (%) 2 (22.2) 11 (25.0) 3 (25.0) 8 (19.0) Secondary osteoporosis, N (%) 2 (22.2) 22 (50.0) 1 (8.3) 27 (64.3) Alcohol ≥ 3 units/day, N (%) 0 (0) 1 (2.3) 0 (0) 1 (2.4) Mean femoral neck t-score (SD) -3.5 (0.3) -1.6 (0.8) -3.3 (0.4) -1.3 (0.7) (1) extreme differences: indicates subjects whose FRAX estimates differ by values which exceed (above or below) the limits of agreement line on the Bland–Altman plots. (2) x indicates DXA-FRAX score minus Web-FRAX without BMD score FRAX is a clinical tool designed to improve the estima- group, but only 28,660 (12.4%) of the validation group [37]. tion of fracture risk by combining some of the most impor- This algorithm is now available in a number of formats, tant determinants in a multivariate algorithm, which should including a web-based calculator and a DXA-based calcula- st be more robust than using any single factor [29, 37–40]. tor [39, 41]. The current web-based tool (31 October 2022) Additionally, the importance of using absolute risk rather includes 87 populations: 18 Asian populations, 36 European than relative risk or a single BMD threshold is an important populations, 19 Middle East and African Populations, 5 advance [26, 29]. The original algorithm was derived from 9 North American populations, 7 Latin American populations, cohorts including 46,340 men (32%) and women (68%) and and 2 Oceania populations, while the DXA-based tool has validated across 11 cohorts including 230,486 men (< 1%) 58 available populations to select from. Previous attempts to and women (> 99%) from 23 countries across the globe [37]. calibrate FRAX for Ireland used national hip fracture esti- BMD was available for 37,305 (80.5%) of the development mates but no patient-level data, and the authors note the 1 3 43 Page 8 of 12 Archives of Osteoporosis (2023) 18:43 Table 6 Comparison of extreme Hip fracture Major osteoporotic fracture differences between DXA- FRAX and Web-FRAX for men x > 5.65% x < − 8.60% x > 6.98% x < − 10.15% Number of subjects 11 10 11 11 Mean age in years (SD) 59.5 (8.8) 82.2 (3.2) 61.9 (7.4) 80.0 (4.5) Mean weight in kg (SD) 70.0 (15.6) 67.5 (12.8) 73.8 (15.4) 69.9 (12.2) Mean height in m (SD) 1.8 (0.1) 1.7 (0.1) 1.8 (0.1) 1.7 (0.1) Mean BMI in kg/m (SD) 22.6 (4.4) 23.2 (3.1) 24.0 (4.9) 23.9 (3.3) Prevalent fracture, N (%) 10 (90.9) 5 (50.0) 10 (90.9) 7 (63.6) Parent fractured hip, N (%) 0 (0) 0 (0) 0 (0) 0 (0) Current smokers, N (%) 3 (27.3) 0 (0) 3 (27.3) 0 (0) Current glucocorticoid use, N (%) 3 (27.3) 7 (70.0) 2 (18.2) 6 (54.5) Rheumatoid arthritis, N (%) 0 (0) 3 (30.0) 0 (0) 2 (18.2) Secondary osteoporosis, N (%) 3 (27.3) 7 (70.0) 2 (18.2) 8 (72.7) Alcohol ≥ 3 units/day, N (%) 3 (27.3) 1 (10.0) 3 (27.3) 1 (9.1) Mean femoral neck T-score (SD) − 2.8 (0.4) − 0.4 (1.7) − 2.8 (0.4) − 0.0 (1.3) (1) extreme differences: indicates subjects whose FRAX estimates differ by values which exceed (above or below) the limits of agreement line on the Bland–Altman plots. (2) x indicates DXA-FRAX score minus Web-FRAX without BMD score inclusion of BMD could be problematic [4]. Today, FRAX important limitations, particularly when examining risk [53, is widely used in the assessment of individuals, despite the 54]. Common errors in the medical literature include inter- lack of validation within a large representative population preting comparisons between two effects without directly [3]. Since the FRAX tool has been incorporated in over 80 comparing them, and over-interpreting non-significant guidelines worldwide [39, 42], there is a need for assurances results [55]. Inevitably, there will be differences between of accuracy and consistency in outputs. measures when different methods are applied; hence, the key Several authors compare the performance of the tool issue is really the quantity of these differences [56]. In our using different calculation methods, with and without BMD, study, the AUC values obtained with and without BMD are and to other risk algorithms, showing variability within and similar in pattern to prior publications whereby the inclusion between populations [33, 37, 41, 43–52]. Some suggest of BMD improved the value. Unlike other studies, the AUC FRAX performs similarly with and without BMD [43, 46, for MOF was greater than the AUC for hip fractures, likely 49, 50], and using different calculation methods [41], while due to the overfitting of the model with a very high fracture others suggest FRAX without BMD is not sensitive enough prevalence, particularly non-hip MOF in our sample. In a to identify those in need of treatment [33, 46–48, 51, 52]. sensitivity analysis where we excluded those with multiple A Japanese study comparing 4 different FRAX calculation fractures or missing fracture sites, this provided a marginal methods for several thousand men and women show they improvement. However, the key aim of the study was to provide similar estimates [41], while a group of Danish examine the within-person die ff rence in FRAX estimates for authors suggests the addition of BMD may be of limited different calculation methods. A more formal analysis of the benefit [43]. In our study, the inclusion of BMD reduces the differences between estimations within individuals displays mean FRAX Ireland estimates for both 10-year risk of HF a far more accurate picture of the size of the problem, and and MOF, for both men and women. More importantly, for where those problems tend to arise. We also show when such some individuals, there were very large differences when differences are more likely to be seen. It would appear from BMD was included in their calculation. Such differences our data that use of the FRAX tool without BMD should could have a significant influence on patient and clinician be interpreted cautiously for individual patients, especially decisions on whether, and how, to intervene or not, and the older patients or those deemed higher risk. downstream clinical consequences for the patient in terms Our study has important limitations. Firstly, these data of benefit, risk, and cost. represent a small sample of a larger dataset, but this analysis Prior studies compare FRAX estimates with and without is an important first step in a multi-step process to examine BMD using ROC (receiver operatic characteristic) curves and understand the validity of FRAX and other tools for and AUC (area under the curve) analyses, or a comparison our population with and without BMD. Secondly, the data of means [37, 44–46, 49–51]. ROC analysis is commonly are cross-sectional in nature, so while we can use the tool used to assess the accuracy of diagnostic testing, but has to estimate risk, and discriminate between those with and 1 3 Archives of Osteoporosis (2023) 18:43 Page 9 of 12 43 without prevalent fracture, we cannot calibrate the results. are many different versions of the FRAX tool in use today, These results are important however as such assessments and our results may not apply to other populations where the with and without BMD are in widespread use in clinical importance of BMD has been clarified or remains unknown. practice today in Ireland. Thirdly, all subjects were referred for a DXA scan for a reason and almost 50% have a prevalent fracture, so these results may not apply to a more general Conclusions population, or those without prior fractures. Current stud- ies are assessing the performance among those with and Significant differences exist in the results of DXA-FRAX without risk factors, and with and without prior fractures and Web-FRAX for Ireland, particularly for men and those in the larger dataset, and longitudinal analyses to calibrate with higher risk estimates so these results should be inter- this and other risk algorithms in a larger cohort. Our larger preted cautiously. Reassuringly, results were similar for dataset is incomplete and has some missing data, but this those deemed at lower risk and for women. These results small subset represents a sample that has been triple-checked support the need for a more formal longitudinal analysis to for the accuracy and completeness of the data for all study calibrate FRAX and other risk tools for our population, with subjects enabling a more robust comparison. Finally, there and without BMD. Appendix Fig. 5 Bland–Altman plots comparing DXA-FRAX to Web-FRAX with BMD for HF by gender 1 3 43 Page 10 of 12 Archives of Osteoporosis (2023) 18:43 Fig. 6 Bland–Altman plots comparing DXA-FRAX to Web-FRAX with BMD for MOF by gender Funding Open Access funding provided by the IReL Consortium References Declarations 1. Compston JE, McClung MR, Leslie WD (2019) Osteoporosis. Lancet 393(10169):364–376. h ttps:/ / doi. or g/ 10. 1 016/ S0 140- Conflict of interest Lan Yang, Mary Dempsey, Attracta Brennan, Bry- 6736(18) 32112-3 an Whelan, Erjiang E, Tingyan Wang, Rebecca Egan, Kelly Gorham, 2. Cosman F, de Beur SJ, LeBoff MS, Lewiecki EM, Tanner B, Ran- Fiona Heaney, Catherine Armstrong, Guadalupe Morote Ibarrola, dall S et al (2014) Clinician’s guide to prevention and treatment of Amina Gsel, Ming Yu, and John J. Carey declare that they have no osteoporosis. Osteoporos Int 25(10):2359–2381. https:// doi. org/ conflict of interest. 10. 1007/ s00198- 014- 2794-2 3. Kanis JA, Norton N, Harvey NC, Jacobson T, Johansson H, Lor- Open Access This article is licensed under a Creative Commons Attri- entzon M et al (2021) SCOPE 2021: a new scorecard for osteo- bution 4.0 International License, which permits use, sharing, adapta- porosis in Europe. Arch Osteoporos 16(1):82. https:// doi. org/ 10. tion, distribution and reproduction in any medium or format, as long 1007/ s11657- 020- 00871-9 as you give appropriate credit to the original author(s) and the source, 4. McGowan B, Kanis JA, Johansson H, Silke C, Whelan B (2013) provide a link to the Creative Commons licence, and indicate if changes Development and application of FRAX in the management of were made. The images or other third party material in this article are osteoporosis in Ireland. Arch Osteoporos 8:146. https:// doi. org/ included in the article's Creative Commons licence, unless indicated 10. 1007/ s11657- 013- 0146-z otherwise in a credit line to the material. If material is not included in 5. Executive HS. Strategy to prevent falls and fractures in Ireland’s the article's Creative Commons licence and your intended use is not ageing population. . HSE Website: Health Service Executive2008 permitted by statutory regulation or exceeds the permitted use, you will June 2008. Report No.: ISBN 978–1–906218–12–6 need to obtain permission directly from the copyright holder. To view a 6. Kelly MA, McCabe E, Bergin D, Kearns SR, McCabe JP, Arm- copy of this licence, visit http://cr eativ ecommons. or g/licen ses/ b y/4.0/ . strong C et al (2020) Osteoporotic vertebral fractures are common 1 3 Archives of Osteoporosis (2023) 18:43 Page 11 of 12 43 in hip fracture patients and are under-recognized. J Clin Densitom. 21. Crandall CJ, Ensrud KE (2020) Osteoporosis screening in younger https:// doi. org/ 10. 1016/j. jocd. 2020. 05. 007 postmenopausal women. JAMA 323(4):367–368. https://doi. or g/ 7. Kelly MA, McGowan B, McKenna MJ, Bennett K, Carey JJ, 10. 1001/ jama. 2019. 18343 Whelan B et al (2018) Emerging trends in hospitalisation for fra- 22. Nayak S, Edwards DL, Saleh AA, Greenspan SL (2015) System- gility fractures in Ireland. Ir J Med Sci 187(3):601–608. https:// atic review and meta-analysis of the performance of clinical risk doi. org/ 10. 1007/ s11845- 018- 1743-z assessment instruments for screening for osteoporosis or low bone 8. Dockery F, Glynn A, Franks K, Carey JJ, O’Gradaigh D, Kenny P density. Osteoporos Int 26(5):1543–1554. https://doi. or g/10. 1007/ et al (2022) Fracture liaison services in Ireland-how do we com-s00198- 015- 3025-1 pare to international standards? Osteoporos Int 33(5):1089–1096. 23. Subramaniam S, Ima-Nirwana S, Chin KY (2018) Performance https:// doi. org/ 10. 1007/ s00198- 021- 06251-4 of osteoporosis self-assessment tool (OST) in predicting osteopo- 9. Walsh ME, Ferris H, Coughlan T, Hurson C, Ahern E, Sorensen rosis-a review. Int J Environ Res Public Health 15(7). https:// doi. J et al (2021) Trends in hip fracture care in the Republic of Ire-org/ 10. 3390/ ijerp h1507 1445 land from 2013 to 2018: results from the Irish Hip Fracture Data- 24. Diem SJ, Peters KW, Gourlay ML, Schousboe JT, Taylor BC, base. Osteoporos Int 32(4):727–736. https:// doi. or g/ 10. 1007/ Orwoll ES et al (2017) Screening for osteoporosis in older men: s00198- 020- 05636-1 operating characteristics of proposed strategies for selecting men 10. Walsh ME, Nerdrum M, Fahey T, Moriarty F (2021) Factors for BMD testing. J Gen Intern Med 32(11):1235–1241. https:// associated with initiation of bone-health medication among older doi. org/ 10. 1007/ s11606- 017- 4153-4 adults in primary care in Ireland. Age Ageing 50(5):1649–1656. 25. Erjiang E, Wang T, Yang L, Dempsey M, Brennan A, Yu M et al https:// doi. org/ 10. 1093/ ageing/ afab0 33 (2021) Utility of osteoporosis self-assessment tool as a screen- 11. Aziziyeh R, Amin M, Habib M, Perlaza JG, McTavish RK, Ludke ing tool for osteoporosis in Irish men and women: results of the A et al (2019) A scorecard for osteoporosis in four Latin American DXA-HIP project. J Clin Densitom 24(4):516–26. https://doi. or g/ countries: Brazil, Mexico, Colombia, and Argentina. Arch Osteo-10. 1016/j. jocd. 2021. 03. 003 poros 14(1):69. https:// doi. org/ 10. 1007/ s11657- 019- 0622-1 26. McCloskey EV, Harvey NC, Johansson H, Lorentzon M, Liu E, 12. Chandran M, Ebeling PR, Mitchell PJ, Nguyen TV, Executive Vandenput L et al (2022) Fracture risk assessment by the FRAX Committee of the Asia Pacific Consortium on O (2022) Harmo- model. Climacteric 25(1):22–28. https:// doi. org/ 10. 1080/ 13697 nization of osteoporosis guidelines: paving the way for disrupting 137. 2021. 19450 27 the status quo in osteoporosis management in the Asia Pacific. J 27. Kanis JA, Cooper C, Rizzoli R, Reginster JY, Reginster JY, Sci- Bone Miner Res 37(4):608–15. https://doi. or g/10. 1002/ jbmr .4544 entific Advisory Board of the European Society for C, Economic 13. Jones AR, Herath M, Ebeling PR, Teede H, Vincent AJ (2021) Aspects of O et al (2019) European guidance for the diagnosis and Models of care for osteoporosis: a systematic scoping review of management of osteoporosis in postmenopausal women. Osteo- efficacy and implementation characteristics. EClinicalMedicine poros Int 30(1):3–44. https://d oi.o rg/1 0.1 007/s 00198-0 18-4 704-5 38:101022. https:// doi. org/ 10. 1016/j. eclinm. 2021. 101022 28. Lewiecki EM, Compston JE, Miller PD, Adachi JD, Adams JE, 14 Maeda SS, Da Silva LLibre R, Arantes HP, de Souza GC, Molina Leslie WD et al (2011) Official Positions for FRAX(R) Bone FFC, Wiluzanski D et al (2021) Challenges and opportunities for Mineral Density and FRAX(R) simplification from Joint Official quality densitometry in Latin America. Arch Osteoporos 16(1):23. Positions Development Conference of the International Society for https:// doi. org/ 10. 1007/ s11657- 021- 00892-y Clinical Densitometry and International Osteoporosis Foundation 15. Lewiecki EM, Binkley N, Clark P, Kim S, Leslie WD, Morin SN on FRAX(R). J Clin Densitom 14(3):226–236. https://d oi.or g/1 0. (2020) Core principles for fracture prevention: North American 1016/j. jocd. 2011. 05. 017 Consensus from the National Osteoporosis Foundation, Osteo- 29 Hans DB, Kanis JA, Baim S, Bilezikian JP, Binkley N, Cauley JA porosis Canada, and Academia Nacional de Medicina de Mex- et al (2011) Joint Oc ffi ial Positions of the International Society for ico. Osteoporos Int 31(11):2073–2076. https:// doi. org/ 10. 1007/ Clinical Densitometry and International Osteoporosis Foundation s00198- 020- 05541-7 on FRAX((R)). Executive summary of the 2010 Position Develop- 16. Lewiecki EM, Binkley N, Morgan SL, Shuhart CR, Camargos ment Conference on interpretation and use of FRAX(R) in clinical BM, Carey JJ et al (2016) Best practices for dual-energy X-ray practice. J Clin Densitom 14(3):171–80. https://doi. or g/10. 1016/j. absorptiometry measurement and reporting: International Society jocd. 2011. 05. 007 for Clinical Densitometry Guidance. J Clin Densitom 19(2):127– 30. Mohammad A, Aamir MU, Mooney S, Coughlan RJ, Carey JJ 140. https:// doi. org/ 10. 1016/j. jocd. 2016. 03. 003 (2014) Appropriateness of referrals to a tertiary referral centre 17. McCloskey E, Rathi J, Heijmans S, Blagden M, Cortet B, Czer- for bone mineral density testing. Ir J Med Sci 183(4):533–537. winski E et al (2020) The osteoporosis treatment gap in patients at https:// doi. org/ 10. 1007/ s11845- 013- 1044-5 risk of fracture in European primary care: a multi-country cross- 31. European Union (Basic safety standards for protection against sectional observational study. Osteoporos Int. https:// doi. org/ 10. dangers arising from medical exposure to ionising radiation) 1007/ s00198- 020- 05557-z Regulations, S.I. No. 256/2022 (2018) 18. Rubin KH, Friis-Holmberg T, Hermann AP, Abrahamsen B, 32. Erjiang E, Wang T, Yang L, Dempsey M, Brennan A, Yu M et al Brixen K (2013) Risk assessment tools to identify women with (2020) The Irish dual-energy X-ray absorptiometry (DXA) Health increased risk of osteoporotic fracture: complexity or simplicity? Informatics Prediction (HIP) for Osteoporosis Project. BMJ Open A systematic review. J Bone Miner Res 28(8):1701–1717. https:// 10(12):e040488. https:// doi. org/ 10. 1136/ bmjop en- 2020- 040488 doi. org/ 10. 1002/ jbmr. 1956 33. Brewer L, Mellon L, Duggan J (2013) Ability of fracture risk 19. Marques A, Ferreira RJ, Santos E, Loza E, Carmona L, da assessment tool and national osteoporosis guideline group guid- Silva JA (2015) The accuracy of osteoporotic fracture risk ance to stratify people appropriately before fracture. J Am Geri- prediction tools: a systematic review and meta-analysis. Ann atr Soc 61(9):1633–1634. https:// doi. org/ 10. 1111/ jgs. 12435 Rheum Dis 74(11):1958–1967. https:// doi. or g/ 10. 1136/ annr h 34. Haroon M, Khan K, Thong L, Ali K, Janjua F (2019) High eumdis- 2015- 207907 prevalence of risk factors for low bone mineral density and esti- 20. Leslie WD, Crandall CJ (2019) Population-based osteoporosis mated fracture and fall risk among elderly medical inpatients: primary prevention and screening for quality of care in osteo- a missed opportunity. Ir J Med Sci 188(2):531–536. https://doi. org/ 10. 1007/ s11845- 018- 1882-2 porosis, Current Osteoporosis Reports. Curr Osteoporos Rep 17(6):483–490. https:// doi. org/ 10. 1007/ s11914- 019- 00542-w 1 3 43 Page 12 of 12 Archives of Osteoporosis (2023) 18:43 35 Erjiang E, Wang T, Yang L, Dempsey M, Brennan A, Yu M MrOS study. Arch Osteoporos 12(1):91. https:// doi. org/ 10. 1007/ et al (2021) How does proximal femur BMD of healthy Irish s11657- 017- 0389-1 adults compare to NHANES III? Results of the DXA-HIP 47. Hamdy RC, Seier E, Whalen K, Clark WA, Hicks K, Piggee Project. Arch Osteoporos 16(1):170. https:// doi. org/ 10. 1007/ TB (2018) FRAX calculated without BMD does not correctly s11657- 021- 01034-0 identify Caucasian men with densitometric evidence of osteo- 36. Carey JJ, Yang L, Erjiang E, Wang T, Gorham K, Egan R et al porosis. Osteoporos Int 29(4):947–952. https:// doi. org/ 10. 1007/ (2021) Vertebral Fractures in Ireland: a sub-analysis of the DXA s00198- 017- 4368-6 HIP Project. Calcif Tissue Int 109(5):534–543. https:// doi. org/ 48. Tremollieres FA, Pouilles JM, Drewniak N, Laparra J, Ribot CA, 10. 1007/ s00223- 021- 00868-7 Dargent-Molina P (2010) Fracture risk prediction using BMD and 37. Kanis JA, Oden A, Johnell O, Johansson H, De Laet C, Brown clinical risk factors in early postmenopausal women: sensitivity of J et al (2007) The use of clinical risk factors enhances the per- the WHO FRAX tool. J Bone Miner Res 25(5):1002–1009. https:// formance of BMD in the prediction of hip and osteoporotic doi. org/ 10. 1002/ jbmr. 12 fractures in men and women. Osteoporos Int 18(8):1033–1046. 49. Leslie WD, Lix LM, Johansson H, Oden A, McCloskey E, Kanis https:// doi. org/ 10. 1007/ s00198- 007- 0343-y JA et al (2010) Independent clinical validation of a Canadian 38. Compston JE, Drake MT (2020) Defining very high fracture risk: FRAX tool: fracture prediction and model calibration. J Bone is FRAX fit for purpose? J Bone Miner Res 35(8):1399–1403. Miner Res 25(11):2350–2358. https:// doi. org/ 10. 1002/ jbmr. 123 https:// doi. org/ 10. 1002/ jbmr. 4134 50. Crandall CJ, Larson J, Cauley JA, Schousboe JT, LaCroix AZ, 39. McCloskey E, Harvey N, Johansson H, Lorentzon M, Liu E, Robbins JA et al (2019) Do additional clinical risk factors improve Vandenput L et al (2022) Fracture risk assessment by the FRAX the performance of fracture risk assessment tool (FRAX) among model. Climacteric 25(1):22–28 postmenopausal women? Findings from the women’s health 40. Kanis JA, Johansson H, Harvey NC, McCloskey EV (2018) A initiative observational study and clinical trials. JBMR Plus brief history of FRAX. Arch Osteoporos 13(1):1–16 3(12):e10239. https:// doi. org/ 10. 1002/ jbm4. 10239 41. Xu G, Yamamoto N, Hayashi K, Takeuchi A, Miwa S, Igarashi K 51. Holloway-Kew KL, Zhang Y, Betson AG, Anderson KB, Hans et al (2020) The accuracy of different FRAX tools in predicting D, Hyde NK et al (2019) How well do the FRAX (Australia) fracture risk in Japan: a comparison study. J Orthop Surg (Hong and Garvan calculators predict incident fractures? Data from the Kong) 28(2):2309499020917276. https:// doi. org/ 10. 1177/ 23094 Geelong Osteoporosis Study. Osteoporos Int 30(10):2129–2139. 99020 917276https:// doi. org/ 10. 1007/ s00198- 019- 05088-2 42. Kanis JA, Harvey NC, Johansson H, Lorentzon M, Liu E, Les- 52. Sambrook PN, Flahive J, Hooven FH, Boonen S, Chapurlat R, lie WD et al (2022) FRAX. In: Pape HC, Kates SL, Hierholzer Lindsay R et al (2011) Predicting fractures in an international C, Bischoff-Ferrari HA (eds) Senior Trauma Patients. Springer, cohort using risk factor algorithms without BMD. J Bone Miner Cham, pp 89–99. https://doi. or g/10. 1007/ 978-3- 030- 91483-7_ 10 Res 26(11):2770–2777. https:// doi. org/ 10. 1002/ jbmr. 503 43. Dhiman P, Andersen S, Vestergaard P, Masud T, Qureshi N (2018) 53. Cook NR (2007) Use and misuse of the receiver operating char- Does bone mineral density improve the predictive accuracy of acteristic curve in risk prediction. Circulation 115(7):928–935. fracture risk assessment? A prospective cohort study in North-https:// doi. org/ 10. 1161/ CIRCU LATIO NAHA. 106. 672402 ern Denmark. BMJ Open 8(4):e018898. https:// doi. org/ 10. 1136/ 54. Obuchowski NA (2005) ROC analysis. AJR Am J Roentgenol bmjop en- 2017- 018898 184(2):364–372. https:// doi. org/ 10. 2214/ ajr. 184.2. 01840 364 44. Fraser LA, Langsetmo L, Berger C, Ioannidis G, Goltzman D, 55. Makin TR, Orban de Xivry JJ (2019) Ten common statistical mis- Adachi JD et al (2011) Fracture prediction and calibration of takes to watch out for when writing or reviewing a manuscript. a Canadian FRAX(R) tool: a population-based report from Elife 8. https:// doi. org/ 10. 7554/ eLife. 48175 CaMos. Osteoporos Int 22(3):829–837. https:// doi. org/ 10. 1007/ 56. Bland JM, Altman DG (1999) Measuring agreement in method s00198- 010- 1465-1 comparison studies. Stat Methods Med Res 8(2):135–160. https:// 45. Hillier TA, Cauley JA, Rizzo JH, Pedula KL, Ensrud KE, Bauer doi. org/ 10. 1177/ 09622 80299 00800 204 DC et al (2011) WHO absolute fracture risk models (FRAX): do clinical risk factors improve fracture prediction in older women Publisher's note Springer Nature remains neutral with regard to without osteoporosis? J Bone Miner Res 26(8):1774–1782. https:// jurisdictional claims in published maps and institutional affiliations. doi. org/ 10. 1002/ jbmr. 372 46. Gourlay ML, Ritter VS, Fine JP, Overman RA, Schousboe JT, Cawthon PM et al (2017) Comparison of fracture risk assess- ment tools in older men without prior hip or spine fracture: the 1 3
Journal
Archives of Osteoporosis – Springer Journals
Published: Mar 20, 2023
Keywords: Osteoporosis; Fracture Risk; Bone Mineral Density; FRAX