Introduction 1 2 3 4 6 1 7 8 4 9 1 8 9 3 10 16 8 17 8 18 20 21 25 Methods Cohorts 10 16 Baseline and outcome variables 2 3 10 16 Fracture ascertainment in the primary cohorts was undertaken by self-report (Sheffield, EVOS/EPOS, Hiroshima) and/or verified from hospital or central data-bases (Gothenburg, CaMos, DOES, Sheffield, EVOS/EPOS, Rochester, Rotterdam). Models used 26 27 Appendix 17 Epidemiology 28 29 30 31 32 Appendix Input and output variables A prior fragility fracture (yes/no) Parental history of hip fracture (yes/no) Current tobacco smoking (yes/no) Ever long-term use of oral glucocorticoids (yes/no) Rheumatoid arthritis (yes/no) Other causes of secondary osteoporosis (yes/no) Daily alcohol consumption of three or more units daily (yes/no) 11 33 40 41 43 44 45 50 51 54 55 58 59 63 Whereas there is strong evidence for the association of these disorders and fracture risk, the independence of these risk factors from BMD is uncertain. It was conservatively assumed, therefore, that the fracture risk was mediated via low BMD, but with a risk ratio similar to that noted in rheumatoid arthritis. From an operational view, where the field for rheumatoid arthritis is entered as ‘yes’, a risk is computed with and without BMD. If the field for other secondary osteoporosis is also filled as ‘yes’ this does not contribute to the calculation of fracture probability. Conversely, where the field for rheumatoid arthritis entered as ‘no’, and the field for secondary osteoporosis is ‘yes’, the same β coefficients as used for rheumatoid arthritis contribute to the computation of probability where BMD is not entered. In the presence of BMD, however, no additional risk is assumed in the presence of secondary osteoporosis, since its independence of BMD is uncertain. 1 Fig. 1 Input and output for the FRAX™ model  64 Results Clinical risk factors 1 2 2 Table 1 Ten-year probability (%) of a major osteoporotic fracture or hip fracture in men and women aged 65 years according to the presence of a single clinical risk factor   Without BMD T-score −2.5 SD Men Women Men Women a Hip a Hip a Hip a Hip No clinical risk factors 4.9 0.8 8.6 1.3 9.8 3.6 12.4 3.0 Parental history of hip fracture 9.3 1.0 16.0 1.7 16.5 3.7 22.1 3.2 Current cigarette smoking 5.1 1.1 9.2 1.9 11.0 5.6 13.7 5.1 Alcohol intake >2 units daily 6.0 1.2 10.4 2.0 12.5 5.4 15.4 4.6 Rheumatoid arthritis 6.8 1.4 11.7 2.3 12.8 5.0 16.1 4.3 Oral glucocorticoids 7.5 1.5 13.7 2.7 15.0 6.1 19.7 5.5 Previous fragility fracture 9.6 1.9 16.4 3.2 16.0 5.9 20.2 5.0 2 a Table 2 2   a Hip fracture 50 60 70 80 50 60 70 80 (a) Men No clinical risk factors 2.8 3.9 5.7 7.2 0.1 0.4 1.3 3.4 Parental history of hip fracture 5.5 7.6 9.1 15 0.2 0.6 2.8 11 Current cigarette smoking 2.8 4.1 5.9 7.5 0.2 0.7 1.8 4.2 Alcohol intake >2 units daily 3.3 4.7 7.1 9.5 0.2 0.7 2.0 5.1 Rheumatoid arthritis 3.7 5.3 8.0 11 0.2 0.8 2.3 5.8 Oral glucocorticoids 4.4 6.1 8.5 10 0.3 0.9 2.4 5.5 Previous fragility fracture 5.8 7.9 11 12 0.5 1.3 2.7 5.2 (b) Women No clinical risk factors 3.5 6.0 11 17 0.2 0.7 2.3 7.0 Parental history of hip fracture 6.9 12 17 31 0.3 0.9 5.0 22 Current cigarette smoking 3.6 6.5 12 19 0.3 1.1 3.4 9.5 Alcohol intake >2 units daily 4.1 7.3 14 22 0.3 1.1 3.6 10 Rheumatoid arthritis 4.7 8.2 15 25 0.4 1.3 4.1 12 Oral glucocorticoids 5.6 9.8 18 26 0.5 1.5 4.8 13 Previous fragility fracture 7.4 12 20 28 0.8 2.1 4.9 11 a 2 2 2 Fig. 2 2  3 2 2 Fig. 3 2 2  Bone mineral density 3 Table 3 2 Age T-score (SD) (years) +1 0 −1 −2 −3 −4 (a) Hip fracture men 50 <0.1 0.1 0.4 1.3 4.7 16.0 60 0.1 0.2 0.6 1.9 5.6 15.6 70 0.2 0.5 1.1 2.6 5.9 12.9 80 0.6 1.1 1.9 3.4 6.1 10.7 90 1.2 1.8 2.5 3.8 5.6 8.3 (b) Hip fracture women 50 <0.1 0.1 0.2 0.8 3.2 11.7 60 0.1 0.2 0.4 1.4 4.4 13.5 70 0.1 0.4 0.9 2.4 6.3 16.2 80 0.5 1.0 2.1 4.4 9.5 19.7 90 1.2 2.0 3.5 5.8 9.8 16.9 a 50 2.5 2.7 3.5 5.1 9.2 20.9 60 3.0 3.5 4.5 6.8 11.5 22.2 70 3.6 4.3 5.5 8.2 12.8 20.9 80 3.2 4.0 5.2 7.6 11.3 16.9 90 3.1 3.9 5.1 6.9 9.7 13.5 a 50 3.0 3.3 3.8 5.2 8.5 17.7 60 4.2 4.8 5.6 7.9 12.6 23.0 70 5.6 6.8 8.4 11.5 18.2 30.4 80 5.8 7.5 10.1 13.8 21.7 34.3 90 5.1 7.1 9.9 13.8 20.0 30.3 a 1 4 Fig. 4 2  5 2 6 2 2 7 Fig. 5 2  Fig. 6 2  Fig. 7 Effect of variations in BMI on 10-year hip fracture probability (%) in women aged 65 years. Probabilities with BMD are computed at a T-score of −2.8 SD. [05Ca074]  Discussion 17 3 Appendix http://www.shef.ac.uk/FRAX/index.htm 18 65 78 18 76 17 79 80 14 81 83 84 88 89 90 The present model has several unique features. FRAX™ uses Poisson regression to derive hazard functions of death and fracture. Such hazard functions are continuous as a function of time, unlike Cox’s regression for which the corresponding hazard functions are zero except at the time points of a fracture or death. There are also several advantages of the Poisson model over logistic regression analysis. Logistic regression does not take account of when a fracture occurred, nor whether individuals without a fracture died or when death occurred. Secondly, for the assessment of 10-year probabilities by logistic regression, the observation period should be for 10 years. Moreover, information longer than the 10-year period cannot be used for analysis. The cost of ignoring information when fractures occur and whether and when deaths occur is on the precision of the estimate. In simulation experiments, the Poisson model gives the same precision as logistic regression with fewer numbers of individuals. In our own simulations in the present context (data on file), the Poisson model gave the same precision as logistic regression with half the number of individuals. Finally, the Poisson model allows adjustments to be made for time trends. The ability to use several Poisson models permits the use of data from different sources to integrate fracture and death hazards, and to calibrate to different countries. 23 25 25 27 35 36 36 3 There are several limitations that should be mentioned. As with nearly all randomly drawn populations, non-response bias may have occurred. The effect is likely to exclude sicker members of society, and may underestimate the absolute fracture risk for example by age. The analyses also have significant limitations that relate to the outcome variables and the characterisation of risk factors. The definition of what was considered to be an osteoporotic fracture was not the same in all cohorts, but the effect of this inconsistency is likely to weaken rather than strengthen the associations that were found. For the hip fracture outcome, the definition was similar in all cohorts, and may explain in part the higher risk ratios associated for this fracture rather than for osteoporotic fracture. Also, the analyses were confined to clinical fractures, and the results might differ from vertebral fractures diagnosed by morphometry or as an incidental radiographic finding. 91 92 93 14 94 95 It should be acknowledged that there are many other risk factors that might be considered for incorporation into assessment algorithms. These include BMD at other skeletal sites, ultrasonography, quantitative computed tomography and the biochemical indices of bone turnover. The available information was too sparse to provide a meta-analytic framework, but they should be incorporated into risk assessment algorithms when they are more adequately characterised. Notwithstanding, the present model provides a mechanism to enhance patient assessment by the integration of clinical risk factors alone and/or in combination with BMD. 96 97 98 99