FAQ
Q: In clinical practice I would prefer to use one year probabilities - why use 10-year fracture probability?
A: In young healthy individuals (with a low mortality) the one year probability is approximately 10% of the 10-year probability. Thus, an individual with a 10-year fracture probability of 40% would have approximately a 1-year probability of 4%. Higher percentage figures are more readily understood by patients and clinicians.
Q: Do the risk factors have the same importance in men, women and different geographic settings?
A: The risk factors work similarly in men and women in different countries in terms of relative risk. However, absolute risk will vary since, at any given age, the absolute risk of fracture and absolute risk of death varies. In addition, risk factors have variable importance depending upon age (e.g. a family history), or on the presence or absence of other risk factors. For example, low BMI is much less of a risk factor when account is taken of BMD.
Q: The clinical risk factors demand a yes or no response. However, two prior clinical fractures carry a greater risk than a single previous fracture. Why is this not accommodated?
A: It is known that dose-responses exist for many of the clinical risk factors. In addition to the number of previous fractures, they include smoking, use of glucocorticoids and consumption of alcohol. The model is, however, based on information that is common to all the cohorts that participated in its creation and such detail is not available. This means that clinical judgement needs to be used when interpreting probabilities. A higher than average number of fractures will carry a higher probability than that displayed.
Q: The clinical risk factors demand a yes or no response. However, high doses of oral glucocorticoids carry a greater risk than an average dose. Why is this not accommodated?
A: It is not, for the reasons indicated in the question above. A higher than average dose of glucocorticoids will carry a higher probability than that displayed. Conversely, a lower than average dose will signify a lower probability. An aid to the adjustment for dose of glucocorticoids is given in Kanis et al, 2010 (see reference list).
Q: A prior vertebral fracture carries a higher risk than a prior forearm fracture. How is this accounted for in the algorithms?
A: It is not, for the reasons indicated in the questions above. It should be noted, however, that a prior morphometric and asymptomatic vertebral fracture carries approximately the same risk as any previous fracture. A clinical vertebral fracture, however, carries a much higher risk (see reference list, Johnell et al 2006).
Q: The markers of bone turnover have been shown to correlate with fracture risk independently of BMD. Can these be used in conjunction with the model?
A: It is correct that high values for indices of bone turnover are associated with a fracture risk independently of BMD. There is, however, no agreement on a reference analyte and insufficient world-wide experience to know how they might be incorporated. The manner in which the results of such tests are interpreted is a matter of clinical judgement.
Q: How is account taken of ethnic minorities?
A: It is not - with the exception of the United States and Singapore where there is sufficient epidemiological information to make the appropriate adjustments.
Q: My country is not represented? What should I do?
A: Use the country for which the epidemiology of osteoporosis most closely approximates your country. Examples of high risk countries are Denmark and Sweden. Low risk countries include Lebanon and China.
New models are expected to be made available in later versions. Lobby your national society for a country specific model or a surrogate model.
Q: Why can't I use the tool to predict fracture risk in a 30 year old patient?
A: The model is constructed from real data in population-based cohorts around the world that have a limited age range. If you enter an age below 40 years, the tool will calculate the probability of fracture at the age of 40 years. You must use your clinical judgement to interpret the risk.
Q: For the clinical risk factors, there is no provision for missing values (i.e. a do not know category) in the program. What should I do?
A: Missing values are not provided for in our program. When calculating the 10 year probability it is assumed that every question (except BMD) can be answered. If you don't have information, for example on family history, you should answer no.
Q: Why not report the probability of all osteoporotic fractures? This would give larger values.
A: Incorporating all osteoporotic fractures is problematic because of limited information on their epidemiology. From Swedish data, the inclusion of other major osteoporotic fractures (e.g. pelvis, other femoral fractures and tibial fractures) would increase the values by about 10 percent (for example, in a patient with a calculated probability of major osteoporotic fractures of 5%, this might be uplifted to 5.5%). Including rib fractures would have a much larger effect. They are, however, difficult to diagnose.
Q: Why not include falls which are a well established clinical risk factor for fracture?
A: Two reasons. The first is that the cohort data used to create the model reported falls in very different ways so that it was not possible to derive a standardized metric. Second, although plausible, pharmaceutical intervention has not been shown to reduce fracture risk in patients selected on the basis of a fall history. It is important that risk assessment models identify a risk that can be reduced by treatment. Note that FRAX is based on the inclusion of individuals who are at all levels of falls risk so , though not an input variable, falls are accounted for in the calculation of FRAX.
Q: Why have you ignored fractures diagnosed on X-ray and focussed on clinical vertebral fracture?
A: A prior morphometric fracture has the same significance as any other prior fragility fracture and can be entered into the FRAX® model. The output does not, however, include the probability of a morphometric fracture. This is a conservative position, since their clinical significance is controversial (other than for risk prediction). Nevertheless, this does not affect who would be eligible for treatment.
Q: How do I decide who to treat?
A: The FRAX® assessment does not tell you who to treat which remains a matter of clinical judgement. In many countries, guidelines are provided that are based on expert opinion and/or on health economic grounds.
Q: What T-score should I input into the FRAX® model?
A: You should input a T-score for the femoral neck derived from the reference standard (the NHANES III database for female Caucasians aged 20-29 years as widely recommended - T (WHO)-score). T-scores from local data bases or ethnic-specific reference ranges will give misleading results. Note that the same reference range is used for men (i.e. the NHANES III database for female Caucasians aged 20-29 years). If you are uncertain about the T-score, input the manufacturer of the DXA device and the BMD result. The T (WHO)-score will be calculated for you.
Q: Can I use the BMD measured at the total hip or lumbar spine rather than the femoral neck BMD?
A: No. The model is constructed from real data in population-based cohorts where femoral neck BMD is available. T-score and Z-score vary according to the technology used and the site measured.
Q: I see patients where the T-score for the lumbar spine is much lower than the T-score at the femoral neck BMD. Will FRAX underestimate the fracture probability in these patients?
A: Yes. But note that FRAX will overestimate the fracture probability in patients where the T-score for the lumbar spine is much higher than the T-score at the femoral neck BMD. An aid for the adjustment for T-score disparities is given in Leslie et al, 2010 (see FRAX Help references).
Q: Fracture rates and life expectancy are changing. How will this affect FRAX?
A: Significant changes will affect the accuracy of the model so that FRAX models need adjustment from time to time.