Section 3: Fracture risk assessment and case finding

Recommendations

Measurement of Bone Mineral Density

1. The risk of fracture increases progressively with decreasing bone mineral density (BMD). Systematic reviews and meta-analyses of observational population-based studies using absorptiometric techniques indicate that the risk of fracture increases approximately two-fold for each standard deviation (SD) decrease in BMD ^28,29; (Evidence level Ia). The gradient of fracture risk varies according to the site and technique used, the person’s age and the fracture type ²⁹; (Evidence level Ia). The predictive value of BMD for hip fracture is at least as good as that of blood pressure for stroke ³⁰; (Evidence level IV).
2. The WHO and the International Osteoporosis Foundation (IOF) recommend that the reference technology for the measurement of BMD is dual-energy X-ray absorptiometry (DXA) applied to the femoral neck, because of its higher predictive value for fracture ^{31, 32}; (Evidence level Ia). DXA measurements of femoral neck BMD are used in FRAX®. The spine is not always a reliable site for risk assessment or for the diagnosis of osteoporosis in older people because of the high prevalence of degenerative changes, which artefactually increase the BMD value. However, a result in an older person showing low BMD is almost always valid and clinically useful, particularly in those people with disproportionately low spine BMD compared to the hip.
3. At the same DXA-measured femoral neck BMD, men and women are at approximately the same fracture risk ^33,34; (Evidence level IIa). Therefore, the recommended reference range, from which femoral neck and total hip T-scores are calculated for men, women and transgender individuals in the US, is that derived from the NHANES III survey for white women age 20-29 years ^{32, 35}.
4. The reference ranges, from which lumbar spine and distal forearm T-Scores are calculated, for both men and women of all ethnicities, are usually those of the manufacturer of the DXA scanner ³⁵.
5. Osteoporosis can be diagnosed on the basis of the BMD T-score measured at the total hip, femoral neck or lumbar spine. However, fracture risk prediction is not improved by the use of measurements from multiple sites ^36,37; (Evidence level IIa). Where hip BMD measurement is not possible for technical reasons, or if the spine is differentially affected, then spine BMD measurements can be used for diagnosis. A diagnosis of osteoporosis can be made based on distal forearm (1/3 radius) T-Score if neither spine nor hip can be reliably measured or interpreted, or if a patient exceeds the weight limit for the DXA table ³⁵; (Evidence level IV).
6. Serial BMD measurement can be used to monitor response to treatment ( see Section 7) ³⁸. Lumbar spine BMD shows the largest treatment-related changes and is the preferred site, although if spinal degenerative changes are marked, BMD at the hip is a better site for monitoring.
7. The validity of BMD measurements depends on good quality control and national (Royal Osteoporosis Society) and international (International Society for Clinical Densitometry) bodies have published standards for the reporting of DXA scans ^{35, 39}.
8. QCT-measured femoral neck areal BMD predicts osteoporotic fractures in men and women and is equivalent to DXA-derived areal BMD ^40-42. Femoral neck and total hip T-scores calculated from two-dimensional projections of quantitative computed tomography (QCT) data are equivalent to the corresponding DXA-derived T-scores. Thus, femoral neck CT X-ray absorptiometry (CTXA) BMD measurements can be included in FRAX ^35,43-45; (Evidence level IIa) ( see Section 4). Other techniques for assessing skeletal BMD, including quantitative ultrasound, have been less well validated than absorptiometric techniques.

Assessment of Clinical Risk Factors

9. The performance characteristics of BMD assessment can be improved by the concurrent consideration of clinical risk factors that operate independently of BMD. Of particular importance is age, which contributes to risk independently of BMD ^{11, 46}; (Evidence level Ia).
10. Additional clinical risk factors have been identified that provide information on fracture risk independently of both age and BMD:
    1. Low body mass index (BMI) is a significant risk factor for hip fracture, but the value of BMI in predicting other fractures is very much diminished when adjusted for BMD ⁴⁷; (Evidence level Ia).
    2. A history of a prior fracture, particularly if sustained from low-trauma and at a site characteristic for osteoporosis, is an important risk factor for further fracture ⁴⁸. The risks are in part independent of BMD ⁴⁹. Fracture risk is approximately doubled in the presence of a prior fracture, including asymptomatic moderate or severe (Grade 2 or 3) morphometric vertebral fractures ^{49, 50}; (Evidence level Ia). The increase in risk is even more marked for more than one vertebral fracture. After a fracture, the risk of subsequent fracture is highest in the immediate post fracture interval (imminent risk) with more than one-third of subsequent fractures over a ten-year time frame occurring within the first year ^51,52; (Evidence level Ic).
    3. A parental history of hip fracture is a significant risk factor that is largely independent of BMD ⁵³; (Evidence level Ia).
    4. Smoking is a risk factor that is in part dependent on BMD ⁵⁴; (Evidence level Ia).
    5. Oral glucocorticoid therapy increases fracture risk in a dose-dependent manner. The fracture risk conferred by the use of glucocorticoids is, however, not solely dependent upon bone loss and BMD-independent risks have been identified ^{55, 56}; (Evidence level Ia).
    6. Alcohol intake shows a dose-dependent relationship with fracture risk. Where alcohol intake is on average two units or less daily, no increase in risk has been identified. Intakes of 3 or more units daily are associated with a dose-dependent increase in fracture risk ⁵⁷; (Evidence level Ia).
    7. There are many secondary causes of osteoporosis (e.g., inflammatory bowel disease, endocrine disorders), but in most instances it is uncertain to what extent an increase in fracture risk is dependent on low BMD or other factors such as the use of glucocorticoids. By contrast, rheumatoid arthritis increases fracture risk independently of BMD and the use of glucocorticoids ⁵⁶; (Evidence level Ia).
    8. Diabetes mellitus (both type 1 and type 2) is associated with an increase in risk of hip and non-vertebral fracture. In type 2 diabetes; a longer duration of disease and insulin use are associated with an increased risk ^{58, 59}; (Evidence level Ia), which is partly independent of BMD ^{60, 61}.
11. The use of combined clinical risk factors alone to predict fracture risk, performs very similarly to that of BMD alone ⁶². The use of clinical risk factors with the addition of BMD is optimal, but BMD measurement can be targeted to those close to the threshold of low/high risk or close to the threshold of high/very high risk (see Section 4).
12. There are many additional clinical risk factors for fracture not included in FRAX, including risks that either act solely by reducing BMD, or have been less well validated, or identify a risk that may not be amenable to particular treatments ^{11, 63}. Liability to falls is an example of the latter where the risk of fracture is high, and treatment with drugs affecting bone metabolism alone may not fully address this risk ⁶⁴.
13. In addition to glucocorticoids, several medications are known to increase hip fracture risk including thyroid hormone excess, aromatase inhibitors for the treatment of breast cancer and androgen deprivation for the treatment of prostate cancer ^65-69; (Evidence level Ia). Thiazolidinediones, used in the treatment of type 2 diabetes also increase fracture risk ^70,71.
14. Several other drugs have been associated with increased fracture risk including antidepressants, antiparkinsonian drugs, antipsychotic drugs, anxiolytic drugs, benzodiazepines, sedatives, h3 receptor antagonists and proton pump inhibitors ^65-69. The extent to which fracture risk is mediated by low BMD, falls risk or other factors, or indeed is definitely causal in each case, is not known. The impact of sex steroids on bone health in transgender individuals is unclear ⁷².
15. Biochemical indices of skeletal turnover have the potential to aid risk assessment but probably play a more immediate role in the monitoring of treatment ^73-75; (Evidence level Ia).

Fracture Risk Assessment Tools

16. The IOF and the WHO recommend that risk of fracture is expressed as an absolute risk, i.e., probability over a ten-year interval ¹¹. The absolute risk of fracture depends upon age and life expectancy as well as the current relative risk. The period of 10 years covers the likely initial duration of treatment and the benefits that may continue if treatment is stopped. Shorter time horizons do not aid the categorisation of risk^{76, 77}.
17. Algorithms that integrate the weight of clinical risk factors for fracture risk, with or without information on BMD, were developed in 2008 by the then WHO Collaborating Centre for Metabolic Bone Diseases at Sheffield. The FRAX tool (www.shef.ac.uk/FRAX) computes the 10-year probability of hip fracture and/or of major osteoporotic fracture. A major osteoporotic fracture is a clinical spine, hip, forearm or humerus fracture. The tool has been externally validated in independent cohorts ^46,78; (Evidence level Ia).
18. QFracture is based on a UK prospective open cohort study of routinely collected data from general practices that takes into account numerous clinical risk factors and estimates the 1 to 10 year cumulative incidence of hip and/or major osteoporotic fracture [http://www.qfracture.org ⁷⁹].
19. The NICE has recommended the use of fracture risk assessment tools (FRAX or QFracture) in the assessment of patients ⁸⁰. Since FRAX and QFracture yield different outputs (probability of fracture accounting for mortality risk in the case of FRAX, and a cumulative risk of fracture in the case of QFracture), the two calculators cannot be used interchangeably. In addition, BMD cannot be incorporated into QFracture estimations. Finally, the NOGG intervention thresholds, recommended by NICE Quality Standards, are based on FRAX probability and thus cannot be used with fracture risk derived from QFracture or other calculators^{78, 81}.
20. The input into FRAX includes, with age and sex, the BMD independent clinical risk factors listed in Table 1. Femoral neck BMD is an optional input. The listed secondary causes are conservatively assumed to be mediated through low BMD and carry no weight when femoral neck BMD is entered into FRAX.
    Table 1: Clinical risk factors included specifically in the FRAX assessment of fracture probability

    N.B. Additional clinical risk factors that should prompt FRAX assessment are listed in Table 4.

    Age

    Sex

    Body mass index (calculated from weight and height in kg/m2)

    Previous fragility fracture, including morphometric vertebral fracture

    Parental history of hip fracture

    Current glucocorticoid treatment (any dose, by mouth for 3 months or more)

    Current smoking

    Alcohol intake 3 or more units daily

    Rheumatoid arthritis

    Secondary causes of osteoporosis including:

    Type I diabetes

    Long-standing untreated hyperthyroidism

    Untreated hypogonadism/premature menopause (

    Chronic malnutrition/malabsorption

    Chronic liver disease

    Non-dialysis chronic renal failure (i.e., CKD 3a – 5)

    Femoral neck BMD

21. FRAX assessment takes no account of prior osteoporosis drug treatment, or of the dose of several clinical risk factors. For example, a history of two prior fractures carries a higher risk than a single prior fracture. A prior clinical vertebral fracture carries an approximately two-fold higher risk than other prior fracture types. Dose responses are also evident for glucocorticoid use and are partially addressed in the NOGG guideline ( see Section 7). Since it is not possible to model all such scenarios within the FRAX algorithm, clinical judgement is needed to interpret FRAX outputs.
22. High and low impact injuries exist on a continuum and the clinical significance of high and low impact fractures is blurred in the context of osteoporosis. Indeed, prior high-trauma fractures are associated with low BMD and future fracture risk to the same extent as fractures without high-trauma ⁴⁸.

23. Although FRAX has a limited input of variables, relatively simple arithmetic procedures are available (Table 2) which can be applied to conventional FRAX estimates of probabilities of hip fracture and major osteoporotic fracture to adjust the probability assessment with knowledge of:

    • High, moderate and low exposure to oral glucocorticoids ⁸²; (Evidence level IIa)
    • Concurrent data on lumbar spine BMD ^{8383, 84}; (Evidence level Ia)
    • Information on trabecular bone score (TBS) ⁸⁵; (Evidence level Ia). TBS values can be entered on the UK FRAX website.
    • Hip axis length ⁸⁶; (Evidence level Ib)
    • Falls history ⁸⁷; (Evidence level IIa)
    • Country of birth ⁸⁸; (Evidence level Ib)
    • Type II diabetes mellitus ⁸⁹; (Evidence level Ib)
    • Recent major osteoporotic fracture (MOF) ⁵²; (Evidence level Ib)

    When applying these FRAX adjustments a suggested increase of x% should be applied as a proportion of the original FRAX score. For example, uplifting the FRAX probability of 30% by 10% gives an adjusted probability of 30 x 1.10 = 33%.

    There is no evidence base available to inform on the accuracy of multiple adjustments. Pragmatically, the adjustment should be made for the most dominant factor, i.e. that which will have the greater impact on the estimated probability; (Evidence level IV).

    Table 2: Approximate adjustments and considerations to probabilities of hip fracture and major osteoporotic fracture to aid the interpretation of FRAX

    * downward adjustment to FRAX probabilities should only be made in the context of a very reliable high lumbar spine BMD measurement and not on the basis of a discordant result due to artefact e.g. from degenerative change
    ^¥ See Section 7: ‘Glucocorticoid-induced osteoporosis’ for further details on glucocorticoid doses and recommendations

    Risk variable	Adjustment to FRAX*	Access
    Medium and high dose exposure to oral glucocorticoids	Medium doses (2.5–7.5 mg daily) are the assumed minimum requirement for FRAX calculation, and the unadjusted FRAX value is used. For high doses (>7.5 mg daily), MOF probabilities are upward revised by about 15% and hip fracture probabilities by 20% ¥	Automatic adjustment available on FRAX website. Kanis et al 2011 82
    Concurrent data on lumbar spine (LS) BMD	Increase/decrease MOF probability by 10% for each rounded T-score difference between LS and FN*	Leslie et al 2011 Johansson et al 2014 83,84
    Trabecular bone score (TBS)	Increase MOF probability by 30% for each standard deviation (SD) decrease in TBS	TBS adjustment can be accessed from the UK FRAX website. McCloskey et al 2016 85
    Hip axis length (HAL)	Increase hip fracture probability by 30% for each SD increase in HAL	Leslie et al 2016 86
    Falls history	Increase MOF and hip fracture probability by 30% for a history of recurrent falls (≥2 falls in the last year)	Masud et al 2011 87
    Country of birth	Use FRAX model for country of birth since individuals retain the risk characteristics of their country of origin	Johansson et al 2015 88 Wändell et al 2021 90
    Type II diabetes mellitus	Enter ‘yes’ in the rheumatoid arthritis input to FRAX	Other adjustments in Leslie et al 2018 89
    Recent MOF	Marked uplift to fracture probabilities (see Section 4h )	Kanis et al 2020 52

24. Although type I diabetes carries a risk of fracture over and above that provided by FRAX, there are yet no empirical data from which to recommend adjustment. In the meanwhile, the same adjustment can be used as for type II diabetes; (Evidence level IV).
25. Additionally, FRAX values have been shown to be largely unaffected by socioeconomic status ⁹¹, variation in body composition ⁹², and chronic renal disease ⁹³; (Evidence level Ib).
26. Adjustments to FRAX probabilities which take into account severity and/or number of vertebral fractures cannot currently be made because of the lack of appropriate empirical data.
27. Risk is best presented to patients numerically using simple frequencies and positive and negative framing e.g., for a 23% risk say ‘100 people like you, over the next 10 years, 23 will break a bone and 77 will not’. Describing risks solely with words, such as ‘You have a high chance of experiencing a fracture’ is ineffective and does not provide patients with the details needed to make an informed decision; it increases risk perceptions, and patients vary in their interpretations of what are low and high risks. It is easier for patients to understand whole numbers and simple frequencies (e.g., 1 in 100) rather than percentages. Graphs and pictograms make numeric information easier to understand and should be used where available ⁹⁴94; (Evidence level IV).

Investigation of osteoporosis and fragility fractures

28. Diagnostic assessment of individuals with osteoporosis should exclude diseases that mimic osteoporosis, identify the cause(s) of the osteoporosis, and include the management of any associated comorbidity. Common investigations are given in Table 3.
    Table 3: Proposed clinical investigations to consider for the investigation of osteoporosis/ fragility fractures.

    ^a Persistent low phosphate or alkaline phosphatase should not be overlooked as this can indicate underlying metabolic bone disease.
    ^b Measure PTH if albumin-adjusted serum calcium ≥2.6mmol/l twice, or if ≥2.5mmol/l twice if primary hyperparathyroidism is suspected ⁹⁵.
    ^c Principally measured to monitor bone turnover in response to anti-resorptive treatment ( see Section 7), CTX reflects bone resorption, P1NP bone formation. CTX is best measured in the morning after an overnight fast.
    Other investigations, for example, bone biopsy and genetic testing for osteogenesis imperfecta, are largely restricted to specialist centres.

    Routine

    Other procedures, if indicated

    Clinical history Physical examination including measurement of height and assessment of thoracic kyphosis Full blood cell count Erythrocyte sedimentation rate or C-reactive protein Serum calcium, albumin, creatinine, phosphatea, alkaline phosphatasea and liver transaminases Serum 25-hydroxyvitamin D Thyroid function tests

    Serum electrophoresis, serum immunoglobulins and serum free light chain assay Plasma parathyroid hormone (PTH)b Serum testosterone, sex hormone binding globulin, follicle stimulating hormone, luteinizing hormone 24-hour urinary free cortisol/overnight dexamethasone suppression test Serum prolactin Serum magnesium if hypocalcaemic Tissue transglutaminase antibodies, +/- endomysial antibodies (coeliac disease screen) Urinary calcium excretion Markers of bone turnover (e.g., CTX, P1NP)c Lateral radiographs of lumbar and thoracic spine or DXA based lateral vertebral imaging Bone densitometry (DXA) if indicated by FRAX assessment and/or required for BMD monitoring Isotope bone scan

Vertebral Fracture Assessment

30. The majority of vertebral fractures do not currently come to medical attention and thus remain undiagnosed ⁹⁶. Moderate or severe vertebral fractures, even when asymptomatic, are strong risk factors for subsequent fracture at the spine and other skeletal sites ^{50,97, 98}; (Evidence level Ia). Vertebral fracture assessment (VFA) should therefore be considered in high-risk individuals, using either lateral lumbar and thoracic spine radiographs or lateral spine DXA imaging ⁹⁹; (Evidence level Ia). The latter delivers a significantly lower radiation dose whilst performing comparably to traditional radiographs ¹⁰⁰.
31. Identification of vertebral fractures on routine radiological images, such as plain abdominal and chest radiographs, performed for other indications, offers the opportunity to detect clinically important osteoporotic fractures.
32. Opportunistic diagnosis of osteoporosis and vertebral fractures is feasible using CT scans acquired for various clinical reasons, since the hip and spine are frequently in the scan field ¹⁰¹; (Evidence level Ia). Vertebral fracture identification from CT using Computer Aided Diagnostics (CAD) can augment and improve standard reporting methods ^102-105; (Evidence level IIb). Reliable CAD methods have high sensitivity, specificity, and accuracy for vertebral fracture detection; (Evidence level IV).

Screening and Case Finding

43. At present there is no universally accepted policy for population-based screening to identify people with osteoporosis. With the recognition that factors in addition to BMD can improve fracture risk prediction, it is possible that screening strategies might be implemented in the future.
44. A trial of screening in the UK used FRAX to target osteoporosis drug treatment to women at high risk of hip fracture. The risk assessment, with subsequent femoral neck BMD measurement and input to FRAX in intermediate/high risk individuals, was conducted in a primary care setting and involved almost 12,500 women aged 70-85 years. Over 5 years, compared to standard clinical care, the screening program reduced the number of hip fractures by 28%. Similar results were observed in a study from Denmark ¹⁰⁶, but with lesser effects observed in a further study in the Netherlands ¹⁰⁷. A meta-analysis of the three trials showed that screening reduced hip fracture risk by 20% ¹⁰⁸108; (Evidence level Ia).
45. In the absence of a screening policy, a case-finding strategy is appropriate where patients are identified because of a fragility fracture or by the presence of other clinical risk factors. There are many clinical risk factors for fracture in addition to those included in FRAX which can be used to trigger fracture risk assessment (see Table 4), including thoracic kyphosis and height loss (> 4cm), either in comparison with recalled young adult height or a documented loss on serial measurements ¹⁰⁹; (Evidence level IIa), and bariatric surgery resulting in malabsorption ¹¹⁰; (Evidence level Ia).
    Table 4: Clinical risk factors for osteoporosis/fractures, not accommodated in FRAX, which should trigger fracture risk assessment.

    ^a Able to be accommodated in FRAX by proxy, by entering ‘yes’ in the rheumatoid arthritis input (see Table 2)

    Thoracic kyphosis

    Height loss (> 4cm)

    Falls and Frailty

    Inflammatory disease: e.g., ankylosing spondylitis, other inflammatory arthritides, connective tissue diseases, systemic lupus erythematosus

    Endocrine disease: e.g., Type I and II diabetes mellitus a, hyperparathyroidism, hyperthyroidism, hypogonadism, Cushing’s disease/syndrome

    Haematological disorders/malignancy e.g., multiple myeloma, thalassaemia

    Muscle disease: e.g., myositis, myopathies and dystrophies, sarcopenia

    Lung disease: e.g., asthma, cystic fibrosis, chronic obstructive pulmonary disease

    HIV

    Neurological/ psychiatric disease e.g., Parkinson’s disease and associated syndromes, multiple sclerosis, epilepsy, stroke, depression, dementia

    Nutritional deficiencies: calcium, vitamin D [note that vitamin D deficiency may contribute to fracture risk through undermineralisation of bone (osteomalacia) rather than osteoporosis]

    Bariatric surgery and other conditions associated with intestinal malabsorption

    Medications, e.g.:

    Some immunosuppressants (calmodulin/calcineurine phosphatase inhibitors)

    (Excess) thyroid hormone treatment (levothyroxine and/or liothyronine). Patients with thyroid cancer with suppressed TSH are at particular risk

    Drugs affecting gonadal hormone production (aromatase inhibitors, androgen deprivation therapy, medroxyprogesterone acetate, gonadotrophin hormone releasing agonists, gonadotrophin hormone receptor antagonists)

    Some diabetes drugs (e.g., thiazolidinediones)

    Some antiepileptics (e.g., phenytoin and carbamazepine)