Using WHO-FRAX to Describe Fracture Risk: Experience in Primary Care

Abstract

Ideally those at highest risk of fracture should be identified prior to fracture occurrence to reduce mortality, morbidity and costs. Case-finding strategies for those at high risk of first fracture or systematic case-finding strategies following fracture are recommended in the UK, rather than population-based screening to identify individuals at high fracture risk. General practices in the UK hold relevant data on individuals beyond fracture history that could allow identification of a wider group of patients at highest risk of fracture. The aim of the paper is to evaluate the feasibility of applying the WHO-FRAX fracture risk calculator to general practice populations using existing recorded data. A cross-sectional study of 2467 women aged 50 years and older (mean 66.2 years, standard deviation = 11.3) registered with two Scottish General Practices with low deprivation (one semi-rural, one urban) was undertaken. Patient data were extracted from the two general practices’ patient information databases and the WHO-FRAX calculator was applied to these data. WHO-FRAX calculation was possible on 1872 patients. Of these, 687 patients were found to have a high fracture risk (risk of major facture ≥15% and or risk of hip fracture ≥3% – 37% of the WHO-FRAX assessed cohort) and should be considered for follow-up. In conclusion, use of the WHO-FRAX calculator using general practice-held data is feasible and can help to identify a patient group at higher fracture risk. Further evaluation and treatments can then be targeted at this group.

Keywords

osteoporosis WHO-FRAX fracture risk primary care clinical risk factors

Introduction

Osteoporosis-related fractures are a major cause of morbidity and mortality.^1–3 One in two women and one in five men will suffer a fracture after the age of 50 years.⁴ The annual incidence of fractures in the UK is estimated at 310,000 with costs of some £1.8 billion in 2000 rising to £2.2 billion by 2020.⁵

The appropriate use of bisphosphonates has been shown to reduce the risk of vertebral, non-vertebral and hip fractures, both in patients who have never fractured before and in those who have already sustained one or more fractures.^6,7 However, the majority of patients who are at high risk of fracture are currently undiagnosed and/or untreated.^8,9

Case-finding tools can be used to select patients who are at a high risk of fracture for referral for dual energy X-ray absorptiometry (DXA) scanning to diagnose osteoporosis. Case finding has been shown to be more cost-effective than population-based screening of bone density.^7,10 However, case finding has been shown to be specific but not sensitive.¹¹ Ideally an evaluation tool needs to be sufficiently sensitive to indicate those needing further assessment of fracture risk and yet specific enough to accurately target those for assessment. An ad hoc case-finding approach has generally been the preferred way to identify patients who might be at high risk of fracture¹² although recently, systematic approaches, such as the development of Fracture Liaison Services, have been shown to be more successful.¹³ It is now becoming clear that while ad hoc case finding is attractive in principle, this is not an effective way to find high-risk patients. If global fracture risks are to be reduced, then more systematic approaches to case finding need to be developed.¹⁴ While Fracture Liaison Services go some way to identifying a population of patients at higher future fracture risk, there remains a large population who have not sustained a recent primary fracture, but who also have a high future fracture risk.¹⁵

Categorizing fracture risk for individual patients in a primary care setting may address the lack of sensitivity of case finding while avoiding the pitfalls of mass (population) DXA scanning.

WHO-FRAX (www.shef.ac.uk/FRAX) is a computer (web) based algorithm which can be used for the assessment of fracture risk. It can be used both when bone mineral density (BMD) data are and are not available. Risks are calculated over a 10-year interval for major fractures (clinical spine, forearm, hip or shoulder) or for hip fracture alone. WHO-FRAX is designed to be used with the patient in consultation with their clinician.

In the UK, general practices have been computerized for many years and have coded data on their patients.¹⁶ If the correct codes can be extracted from practice population data, then the WHO-FRAX algorithm could be applied to individuals and allow identification of those at high risk of fracture. Further investigation or treatment might then be considered in these patients.

We conducted a study to investigate whether data currently available to general practices are adequate to allow a WHO-FRAX calculation and therefore identify individuals at risk of fracture.

Methods

Two general practices were approached to carry out the study. A semi-rural practice and an urban practice, both with low deprivation (Scottish Index of Multiple Deprivation Score <8%).¹⁷ Both practices have been computerized for at least five years and have scored maximal points in the ‘Qualities and Outcomes’ Framework (QOF) for general practice, suggesting good administration and data recording. Both practices use GPASS (General Practice Administration System for Scotland; www.gpass.scot.nhs.uk), the government supplied and currently the most frequently used GP software in Scotland.

Data (which is used in the WHO-FRAX risk calculator) were extracted from both practices for women aged 50 years and above. These data are shown in Table 1.

Table 1

Data extracted from practices

Gender	Female
Date of birth	Date
Height	cm
Weight	kg
Smoking status	Never, ex or current
Alcohol consumption	Amount
Premature menopause	Yes or No
Past fracture	Site and date
History of fragility fracture	Yes or No
Family history of maternal hip fracture	Yes or No
Rheumatoid arthritis	Yes or No
Inflammatory bowel disease	Yes or No
COPD diagnosis	Yes or No
Type 1 diabetes mellitus	Yes or No
Chronic liver disease	Yes or No
Coeliac disease	Yes or No
Ovarian hypogonadism	Yes or No
Osteogenesis imperfecta	Yes or No
Corticosteroid prescribed	Yes or No
DXA scan recorded	Yes or No
Osteoporosis diagnosis	Yes or No
Bisphosphonate prescription issued	Yes or No

COPD, chronic obstructive pulmonary disease

The extracted data were then manually entered into the WHO-FRAX calculator for each individual, to calculate the 10-year major fracture risk and hip fracture risk.

The records of 2464 female patients aged over 50 years were reviewed. ‘High risk’ of fracture was defined in this study (as per the Greater Glasgow and Clyde Osteoporosis Service) as a greater than 15% risk of major fracture¹⁸ or greater than 3% risk of hip fracture in the next 10 years.¹⁹ Where a choice is to be made, major fracture risk is preferred to hip fracture when considering treatment thresholds.²⁰

Results

Collation of data within practices

The two practices had similar levels of data recording and disease prevalence, and thus the data were combined (Table 2).

Table 2

Absolute number and percentage of individuals with information recorded

Summary Table	Number	Percentage	Expected prevalence
Total patients (women over 50)	2464
Date of birth	2464	100
Height	1963	80
Weight	1978	80
Smoking status	1980	80
Alcohol consumption	1558	63
Past history/history of fragility fracture	466	19.1	24.3%²¹
Corticosteroid prescribed	149	6	5.0%²²
Osteoporosis diagnosis	179	7.3	25.0%²³

WHO-FRAX includes a section for conditions associated with secondary osteoporosis and with this in mind, the practices’ prevalences were in keeping with the national expected prevalences for the conditions type 1 diabetes mellitus,²⁴ rheumatoid arthritis,²⁵ inflammatory bowel disease,²⁶ chronic liver disease,²⁷ chronic obstructive pulmonary disease²⁸ and coeliac disease^29,30 (data not shown). Corticosteroid script prevalence was near the national recorded prevalence.²² The prevalence of identified osteoporosis was in keeping with expected recorded figures,³¹ but lower than the true disease burden.²³ The prevalence of previous fractures was in keeping with national figures²¹ (19% versus an expected prevalence of 24%).

The national prevalences of premature menopause and ovarian hypogonadism³² are low, but it appeared that neither practice recorded these data. Osteogenesis imperfecta is a rare condition³³ and neither practice contained an individual with this condition. The national prevalence of a history of maternal hip fracture is 6%,³⁴ but these data were not recorded by either practice.

The contemporaneous nature of the recorded demographic data was assessed (Figure 1).

Figure 1

Date of recording demographic data

There were 592 individuals (24%) with no height and/or weight recorded. No WHO-FRAX calculation was performed on these individuals. The remaining 1872 (76%) proceeded to have a WHO-FRAX calculation performed.

Patients identified as high risk of future fracture

Of the 1872 individuals assessed, 513 (27%) had both a 10-year risk of major fracture of >15% and a >3% risk of hip fracture. A further 171 (9%) individuals had a 10-year hip fracture risk of >3% and another three individuals (0.2%) had a 10-year risk of >15% for major fracture. In total, there were 687 patients (37% of the measured cohort) who should be considered for follow-up/further investigation.

Of those in whom WHO-FRAX assessment was possible, 28% (21% of the study cohort) had a 10-year risk of major fracture of greater than 15% (Figure 2).

Figure 2

WHO-FRAX calculated 10-year major fracture risk

Of those in whom WHO-FRAX assessment was possible, 37% (28% of the study cohort) had a 10-year risk of hip fracture greater than 3% (Figure 3).

Figure 3

WHO-FRAX calculated 10-year hip fracture risk

Of the 687 individuals who were identified as having a 10-year risk >15% for major fracture or a >3% risk of hip fracture, only 255 individuals (37%) had a fracture history.

Discussion

In this study the WHO-FRAX calculator identified 37% of the patients studied as having a high risk of major fracture, hip fracture or both. The prevalence of identified and recorded osteoporosis (by DXA scanning) in our cohort was 7.3%. This is in keeping with previous studies,³¹ but contrasts with the expected disease burden of 25%.²³ Therefore, there is a large group of patients (29.7%) who are at a sufficiently high risk of fracture to justify further investigation and/or treatment. These patients are not being identified and assessed at present.

In theory there is sufficient data held in primary care computerized information systems to identify those at high risk of fracture using WHO-FRAX. Our study's strength demonstrates how WHO-FRAX can readily be applied to patients in general practice, using information already available. Interrogation of a database can only include coded conditions, whereas a face-to-face consultation might enable additional WHO-FRAX parameters to be identified. Any data not considered (when using coded information) will underestimate rather than overestimate fracture risk. It follows that the patients with the highest risk would be identified if WHO-FRAX was used as a screening tool.

We found that the two practices in this study had sufficiently accurate coded data to enable WHO-FRAX to be used as a screening tool. However limitations existed. Most general practices take a pragmatic approach to coding data and rather than coding all data, only code information that may be useful or affect clinical care. Of the cohort in this study, 24% could not be assessed due to missing data. In general, smoking, alcohol history, height and weight data are all well recorded, but these data are often not contemporary. This limited the accuracy and ability to perform the WHO-FRAX calculation.

Although an ad hoc case-finding strategy has been shown to be more cost-effective than screening of bone density^7,10; the opportunistic use of case-finding tools, such as WHO-FRAX or other such tools,^11,35 appear to be used rarely. In day-to-day busy practice, such screening is not practical. A current limitation of using the WHO-FRAX algorithm for screening is that it can only be applied individually, making ‘en mass’ application to practice data impossible. Other ways of using the WHO-FRAX tool for groups of patients would need to be developed.

The presence of a fracture history is a risk for future fracture. However, we have demonstrated that identification of these data alone would miss a sizeable cohort (the majority in this study) of those at risk of future fracture and most notably those at risk of their first fracture.

Risk calculators other than WHO-FRAX have been developed. QFractureScores³⁶ is a newly released fracture risk calculator that might provide an alternative to WHO-FRAX. However, its use in practice has yet to be determined and there are concerns with its construction.³⁷

Both practices in this study had low deprivation status. In practices with higher deprivation, the quality of coding of data is likely to be similar as in this work. However, if the actual data differed, the levels of fracture risk could be different. A lower socioeconomic status has been shown to increase hip fracture incidence³⁸ and also to be associated with lower BMD assessment rates.³⁹

The WHO-FRAX derived fracture probability at which BMD testing or intervention is required, remains controversial. There appears to be wide variation in recommendations.⁴⁰ In the UK it is estimated to be cost-effective to offer intervention at a greater than 7% WHO-FRAX major risk of fracture over 10 years.²⁰ On the other hand, the National Osteoporosis Foundation (USA) recommends that intervention is offered where there is a WHO-FRAX risk of major fracture of greater than 20% over 10 years and hip fracture greater than 3% over 10 years.⁴⁰ Clearly a cost-effective intervention is different from a clinically appropriate intervention level.

What is clinically appropriate remains uncertain. In this study we opted for a greater than 15% level of WHO-FRAX major risk of fracture¹⁸ and/or a >3% risk of hip fracture¹⁹ to identify patients at sufficiently high risk to justify further investigation. These levels are those adopted by the Greater Glasgow and Clyde Osteoporosis Service. Assessment of patients at risk requires measurement of BMD, a risk factor which can be modified by drug therapy.^41,42 BMD is routinely used as an essential prerequisite for identifying patients who might benefit from antiosteoporosis drug treatment.^43,44 Our local policy is not to start treatment on the basis of risk alone, but to refer patients who have ‘high-risk’ WHO-FRAX scores for BMD assessment. This allows assessment of ‘modifiable risk,’ i.e. that part of the risk profile that can be altered by bisphosphonate therapy.

Some groups advocate starting anti-osteoporosis treatment in individuals presenting after fracture without further BMD assessment.^45,46 In this present study, patients presenting with fracture accounted for a total of 19% of the population assessed. It is clear, however, that those who present after a low trauma fracture can be considered at ‘high risk’ for a further event over a 10-year period. This is especially true in those under the age of 60 years at the time of their index fracture. Furthermore, even those individuals who are shown to be at high risk of a further fracture using WHO-FRAX may not have low BMD. In this situation, it may not be possible to modify their high-risk status by bisphosphonate therapy.

Currently the WHO-FRAX algorithm can only be applied on the website one patient at a time. It is easy to see that if the algorithm was applied ‘en-mass’ to data held in primary care, it would allow rapid screening and identification of individuals who might require further bone health assessment. The use of the WHO-FRAX calculator as we have described, may not be an entirely accurate use of the tool and may underestimate fracture risk. We appear to have identified a cohort figure similar to the expected prevalence for osteoporosis. Fracture risk and osteoporosis are not necessarily the same entities but there will be considerable overlap between the two groups.

WHO-FRAX's use has been considered in primary care,⁴⁷ but not yet been applied as a screening tool for fracture risk. However, we have demonstrated that existing data held in primary care can be used to screen rapidly and easily. In well recording practices the application of WHO-FRAX, despite some limitations, can help to identify patients at greatest risk. In this way the missing cohort of those at risk of fracture may be more readily identified and searched for on a regular basis, thereby reducing future fracture related morbidity, mortality and costs.

Funding: Funding for the study was provided through an unrestricted grant from Proctor and Gamble Pharmaceuticals Ltd.

Ethical approval: Not applicable.

Guarantor: Fergus R MacLean.

Contributorship: All authors contributed to the concept, design, data collection, analysis and article completion.

Footnotes

Competing interests: The authors declare they have no competing interests.

Acknowledgements

We thank Prof DH Lawson for stimulating the project and reviewing the manuscript. We also thank Dr D Skelton (School of Health, Glasgow Caledonian University) and Dr PR MacLean for their advice on the manuscript.

References

Center

, Nguyen

, Schneider

, Sambrook

, Eisman

JA.

Mortality after all major types of osteoporotic fracture in men and women: an observational study.

Lancet 1999; 353: 878–82

Tosteson

, Gottlieb

, Radley

, Fisher

, Melton

III . Excess mortality following hip fracture: the role of underlying health status. Osteoporos Int 2007; 18: 1463–72

Randell

, Nguyen

, Bhalerao

, Silverman

, Sambrook

, Eisman

JA.

Deterioration in quality of life following hip fracture: a prospective study.

Osteoporos Int 2000; 11: 460–6

van Staa

, Dennison

, Leufkens

, Cooper

Epidemiology of fractures in England and Wales.

Bone 2001; 29: 517–22

Burge

, Worley

, Johansen

, Bhattacharyya

, Bose

The cost of osteoporotic fractures in the UK: projections for 2000-2020.

J Med Econ 2001; 4: 51–62

Wells

, Cranney

, Peterson

Alendronate for the primary and secondary prevention of osteoporotic fractures in postmenopausal women.

Cochrane Database Syst Rev 2008; CD001155

Kanis

, Burlet

, Cooper

European guidance for the diagnosis and management of osteoporosis in postmenopausal women.

Osteoporos Int 2008; 19: 399–428

Nguyen

, Center

, Eisman

JA.

Osteoporosis: underrated, underdiagnosed and undertreated.

Med J Aust 2004; 180 (Suppl): S18–22

Eisman

, Clapham

, Kehoe

Osteoporosis prevalence and levels of treatment in primary care: the Australian BoneCare Study.

J Bone Miner Res 2004; 19: 1969–75

10.

Kanis

, Johnell

Requirements for DXA for the management of osteoporosis in Europe.

Osteoporos Int 2005; 16: 229–38

11.

Verdijk

, Romeijnders

, Ruskus

, van der

, Pop

VJ.

Validation of the Dutch guidelines for dual X-ray absorptiometry measurement.

Br J Gen Pract 2009; 59: 256–60

12.

Royal College of Physicians. Osteoporosis: Clinical Guidelines for Prevention and Treatment. Royal College of Physicians. 1999. ISBN: 978-1860-160970. See http://bookshop.rcplondon.ac.uk/details.aspx?e=136 (last checked 20 November 2011)

13.

McLellan

, Gallacher

, Fraser

, McQuillian

The fracture liaison service: success of a program for the evaluation and management of patients with osteoporotic fracture.

Osteoporos Int 2003; 14: 1028–34

14.

McLellan

, Reid

, Forbes

Effectiveness of Strategies for the Secondary Prevention of Osteoporotic Fractures in Scotland.

NHS Quality Improvement Scotland 2004. See http://www.healthcareimprovementscotland.org/default.aspx?page=12406 (last checked 20 November 2011)

15.

Moayyeri

The importance and applications of absolute fracture risk estimation in clinical practice and research.

Bone 2009; 45: 154–7

16.

, Chan

, Wood

, Hague

, Valentin

, Van

VJ.

Quality and variability of osteoporosis data in general practice computer records: implications for disease registers.

Public Health 2005; 119: 771–80

17.

Scottish Index of Multiple Deprivation 2009: General Report. 2009. See http://www.scotland.gov.uk/Publications/2009/10/28104046/0 (last checked 20 November 2011)

18.

Borgstrom

, Strom

, Coelho

The cost-effectiveness of risedronate in the UK for the management of osteoporosis using the FRAX(R).

Osteoporos Int 2009; 21: 495–505

19.

Tosteson

, Melton

III , Dawson-Hughes

Cost-effective osteoporosis treatment thresholds: the United States perspective.

Osteoporos Int 2008; 19: 437–47

20.

Kanis

, McCloskey

, Johansson

, Strom

, Borgstrom

, Oden

Case finding for the management of osteoporosis with FRAX - assessment and intervention thresholds for the UK.

Osteoporos Int 2008; 19: 1395–408

21.

Brankin

, Mitchell

, Munro

, Gallagher

What is the prevalence of post-menopausal fragility fracture?

J Bone Miner Res 2005; 20: S265

22.

Kanis

, Johansson

, Oden

A meta-analysis of prior corticosteroid use and fracture risk.

J Bone Miner Res 2004; 19: 893–9

23.

Scottish Intercollegiate Guidelines Network. Management of Osteoporosis. SIGN guideline 71. 2003. See http://www.sign.ac.uk/guidelines/fulltext/71/index.html (last checked 20 November 2011)

24.

Diabetes UK. Diabetes in the UK 2004. http://www.diabetes.org.uk/Professionals/Publications-reports-and-resources/Reports-statistics-andcase-studies/Reports/Diabetes_in_the_UK_2004/2004 (last checked 20 November 2011)

25.

Symmons

, Turner

, Webb

The prevalence of rheumatoid arthritis in the United Kingdom: new estimates for a new century.

Rheumatology (Oxford) 2002; 41: 793–800

26.

Thomas

, Millar-Jones

, Rhodes

, Roberts

, Williams

, Mayberry

JF.

Incidence of Crohn's disease in Cardiff over 60 years: 1986-1990 an update.

Eur J Gastroenterol Hepatol 1995; 7: 401–5

27.

Green

, Flamm

AGA technical review on the evaluation of liver chemistry tests.

Gastroenterology 2002; 123: 1367–84

28.

Nacul

, Soljak

, Meade

Model for estimating the population prevalence of chronic obstructive pulmonary disease: cross sectional data from the Health Survey for England.

Popul Health Metr 2007; 5: 8

29.

Johnston

, Watson

, McMillan

, Sloan

, Love

AH.

Prevalence of coeliac disease in Northern Ireland.

Lancet 1997; 350: 1370

30.

West

, Logan

, Hill

PG.

Seroprevalence, correlates, and characteristics of undetected coeliac disease in England.

Gut 2003; 52: 960–5

31.

Brankin

, Mitchell

, Munro

Closing the osteoporosis management gap in primary care: a secondary prevention of fracture programme.

Curr Med Res Opin 2005; 21: 475–82

32.

Gosden

, Treloar

, Martin

NG.

Prevalence of premature ovarian failure in monozygotic and dizygotic twins.

Hum Reprod 2007; 22: 610–5

33.

Hackley

, Merritt

Osteogenesis imperfecta in the neonate.

Adv Neonatal Care 2008; 8: 21–30

34.

Kanis

, Johansson

, Oden

A family history of fracture and fracture risk: a meta-analysis.

Bone 2004; 35: 1029–37

35.

Sandhu

, Nguyen

, Center

, Pocock

, Eisman

, Nguyen

TV.

Prognosis of fracture: evaluation of predictive accuracy of the FRAX algorithm and Garvan nomogram.

Osteoporos Int 2009; 21: 863–71

36.

Hippisley-Cox

, Coupland

Predicting risk of osteoporotic fracture in men and women in England and Wales: prospective derivation and validation of QFractureScores.

BMJ 2009; 339: b4229

37.

McCloskey

Limited Application of QFracture.

BMJ 2009. See http://www.bmj.com/rapid-response/2011/11/02/limitedapplication-qfracture (last checked 20 November 2011)

38.

Zingmond

, SooHoo

, Silverman

SL.

The role of socioeconomic status on hip fracture.

Osteoporos Int 2006; 17: 1562–8

39.

Demeter

, Reed

, Lix

, MacWilliam

, Leslie

WD.

Socioeconomic status and the utilization of diagnostic imaging in an urban setting.

CMAJ 2005; 173: 1173–7

40.

Dawson-Hughes

, Looker

, Tosteson

, Johansson

, Kanis

, Melton

III . The potential impact of new National Osteoporosis Foundation guidance on treatment patterns. Osteoporos Int 2009; 21: 41–52

41.

Ralston

, de'Lara

, Farquhar

, Gallacher

, Hannan

, McLellan

AR.

NICE on osteoporosis. Women over 75 with fragility fractures should have DEXA.

BMJ 2009; 338: b2340

42.

Oyen

, Gjesdal

, Brudvik

Low-energy distal radius fractures in middle-aged and elderly men and women-the burden of osteoporosis and fracture risk: a study of 1794 consecutive patients.

Osteoporos Int 2009; 21: 1257–67

43.

Cummings

, Black

, Thompson

DE.

Effect of alendronate on risk of fracture in women with low bone density but without vertebral fractures: results from the Fracture Intervention Trial.

JAMA 1998; 280: 2077–82

44.

McClung

, Geusens

, Miller

PD.

Effect of risedronate on the risk of hip fracture in elderly women. Hip Intervention Program Study Group.

N Engl J Med 2001; 344: 333–40

45.

National Osteoporosis Guideline Group. Osteoporosis Clinical Guideline for Prevention and Treatment. See http://www.shef.ac.uk/NOGG/NOGG_Executive_Summary.pdf (last checked 20 November 2011)

46.

National Institute For Clinical Excellence. Alendronate, Etidronate, Risedronate, Raloxifene, Strontium Ranelate and Teriparatide for the Secondary Prevention of Osteoporotic Fragility Fractures in Postmenopausal Women. London: NICE, 2008

47.

Siris

, Baim

, Nattiv

Primary care use of FRAX: absolute fracture risk assessment in postmenopausal women and older men.

Postgrad Med 2010; 122: 82–90