Do we need another Osteoporosis risk tool? Making the case for DXA-MAP

Introduction

Osteoporosis is as one of the most prevalent non-communicable diseases globally^1,2, representing a major source of morbidity, mortality, and healthcare expenditure^3–6. Characterised by reduced bone mineral density (BMD) and increased fracture risk^7–9, osteoporosis affects approximately 300,000 to 500,000 Irish people^5,10. Ireland has one of the highest relative burdens of osteoporosis-related disease in Europe⁵. Due to rising life expectancy, this can only worsen, leading to a substantial burden on the Irish healthcare system⁷. Globally, nearly 200 million fractures occurred in 2019, many of which were likely osteoporosis-related especially among aging populations^7,11. Over 4 million osteoporotic fractures were recorded in Europe in the same year, with direct healthcare costs approximating €60 billion⁷. In Ireland, the financial burden is similarly considerable, with hip fractures alone accounting for the majority of osteoporosis-related healthcare spending⁷. Fragility fractures, especially of the hip, vertebrae, and pelvis, are the most serious consequence of untreated osteoporosis, often resulting in chronic pain, reduced quality of life, disability, loss of independence, and higher risk of mortality^{3–5,12–15}. In 2019, osteoporotic fractures in Ireland accounted for approximately 2% of total healthcare spending^9,15. However, the true cost is likely much higher, potentially twice that amount, as indirect costs are not taken into account. Audits at Galway university hospital show that <20% of fracture related medical admissions included the terms ‘fracture’ or ‘osteoporosis’ in the hospital discharge diagnoses. Despite the significant economic and social burden of osteoporosis, it remains a low priority for many governments, including the Irish government, which doesn’t recognise the need for, or support a formal osteoporosis programme¹⁶.

The absence of a national screening programme in Ireland has resulted in fragmented and poor quality care in diagnostic and treatment practices^15,17. Although Dual-energy X-ray absorptiometry (DXA) remains the gold standard for measuring bone mineral density (BMD) and diagnosing osteoporosis^2,18–20, it is often underused in high-risk groups²¹ and overused in low-risk groups,. Meanwhile, the quality of scan acquisition and reporting vary widely in Ireland today^22–24.

Current guidelines recommend DXA testing for all men and women with a fragility fracture and those with major risk factors for fractures or those who are deemed at high risk for fracture, or where the results could influence treatment²⁵. In a recent study of 3,474 Irish subjects referred for a DXA scan, and 2,090 Irish patients who completed a DXA scan, the majority of patients with osteoporosis or at high risk of fracture did not receive appropriate osteoporosis treatment, while inappropriate prescribing remained substantial (journal paper forthcoming). The undertreatment of those at high-risk and the overtreatment of those at low-risk, undermines the cost-effectiveness and credibility of osteoporosis management programmes¹⁶. Clinical decision-making is further complicated by the absence of national osteoporosis threshold guidelines and poor agreement between existing criteria. Although current clinical guidelines recommend osteoporosis treatment for those with specific risk factors, many eligible patients remain untreated. A recent study comparing five commonly used treatment criteria in 1,254 adults aged 40–90 years found that while 82% met at least one treatment threshold, only 9 individuals (<1%) met all. Eligibility varied widely depending on the criteria used, with agreement between thresholds being markedly inconsistent. The findings underscore the urgent need for harmonised, evidence-based thresholds to help close the osteoporosis treatment gap (journal paper forthcoming).

Despite some debate about who should be screened for osteoporosis by DXA scanning, robust evidence support their use as being clinically and cost effective in primary prevention^2,26–28. Several clinical algorithms are available to help clinicians identify those most likely to benefit from DXA scanning or those who are at risk for osteoporosis, with or without incorporating DXA results^2,29. Fracture risk tools such as FRAX®, Garvan, and OSTi (Osteoporosis Self-Assessment Tool index) can be used to calculate individual risk, however each varies in complexity, population applicability, and predictive power. Although several effective screening tools and treatment options exist, their use remains limited in clinical practise²⁰. A systematic review concluded that no single tool significantly outperforms others²⁹. This conclusion is further supported by clinician studies, which highlight ongoing challenges in the adoption of risk tools; challenges relating to limited familiarity, lack of confidence in interpreting the results and uncertainty in applying said results in clinical decision-making and patient care^30–32. One of our authors, who reviews all DXA requests for DXA scans in our hospital over the past 17 years, noted that <1% of these referrals included the results of any of these algorithms. Empirical evidence from Ireland suggests that clinicians are often unaware of these tools, what they actually mean, or how they should be used in practise.

Compounding these challenges, clinician-patient communication remains a significant barrier to effective risk assessment and management. Clear risk communication is considered essential in primary care^31,33, and a critical component of patient-centred osteoporosis management^34,35. The RICO (Risk Communication in Osteoporosis) study demonstrated that patients often struggle to interpret numerical scores or abstract statistics^20,34, preferring visual representations of their risk level³³. Despite the availability of validated and widely used tools like FRAX®, Garvan, and OSTi, the format and delivery of risk information ‘falls short’. Research shows that the manner in which the information is presented, is just as important as the content itself in ensuring effective patient understanding and informed decision-making³⁴.

Methods

In order to build the case for another osteoporosis risk tool, this paper presents findings from four larger studies that are currently being prepared for publication. They include: (1) an audit of all DXA referrals in our centre, since 2021; (2) retrospective model development and validation of the DXA-MAP tool, (3) a survey of general practitioners (GPs), and (4) a series of public engagement pilot studies.

Quantitative data from the pilot study and surveys were analysed using descriptive statistics. Differences between participant groups were explored using appropriate statistical tests, including the Wilcoxon rank-sum test, Chi-squared test, and Fisher’s exact test. Model performance in the development and validation phases was assessed using standard classification metrics (e.g., AUC-ROC curves). All calculations were computed using R (Version 4.4.2).

Formal ethical approval was obtained from the Clinical Research Ethics Committee, University of Galway (Reference:C.A. 2019). An amendment for an extension was submitted and granted until May 2024. Ethics was also approved by the Sligo Research Ethics committee (Reference: Application no. 660).

Results

Both OST and DXA-MAP models demonstrate robust and comparable performance in discriminating between those with and without osteoporosis among the internal and external validation datasets. In internal validation, both models achieved similar AUCs, with balanced sensitivity and specificity. The DXA-MAP model had a slightly higher AUC compared to OST. In external validation, both the OST and DXA-MAP models showed slightly reduced performance, with AUCs in the range of 74.3% – 74.7%, and maintained good sensitivity and specificity. In terms of predictive accuracy, there was no significant difference between the OST and DXA-MAP models on external validation of the lowest site (i.e. the lowest of the FN and TH site for each person as per ISCD criteria), p=0.16, DeLong test. These results reinforce earlier findings from our group on the utility of OST as a simple and effective screening tool¹⁷. Although OST is widely used globally, and easy to calculate, our clinicians report not knowing about it, when to use it, the need to perform a calculation and difficulty in interpreting the OST score as barriers to its use in practice. Our DXA-MAP tool was developed to not only estimate the probability of osteoporosis, but to address these challenges of use, interpretation and effective communication, while offering comparable predictive performance.

During Medical Grand Rounds in our hospital, 26 healthcare professionals voluntarily participated (<50% of attendees) in an online polling tool. They were asked whether they used risk assessment tools to decide whether or not a patient needed a DXA scan. 35% said “sometimes,” 19% said “always” and 15% said “no.” When asked whether they would consider using a risk assessment tool if they knew which one to use, 31% responded “yes.” Finally, when presented with a validated tool based on Irish data that takes less than 30 seconds to use (i.e. the DXA-MAP tool), nearly 90% said they would use it, while 4% were unsure.

Results from the public engagement pilot studies indicated that both male and female participants reported high levels of concern about osteoporosis. Among males, worry was more concentrated in the highest worry category, whereas females showed a more evenly distributed pattern across worry levels. However, this difference was not statistically significant (Wilcoxon p = .38). Perceived risk of fracture increased with age, with the most notable rise observed between the 50–59 and 60–69 age groups. Public participants who had previously undergone a DXA scan were generally older and more likely to self-report worry about osteoporosis. Statistically significant differences were found between public participants and healthcare professionals in perceived fracture risk (p = .045) and levels of worry about osteoporosis (p = .005), with the public cohort reporting higher levels in both.

Following the piloting of the DXA-MAP tool to members of the public and healthcare professionals, the feedback indicated strong acceptance across both cohorts. Public participants reported that presenting the probability of osteoporosis with a 95% confidence interval improved their understanding of personal risk and provided reassurance, a factor which is considered important for long-term adherence to osteoporosis prevention and treatment plans. These findings align with existing literature advocating for patient-centred approaches to risk communication, particularly in chronic conditions such as osteoporosis, where early intervention is often delayed due to confusion, miscommunication, or system-level barriers⁶. Meanwhile, healthcare professionals noted that the DXA-MAP tool could save time during consultations and reduce confusion in communicating risk. All participants agreed that the DXA-MAP tool held potential for bridging the gap between screening, diagnosis, and actionable patient engagement. The risk communication strategy adopted in the DXA-MAP tool reflects five of the eleven key components of effective communication³³: presenting risk numerically, conveying uncertainty (through confidence intervals), using visual formats (e.g., charts with arrows), providing evaluative labels (e.g., high risk, medium risk, low risk), and personalising probabilities. Regarding worry about osteoporosis before and after using the DXA-MAP tool, results showed that worry levels decreased significantly, particularly among male participants and members of the public. Gender differences were observed: females were more likely to perceive themselves at risk of fracture and showed a stronger alignment between perceived risk and worry, whereas males frequently reported high levels of worry despite low perceived risk. Members of the public reported higher levels of worry and perceived risk compared to healthcare professionals; however, this difference reduced following use of the DXA-MAP tool. Overall, the DXA-MAP tool was perceived as useful and reassuring across all participants, supporting its potential as a clinician-patient communication aid to reduce anxiety, and enhance understanding of osteoporosis risk.

Discussion

Ireland has one of the highest rates of osteoporosis in Europe, a rate which is expected to rise as the population ages. However, in contrast to other European countries, Ireland has neither a national policy programme nor formal recognition of professional training or standards to address this massive public health issue. This presents an impending public health challenge that requires a targeted response. While DXA scanning remains the gold standard for diagnosing osteoporosis, low bone mineral density and osteoporosis in people without prior fractures, particularly following the World Health Organisation’s introduction of densitometric criteria (e.g., T-score ≤ –2.5), access to DXA services remains uneven; this has contributed to both under and over treatment. However, in Ireland today, access to quality DXA remains limited, with widespread issues of inappropriate scanning and poor quality DXA reporting. This represents a growing concern for effective osteoporosis diagnosis and management, especially given that many fractures could be prevented through appropriate assessment and treatment, potentially reducing both patient harm and healthcare costs. Some of these savings could be achieved by reducing over-testing and inappropriate testing and treatment among younger, healthier individuals and providing clinicians with a clear, simple, evidence-based tool to help decide who needs a screening DXA scan.

While several risk assessment tools are available i.e., FRAX®, Garvan, there is currently no single gold standard. A systematic review concluded that no single tool significantly outperforms others²⁹. Furthermore, many tools have not been validated for our population with the exception of the OST index (OST)¹⁷. Despite OST being validated for the Irish population and being easy to use, it is underutilised in clinical practice. One of the authors reported that among more than 10,000 DXA referrals received since publishing OST validation for the Irish population¹⁷, not a single referral has included an OST score. Whilst the algorithm is simple (weight-age)*.2, the main obstacles to its use are; (1) clinicians are unaware of it, (2) clinicians are reluctant to perform calculations in practise, (3) clinicians don’t know how to interpret the result i.e. a number without a context and (4) clinicians struggle to interpret and communicate risk scores in a way that is both meaningful and accessible to patients. Studies on clinician-patient communication highlight the need for clearer, more accessible ways of communicating risk, with patients expressing a preference for visual formats and plain language that contextualise risk scores^33–36. From a clinician context, the absence of national guidelines and the under utilisation of existing tools due to challenges in use and in the interpretability of results, has resulted in the limited integration of such risk assessment tools into practice.

The DXA-MAP tool was developed in response to all of the afore-mentioned challenges. While internal and external validation demonstrated that the predictive performance of the DXA-MAP model was comparable to OST, the goal was not to replace OST, but to enhance its practical utility. The advantage of the DXA-MAP tool lies in its improved usability and ability to support more effective risk communication, as it was designed with a focus on ease of use, simplicity, accessibility and visual communication. Importantly, it addresses the limitations reported by both clinicians and patients. Results are presented in multiple ways, including quantitatively as a percentage with a 95% confidence interval, visually (e.g., arrows on charts), and with qualitative risk labels (e.g., high risk, medium risk and low risk), reflecting the reality that individuals understand and interpret numbers differently³². The numerical presentation of osteoporosis risk aligns with studies which show that patients prefer absolute risk formats (e.g., ‘your probability of osteoporosis is 25%’) over relative risks (e.g., ‘there is a 25% increase in your probability of osteoporosis’)³⁰. Pilot data for our DXA-MAP tool from members of the public show it improves their understanding of risk and provides some reassurance. Clinicians note it has potential to save them time and simplify risk discussions during consultations. The value of the DXA-MAP tool was further supported by significant improvements in understanding and decreased levels of worry, especially among the wider public.

In summary, current osteoporosis risk tools are rarely used in practice in Ireland because they are not well understood and do not communicate risk in a simple, clear understandable format. Our findings indicate that healthcare professionals and members of the public want a tool that is simple, accessible, easy to use and clear. A novel DXA-MAP tool shows robust scientific validity and preliminary feedback suggests results help explain and understand risk and could greatly improve osteoporosis clinician-patient risk communication.

Conclusion

Despite the availability of an array of risk tools (some of which are validated for our population), their uptake in clinical practice in Ireland remains limited. One of the primary barriers to their use, as reported by clinicians, is a difficulty in interpreting risk scores and effectively communicating this risk to their patients. At the same time, patients report that they are rarely provided with personalised, easy-to-understand risk information. In response to this gap, we developed the DXA-MAP tool, with the aim of providing not just another algorithm, but a practical solution to improve the interpretation and communication of results. Internal and external validation demonstrate the DXA-MAP model performs comparably to OST, with no statistically significant differences observed across the external validation dataset. However, the main advantage of the DXA-MAP tool lies in its design whereby risk is communicated with contextual relevance through a multi-modal approach i.e. quantitatively, qualitatively and visually. This multi-modal approach appears to enhance clinician and patient confidence and support shared decision-making, and understanding of risk³². In summary, there is a compelling case for integrating the DXA-MAP tool into routine clinical workflows, to bridge the gap between evidence and action, and to support effective clinician-patient communication.

Ethics approval

All study procedures involving human participants were conducted in accordance with the ethical standards as outlined in the Declaration of Helsinki.

Formal ethical approval was obtained from the Clinical Research Ethics Committee, University of Galway (Reference:C.A. 2019, Chair: Prof. Peter McCarthy). An amendment for an extension was submitted and granted until May 2024. Ethics was also approved by the Sligo Research Ethics committee (Reference: Application no. 660. Chair: Dr. Miriam Sullivan).

Professional and Public engagement events were conducted as part of informal engagement activities for the purposes of understanding: (1) Healthcare concerns about risk tools, and (2) concerns to members of the public and healthcare professionals on osteoporosis risks in addition to feedback on the usability and impact of the DXA-MAP tool. These interactions were not submitted for formal ethical approval, as they did not involve the collection of identifiable personal data, posed minimal risk, and were considered part of routine service evaluation and public involvement. All participants were provided with a clear explanation of the study aims and procedures and gave verbal informed consent prior to voluntary participation. Responses were always fully anonymised.