The objective of this study was to assess the cost-effectiveness of IBEX BH as an opportunistic screening tool of fracture risk determined from a radiograph of the distal forearm in men and women in the UK compared with current usual care. The analysis was conducted using a health economic model, consisting of a decision tree and a Markov simulation model following men and women aged 50 and older with forearm radiograph and potential subsequent osteoporosis treatment. Opportunistic screening with IBEX BH was assumed to entail higher proportion of patients treated for osteoporosis compared with current usual care (26.9% vs. 6.1% in fractured patients, 17.6% vs. 2.7% in non-fractured patients).
Averaging cost-effectiveness results over women and men, with and without baseline fracture, age group and baseline BMD T-score, IBEX BH lead to 0.013 additional QALYs and cost reduction of £109 per patient. The analysis included additional costs of screening related to increased number of GP referrals and pharmaceutical treatments, but not costs related to IBEX BH software. The analysis may be updated in the future when the investment for the NHS per patient has been determined.
Sensitivity analyses showed that the results were robust when varying several parameters in the model. Results were most sensitive to treatment length and time horizon. Fracture risk and cost-effectiveness are highly dependent on age, sex, and other clinical risk factors including aBMD and fracture prevalence. Separate analyses over risk factors were conducted, showing that, incremental QALYs varied between 0.001 in patients with higher T-score (-1.0 or higher) without fracture at baseline to 0.031 in eldest women aged 70 + with fracture and lower T-score (-4.0 or lower). Screening with IBEX BH was cost-effective in all patient groups (combinations of risk factors) at willingness-to-pay threshold £30,000, and cost-saving in most cases.
The operating point on the ROC curve provided in the OFFER1 study (13) was chosen to match the high sensitivity of FRAX (0.93) and a specificity of 0.89 in the base case analysis. While a higher proportion treated leads to additional cost-savings and QALYs gained, it is desirable to not overwhelm down-stream services like DXA, which is already troubled with long waiting times. At lower sensitivity, IBEX BH strategy was associated with a lower QALY gain than the base case and less cost-savings (shown in Supplementary material) in terms of avoided fracture costs but with less intervention costs inflicted on healthcare. The model simulated a heterogenous patient population which is representative of individuals undergoing wrist X-ray. Cost-effectiveness of screening and osteoporosis treatment is, as shown in this paper and in many previous health economic analyses, dependent on prevalence of risk factors and fracture risk (18). This is also reflected in the intervention threshold described by for example NOGG’s guidelines. In this analysis, the ICER in younger women and men (around 50–60 years) without prior fracture was around £20,000–25,000 meaning that screening in this group is cost-effective at a willingness-to-pay threshold of £30,000 but not £20,000. Therefore, an age-dependent threshold for opportunistic screening may be warranted. A sensitivity analysis was conducted where we averaged cost-effectiveness results on the basis of NOGG intervention thresholds and the results indicate that base case results were robust to different weighting methods.
Our findings can be compared with previous analyses of other types of fracture prevention programs in the UK. In an analysis in 2009 by the Department of Health, an estimated £290,708 over a 5-year period in NHS acute and community service was saved by introducing an FLS and treating 77% of 767 patients with hip, vertebral, wrist and humerus fractures (44). This corresponded to a £8.5 million cost-saving on the national level over five years. McLellan et al. conducted an economic evaluation of West Glasgow FLS in 2011 in which FLS increased treatment rate after fragility fracture to 69% from 19% vs. usual care, saving 18 fractures, 22 QALYs and £312,000 fracture cost reductions per 1,000 patients (45). Turner et al. reported a cost-effectiveness analysis of a screening program in women aged 70–85 in the UK who were randomised to either usual care or screening with FRAX and potential BMD measurement. In the screening strategy, 15% received osteoporosis treatment vs. 4% in the usual care strategy. Over a 5-year period, number of QALYs was numerically but not statistically significantly higher in the screening group vs. usual care (0.0237, 95% confidence interval − 0.0034 to 0.0508) and fracture costs were reduced by around £42 (46, 47). Differences between our study and previous analyses in QALYs gained and fracture costs avoided may be explained by differences in prevalence of different clinical risk factors in patient population.
Simplifications are always necessary in health economic modelling leading to some uncertainties. In this model, uncertainties mainly related to decision tree data inputs. The model simulated a heterogenous population in which care pathway probabilities were averaged and sensitivity and specificity were independent of risk factors (sex, age, T-score, and prior fracture). A limitation with this approach is that in reality, referral and treatment probabilities differ among fracture risk profiles. There was a lack of data on patients who had been referred for fracture risk assessment but did not receive DXA or osteoporosis treatment. We assumed that 100% of patients who did not have DXA after wrist DR but got referred for a bone health assessment received treatment. The referral and treatment probabilities were however calibrated such that the proportion treated matched the proportions in the hospital data. A consequence of this modelling approach is that increasing specificity of IBEX BH did not lead to the expected increase in cost savings of avoiding referrals and DXA in those who did not need treatment, which is a limitation of the model. The analysis may be updated in the future when more detailed information on treatment pathway becomes available. The drug distribution and fracture risk reduction for osteoporosis treatment was simplified and reduced to alendronate, risedronate, zoledronate, denosumab, raloxifene, and teriparatide, based on overall distribution in a large UK sample. Additional pharmaceutical treatments for osteoporosis are available in the UK but were not included in the model due to lack of data on usage. The model construct has some limitations. The hierarchical structure may to some extent underestimate the number of less-severe fracture types, most notably wrist fractures as they are at the bottom of the hierarchy. The cohort approach does not allow tracking of patients, entailing that quality of life and cost impact of multiple fractures are not included.
An individual-state simulation approach could have addressed these uncertainties in the model construct, but such models are burden by first-order uncertainty introducing random noise that can distort result interpretation. Another limitation relates to cost of implementation and quality of life impact of screening. Screening has been shown to have a small to moderate negative impact on quality of life in other disease areas such as cancer (48). The impact in osteoporosis screening has not, to our knowledge, been quantified before, but could be non-negligible. Research has been made into the acceptability of opportunistic screening with IBEX BH showing that patients and the public were generally positive and accepting of the product (49). Despite the inevitable simplifications of the model, the findings are robust as demonstrated by the extensive sensitivity analysis. The cost-effectiveness analysis is based on a published modelling framework used and validated in several previous studies, and the results are in line with what could be expected based previous cost-effectiveness analyses of screening strategies in the UK.
Opportunistic screening during routine wrist radiograph could be a cost-effective instrument in addressing the osteoporosis treatment gap. Most available fracture prevention programs will only identify patients after they have suffered a fracture. At the same time, provision of prevention programs and access to DXA is low and unequally distributed. Around 25% wait longer than 6 weeks for DXA, with large variation across UK regions (up to 70% wait longer than 6 weeks in some regions) (50). The product addresses several of current challenges, by providing early identification, integration with existing care pathways and healthcare equipment requiring no additional imaging or appointments to patients, and a clear benefit to patients at risk of suffering fragility fractures.