U.S. flag

An official website of the United States government

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Davis S, Martyn-St James M, Sanderson J, et al. A systematic review and economic evaluation of bisphosphonates for the prevention of fragility fractures. Southampton (UK): NIHR Journals Library; 2016 Oct. (Health Technology Assessment, No. 20.78.)

Cover of A systematic review and economic evaluation of bisphosphonates for the prevention of fragility fractures

A systematic review and economic evaluation of bisphosphonates for the prevention of fragility fractures.

Show details

Chapter 6Discussion

Statement of principal findings

Principal findings: clinical effectiveness

A total of 46 RCTs were identified that provided data for the clinical effectiveness systematic review. Alendronic acid was compared with placebo in 17 RCTs. A daily dose of 2.5 mg of oral ibandronic acid (dose no longer licensed) was compared with placebo in three RCTs and with i.v. administration in one RCT. Daily administration of 2.5 mg of oral ibandronic acid was compared with 150 mg per month of oral ibandronic acid administration in one RCT. Risedronic acid was compared with placebo in 12 RCTs and zoledronic acid was compared with placebo in four RCTs. One RCT compared alendronic acid with 150 mg per month of oral ibandronic acid, five RCTs compared alendronic acid with risedronic acid, one RCT compared zoledronic acid with alendronic acid and one RCT compared zoledronic acid with risedronic acid. The maximum trial duration was 48 months.

The risk of bias associated with the included RCTs was assessed using the Cochrane risk-of-bias instrument. An attrition bias of 10% across treatment groups was evident for 29 (63%) of the included RCTs. Five trials were reported as either open label or single blind and were considered at high risk of performance bias. Blinded outcome assessment was reported by only 13 (28%) trials.

The outcome measures prespecified in the final NICE scope23 were addressed by the included trial evidence to varying degrees. Femoral neck BMD was the most widely reported outcome and fracture was the second most widely reported outcome. AEs were reported by the majority of included trials. Across the included trials there was limited reporting on outcomes of compliance (adherence and persistence), hospitalisation and service use, and quality of life.

A total of 27 RCTs provided suitable fracture data for inclusion in the fracture NMA: nine compared alendronic acid with placebo; two compared 150 mg per month of oral ibandronic acid with placebo; one compared 2.5 mg per day of oral ibandronic acid with placebo; nine compared risedronic acid with placebo; three compared zoledronic acid with placebo; one compared alendronic acid with risedronic acid; one compared 150 mg per month of oral ibandronic acid with alendronic acid; and one compared zoledronic acid with risedronic acid.

A total of 35 RCTs provided suitable femoral neck BMD data for inclusion in the BMD NMA: 12 compared alendronic acid with placebo; one compared 2.5 mg per day of oral ibandronic acid with placebo; one compared 150 mg per month of oral ibandronic acid with placebo; one compared 2.5 mg per day of oral ibandronic acid with 3 mg every 3 months of i.v. ibandronic acid; one compared 2.5 mg per day of oral ibandronic acid with 150 mg per month of oral ibandronic acid; 10 compared risedronic acid with placebo; four compared zoledronic acid with placebo; three compared alendronic acid with risedronic acid; one compared alendronic acid with 150 mg per month of oral ibandronic acid; and one compared zoledronic acid with risedronic acid.

Bone mineral density may be considered a surrogate for fracture outcomes. Analysis of the femoral neck BMD data was of interest in order to confirm the direction of treatment effects. As more studies presented data on femoral neck BMD than any of the individual fracture outcome types, the network analysis also provides more information for assessing treatment effect modifiers.

All treatments were associated with beneficial effects on fractures and femoral neck BMD relative to placebo. HRs for fractures varied from 0.41 to 0.92 depending on treatment and fracture site. For vertebral fractures and percentage change in femoral neck BMD the treatment effects were also statistically significant at a conventional 5% level for all treatments. Pairwise comparisons between treatments indicated that no active treatment was statistically significantly more effective than any other active treatment for fracture outcomes. For vertebral fractures and percentage change in femoral neck BMD, the greatest effect was for zoledronic acid, although in general the ranking of treatments varied for the different outcomes, with the treatments providing broadly similar effects. There was no evidence to suggest different treatment effects according to age or sex.

Assessment of vertebral fractures was based on both clinical and morphometric fractures. Ideally, the effect of assessment method would be assessed through metaregression; however, data for clinical fractures were limited. An analysis of the studies reporting clinical fractures did not provide any evidence to suggest differential treatment effects according to assessment method, although the evidence was limited.

The main analyses were based on a class-effects model such that the bisphosphonates are assumed to be related but not identical. The treatment effects estimated using the class-effects model were broadly similar qualitatively (i.e. direction of effect) and quantitatively (i.e. magnitude of effect) to those estimated using the standard random-effects model but with the treatments effects in the class-effects model shrunk towards the overall bisphosphonate treatment effect. The qualitative effects of treatment (i.e. direction of effect) were the same for the majority of outcome types and treatments from the class effects and standard random-effects models with the exception of zoledronic acid (hip fractures), 150 mg per month of oral ibandronic acid (hip and wrist fractures) and 2.5 mg per day of oral ibandronic acid (non-vertebral fractures). Although the point estimates changed from being relative increases in effect in the standard random-effects model to relative decreases in effect in the class-effects model, there was considerable uncertainty about the true effects as reflected in the CrIs.

Non-vertebral fractures are used as proxy for fractures of the proximal humerus, as this outcome is not commonly reported. Two studies presented results for proximal humerus fractures, both considering the effects of risedronic acid against placebo.70,85 A random-effects meta-analysis of these two studies provided a HR of 0.45 (95% CrI 0.13 to 1.41), which was greater than that estimated for non-vertebral fractures but with considerably more uncertainty.

There were no statistically significant differences between treatments in the incidence of upper GI events associated with any oral bisphosphonate (alendronic acid, risedronic acid or ibandronic acid) compared with placebo when data were pooled across RCTs for each bisphosphonate. However, evidence from one RCT indicated a significantly higher risk of upper GI events in men receiving risedronic acid than in those receiving placebo. Where reported across the RCTs, treatments were prescribed in accordance with the SmPC for oral bisphosphonates to minimise gastric irritation. There was no evidence of significant differences between treatments in mortality across the RCT evidence when data were pooled by bisphosphonate. However, evidence from one RCT indicated that the proportion of men and women dying following hip fracture was significantly higher among those receiving placebo than among those receiving zoledronic acid. There was also no evidence of significant differences between treatments in participants withdrawing because of AEs across the RCT evidence when data were pooled by bisphosphonate. However, evidence from one RCT indicated that the proportion of men withdrawing because of AEs was significantly higher in the alendronic acid group than in the placebo group.

In agreement with the SmPC, there was evidence of influenza-like symptoms associated with zoledronic acid. There was no statistically significant difference in the incidence of atrial fibrillation associated with zoledronic acid compared with placebo (one RCT) or risedronic acid (one RCT). There was no statistically significant difference in the incidence of bone pain associated with zoledronic acid compared with placebo (one RCT) or alendronic acid (one RCT). There was evidence that the risk of eye inflammation in the first 3 days following drug administration was significantly higher for zoledronic acid than for placebo (one RCT). Single RCT evidence indicated no statistically significant difference between zoledronic acid and placebo in the incidence of stroke over 36 months. No RCT evaluating zoledronic acid reported any case of spontaneous osteonecrosis of the jaw in any treatment group during the trial period.

Adverse events of hypocalcaemia and atypical femoral fracture were not reported outcomes by any RCT of any bisphosphonate.

A summary of evidence from systematic reviews that include observational data indicates that the rates of GI toxicity associated with alendronic acid, risedronic acid and oral ibandronic acid are similar to that observed in placebo-treated participants. However, prescription event monitoring study data suggest a high level of reporting of a number of conditions in the first month of therapy with alendronic acid or risedronic acid, particularly those affecting the upper GI tract. Retrospective cohort data also suggest that switching patients who are stabilised on risedronic acid to alendronic acid is associated with an increased risk of GI AEs. Zoledronic acid may be compromised by renal toxicity, and myalgias and arthralgias are evident in the acute phase following i.v. administration. Intravenous bisphosphonates, especially zoledronic acid, are more likely to predispose patients to osteonecrosis of the jaw; however, in addition to bisphosphonate use, there appears to be several other factors involved in the development of osteonecrosis of the jaw (e.g. dental trauma). There is an increased risk of atypical fracture among bisphosphonate users; however, events are rare and long-term bisphosphonate therapy might not be a prerequisite for development of atypical fractures. Moreover, the use of glucocorticoids and proton pump inhibitors is a potentially important risk factor for atypical fracture. Bisphosphonates are associated with serious atrial fibrillation, but heterogeneity of the existing evidence and a paucity of information on some agents preclude any definitive conclusions with respect to risk. The review evidence for the use of bisphosphonates and oesophogeal cancer is equivocal.

Evidence for persistence and adherence reported by RCTs was very limited. Where reported, high levels of compliance, reported as a pill count, were evident over the trial duration. A summary of evidence from systematic reviews including observational data indicates that, although patients using weekly bisphosphonate medication follow their prescribed regimens better than those using daily therapy, overall compliance and persistence rates are suboptimal for postmenopausal women receiving bisphosphonate therapy for the treatment of osteoporosis. Furthermore, one-third to one-half of patients, including men being treated with bisphosphonates for osteoporosis, do not take their medication as directed.

With the exception of the RCTs evaluating bisphosphonates in steroid users, the majority of trials included in the clinical effectiveness systematic review typically excluded patients with underlying conditions or receiving medications that affect bone metabolism. Furthermore, patients with history of, or receiving medication for, upper GI tract disorders were also excluded by the majority of included trials. Therefore, the effects of alendronic acid, ibandronic acid, risedronic acid and zoledronic acid are unknown in these populations.

Principal findings: cost-effectiveness

The de novo economic model estimates that a strategy of no treatment is predicted to have the greatest net benefit for patients with an absolute risk of < 1.5% when using QFracture to estimate absolute risk and valuing a QALY at £20,000. Alendronic acid is predicted to have the maximum INB from 1.5% to 7.2% and risedronic acid is predicted to have the maximum INB from 7.2% upwards. However, the absolute costs and QALY gains are small in patients with low absolute risk, and the PSA suggested that there is considerable uncertainty regarding whether or not no treatment is the optimal strategy until the QFracture score is approximately 5.5% (the mean absolute risk for the eighth risk category for QFracture).

The mean INBs for oral bisphosphonate treatment (alendronic acid, risedronic acid or ibandronic acid) compared with no treatment were positive across all FRAX risk categories. An exact threshold for the absolute risk at which the INB became positive was therefore not available, but the minimum FRAX score in the modelled population was 1.2% and the lowest risk category had a mean absolute risk of 3.1%. Oral ibandronic acid is predicted to have the highest INB compared with no treatment up to 8.6%, with alendronic acid having the highest INB from 8.6% to 38.5% and risedronic acid having the maximum INB > 38.5%. The PSA suggested that there was a low probability of the no-treatment strategy being optimal across all FRAX risk categories when valuing a QALY at £20,000. However, the PSA also demonstrated that there is considerable uncertainty regarding the optimal bisphosphonate treatment with all of the oral bisphosphonates having reasonably similar probabilities of having maximum INB across most of the FRAX risk categories.

Intravenous bisphosphonates (ibandronic acid and zoledronic acid) were predicted to have lower INBs than oral bisphosphonates across all levels of absolute risk when estimated using either QFracture or FRAX. In the highest-risk categories the ICERs for i.v. ibandronic acid and zoledronic acid compared with oral bisphosphonates were consistently > £50,000 per QALY even though the base-case analysis assumed longer durations of persistence for i.v. bisphosphonates than oral bisphosphonates. Although the mean INB compared with no treatment for i.v. ibandronic acid did become positive at very high levels of absolute risk when using QFracture, the results when using FRAX went in the opposite direction. This may be because of the small number of patients and parameter samples informing the estimates at high levels of absolute risk which makes these estimates more uncertain.

The results appeared to be broadly similar across the majority of the structural sensitivity analyses which examined the application of alternative data or assumptions. The results were more favourable to treatment when assuming full persistence with treatment for the intended treatment duration (3 years for zoledronic acid and 5 years for all other bisphosphonates) or when assuming no AEs. The sensitivity analysis examining an AE rate of 30% in the month following initiation of oral bisphosphonate therapy showed that the cost-effectiveness of oral bisphosphonates is very sensitive to the rate of AEs experienced. The INBs compared with no treatment fell below zero (when valuing a QALY at £20,000) for all 10 QFracture risk categories and for all but the highest FRAX risk category when assuming an AE rate of 30% in the first month of oral bisphosphonate treatment.

The structural sensitivity analyses, which varied the way in which the fracture risk was estimated, showed results that were broadly similar for QFracture but slightly less favourable for FRAX, which brought the cost-effectiveness estimates from the QFracture and FRAX model closer together for patients with similar mean absolute risk. We would expect from the way the model is structured that the threshold for cost-effective treatment would be broadly similar across the two risk scores but in the base-case scenario the INBs of bisphosphonates compared with no treatment were higher for FRAX than QFracture for risk categories with similar absolute fracture risk. The fact that the results are similar in these particular structural sensitivity analyses suggests that the base-case analysis may have overestimated the proportion of fractures occurring at the hip for the FRAX model. We suspect this is because we have assumed that the proportion of major osteoporotic fractures which occur at the hip is the same across both risk tools but, in fact, the proportion is lower for FRAX than QFracture in 64% of the modelled population.

Some of the difference in the INBs estimated for QFracture and FRAX risk categories with similar levels of absolute fracture risk may also have occurred because of the different risk scores selecting patients with different characteristics who have different consequences of fracture into risk categories with a similar absolute fracture risk.

Strengths and limitations of the assessment

The clinical effectiveness systematic review was based on rigorous methods, with comprehensive searches for evidence, a good level of consistency between reviewers in study selection and double-checking of data extraction. A formal assessment of methodological quality of included trial was undertaken. Attrition of ≥ 10% across treatment groups was evident for 63% of the included RCTs.

Fracture data were reported for 27 (59%) of the 46 included RCTs and femoral neck BMD data were reported for 35 (76%). However, for fracture data there was variability across the included trials in the skeletal fracture site evaluated, the most frequently evaluated being vertebral fracture. In addition, femoral neck BMD was summarised in study reports as the percentage change from baseline, which is a relative measure of treatment effect and tends to have poor statistical properties. Ideally, for a continuous outcome measure assessed at baseline and post treatment, we would work with the post-treatment response adjusted for baseline in an analysis of covariance.

Network meta-analyses were used to synthesise the evidence to permit a coherent comparison of the efficacy of interventions in terms of fracture and femoral neck BMD. An assumption of the model is that the studies are exchangeable in the sense that we would be prepared to treat any patient in the population with all of the treatments. However, not all treatments are licensed in all patient populations, which means that the studies are not exchangeable, although the analysis follows the scope defined by NICE.23

Adverse event data were widely reported and also supplemented by review evidence of observational data. However, evidence for compliance and concordance was limited in the RCT evidence base and, where reported, was reported as being assessed mainly through pill counts. Evidence for compliance and treatment persistence was limited to review evidence of observational data.

In summary, fracture, BMD and AE data were widely reported. However, these data were limited as there was variability across the included trials in the skeletal fracture site evaluated with a limited number of trials reporting data for the hip. Summary statistics for BMD were not provided by all trials and were extracted from graphical representations. Furthermore, the majority of RCTs were placebo-controlled trials with a limited number of head-to-head comparison trials.

Although the search strategy for this assessment report was comprehensive, the possibility of a publication bias cannot be discounted. A formal assessment of publication bias was not undertaken.

The majority of included trials typically excluded patients with underlying conditions or receiving medications that affect bone metabolism. Furthermore, patients with a history of or who were currently receiving medication for upper GI tract disorders were also excluded by the majority of included trials; therefore, the effects of alendronic acid, ibandronic acid, risedronic acid and zoledronic acid are unknown in these populations.

None of the consultee submissions included a de novo economic evaluation and none of the published economic evaluations compared all five bisphosphonate treatment regimens specified within the scope of this appraisal in a fully incremental analysis as required by the NICE reference case.

The patient-level simulation approach used in the Assessment Group’s model allowed the distribution of patient characteristics to differ across the risk categories providing estimates of cost-effectiveness that have taken into account the differing consequences of fracture in patients with different characteristics. However, the DES modelling approach provides a stochastic estimate of the costs and QALYs gained. We therefore needed to simulate a large number of patients to obtain stable estimates of the cost and benefits of treatment. This was particularly true in the lower-risks groups in the base-case scenario where we reduced the treatment duration to reflect evidence from observational studies on the duration of persistence with bisphosphonate treatment. In order to obtain stable estimates of the costs and QALYs at differing levels of absolute risk, we had to group the patients into broad risk categories. A full incremental analysis has been conducted for each risk category and CEACs have also been provided allowing the uncertainty in the cost-effectiveness to be assessed at different levels of absolute risk. We have also used a non-parametric regression to estimate the relationship between INB and absolute risk across the whole population eligible for risk assessment in CG146.16 From this we have identified treatment thresholds for each treatment for both QFracture and FRAX.

The model generally adheres to the NICE reference case and fully addresses the decision problem set out in the final NICE scope. In particular, the modelling approach used allows intervention thresholds to be linked to absolute risk measured using the two risk assessment tools recommended in CG146 as specified in the scope.23 However, in order to provide a single intervention threshold for each treatment that could be applied across the whole population, we had to assume that all of the bisphosphonate treatment strategies were viable treatment options across all patients eligible for risk assessment within CG146. This would not be true if the licensed indications for each intervention were to be strictly applied.

The de novo economic model is underpinned by a NMA across all drug options which provides a coherent synthesis of the evidence within a single model. Where appropriate and possible, systematic search methods have been used to identify evidence to inform the model’s parameters (efficacy evidence and HRQoL). However, it was not feasible to conduct a full systematic review to identify evidence to inform all model parameters and, therefore, published cost-effectiveness models and published systematic reviews were used to identify appropriate sources of evidence for some model parameters. It is possible that if we had searched systematically we may have found other more appropriate data sources for some model parameters but it is not possible to say whether or not this would have changed the model results.

The main limitations of the economic analysis relate to the assumptions required to populate the model given the data available. In particular, several assumptions were necessary to generate estimates of time to fracture for each fracture type from the estimates of absolute risk provided by the QFracture and FRAX tools.

Uncertainties

Although differential effects were found when comparing the bisphosphonates with placebo, and the effects of the bisphosphonates were generally similar, there was uncertainty about the true treatment effects and some evidence of heterogeneity in treatment effects between studies.

It is uncertain whether or not the cost-effectiveness of bisphosphonate treatment at a particular level of absolute fracture risk would be similar for patients who have been assessed using the FRAX algorithm for patients with known BMD.

Other relevant factors

Although the mean INBs for treatment with oral bisphosphonates are positive at low levels of absolute risk, it is important to note that the absolute costs and benefits are small and the no-treatment strategy has a reasonable probability of being optimal until the QFracture score is above approximately 5.5% (the mean absolute risk for the eighth risk category for QFracture). Therefore, it is possible that patients and clinicians may not consider treatment worthwhile in the lowest-risk patients even though it may be cost-effective.

Copyright © Queen’s Printer and Controller of HMSO 2016. This work was produced by Davis et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Included under terms of UK Non-commercial Government License.

Bookshelf ID: NBK390985
Contents

Views

  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (19M)

In this Page

Other titles in this collection

Recent Activity

ClearTurn OffTurn On

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...