Protocol for a trial-based economic evaluation analysis of a complex digital health intervention including a computerised decision support tool: the iFraP intervention

Illustrative case: iFraP intervention rationale and development

Osteoporosis is a disease that compromises bone structure, making bones weak and more likely to break.¹⁶ The drug treatments for osteoporosis revolve around fracture risk reduction through bone strengthening. There are different types of osteoporosis medicine given either by tablet or injection (including bisphosphonates, RANK ligand inhibitor; and anabolic agents such as recombinant parathyroid hormone and anti-sclerostin monoclonal antibodies) with calcium and/or vitamin D supplements acting as treatment adjunct.¹⁷

For osteoporosis, despite the considerable disease burden and the numerous treatment options, there is still a considerable treatment gap¹⁸ (i.e. proportion of people in whom osteoporosis medicine is recommended by clinical guidelines, but do not receive treatment). The treatment gap arises because people at risk are either not identified or offered treatment, or are offered treatment, but decide not to take it. The care gap is a newly proposed term that emerged during the development of the iFraP intervention to address the implicit assumption that all patients recommended medicine should take it regardless of their preferences.¹⁸ This term frames the gap (and possible solutions) as a clinician issue rather than a patient issue, and emphasizes the importance of clinician behaviours to facilitate informed, shared decision-making between healthcare professionals (HCPs) and patients. The reasons behind osteoporosis undertreatment are multiple, including: patients’ concerns about long term bisphosphonates efficacy¹⁷ and safety^19,20; HCPs’ and patients’ uncertainty about treatment action mechanism¹⁹; and, overconfidence with routinely used risk calculators (under certain circumstances these have been associated with a risk of underestimating the fracture risk).²¹ To an extent, promoting better patient-clinician communication on bone health and treatment options could contribute to address some of these challenges.

Some existing osteoporosis DSTs aim at improving patient-clinician communication. However, these have not met the required international quality criteria and have as yet, failed to demonstrate effectiveness at improving drug initiation and/or persistence (collectively described as adherence).^22,23 Building on these findings, our team developed the Improving uptake of Fracture Prevention drug treatments (iFraP) intervention, a complex DHI, featuring a CDST (i.e., iFraP DHT) with an interactive representation of individual fracture risk; information about risk factors for osteoporosis and bone density; and, accessible explanations of bone health and treatment risk/benefit profile. In addition, a training package for clinicians on how to use the tool and consultation skills is provided, along with additional resources from the CDST to refer to after the consultation. For simplicity, this enhanced consultation will henceforth be referred to as iFraP-consultation.

A detailed description of iFraP’s development is provided elsewhere.¹⁵ Briefly, iFraP’s overall conceptualisation and development followed the MRC complex intervention development framework²⁴ and included: i) an evidence synthesis and evaluation of the available online patient information,²⁵ assessed using a modified version of the International Patient Decision Aids (IPDAS) checklist²⁶; ii) an assessment of existing decision tools in osteoporosis²³; iii) a Delphi study to inform the model content of the consultations²⁷ and iv) focus groups and individual interviews with patients and health professionals.²⁸ iFraP was developed collaboratively, involving bone specialists, nurses, people with lived experience, GPs, behavioural psychologists, representatives from Health Literacy UK, the Royal Osteoporosis Society, academics, and health economists.

iFraP trial outline

The iFraP team designed the trial to evaluate the effect of the iFraP intervention (i.e. an iFraP consultation delivered as part of Fracture Liaison Services (FLS)) on patient-reported ease in decision-making about osteoporosis medicine. The iFraP trial was designed to examine the experience of care and clinical effectiveness, and to assess the within-trial cost-effectiveness and RVoI of the iFraP intervention compared to usual FLS care. FLS are services which enact secondary fracture prevention and assess and diagnose people at high risk of osteoporosis (i.e. people aged 50+ who have recently sustained a fragility fracture (i.e. a fracture sustained following a fall from standing height or less)). The iFraP trial is a multicentre, two-arm, individually randomised RCT, with parallel process evaluation and health economic evaluation. Participants are randomised in a 1:1 ratio, using blocked randomisation stratified by FLS. The planned sample size is 380 patients who have sustained a fragility fracture, have been referred to FLS; and, are scheduled to receive an FLS consultation.²⁹

Study population

Adults at increased risk of osteoporosis who are eligible for an FLS consultation (i.e. people aged 50+ years who have had a previous fragility fracture). As part of their FLS consultation, some of these patients receive an osteoporosis diagnosis, and some receive a recommendation to initiate osteoporosis treatment.

Study centres

Four FLS study sites: Oxford, Portsmouth, Stoke-on-Trent, and Wolverhampton.

Intervention and comparator

The intervention (iFraP-consultation delivered in FLS) is a consultation delivered by FLS HCPs with the aid of a dynamic, interactive, and patient-tailored CDST to communicate individual fracture risk, bone health information and treatment recommendations to patients. The clinicians in the intervention arm will receive a dedicated training course and will partake in a 3-hour role play session with experts to familiarize themselves with the new consultation. Lastly, additional information resources will be given to both the patient and their GP after the consultation.

The comparator is usual FLS care within the NHS, which does not involve CDST-supported discussions. FLS HCPs in the comparator arm will not have access to iFraP’s training and resources.

Recruitment

Recruitment began on 31^st March 2023 and is due to end on 28^th November 2024.

Type of economic evaluation

Within-trial cost-effectiveness analysis of the iFraP intervention compared with usual FLS care and RVoI analysis.

Study perspective

The perspective for the analysis will be that of the English National Health Service (NHS) and Personal Social Services.

Costing year

The costing year will be 2023. Main sources will be the 85^th volume of the British National Formulary (BNF) (Covering the March–September 2023 period) and the 2023 edition of the Unit Costs of Health and Social Care manual (PSSRU).

Time horizon

Due to the within-trial nature of the analysis, we will employ a three-month time horizon, equivalent to the length of the trial follow-up.

Discount rate

Given that our time horizon will cover less than a year, we will not discount costs or health outcomes. This is equivalent to applying a 0% discount rate.³⁰

Health outcomes

The EQ-5D-5L questionnaire is a standardised tool developed by the EuroQoL group, used to estimate generic health-related quality of life (HRQoL).³¹ In the questionnaire, respondents are asked to rate their severity level in the five areas of life (dimensions) of mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. In this version of the questionnaire, there are 5 possible answers to each question, equating to (1) no problems, (2) slight problems, (3) moderate problems, (4) severe problems, or (5) inability in the domain. The final profile of the respondents is a sequence of 5 digits (e.g. 12322), corresponding to the answers in the 5 dimensions. Patient participants will self-report their health status at baseline, 2-week follow-up, and 3-month follow-up.

Resource use and costs

There are four main categories of costs: health care (services) utilisation, medications use, iFraP intervention, and medication side effects. Self-reported healthcare utilisation and medication use data will be collected via patient questionnaire administered at baseline and 3 months. Osteoporosis medication initiation and persistence (i.e. adherence) will be cross-referenced with hospital medical record review data. Further details of this can be found in the clinical protocol for the iFraP trial.³²

Intervention cost

Micro-costing will be used to estimate the unit cost associated with the development and delivery of the iFraP intervention. This will include decision support tool development cost and training costs. The consultation cost for the iFraP- and usual-FLS consultation will not be estimated as they are equivalent across study arms.

Bayesian rapid value of information analysis (RVoI)

This analysis will evaluate the expected value of perfect information, which captures the hypothetical value of simultaneously eliminating all uncertainty from the economic evaluation. Our analysis will be conceptualized as RVoI, in a framework similar to the one used in research prioritization.³³

Handling missing data

After examining the data, we will evaluate eventual missing data strategies to implement, following the recommendations presented in Faria et al.³⁴ Specifically, we expect to see a relatively low amount of missing data at baseline, making it possible to impute the missing values by using the group mean for any missing covariate. We will test the assumption of missing-at-random mechanism as described in Cro et al.³⁵ and then inspect the missingness pattern in covariates and outcomes. If a monotonic pattern is found in the missing data over time, we will rely on Inverse Probability Weighting (IPW). If not, we will use multiple imputation (MI) methods.

From EQ-5D-5L to utilities

To estimate utility scores, we will follow NICE recommendations³⁶ and convert EQ-5D-5L scores into EQ-5D-3L ones using van Hout et al.³⁷ mapping function.

Within-trial cost-effectiveness analysis

The cost-effectiveness analysis will estimate Incremental Cost effectiveness Ratio (ICER) and compare it with recommended cost-effectiveness thresholds.³⁸ Utility values will be described using beta-based regression models to manage the multimodality and limited acceptable range of values.³⁹ Costs will be modelled using Generalized Linear Models (GLM), to handle the characteristic skewness of cost data, as suggested in Mihaylova et al.⁴⁰ To account for the correlation between costs and effectiveness, the incremental costs and effectiveness will be estimated by bootstrap techniques, which will also be used to estimate the relevant confidence intervals.^41,42 Analyses will be conducted on the statistical software R.

Bayesian rapid value of information analysis (RVoI)

In brief, the RVoI evaluation revolves around describing the intervention’s net benefit without specifying a full economic model, instead comparing only the relative effectiveness of the intervention and its costs. Then, the uncertainty in relative effect and baseline event rates are described by suitable statistical distributions, constructed either from to the literature or by expert elicitation. Afterwards, a large number of individual values (realisations) are sampled from said distribution to estimate the health benefit distributions. The analysis of the curves enables the estimation of the consequences of uncertainty for the intervention. Following the procedures described in Glynn,³³ we present in Table 1 the minimum evidence required to perform a RVoI analysis.

Strengths

This paper outlines the protocol for designing a within-trial economic evaluation of a DHI (specifically, a CDST). This within-trial evaluation is expected to be associated with high internal validity and to minimise the impact of selection bias on our estimates of uptake of osteoporosis medicine (primarily oral bisphosphonates). To explore the decision uncertainty associated with the within-trial cost-effectiveness of iFraP to enhance osteoporosis medicine adherence we will conduct a RVoI analysis. An electronic search of trial-based economic evaluation of DHIs protocols identified only two protocols and neither of them included a within-trial RVoI analysis.^44,45 Similarly, no RVoI analysis was found in published protocols of within-trial cost-effectiveness analyses of DSTs.^46,47

Limitations

Evaluation of the cost-effectiveness of improving uptake and adherence with osteoporosis medicine presents both a short-term and a lifetime decision problem. The within-trial cost-effectiveness analysis described in this protocol will allow us to robustly explore the short-term impact of the iFraP intervention on osteoporosis medicine uptake. Estimating the value associated with improving adherence with osteoporosis medicine over a lifetime horizon in England would require a model-based economic evaluation. Funding restrictions however, limited our ability to design and implement such an analysis. To explore iFraP’s long-term impact on adherence with osteoporosis medicine, future research may focus on reanalysing existing model-based economic evaluations of osteoporosis medicine using uptake and adherence data from the iFraP trial. A long-term model would facilitate exploration of the value of conducting further primary research using Bayesian value of information analysis (VoI).