Health economics analysis for FAME: In younger adults with unstable ankle fractures treated with close contact casting, is ankle function not worse than those treated with surgical intervention?

Every day approximately 170 people sustain an ankle fracture in the UK. They may experience pain and physical impairment for several months and years after injury, either through the index injury or from complications of treatment. Prolonged work absence, chronic pain, psychological distress, and later post-traumatic arthritis are all commonly reported.

This study covers only non-complex fractures, which include a wide range of injuries over a wide age range and for which the mechanisms of injury vary substantially. Treatment options are wide, which presents an important challenge to the NHS. Many non-complex fractures get better with minimal clinical intervention; however, they can also appear minor and be easily missed, but lead to potential poor long-term outcome. In the case of non-complex ankle fracture, the broad aim of ankle fracture treatment is to maintain the alignment of the joint whilst the fracture heals and to reduce the risks of problems such as stiffness. More severe injuries to the ankle are routinely treated surgically. However, even with advances in surgery, there remains a risk of complications; for patients experiencing these, the associated loss of function and quality-of-life is considerable. Non-surgical treatment is an alternative to surgery and involves applying a cast carefully shaped to the patient’s ankle to correct and maintain alignment of the joint; the key benefit being a reduction in the frequency of common complications of surgery. The main potential risk of non-surgical treatment is a loss of alignment with a consequent reduction in ankle function.

Fractures that are judged to be unstable are usually treated surgically with the aim of correcting and then stabilising the alignment of the ankle bones in an attempt to ensure good ankle function once the fracture has healed. Even with advances in surgery, there remains a risk of complications. Many of these complications are related to the surgical treatment – failure of bone healing (1%), wound breakdown (9.1%), metal implant failure (1.7%) or irritation from implants requiring removal (1.3%) and infection (2.7%). For those people experiencing complications, the functional loss and decline in quality-of-life are still experienced months and sometimes years after injury.

Non-surgical treatments have the key benefit of avoiding the risks of surgical complications. For example, close contact Casting (CCC) involves applying a cast, carefully shaped to the patient’s ankle, to correct and maintain alignment of the joint through external support. This avoids the need for incisions in the skin and implantation of metalwork, thereby reducing the risk of wound complications, infection, and irritation from implants. The concern with non-surgical treatment, where the opportunity to directly and anatomically realign and fix the bones of the ankle is not realised, is that it may yield inferior outcomes compared with surgery.

However, there is increasing recognition across other orthopaedic conditions that perfect anatomical reconstruction of the bones does not necessarily correlate with improved functional outcomes. The clinical uncertainty here lies in whether non-surgical treatment can yield similar outcomes compared with surgical treatment.

Moreover, 60% of ankle fractures occur in adults less than 60 years of age. The majority of these fractures in younger adults will be treated non-operatively with a standard plaster cast or walking boot. Forty per cent, however, are more severe, and currently treated with an operation; representing around 14,000 surgically treated fractures per annum in the UK. Younger adults typically have a higher functional demand and may have a greater risk of developing late post-traumatic arthritis. It is reasonable to expect that treatments may yield different outcomes in this younger population and that the findings of previous studies may not be generalisable.

Opinion is genuinely divided amongst trauma and orthopaedic surgeons in how best to manage unstable ankle fractures. To University of Oxford's knowledge, there are no existing trials comparing CCC with surgical treatment of unstable ankle fractures in younger adults. There are compelling reasons to believe that outcomes and resource use will be different in younger, working-age adults compared with older people. The risk of complications following surgical treatment in younger, fitter adults may well be lower and poor outcomes therefore less frequent; equally, productivity losses associated with work absence may substantially influence cost-effectiveness in this working-age population. With this substantial instance of disease, and uncertainty in the clinical and cost-effectiveness of the technologies, high quality evidence is required to determine whether the drawbacks of surgical management of ankle fracture are balanced by any improvement in functional outcomes in younger adults and a need to definitively test if non-surgical management can produce similarly acceptable outcomes as surgical management in adults aged 60 years and younger.

The clinical and cost-effectiveness of surgical management of unstable ankle fractures in younger adults was a ‘Top 5 research recommendation’ in the recent guidance by the National Institute for Health and Care Excellence (NICE) and identified as a priority at the joint Royal College of Surgeons and The National Institute for Health Research (NIHR) Research Prioritisation Exercise 2017.

The Fractured Ankle Management Evaluation (FAME) study aims to determine whether ankle function, four months after treatment in patients with unstable ankle fractures treated with close contact casting (an alternative to surgery which uses less padding than a traditional cast and sets the bones by being a close anatomical fit), is not worse than in those treated with surgical intervention, which is the current standard-of-care. The study is conducted as a randomised clinical trial. The overarching objective of this study is to investigate the difference in ankle function, the risk of late complications and comparative cost-effectiveness between the trial treatment groups over five years.

The objectives that are addressed through this data processing are:

1. To assess the longer-term incidence of complications of the two treatments under investigation

2. To validate patient-reported hospital healthcare use collected during the trial against data collected from NHS Hospital Episode Statistics

3. To assess the longer-term cost-effectiveness of close contact casting (CCC) compared with surgery in the patient population of the trial

A randomised clinical trial is the best method to compare treatments to guide the care of patients. Randomisation will be used to produce two groups of patients: those who undergo internal fixation of their fracture, and those who undergo close contact casting (CCC). Patient follow-up will extend to 5 years post-treatment via hospital records. The trial is funded by the National Institute for Health Research (NIHR) Health Technology Assessment, project reference NIHR127273. The project is not part of a wider project, collaboration, or associated work.

Adult patients, aged 60 years and younger, with unstable ankle fractures are identified in daily trauma meetings and fracture clinics and approached for recruitment prior to their treatment. The study includes only adult participants since injuries in children behave differently and should not be grouped with all injuries. Older adults (above 60 years of age) are not included since we already have a trial for this age group (AIM) that has addressed the role of the treatments in this older group with different injury patterns. Treatments are performed in trauma units across the UK by a wide range of surgeons. Details of the surgical treatment, including how the operation is done, implant choice and the recovery programme afterwards is at the discretion of the treating surgeon.

The non-surgical treatment is close-contact casting performed under anaesthetic, a technique which has gained in popularity since the publication of the Ankle Injury Management (AIM) trial. Eight hundred and ninety (445 per arm) participants will be randomly allocated to surgical or non-surgical treatment. Data regarding ankle function, quality-of-life, complications, and healthcare related costs will be collected at eight weeks, four and twelve months and then annually for five years following treatment. The primary outcome measure is patient-reported ankle function at four months from treatment.

The trial team will aim to recruit 890 patients over a 24-month period, from more than 26 hospitals within the UK. Recruitment started in December 2019, and as of July 2022, 580 people have been recruited with recruitment ongoing. Access to data from hospitals not in England will be requested from the relevant data sources (e.g.: ISD Scotland). Every participant for whom data are requested will provide prospective consent to access their personal data.

The data subjects will be the trial participant cohort who fulfil all eligibility criteria as defined below:

Inclusion Criteria

• Patient is able and willing to give informed consent for participation in the trial, and

• Patient is aged 18 to 60 years inclusive with an unstable ankle fracture and

• who in the opinion of the treating surgeon may benefit from surgical treatment with internal fixation.

Exclusion Criteria

The patient may not enter the study if ANY of the following apply:

• The fracture is open.

• The fracture is complicated by local tumour deposits.

• The injury is an isolated fracture of the medial malleolus (The injury only involves the inner side of the ankle joint).

• The index injury occurred more than 14 days prior to recruitment.

• They are unable to adhere to trial procedures.

• Previous randomisation in the current trial.

The University of Oxford relies on GDPR Article 6(1)(e) as the lawful basis for processing the data within this Agreement. There is public interest for patients, healthcare staff and the NHS, as this research will decrease uncertainty and allow standardisation of care and promotion of development of pathways for more efficient and cost-effective care.

The Agreement requires processing of special category data and relies on Article 9(2)(j) as a lawful basis for processing data. Data for this project has been minimised to ensure researchers only have access to the data they require to carry out the statistical and scientific processing of the data and to meet the purpose of the project for which there is a public interest. There is public interest for patients, healthcare staff and the NHS, as this research will decrease uncertainty and allow standardisation of care and promotion of development of pathways for more efficient and cost-effective care. There has been consideration given to any potential ethical or moral issues, and as such an NHS Research Ethics Approval has been granted for this study and all participants have prospectively consented to share their data in line with the processing described in this Data Sharing Agreement (DSA).

NHS Digital datasets to be used include HES Admitted Patient Care, Outpatients, Accident and Emergency and Critical Care, as well as Civil Registration - Deaths and ECDS. Each of the requested datasets provide distinct data that are not available elsewhere and are required to determine outcome necessary to answer the research questions described in the objectives.

Data will be required at the level of the participant in order to construct an adequately explanatory statistical model to address the research questions; all data will be deidentified prior to transfer to University of Oxford. Only consented trial participants in England will be included in the requested cohort. The study team have carried out a multicentre trial in order that the results are generalisable to NHS practice and therefore require data from participants across England.

Not all the study team are employees of Oxford. However, only substantive employees of University of Oxford will be allowed access to any data from NHS Digital.

The research team is minimising data requested to only the trial participant cohort; data for each participant in the cohort will be requested for the period that they are involved in the trial follow-up (up to five years).

In planning the study with patient representatives and gaining NHS Research Ethics approval, the team explored alternative means to address these objectives and the chosen approach was considered both proportionate and appropriate.

This study was developed through collaboration with several stakeholders, most importantly patient representatives. Collaborating stakeholders contributing to the trial development included orthopaedic surgeons, physiotherapists, occupational therapists, trainees in orthopaedic surgery, and patient representatives. The study design was also supported by the British Orthopaedic Association and the UK Orthopaedic Trauma Society. Patient and Public Involvement (PPI)s confirmed the appropriateness of the outcome instrument to measure effectiveness in the defined clinical setting; determined the exact nature of the health technologies; and drafted and revised the text of the project application. PPIs are also involved in the dissemination of the main results of the study.

The University of Oxford's patient representatives will lead dissemination to the patients and carers directly through their extensive network of patient advocacy organisations, which include the Arthritis Research UK (ARUK) Centre for Epidemiology, Wales Centre for Primary and Emergency Care (Including Unscheduled) Care research (PRIME) and the Oxford Link and other local interface organisations. They will help generate a plain language summary for patients and the public.

University of Oxford is the sole data controller who also processes the data for the purposes described in this Agreement.

The University of Oxford is working with other NHS and research organisations to deliver this study:

• University of Bristol, Musculoskeletal Research Unit

• South Tees Hospitals NHS Foundation Trust

• University of Warwick, Clinical Trials Unit

Members of the study team from these organisations will only have access to aggregated data for the purposes of contributing to the oversight, analysis and reporting of the study

The Chief Investigator is substantively employed by Queen Mary University London (QMUL) but is an honorary departmental associate of trauma surgery at the University of Oxford. QMUL have no role in FAME. Co-investigators named in the protocol associated with University of Bristol and University of Warwick will have input on reviewing and approving the health economic analysis plan (HEAP) and interpreting the aggregated results of the analysis, but they will not have any access or involvement in determining the purpose, processing the data, acting on behalf of the data controller, or carrying out any operation on the requested data. University of Oxford only will act in the capacity of data controller to conduct and implement the health economics analysis plan. The trial steering committee (TSC), which includes independent members, provides overall supervision of the trial on behalf of the funder. The data safety and monitoring committee (DSMC) is a group of independent experts external to the trial who assess the progress, conduct, participant safety and, if required critical endpoints of a clinical trial. TSC, and DSMC will not have access to NHS Digital data.

A file of unique identifiers and patient-level identifiers (NHS number, date of birth, postcode and gender) will be sent from the University of Oxford to NHS Digital. The trial team will aim to recruit 890 patients over a 24-month period, from more than 26 hospitals within the UK. Recruitment started in December 2019, and as of July 2022, 580 people had been recruited with recruitment ongoing. The identifiers for newly recruited participants will be sent the NHS Digital to update the cohort used for linkage throughout the length of this Agreement.

NHS Digital will link data (Outpatients (HES OP), HES Accident and Emergency (AE), HES Critical Care (CC), HES Admitted Patient Care (APC), Emergency Care Data Set (ECDS)) alongside Civil Registration Data (CRD) for each patient identified in the cohort using the matching data file (containing NHS number, date of birth, gender and postcode) to the unique identifier. The HES data will be at patient-level and deidentified. The deidentified HES, ECDS and CRD data, which will include special category health data, with the linked study ID will be sent to the University of Oxford not including patient-level identifiers, postcode, date of birth, gender and NHS number.

There will be no subsequent flows of data.

The processing organisation is University of Oxford.

The dataset will be transferred to the University of Oxford via Secure Electronic File Transfer Service (SEFT) and securely stored and processed as described below. One processing/storage site will be used: Botnar Research Centre. Within the University of Oxford Big Health Data Group, data is held in the Secure Computing Room, which is strictly access-controlled and supported by comprehensive Information Governance Policy and procedure. Data are encrypted and stored in a safe. Data are accessed on non-networked computers.

The trial team will initially prepare the linkage file as described above. On receipt of the linked, deidentified data from NHS Digital the trial team will carry out the following processing described in the trial protocol:

A prospective economic evaluation at 12 months, conducted from an NHS and personal social services perspective, will be integrated into the trial design. All economic analyses will be performed following the statistical analysis plan by analysing groups, both according to the treatment participants received (per-protocol basis) and according to the treatment they were assigned at the time of randomisation (intention-to-treat basis). Given that this economic evaluation will be conducted alongside a trial intending to test whether close contact casting is not worse than surgery (non-inferiority) as opposed to testing whether it is better than, findings from the per-protocol analysis are expected to be more conservative. The economic evaluation will estimate the difference in the cost of resource inputs used by participants in the two arms of the trial, allowing comparisons to be made between the surgical and non-surgical treatment of unstable ankle fractures in adults aged 60 years or less and enabling costs and consequences to be compared.

Consequences of interest will be quality-adjusted life years (QALYs) (a measure of a person’s health which combines both their length of life as well as the quality of their life) at 12 months and clinical primary outcome of the Olerud-Molander Ankle Score (OMAS) score (reporting how well the ankle works after the injury) at four months. All other secondary outcomes will also be reported in a cost-consequences table and follow NICE guidelines. Given the importance of returning to work and usual activities to the younger patients with ankle fracture, University of Oxford will separately report productivity losses from paid and unpaid work and need of informal care.

Resources used to deliver the treatment in both arms will be valued using a macro-costing approach when possible, using department of health and social care reference costs for secondary care resources, unit costs for health and social care for community resources, average weekly earnings for productivity losses and patient self-reported expenses. Costs will be reported grouped by secondary care resource use, community-based resource use (including primary and social care) and productivity losses (including lost time off-work, leisure and informal care). The aggregate health and social care resource use and costs at 12 months will also be reported. Costs and QALYs will be estimated using regression analyses controlling for baseline scores and trial stratification variables.

In the economic analyses, given the number and nature of resource use data collection methods and time-points, University of Oxford expect the amount of missing data to be considerable. Multiple imputation methods will be used to substitute those missing data values with estimates based on the pattern of responses of those who did answer, both in the per protocol and intention to treat (ITT) analyses. Cost categories and health outcomes will be jointly imputed if computationally feasible and the imputed primary outcome data estimates will be used for the statistical analysis. To calculate costs, HES data will be used to identify the instances of inpatient stays, outpatient consultations, and A&E visits by trial participants as measures of healthcare resource use, which will be costed based on estimated Healthcare Resource Groups (HRGs). This will allow for a comparison of long-term clinical outcomes such as complications (also identified in HES) and costs associated with close contact casting (CCC) with regard to surgical treatment.

The results of the economic evaluation will be reported both in a table showing the cost and the effectiveness as well as in representative figures (cost-effectiveness planes). The outcome of the cost-effectiveness is cost per QALY gained (incremental cost-effectiveness ratio - ICER) of CCC compared with surgery in the patient population of the trial. The ICER is calculated based on the equation: ICER= Incremental costs / Incremental QALYs.

For healthcare interventions to be considered cost-effective (i.e., providing sufficient value for money to merit their provision) in England, they should have an incremental cost per additional QALY of less than £20,000–£30,000. The ICER will therefore provide a measure of the value for money of using close contact casting compared to surgery for patients with unstable ankle fractures. The trial team will derive incremental net monetary benefit (NMB) statistics. This is calculated by subtracting the cost from the monetary equivalent of the health benefits, estimated applying a standard exchange of QALYs by GB£ threshold used by NICE of £20,000 and £30,000 per QALY but also a lower threshold of £10,000/QALY. The team will replicate the analysis on a very large number of datasets (~1,000) made up of trial participants picked at random, allowing for participants to be picked more than once for each dataset (called non-parametric bootstrap estimation) to reflect the possibility that real-world samples could have a different distribution of participants to the trial sample. This will allow to derive 95% confidence intervals (CIs) for mean cost differences between the trial groups and to calculate 95% (bootstrapped) CIs for incremental net monetary benefit statistics.

A series of sensitivity analyses will be undertaken to explore the implications of uncertainty around the costing and methodological assumptions on the incremental net monetary benefit statistics and to consider the broader issue of the generalisability of the study results. One such sensitivity analysis will involve adopting a societal perspective for the economic evaluation, which will incorporate direct costs to trial participants, informal care provided by family and friends and productivity losses.

These analyses will not be reported in the initial publications of the trial results (limited to one year follow-up) but will be reported in a separate publication at the end of the five-year follow-up period.

The long-term follow-up data will be collected to achieve three objectives: firstly, longer-term clinical effectiveness of the two treatments under investigation will be assessed; secondly, University of Oxford will validate patient-reported hospital healthcare use collected during the trial against data collected from HES; and finally, University of Oxford will assess the 5-year cost-effectiveness of CCC compared with surgery.

Inpatient, outpatient and A&E attendances during the 12-month trial duration will be compared to what was reported by patients through clinical reporting forms (CRFs). Relevant administrative database records will be identified using Office of Population Censuses and Surveys version 4 (OPCS-4) and International Classification of Diseases (ICD-10) codes corresponding to the CRF wording used for collection of hospital resource use. As the validation will compare the number of attendances to the various hospital services, missing data will not be a problem as a record of hospital attendance in HES will be the only marker required to identify the use of resources.

For the purposes of the trial analyses the trial team will only process linked, pseudonymised data. In order that this dataset can be created, identifiable data will be provided to NHS Digital for data linkage. A bespoke trial cohort will be generated from the trial database and sent to the data controller containing participant health service number, date of birth, gender and postcode as well as a unique trial identifier for linkage. NHS Digital will link the cohort to the relevant civil register of deaths and administrative databases in their jurisdiction.

The data will not be linked to any other data and only the linkages described are permitted under this Agreement. All outputs will be aggregated with small numbers suppressed in line with the HES analysis guide to reduce the risk of re-identification. No attempt will be made in the processing to re-identify individuals.

As mentioned before, data received from NHS Digital will be kept in a designated, locked, secure data processing office with standalone computers, where no other external data including those of the trial can be brought in or processed. This guarantees that NHS Digital and trial datasets will be kept completely separate and never linked back.

Data processing will only be carried out by substantive employees of the University of Oxford who have been appropriately trained in data protection and confidentiality.

All organisations party to this Agreement must comply with the Data Sharing Framework Contract requirements, including those regarding the use (and purposes of that use) by Personnel (as defined within the Data Sharing Framework Contract, i.e., employees, agents and contractors of the Data Recipient who may have access to that data).

Data will only be accessible to these employees in a designated, locked, secure data processing office with standalone computers in accordance with the data security policies of the University of Oxford, Big Health Data Group (NDORMS).

At the conclusion of this study, the University of Oxford will hope to have provided the most robust evidence available to infer whether ankle function in patients with unstable ankle fractures treated with close contact casting, is not worse than in those treated with surgical intervention. Patients and members of the public will help design a publicity strategy so that the results of the study are distributed outside of the routine scientific literature.

A report will be produced, which will inform the full update of fractures (non-complex): assessment and management guidance (NICE Guidance NG38) for healthcare professionals in 2027. Plain English outputs will include papers and web and blog media.

All outputs will be aggregated with small numbers suppressed in line with the HES analysis guide. Data will be aggregated and presented at the level of the randomised treatment arm. Applicable data within cells will be suppressed if they are small values to reduce the risk of re-identification.

No participant-level data falling under this Agreement will be shared with any third-party.

The dissemination strategy will consist of three strands. The first will ensure that patients and the public are informed of the trial results; the second will engage practitioners and health-care providers, and the third will inform national guideline and policymakers.

Patients, patient advocacy groups & members of the public:

Patient representatives will lead dissemination to the patients and carers directly through their extensive network of patient advocacy organisations. They will help generate a plain language summary for patients and the public. This document will be available in paper copy, podcast and as a blog. An abstract will be submitted to the biannual “Involve” Conference. Posters will also be prepared with the Patient and Public Involvement (PPI) team for inclusion at any workshop or conference where relevant PPI is being discussed. In addition, to disseminate directly to study participants, findings will be more widely available locally through posters in appropriate outpatient rooms and liaising with identified service user groups.

Health care providers: the trial team will work with the Oxford National Institute for Health Research Biomedical Research Centre (NIHR BRC) and Collaboration for Leadership in Applied Health Research (CLARHC) media teams to maximise the reach of the press and publicity outputs from this study. University of Oxford have costed the application to include five free-to-access publication in the mainstream literature. The final results will be submitted for presentations at annual meetings of the British Orthopaedic Association (BOA) and the Orthopaedic Trauma Society (OTS). University of Oxford will present the findings to the entire NHS via the NHS national electronic Library for Health (NHS Evidence). International ‘reach’ of the published research findings will be supplemented by presentations at high visibility meetings such as the Orthopaedic Trauma Association (OTA) Annual Meeting (United States) and BOA Annual Congress (Europe). In addition, University of Oxford are developing complementary systems incorporating non-traditional media. The chief investigator has been developing an enhanced web presence through blogging on the leading UK trauma and orthopaedic websites. These blogs engage both trauma and research communities. They have been very successful and have provided a means for rapid dissemination. University of Oxford plan to expand this activity into additional subject-specific and general blogs such as the British Medical Journal (BMJ).

National guidelines: University of Oxford will use their established network involvement to disseminate these research findings. These include the NIHR Clinical Research Network, and specialist interest groups (BOA/OTS/OTA/ European Federation of National Associations of Orthopaedics and Traumatology (EFORT)). University of Oxford will alert the relevant NICE standing committee to the results of the trial by notifying their surveillance team.

The study team is due to report in March 2023 and inform the full update to NICE Guidance NG38 in 2027.

The outcomes of this project are hoped to lead to updating NICE guidance on non-complex fractures.

This is highly likely to be the only definitive trial addressing the research question and it is the therefore expected to directly influence national clinical practice within five years. These outputs would be expected to directly benefit patients, healthcare staff and the NHS by decreasing uncertainty, allowing standardisation of care and promoting the development of pathways for better care by providing greater information and guidance regarding the treatment of non-complex fractures.

In addition, the health economic evaluation results on the cost-effectiveness of both treatment strategies will provide decision-makers with evidence to inform commissioning decisions. In this way, University of Oxford expect the results of this trial to improve value-for-money of treatments for ankle fracture in the NHS.

The realisation of benefits will begin immediately once the outputs are delivered. The study is not in support of a PhD/post graduate research study.

This is a new study and there are therefore no yielded benefits to report.