Evaluation of the effectiveness and cost-effectiveness of lightweight fibreglass heel casts in the management of ulcers of the heel in diabetes: a randomised controlled trial

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 09/01/53. The contractual start date was in February 2012. The draft report began editorial review in March 2016 and was accepted for publication in December 2016. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

none

Copyright statement

© Queen’s Printer and Controller of HMSO 2017. This work was produced by Jeffcoate 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.

Protocol amendment

A major amendment to the protocol was approved on 29 March 2012. Prior to that date, if there was a protocol violation from non-use of the heel cast by a participant allocated to the intervention arm, the participant was withdrawn and took no further part in the study. After the amendment took effect, such a protocol violation was noted but the participant was not withdrawn. On the same date, the protocol was amended such that adverse events (AEs) were no longer logged, and safety was documented only by the use of adverse device effects (ADEs) and serious adverse device effects (SADEs). The number of participants who were randomised before and after this protocol amendment was 133 and 396, respectively. The amended protocol has been published. 1

Participants

People with diabetes mellitus and ulcers of the heel attending specialist centres in the UK were screened for inclusion in the study. The ulcers of the heel were required to be of National Pressure Ulcer Advisory Panel/European Pressure Ulcer Advisory Panel (NPUAP/EPUAP) grades 2, 3 or 4 (Box 1); to be affecting the skin below the malleoli but overlying the calcaneum inferiorly, posteriorly, medially or laterally; and to have been present for at least 2 weeks. The eligibility criteria, shown in Box 2, were chosen to be as inclusive as possible; neither soft tissue infection nor mild or moderate degrees of limb ischaemia were contraindications to inclusion. Those who met the criteria for inclusion and who provided informed written consent were recruited and randomised to either the intervention arm, that is, to receive treatment with a lightweight, moulded, fibreglass heel cast together with continuing usual care, or to the control arm, that is, to continue with usual care alone (Box 3). If a person had more than one ulcer that fulfilled the selection criteria, only one ulcer (generally the largest and most clinically significant) was selected as the index ulcer for the purpose of the study. All data were entered into an electronic case report form (eCRF), which also incorporated the randomisation tool.

Baseline clinical assessment

Intervention

Both groups received usual care delivered by a specialist service designed for the management of ulcers of the foot in diabetes mellitus. In addition to usual care, those in the intervention group had a lightweight, fibreglass cast moulded to their heel. Each cast was lined and flexible at the edges but reinforced over the ulcerated area and was intended to be thin enough to be worn inside footwear or another offloading device. It was applied over the primary dressing and a single protective layer of Softban^® (BSN Medical, Hull, UK) or equivalent bandage. The cast was then held in place by bandaging. The choice of primary dressing and any other secondary dressing was at the discretion of the participating centre. The cast was removable and was reapplied over the primary dressing with each dressing change until it needed to be replaced when it was soiled, lost or otherwise unusable. The design of the cast was based on that reported by the group originally reporting a good outcome with the use of such casts,16 and all researchers involved in the clinical management of ulcers at each centre were trained by a single specialist podiatrist in how to make heel casts according to the study specific procedure. Researcher performance was assessed every 6 months by a designated competent researcher, whose own performance was also assessed twice over the course of the study.

Study conduct

Following randomisation, participants were asked to attend the specialist clinic for review every 2 weeks. The ulcer was cleaned with local sharp debridement if required. It was then dressed using the primary dressing of choice of the participant’s usual carer. Participants in the intervention group then had their heel cast repositioned (a new cast having been made if necessary) and held in place with a bandage as described above. Any intercurrent infection, deterioration or other clinical problem was treated in accordance with usual practice.

If the ulcer was thought to be healed, it was checked by an observer who was blind to randomisation group and it was reviewed after a further 2 and 4 weeks. If the ulcer broke down within 4 weeks, the person was asked to continue in the study, attending the clinic every 2 weeks. If healing was confirmed 4 weeks after initial healing, the person was asked to attend only the research visits at week 12 (assuming that this had not already passed) and week 24. If an ulcer was first judged to be healed at week 22 or 24, the participant remained in the study until healing had or had not been confirmed by a blinded observer after 4 weeks. A participant’s failure to attend two consecutive fortnightly visits or a total of three visits during the 24-week follow-up period was regarded as a protocol violation necessitating that person’s withdrawal from the study. Failure of those in the intervention group to use the heel cast for > 7 consecutive days or for a cumulative total of > 14 days was regarded as a protocol violation, but – following the protocol amendment of 29 March 2012 (see Protocol amendment) – did not necessitate their withdrawal from the study.

Withdrawal

Participants were removed from the study if they (1) withdrew consent to participate, (2) lost capacity, (3) were recruited in error, (4) did not attend the required number of follow-up visits or (5) were lost to follow-up. Prior to the protocol amendment in March 2012, participants randomised to the intervention arm were also withdrawn if they violated the protocol with regard to use of the cast (see Protocol amendment).

Objectives

The primary objective was to determine whether or not the use of a heel cast in addition to usual care was associated with a higher incidence of healing at or before 24 weeks than usual care alone.

Outcomes

Sample size

The expected percentages of healed ulcers at or before 24 weeks in the control group and the treatment group were 40% and 55%, respectively. With two-sided significance level of 5% (α = 0.05), power = 80% and estimated non-collection of primary outcome data of 25%, a total of 496 patients to be randomised was required in order to achieve 186 in each group for the primary analysis.

Randomisation

Randomisation was stratified by ulcer grade (NPUAP/EPUAP grade 2, 3 or 4; see Box 1) and by ulcer area (25–100 mm² or > 100 mm²), using randomly permuted blocks of randomly varying size. Randomisation was undertaken by Nottingham Clinical Trials Unit, using a secure web-based system to ensure allocation concealment. This system was embedded within the eCRF, and, once the clinical data were entered at the time of the randomisation visit, the allocation was revealed immediately to the researcher online.

Blinding

Owing to the nature of the intervention, it was not possible to blind either the participant or the clinical researcher. However, all of the confirmation assessments of healing and laboratory measures were undertaken by an observer who was blind to randomisation group. In addition, all images of ulcer tracings used to document change in ulcer area were measured by a single, central blinded observer.

Statistical analysis

The analysis and reporting of the trial was in accordance with the Consolidated Standards of Reporting Trials (CONSORT) guidelines. 23 Analyses were detailed in a statistical analysis plan (SAP), which was finalised prior to database lock but after publication of the trial protocol. Participants were analysed according to their randomised allocation regardless of their adherence to their allocated intervention and without imputation for missing data for all primary and secondary outcomes [in keeping with the intention-to-treat (ITT) principle]. All analyses were conducted using Stata^® version 13.1 (StataCorp LP, College Station, TX, USA).

Health economic analysis

The health economic evaluation originally set out to estimate the cost-effectiveness of the use of lightweight heel casts versus usual care by developing a decision-analytic model across a series of time horizons that reflect the management of patients with diabetic ulcers of the heel. A health economic analysis plan was constructed in accordance with the trial SAP as follows:

1. Carry out a within-trial analysis of cost-effectiveness (including cost–utility) of the use of lightweight heel casts versus usual care.

2. Depending on the within-trial results, carry out an analysis of the longer-term cost-effectiveness over agreed time horizons with a budget impact analysis for 1-year and 5-year periods, which will compare the costs to the NHS of the use of heel casts plus usual care with usual care alone in the management of ulcers of the heel in diabetes mellitus.

In accordance with the NICE reference case,24 the perspective adopted was a UK NHS and Personal Social Services (PSS) one. The primary end point for the economic analysis was an incremental cost per quality-adjusted life-year (QALY), and the secondary end point was an estimation of incremental cost per percentage of additional healed ulcers at 24 weeks. Base case results were calculated using a series of sensitivity analyses, undertaken to assess the degree to which variations in parameter estimates would affect the relative cost-effectiveness ratios. Bootstrapping was undertaken to address the uncertainty associated with point estimates of costs and outcomes, and to produce CIs for the incremental cost-effectiveness ratio (ICER). Cost-effectiveness acceptability curves (CEACs) were generated to give a range of cost-effectiveness thresholds for decision-makers. A detailed description of the health economics methods is provided in Appendix 4, including the justification for the health economic analysis being constrained to a within-trial analysis.

Safety end points

The study population was one that was relatively older and that had a high prevalence of comorbidities, including renal, cardiac and cerebrovascular diseases. Moreover, ulcers of the foot may worsen, potentially resulting in hospital admission. Although no specific safety issues were foreseen with the use of the heel cast, significant events were listed among the secondary outcome measures. Other unexpected ADEs were recorded and, if considered serious (SADEs), were reported to the sponsor in accordance with the principles of GCP.

Adverse events

Adverse events were documented only in the period prior the protocol amendment dated 29 March 2012. Those AEs that were documented were mentioned either spontaneously or in response to questioning. The clinical researcher assessed the causal relationship of the AE to the use of the fibreglass heel cast (if used). In addition, AEs were rated according to severity – mild, moderate or severe – using predefined criteria.

Adverse device effects

Following the change to the protocol on 29 March 2012, data were collected on ADEs and SADEs. An ADE was regarded as any untoward and unintended consequence of the use of the device, including any effect resulting from insufficiencies or inadequacies in the instructions for its use and any effect resulting from user error or accident. The researcher was required to grade the severity of any ADE as ‘mild’, ‘moderate’ or ‘severe’ and to assess the causal relationship of the ADE to the device as ‘none’, ‘possible’ or ‘probable’ using prespecified criteria. Foreseeable ADEs were those related to worsening of the clinical state of the ulcer and were reported as secondary outcomes. These included increase in ulcer area, infection, major and minor amputation, ulcer recurrence and secondary ulceration on either foot. Other foreseeable patient-related AEs were also recorded as secondary outcomes, including an increase in pain, worsening mood or function, hospital admission (relating primarily to the heel ulcer) and death from pre-existing medical conditions.

Serious adverse device effects

These were defined as any ADE that resulted in death, life-threatening illness or injury, hospitalisation or additional medical or surgical intervention. The causal relationship between the SADE and the use of the device was documented as ‘none’, ‘possible’ or ‘probable’ using prespecified criteria.

Patient-related AEs that were unrelated were also recorded as secondary outcomes. These included an increase in pain, worsening mood or function, hospital admission (relating primarily to the heel ulcer) and death from pre-existing medical conditions.

Participant flow

The details of participant flow are shown in Figure 1. From 1915 people who were screened at participating specialist units and who had diabetes mellitus complicated by heel ulcers, 509 participants were recruited. Two hundred and fifty-three were randomised to continue with usual care (control), and 256 were randomised to the intervention arm (although the actual use of a heel cast was later recorded in only 244). A total of 84 participants (16.5%) did not complete the study (44 in the intervention group and 40 in the control group), but a primary outcome was available for the remaining 425 (212 in the intervention group and 213 in the control group).

FIGURE 1.

Consolidated Standards of Reporting Trials diagram.

Recruitment

Recruitment started in April 2011 and continued until September 2014. Follow-up was completed in March 2015. Specialist NHS centres routinely involved in the management of diabetic foot disease were included progressively during the active phase of the study. The total number of involved centres was 35 (Table 1).

Baseline data

Study quality

Protocol violations

The numbers of protocol violations leading to failure to complete the study are given in the CONSORT diagram (see Figure 1). The details of all protocol violations are listed in Appendix 6.

Withdrawals

The total withdrawals are given in the CONSORT diagram (see Figure 1). The detailed reasons for withdrawal are listed in Appendix 7.

Primary outcome

Ninety-four out of 212 (44%) ulcers healed in the intervention group, compared with 80 out of 213 (37%) in the usual care group (OR 1.42, 95% CI 0.95 to 2.14; p = 0.088). The risk difference was 8% (95% CI –1% to 17%; p = 0.087) (Table 11). These data suggest that any effect of the intervention is less than that used in designing the trial (risk difference 15%). The possibility of no effect or even a very small harmful effect cannot be ruled out using conventional 95% CIs. Further adjustment for sex and baseline mobility type (due to baseline imbalance) had minimal impact and resulted in OR 1.42 (95% CI 0.94 to 2.15).

Secondary outcomes

Time to healing

The Kaplan–Meier curve for time to healing in the two groups is illustrated in Figure 2. The Cox regression analysis suggested some evidence that participants in the intervention arm healed faster than those in the usual care arm (adjusted hazard ratio 1.30, 95% CI 0.97 to 1.75; p = 0.083).

FIGURE 2.

Kaplan–Meier plot of time to healing.

Change in ulcer area

The mean ulcer area in each of the two groups is plotted against time in Figure 3. The mean change from baseline is shown in Figure 4. It should be noted that these plots include all participants who remained unhealed at each time point.

FIGURE 3.

Mean (95% CI) ulcer area over time for unhealed ulcers.

FIGURE 4.

Change of ulcer area from baseline for unhealed ulcers.

Ancillary analyses

Serious adverse events, adverse events, serious adverse device effects and adverse device effects

Data on serious adverse events (SAEs) and AEs were collected until the time of the protocol amendment (29 March 2012). There was no clear evidence of a difference in the incidence of AEs between the two groups, as shown in Table 21. Six AEs were, however, judged to be ‘probably related’, and five were judged to be ‘possibly related’ to the use of the device. Two AEs were ‘possibly’ or ‘probably’ attributed to the use of the device in the control group, in which the device was not used. The reason for this attribution by the clinical researchers is not clear. Similarly, there was no clear evidence of a difference between groups in the incidence of SAEs, and only one episode was judged ‘possibly related’ to the use of the device. Full details of AEs and SAEs are listed in Appendix 7.

Data on ADEs were collected from 29 March 2012 and are summarised in Table 22. A total of 18 ADEs were recorded in the control group, in which the participants were not exposed to the device. There were 40 ADEs in the intervention group, of which only 25 were judged to be related to the use of the device. There were no SADEs in either the intervention or the control groups. The full details of ADEs that occurred during the study are listed in Appendix 8.

Health economic analysis

One-way sensitivity analysis

Incremental cost-effectiveness ratios were recalculated based on the lower and upper 95% CIs for the intervention mean cost differences reported. The only parameter change that estimated that the lightweight heel cast plus usual care was more cost-effective (£9057.89 per QALY gain) than usual care alone occurred when the lower-bound estimate of the total mean cost of the intervention was used. Within the PP analysis, the ICER was £5164.91 per QALY gain for the lightweight heel cast plus usual care versus usual care alone.

Figure 5 represents the cost-effectiveness plane for all patients of estimates resulting from bootstrapping. The vertical axis of the plane shows the incremental total costs, and the horizontal axis shows the QALY gains. This illustrates how the lightweight heel cast plus usual care is dominated by usual care alone, with the majority of results occurring in the north-west quadrant of the plane.

FIGURE 5.

Cost-effectiveness plane for all patients of estimates resulting from bootstrapping.

The CEAC for the base case is shown in Figure 6. The CEAC shows the probability (expressed as a percentage) of the use of lightweight heel cast plus usual care being a cost-effective intervention against different willingness-to-pay (WTP) values. Based on a NICE threshold of < £20,000 per QALY gain, the lightweight heel cast plus usual care would be 5.2%, with a negative net benefit of –£106 (95% CI –£241 to £19) when a QALY is valued at £20,000.

FIGURE 6.

Cost-effectiveness acceptability curve, sensitivity analysis: probability of the HEEL cast being cost-effective.

Limitations

A decision was made to base this pragmatic trial on a comparison between standard care alone and standard care plus the use of heel casts, because the intention was to determine whether or not the inclusion of heel casts in routine practice in expert centres resulted in clinical improvement. This meant that the comparator treatment was deliberately chosen as the usual standard of care in the centres selected on the basis of their specialist interest in the management of foot disease in diabetes mellitus. And even though the criteria of care were agreed (see Box 3), there was considerable variation documented in the use of offloading. 15 This was, however, assumed to be a reflection of current expert practice in UK. It should also be noted that there is currently no evidence to suggest that ulcers of the heel should be managed with non-removable offloading.

An error that was not detected during the course of the study was the failure of some of the clinical researchers to correctly report the results of the ABPI measurements. However, this had no impact on the conduct of the study or on the interpretation of the results. Another error was the failure to document the prevalence of infection in the two groups at the time of randomisation, even though it was recorded at every other visit, and the use of antibiotics was recorded when appropriate. There is no reason to suspect that the distribution of infection differed between the two groups.

There were minor problems relating to the stratification used at the time of randomisation; however, these were detected early and remedied. Very occasional discrepancies found between the area assessed by researchers at the time of randomisation and the area later determined by a central blinded observer may be attributable to the less precise methods of measurement used by the researchers. This resulted in a total of 24 participants being recruited in error because the ulcer area at baseline was ultimately judged to be < 25 mm².

The results of the trial suggest that the incidence of AEs was greater in the intervention group than in the usual care group. It should, however, be noted that AEs (which were recorded only up until the protocol change in 29 March 2012) were recorded by researchers who were not blind to randomisation group; therefore, the data are susceptible to bias. Moreover, none of the observed differences was statistically significant.

Ancillary analysis

An ancillary analysis was also undertaken to determine whether or not the reduction in cross-sectional area (with ulcers from both the intervention and control arms combined) by 4 weeks was associated with eventual healing within 24 weeks, as has previously been reported for other types of ulcer of the foot in diabetes mellitus. 25 There was strong evidence of an association, but ROC analysis failed to identify any clear threshold value for change in ulcer area at 4 weeks that predicted eventual healing. This issue would benefit from further research that may inform patient counselling and influence management decisions.

Health economic analysis

There were small differences in costs between the intervention and usual care arms, but none was statistically significant. When costs were adjusted, the magnitude of differences in cost became smaller. The results of the incremental cost per QALY analysis did not provide evidence of cost-effectiveness. When the results of the incremental costs were expressed per percentage increase in likelihood of achieving an additional healed ulcer, there were differences between the adjusted and unadjusted analyses. With further examination of the impact on changes in parameter variation around the adjusted base case, there was wide variation in the results. This uncertainty should be clearly noted in any interpretation of these results, particularly in the light of the clinical effectiveness results.

Because of the complex nature of diabetes mellitus, it is possible that other factors played a part in the resources consumed, something that has been identified as an issue in accurately costing the inpatient management of diabetic foot ulcers. 26 Some of the challenges encountered in obtaining accurate data on resource utilisation must also be acknowledged. Although a patient diary allowed for timely data collection, this method is, inevitably, rather crude; however, any resultant errors in assigning costs will have applied to both study groups. It was not possible to explore the costs beyond those to the NHS (e.g. costs to the patient, loss of productivity) because of limitations of the data collected. One interesting finding is that the baseline utilities in both groups appeared to be much lower, and the utility gains smaller, than in other published estimates of diabetes mellitus models and cost–utility evaluations. 27 Overall, however, the results of the health economic analysis did not provide evidence to substantiate the cost-effectiveness of using lightweight heel casts in the management of heel ulcers in diabetes mellitus.

Patient and public involvement

The authors endorse the principles of patient and public involvement, and users/members of the public have been involved at every stage of this project from the planning, the application for funding and the trial conduct to the review of the results.

Contributions of authors

The study was planned by William Jeffcoate (Diabetologist and Clinical Triallist) and Frances Game (Diabetologist and Clinical Triallist). Patricia Price (Pro-Vice Chancellor, University of Cardiff and Clinical Triallist) prepared the original SAP.

Ceri Phillips (Health Economist) prepared the original health economics analysis plan.

William Jeffcoate, Frances Game, Patricia Price, Alison Musgrove (Research Podiatrist) and Ceri Phillips were co-applicants for funding of the study by the Health Technology Assessment programme.

The SAP was later revised by Alan Montgomery (Professor of Medical Statistics and Clinical Trials), Lucy Bradshaw (Statistician) and Wei Tan (Statistician).

The conduct of the trial was managed by William Jeffcoate, Frances Game and Vivienne Turtle-Savage (Trial Manager).

Statistical analysis was undertaken by Wei Tan, Lucy Bradshaw and Alan Montgomery.

Health economic analysis was undertaken by Deborah Fitzsimmons (Professor of Health Outcome Research) and Angela Farr (Researcher in Health Economics), with the assistance of Thomas Winfield (Researcher in Health Economics).

The first draft of the manuscript was prepared by William Jeffcoate, and all authors reviewed and corrected this and subsequent drafts and approved the final version.

Publications

Data sharing statement

The authors are keen to share anonymised data with scientific colleagues. Available data can be obtained by contacting the corresponding author.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health.

Introduction

This appendix describes in detail the health economic objectives and methods used for the health economic evaluation. A summary of the health economic methods used to evaluate the cost-effectiveness and cost implications of lightweight fibreglass heel casts was provided in the published study protocol. 1 This was supplemented by the development of a health economic analysis plan, developed alongside the SAP. This appendix also provides the rationale for a major amendment to the published health economic protocol, in which longer-term evaluation of the cost-effectiveness of lightweight fibreglass heel casts and budget impact analysis were not undertaken. As this substantial amendment to the published protocol could be questioned, a full rationale for this is presented.

Summary of the decision problem

As discussed in Chapter 1, ulcers of the heel can cause substantial pain and suffering in people affected with diabetes mellitus. Diabetic foot ulcers are extremely costly and have been estimated to account for 0.7% of the total UK NHS budget. With the consequences of unhealed and/or recurrent ulcers leading to further health interventions (including amputation), establishing the cost-effectiveness and clinical effectiveness of new interventions is warranted. Although other health economic analyses have assessed the cost-effectiveness of a range of orthotic devices within the context of patient management strategies,28,29 the orthotic devices assessed have largely been intended for use as preventative strategies across the spectrum of risk categories for developing an ulcer of the foot, as opposed to being specifically intended for use in the management of an established ulcer of the heel.

A rapid review of the health economic evidence, completed as part of the preparatory stages of the health economic analysis, failed to locate specific studies of the cost-effectiveness of lightweight fibreglass heel casts specific to the patient population described within the trial protocol and setting (either the UK NHS or another health-care system). Thus, a health economic analysis, alongside a prospective randomised controlled trial, can provide evidence to support decision-makers in the assessment of whether or not the use of lightweight fibreglass heel casts is a beneficial intervention in the management and care of people with a diabetic ulcer of the heel within the UK NHS setting.

Original protocol intentions

In the published protocol,1 it was stated that cost-effectiveness would be assessed by developing a decision-analytic model to estimate costs and health outcomes, including the percentage of healed ulcers and QALYs gained. ICERs and cost–utility ratios would be generated for a series of time horizons (including a lifetime perspective) that reflect the management of people affected by diabetic ulcers of the heel. It was also stated that a budget impact analysis would be undertaken.

Summary of the major protocol change to the published protocol

Based on the clinical effectiveness results, the plan was to undertake further modelling to assess the longer-term costs and outcomes of the intervention compared with usual care. The proposed structure was based on a Markov model with a series of health states to represent healed and unhealed target ulcer, amputation and death, with the recently published NICE economic evaluation of the cost-effectiveness of providing custom orthotic footwear to patients at low, moderate and high risk of developing foot ulcers identified as a possible model for adaption. 26 The adaption proposed would focus on the trial population as the entry point, that is, those who are already affected, with the trial results being used as the primary source to estimate required inputs, such as healing time and number of unhealed ulcers. Longer-term impact would be considered on the basis of the trial results, published literature as part of a rapid review of the literature (including assessment of the evidence used in the recent NICE model26) and in discussion with the trial team. A sensitivity analysis was undertaken to account for the uncertainty in key parameters (one-way sensitivity analysis) and a probabilistic sensitivity analysis was undertaken to quantify the joint uncertainty in parameter estimates. The aim of undertaking this longer-term modelling was to capture a more accurate estimation of the cost-effectiveness based on a more realistic explanation of the clinical pathway and to capture health states (and associated costs/utilities) that may not have been experienced owing to the short duration of the trial.

Prior to the further development of the model and running of the analysis, the feasibility of undertaking this longer-term modelling was discussed with the Trial Steering Committee and trial team after initial consideration of the main clinical results. The evaluation of further long-term modelling was tested against the feasibility that (a) the intervention would be able to demonstrate clinical effectiveness in the short term, (b) the trial results would provide a realistic estimation of health states attributed to the intervention (different healing rates/amputation, etc. , and the additional costs and time that these would incur) and (c) the utilities derived as a result of the intervention could be confidently estimated over a longer period of time.

However, in the light of the clinical effectiveness findings, a decision was made to focus on reporting the within-trial analysis only. The specific reasons were:

1. On the basis of the trial, there was no evidence that the use of a heel cast could be seen as clinically effective.

2. The results of the within-trial health economic analysis showed no evidence of cost-effectiveness, based on the results of the incremental cost per QALY analysis.

3. With a lack of evidence of short-term benefits from using the heel cast, there was no clinical evidence or opinion that these would or could translate into longer-term benefits.

The justification for undertaking a budget impact analysis was also discussed. Based on the trial findings, and given the lack of support for the use of lightweight heel casts within the NHS, the Trial Steering Committee and the trial team decided that the budget impact analysis would be an arbitrary one.

Methods

Collection of resource utilisation and costs

Two main costs were assessed: first, the implementation costs associated with the manufacture and use of the lightweight heel cast device, and, second, the costs associated with health-care utilisation during the trial period. In addition, the assessment of costs associated with time away from home or work as a result of the management of the heel ulcer were included as part of the data collection.

All assumptions made in the calculation of costs were verified by the trial team to ensure that they were appropriate and commensurate with the trial protocol and/or standard clinical practice. Only resource usage and associated costs, as part of the management of the target diabetic heel ulcer, were considered.

Calculating the unit cost of revascularisation therapy

As the type, complexity and mode of hospital delivery of revascularisation treatment was unknown, the weighted average cost of open procedures on blood vessels of the lower leg was calculated as shown in Table 33.

Calculating the unit cost of falls relating to the foot with the target ulcer that resulted in a hospital admission

As the details of injuries sustained due to falls were not clearly captured and the trial population was considered to be more susceptible to injury due to a fall, non-elective long-stay Healthcare Resource Group codes were used to calculate a weighted average unit cost as shown in Table 34.

Costing methodology

Hospital admission rates were taken from the secondary outcomes data, and more details regarding the nature, complexity and length of stay of each hospital admission for amputations, revascularisation, falls and ‘other’ reasons (primarily related to systemic infections) was sought from the SAE files. However, the SAE data collection was largely incomplete for patients who had reported a hospital admission and had not been routinely collected. Little information was collected describing the type, the complexity or the mode of delivery (i.e. elective, non-elective long/short stay) of a revascularisation procedure. Therefore, the ‘total Healthcare Resource Group activity’, combining all procedures, was used to calculate the weighted average. It was assumed that falls were accidental injuries and that hospital admission was non-elective in nature. A weighted average cost of non-elective long stay admissions was calculated using full consultant episodes.

For this reason, already compiled unit costs from the literature and national reference costs were used to cost hospital admissions (see Table 32). For every trial participant, treatment costs were calculated by multiplying each individual resource use by an appropriate unit cost. These costs were added to produce a total cost for each participant during the trial. These totals were then summated for each trial arm.

Outcomes used in the economic analysis

Two outcomes were used in the evaluation of cost-effectiveness, (1) the percentage of healed ulcers at or before 24 weeks and (2) the EQ-5D-3L, in order to generate QALYs.

Health economic analysis

Sixty-five cases

In 65 cases, ABPI was not recorded but was likely to have been > 1.4 because the reading was said to be too high or the vessel non-compressible, calcified or equivalent.

Twenty-nine cases

In 29 cases, ABPI could not be measured because of pain, ulceration, fragile skin, wound dressing or oedema.

Fifty-eight cases

In 58 cases, no clear clinical reason was provided.

Implementation of the intervention and costs

The implementation of the intervention incurred initial costs of training podiatrists in each participating centre and of producing the lightweight heel cast device for each patient, in addition to the cost of the device itself. The training costs were estimated from discussion with the trial team, based on the estimation that 30 minutes of training delivered by a specialist podiatrist would be required at each of the 35 centres, with two podiatrists in each centre receiving the training. that 30 minutes of training delivered by a specialist podiatrist would be required at each of the 35 centres, with two podiatrists in each centre receiving the training. The initial cost of producing the first heel cast for each participant was calculated for all participants in the intervention group.

During the trial, participants in the intervention group may also have received a subsequent Heels device owing to damage or other reasons. The additional costs of subsequent Heels devices were captured from the patient diaries, and thus relied on participants self-reporting that they had received a new one. In the 115 participants who reported receiving at least one replacement device, a wide range in number of replacement devices was reported (median, 3; range 1–24), see Table 36. The total number of replacement devices (n = 513), with podiatrist time included, was used to estimate the costs of replacement.

The implementation costs for the intervention are summarised in Table 37. The mean implementation cost for the intervention was estimated to be £61.18.

Life span of the first heel cast applied at baseline/week 0

The number of days between the first heel cast being applied to a patient’s heel ulcer at baseline (week 0) and the first date a patient reported that a new heel cast had been made (alongside the date of a dressing change) was examined, and the results are displayed in Table 38, which shows the number of days before a new heel cast was reported by any patient with no differentiation between grade of ulcer. In Table 39, the number of days reported when a new heel cast was applied is examined by the grade of ulcer. The median number of days before a new heel cast was reported was 14 for all grades of ulcers. The mean number of days was 30 (mean days for grade 2 ulcers = 27, grade 3= 32.5, grade 4 = 26). After this initial reporting of receiving a new heel cast, it is difficult to tell if subsequent reports of new heel casts are a reflection of a patient continuing to receive necessary new heel casts, or if a patient is continuing to report they have a new heel cast – the same one. In the majority of cases, there is continued reporting of new heel casts every 2 weeks, which has likely led to an overestimation of the number of replacement heel casts required.

Summary of resource utilisation and costs associated with HEELs in primary and outpatient care

Health-care consultations (primary care and hospital outpatient visits)

Aside from reporting the mode of dressing change for the target heel ulcer, patients were asked if they had consulted any other health-care practitioners regarding the care of their heel ulcer. Table 43 summarises the health-care consultations reported by participants for a consultation regarding their target heel ulcer. These consultations may have occurred in the same week as a dressing change performed by another health-care practitioner (such as a district nurse or a podiatrist), or may have occurred without a dressing change being reported. Many patients reported a dressing change and a consultation with the same health-care professional during one week. It was assumed in such cases that the dressing change and the consultation (for example with a district nurse or a podiatrist) occurred at the same time. In such cases, a unit cost was applied only once to avoid double costing the same appointment. For outpatient appointments, if a patient reported seeing a ‘hospital doctor,’ this was assumed to be at a consultant-led outpatient clinic, and so the consultant-led outpatient clinic unit cost was applied. If a patient reported attending a hospital clinic without describing the health-care professional seen, then the generic cost for an outpatient appointment was applied. In total, 105 patients reported an outpatient visit, with 93 (89%) reporting seeing a ‘hospital doctor’ and 12 (11%) not specifying who they saw. Of the participants who reported attending a health-care consultation, participants receiving usual care reported more hospital outpatient consultations than those receiving the intervention, but this difference was not statistically significant. There were similar numbers of primary care contacts across the different categories, with no statistically significant differences.

The costs of primary care and outpatient consultations are reported in Table 44. The number of outpatient visits in which the clinic lead was unspecified was higher in the usual care group, which may have led us to underestimate usual care costs in this area. However, even so, costs related to outpatient consultations were higher in the usual care group than in the intervention group. Overall, mean cost per participant was slightly higher in the usual care group than in the intervention group (£1182.00 vs. £1072.00), but the difference was not a statistically significant; the mean cost difference was £110.26 (95% CI –£101.10 to £321.63; p = 0.306).

Hospital admissions and associated costs

Following our costing methodology (see Appendix 4), the total number of hospital admissions costed was 43 (17%) in the usual care group, compared with 56 (22%) in the HEELs group. An overall summary of the number and cost of hospital admissions is reported in Table 45. Reasons for admissions were amputation (minor/major), revascularisation, falls and other less defined reasons reported as specifically related to the target heel ulcer and were, in the majority of cases, accompanied by systemic infection. Many patients reporting hospital admissions spent more than one spell in hospital during the trial.

Costs of medications (antibiotics)

Of the 509 patients, 209 patients (107 in the usual care group and 102 in the intervention grou) reported receiving prescribed medication for an infection related to the target heel ulcer. For those prescribed medication, the mean cost of antibiotics was £71.09 in the usual care group and £42.56 in the intervention group, but the difference was not statistically significant (£28.53, 95% CI –£14.63 to £71.68; p = 0.194. Table A9.10 below provides a descriptive analysis of the medications prescribed for infections related to the target foot ulcer (Table 46).