Treatment of severe ankle sprain: a pragmatic randomised controlled trial comparing the clinical effectiveness and cost-effectiveness of three types of mechanical ankle support with tubular bandage. The CAST trial

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 01/14/10 from route sheet. The contractual start date was in November 2002. The draft report began editorial review in July 2006 and was accepted for publication in July 2007. 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

© 2009 Queen’s Printer and Controller of HMSO. This monograph may be freely reproduced for the purposes of private research and study and 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: NCCHTA, Alpha House, Enterprise Road, Southampton Science Park, Chilworth, Southampton SO16 7NS, UK.

2009 Queen’s Printer and Controller of HMSO

Background

Acute ankle sprain is one of the most common conditions seen in UK emergency departments, accounting for between 3% and 5% of all UK emergency department attendances1 and approximately 5600 injuries each day. 2 The majority of ankle sprains involve the lateral ligament complex3 (Figure 1), accounting for one-quarter of all sports injuries. 4 The injury is painful and incapacitating and, unless the injury is minor, weight bearing is difficult to tolerate. Lateral ligament sprains are widely viewed as being uncomplicated and self-limiting. However, several studies have shown that, although the acute symptoms resolve, residual symptoms can persist for months or even years after the initial injury. 5 One long-term follow-up study6 showed that, 7 years post injury, 32% of subjects experienced chronic complaints of pain, swelling or recurrent sprains. Early effective treatment is not only crucial to promote a speedy resolution of acute symptoms but also an important feature in limiting the chronicity of the injury. 5

FIGURE 1.

Lateral ligaments of the ankle.

Classification of ankle sprains

Sprains of the lateral ligament complex of the ankle vary in severity. The anterior talofibular ligament (ATFL) is always first to be injured, and in severe sprains the calcaneofibular ligament (CFL) is also disrupted (see Figure 1). A classification of the severity of ankle sprains has been developed by Crichton et al. 7 (Table 1).

Diagnosis of the grade of sprain in the emergency department is usually guided by the presenting clinical features. Previous trials have described diagnosing the grade of injury using stress radiography,8–10 arthrography11,12 or both. 13,14 Such diagnostic methods are rarely used clinically, making their results difficult to generalise into the clinical setting. Other studies have used talar tilt and anterior drawer tests to indicate the severity of the injury. 15–17 These tests are more easily incorporated into an elective orthopaedic examination, but are difficult to perform accurately in the first few hours following a sprain, when there is gross swelling and the tests themselves induce pain. Evidence shows that it is difficult to gauge the severity of sprain accurately in the acute situation,18 and so the degree of severity is usually assessed based on the ability to weight bear and the extent of the pain, swelling and bruising.

Treatment of grade I and minor grade II injuries is generally conservative. Traditional teaching advocates ice, compression and elevation of the affected limb together with anti-inflammatory medication in the early phase,19 although there is less consensus on the value of initial rest or early mobilisation. Optimal treatment for severe sprains (grade III), however, remains unclear. Treatments described include no intervention, physiotherapy, different types of brace and supports, immobilisation and surgical repair of the ligaments. Recent systematic reviews highlight a lack of good-quality evidence to aid clinical decision-making in managing these injuries. 2,7

Review of treatment of ankle sprains

Context of the CAST trial

Three Cochrane reviews of different treatments for lateral ankle ligament injuries have been conducted,21,28,29 but none has been updated since 2002. To provide a context for the CAST trial, and to identify any new research since 2002, the search strategies of the Cochrane reviews of immobilisation versus functional treatment28 and of different functional treatments29 were rerun. Any new studies found were added to the results of the Cochrane review and the meta-analysis was repeated. The search strategy is detailed in Appendix 1.

Results of the literature review

Trial flow

A flow chart summarising the study selection process is shown in Figure 2.

FIGURE 2.

Study selection process. RCT, randomised controlled trial.

Study characteristics

The Cochrane review28 of immobilisation versus functional treatment identified 20 trials10,15,16,31,43–49,51–57 enrolling 2120 participants. Of these, seven trials compared immobilisation with the use of an ankle brace,15,16,47–51 five trials compared immobilisation with an elastic bandage,43–45,52,53 four trials compared immobilisation with strapping14,51,54,55 and five trials compared cast immobilisation with treatment using a softcast or wrap. 8,10,46,56,57 The rerun of the search strategy identified only one new trial fulfilling the inclusion criteria,31 comparing immobilisation with strapping. The trial enrolled 121 participants and used outcomes of pain, subjective instability and recurrent injury.

In the trials reported, participants were likely to be young (< 50 years) and trials tended to include a higher percentage of males. Seven of the Cochrane trials were initially deemed to be of high quality (i.e. scoring higher than 50% on the quality assessment tool). 8,10,44,46,48,52,54 After additional information was retrieved from the authors, three further trials were categorised as high quality. 45,50,56 The new study31 was assessed for quality using the same tool by two independent reviewers (RN and SG) and was deemed to be of high quality.

The Cochrane review of different functional treatments29 identified nine trials33,47,51,58–63 enrolling 892 participants; however, none of the trials used the same intervention comparisons. Of these nine trials, four compared elastic bandage and strapping,59,60,62,63 four compared elastic bandage with a semi-rigid ankle support,33,47,61,63 one compared elastic bandage with strapping,63 two compared strapping with semi-rigid ankle support,51,63 two compared strapping and lace-up ankle support58,63 and one compared semi-rigid with lace-up ankle support. 63 The new study34 compared semi-rigid ankle support with elastic bandage, with a primary outcome of ankle joint function.

The mean validity score of the nine trials reported was 10.8 points (range 5–15 points, maximum attainable 22 points); following retrieval of additional information from authors the mean validity increased to 10.9 (SD 2.9). An additional study,34 which enrolled only 50 participants and reported results for only 35 patients, was deemed to be of insufficient quality for inclusion; it was also assessed for quality using the same tool by two independent reviewers (RJ and SG) (validity 10.5 points). The existing Cochrane review results, therefore, have not been amended.

Quantitative data synthesis

For the review of immobilisation versus functional treatment, data were extracted for the outcomes of return to pre-injury level of work and sports activities, pain, subjective instability and recurrent injury. The new study identified from the rerun of the Cochrane literature search was added to the studies in the Cochrane review28 and the meta-analysis was repeated.

Meta-analysis: pain

The new study31 was added to the nine studies of the Cochrane review that reported whether the patient was experiencing pain after treatment. This is the only parameter for which the results of the original Cochrane analysis were changed by the addition of the new study (Figure 3). Four studies reported short-term results (6 weeks),8,15,31,47 six reported intermediate results (6 weeks to 12 months)8,15,31,47,53,56 and six reported long-term results (> 12 months). 8,10,31,46,55 The original Cochrane analysis showed no significant differences at any follow-up time point; addition of the new study altered these results showing that significantly fewer patients reported pain in the functionally treated group at both the short time point (3 months) (RR 1.6, 95% CI 1.20–2.08) and intermediate time point (12 months) (RR 1.6, 95% CI 1.08–2.40) compared with the immobilisation group. It is however possible that such change in updated meta-analyses can have occurred by chance.

FIGURE 3.

Meta-analysis of effects of immobilisation vs functional treatment on pain, including the new study. 31

Rationale for the CAST trial

The updating of these two Cochrane reviews shows clearly that there is a lack of high-quality evidence to support clinical decisions about which type of treatment is best in the management of severe ankle sprains. Present evidence suggests that functional treatment is better than either immobilisation (for longer than 3 weeks) or no treatment. There remains a need for a well-conducted and adequately powered RCT of the clinical effectiveness and cost-effectiveness of different clinical approaches. Our survey of current practice20 was used to inform decisions about the design of this trial. We selected three treatments to compare with tubular bandage, which between them represented one method of functional treatment (Aircast brace), one immobilisation method (below knee cast) and one intermediate method [Bledsoe^® boot (Bledsoe Boot Systems, Grand Prairie, TX)]. The different methods also allowed varying degrees of ankle movement: full immobilisation (below knee cast), flexion/extension only (Aircast brace) and full range of movement (tubular bandage). In addition, three of the devices allowed the patient to exercise their ankle (tubular bandage, Bledsoe boot, Aircast brace) whereas one allowed no exercise regime (below knee cast).

The trial was designed to be pragmatic in nature and to reflect a model of practice used in the majority of UK hospital emergency departments. It included an analysis of the cost-effectiveness of the different therapies.

Trial design

The Collaborative Ankle Support Trial (CAST) was a pragmatic multicentred RCT with blinded assessment of outcome. It was designed to estimate the clinical effectiveness and cost-effectiveness of three different types of mechanical support in the treatment of severe ankle sprains compared with tubular bandage [Tubigrip^® (AliMed, Dedham, MA) was used in this study].

People attending selected emergency departments in England who had sustained a sprain of the ankle (ligamentous injury) and were unable to weight bear were identified at the time of presentation. The inability to weight bear was used as a proxy for a diagnosis of a grade II or III ankle sprain as classification is not possible in the acute stage. Potential participants were invited to join the trial and to attend a follow-up clinic 2–3 days following injury where they were randomised to one of four treatment arms: (1) tubular bandage, (2) Bledsoe boot, (3) Aircast ankle brace or (4) below knee cast.

Tubular bandage was chosen as the reference treatment following results of the national survey undertaken immediately before this study showing it to be the most common treatment. 20,64 The Bledsoe boot, costing £212.68, is many times more expensive than tubular bandage and its clinical effectiveness is yet to be proven. The cast was a conventional below knee walking cast and was applied for 10 days. The Aircast brace (£38.19) was chosen from the range of ankle supports available.

Research physiotherapists recruited and randomised participants at the follow-up clinic and undertook baseline assessments. Appropriately trained health professionals applied interventions to a defined standard. All other treatments were standardised and included ice, elevation, crutches and pain-relieving medications if needed. We felt that withdrawal of these treatments would be inappropriate as they constitute normal and accepted care. Outcomes were measured to reflect short- and longer-term recovery and possible complications. A health economics analysis was included.

Method of application

A double layer of tubular bandage was applied from the level of the tibial tuberosity to the base of the toes. Sizing of the tubular bandage was undertaken as per the manufacturer’s instructions. Patients were instructed to remove the tubular bandage at night.

The Bledsoe boots were sized and applied according to the manufacturer’s instructions (supplied with each device; available at www.bledsoebrace.com) with advice to remove at night.

The Aircast splint was sized and applied according to the manufacturer’s instructions (supplied with each device; available at www.aircast.com). The below knee cast was applied from the level of the tibial tuberosity to the base of the toes. A layer of Tubinette^® (Mölnlycke Health Care, Göteborg, Sweden) and a layer of padding were applied under a complete synthetic non-flexible cast.

Objectives

The trial had two objectives:

1. to estimate the clinical effectiveness of three different methods of ankle support (below knee cast, Aircast ankle brace and Bledsoe boot) compared with tubular double layer (tubular bandage) in terms of the recovery of function (primary outcome); the recovery of normal occupation, including return to normal work, study, caring or other activities (secondary outcome); and avoidance of residual symptoms including recurrent instability, lasting limitation of physical activity and need for further medical, rehabilitation or surgical treatment (secondary outcomes)

2. to measure the cost-effectiveness of each strategy, including treatment and subsequent health-care costs.

Participants

Participants were recruited from the emergency departments of eight collaborating hospitals: John Radcliffe Hospital, Oxford; Frenchay Hospital, Bristol; Coventry and Warwickshire Hospital, Coventry; Hospital of St Cross, Rugby; Heartlands Hospital, Birmingham; Alexandra Hospital, Redditch; Solihull Hospital, Solihull; and Warwick Hospital, Warwick.

Trial procedure

Recruitment

A standard approach was instituted across all participating hospitals. People who attended emergency departments with an ankle sprain were assessed for injury severity by emergency department medical staff who completed a standard proforma (Appendix 2), which recorded details such as the ability to weight bear, talar tilt and anterior draw test to assess stability. Radiography was used to exclude fracture when this was clinically indicated and as guided by the Ottawa guidelines. 65 People able to weight bear and those with a fracture were not eligible for the trial and were managed in accordance with normal practice. All those who could not weight bear (proxy for grade II and III injuries) were referred to a follow-up emergency department clinic, which reflected national normal practice reported in the survey undertaken20 (Figure 4). Emergency department staff were asked to provide participants with information, a letter inviting them to join the trial (Appendix 3) and a patient information sheet (Appendix 4). Between the time of presentation and the clinic all participants had identical self-treatment consisting of analgesia, ice and elevation combined with an exercise regime as tolerated. Patients were advised to mobilise as tolerated and were provided with crutches as appropriate. Written advice was given to all participants (Appendix 5).

FIGURE 4.

Flow chart of participant progress. ED, emergency department.

The follow-up clinic was 2–3 days after emergency department attendance. This delay was introduced for three reasons. First, a period of elevation and ice is widely advocated to reduce swelling and one of the treatments (the below knee cast) is contraindicated in the first few days because of the risk of developing a compartment syndrome if swelling increases. Second, it is difficult to accurately assess injury severity at initial presentation. The symptoms associated with grade I sprains resolve rapidly and the time delay allowed additional assessment of injury severity. Last, the delay allowed participants sufficient time to consider their involvement in the trial. If clinic attendance was not possible within 7 days of injury the patient was not eligible.

A record was made of all people who had an ankle sprain and were non-weight bearing at presentation in the emergency department to monitor referral and attendance rates at the follow-up clinic. At the follow-up clinic participants were given a full verbal explanation of the trial by the trial research physiotherapists. Informed consent was obtained from eligible and willing participants (Appendix 6). The reasons for declining to participate in the trial were recorded, along with age, sex, ethnicity and severity of injury of all people approached. The research physiotherapist performed a short baseline interview (Appendix 7) to ensure eligibility for randomisation and, after randomisation, the participant completed the baseline questionnaire (Appendix 8). This collected details that included date of birth, age, sex, body mass index (BMI), ethnicity and an assessment of pre-injury abilities including usual levels of mobility, engagement in sporting activity and usual occupation and employment (including hours worked and type of work undertaken), together with a brief examination of weight-bearing status (measured using weighing scales) and pain. Pre-injury quality of life was measured using the EuroQol 5 dimensions (EQ-5D). 66

Follow-up

Handling withdrawals

Participants were free to withdraw from the trial at any time. When possible, reasons for withdrawal were ascertained, including any potential dissatisfaction with the treatment proposed or provided.

Loss to follow-up

Loss to follow-up has been a problem in previous trials of ankle sprain management, particularly when these have involved participants attending follow-up research clinics at the hospital. 67 Postal questionnaires were used in an attempt to minimise loss to follow-up, and a system of reminder letters and telephone calls (Figure 5) was instituted to follow up those who did not return their questionnaires. Term-time and holiday addresses were obtained from those participants who were students. If, following completion of the measures shown in Figure 5, participants still did not return questionnaires they were telephoned by a researcher and asked to give answers to key questions over the phone. This enabled us to collect core outcome data for analysis.

FIGURE 5.

Flow chart of method used to maximise questionnaire return for CAST.

Questionnaires

The layout of questionnaires was designed to minimise the possibility of systematic missing responses, for example avoidance of single-sided photocopying and other recommendations for increasing the return rate. 68

Resource use

For data relating to resource use and complications we obtained consent at the beginning of the study to access participant records. This enabled additional information to be gained when participants were lost to follow-up and was used to assess whether data were missing at random and to model the effects of missing data.

Pilot study

Sample characteristics

The baseline characteristics of the sample are shown in Table 9. As anticipated from previous studies there was a greater proportion of men (58%) than women (42%). The mean age of participants was 30 years (SD 10.8, median 27, range 16–72). The majority of the sample was of Caucasian ethnic status (94%), with 68% having O levels/General Certificates of Secondary Education (GCSEs) or A levels as their highest educational qualification, 80% employed for over 25 hours per week and 82% spending greater than 4 hours per day on their feet.

The mean height of participants was 1.7 m (SD 0.1 m) and the mean weight was 78.6 kg (SD 15.4 kg), resulting in a mean BMI of 26.3 kg/m2 (SD 5.2 kg/m2).

Almost half of the participants (49%) had experienced an ankle injury previously and 9% of the sample reported recurrent sprain (defined as a previous sprain on three or more occasions, with the most recent incident being within the last year). Pre-existing injuries were generally only of mild to moderate severity and symptoms were intermittent.

Randomisation

Randomisation groups were generally well matched in terms of age, sex, educational level and baseline symptom profile and injury characteristics. There was a slightly larger number of males in the below knee cast group. All analyses have been adjusted for age, sex and baseline score. The median duration between time of injury and application of treatment was 3 days (interquartile range 2), with no differences between the groups.

Baseline injury characteristics

At baseline FAOS ranges indicated that nearly all participants were unable to participate in sports activities, experiencing significant loss of confidence in the ankle, had at least moderate or very frequent pain, and were experiencing at least moderately severe or frequent symptoms including swelling, bruising, instability and inability to weight bear (Table 10). Over 75% of participants had scores of less than 50 points for the pain, symptoms, sport and QoL subscales.

Nearly all participants reported at least some difficulty or frequent problems in a range of basic self-care activities of daily living. Mobility, as quantified by both the FLP and FAOS, was significantly impaired. In total, 75% of participants reported scores of 30 points or higher on the FLP ambulatory scale, and nearly all participants (97%) reported at least some problems in getting about.

In comparison with population norms, SF-12 physical performance scores indicated significant impairment. Mental health scores were consistent with age- and sex-adjusted population means. VAS pain scores in all four groups were very similar, with more pain present on weight bearing than at rest. Over 75% of participants had scores well below the population norm as defined by the EQ-5D measure of general health.

Data collection and completeness of follow-up

Adverse events and serious adverse events

There were seven adverse events in total, of which two were non-serious and five were serious. The two non-serious adverse events were both cases of cellulitis treated with antibiotics, one each in the Aircast brace and Bledsoe boot groups. The five serious adverse events were all either suspected/confirmed deep vein thrombosis (DVT) or pulmonary embolism. These events are not unexpected in this injury and following immobilisation of any type. There were two each in the tubular bandage and Aircast arms and one in the below knee cast arm. Of these, three were possibly related to the trial treatment, one was unrelated and one was of unknown relatedness (the participant had a previous history of recurrent DVT and was taking warfarin but was classified as eligible for the trial and randomised in error). Table 13 gives full details of the individual cases.

Recovery and estimates of treatment effectiveness

Figures 8 and 9 illustrate recovery (unadjusted medians) by randomisation group for the primary and secondary outcomes. Descriptive statistics are given in Appendix 13.

FIGURE 8.

Primary outcome recovery curves for the full 9-month follow-up period. Graphs (a)–(e) show the five FAOS subscales – pain, other symptoms, activities of daily living (ADL), sport and recreation and quality of life (QoL) respectively – and graph (f) shows the FLP ambulatory scale. Figures are unadjusted medians. The FAOS uses a scale of 0–100, in which 0 = extreme symptoms and 100 = no symptoms, whereas the FLP ambulatory subsection uses a score of 0–100, with a lower score being better. (Tables of means, medians and quartiles are given in Appendix 13.)

FIGURE 9.

Secondary outcome recovery curves for the full 9-month follow-up period. SF-12 physical (a) and mental (b) component scores (each scale 0–100 with population norm of 50). (c) The EQ-5D (scale –0.594 to 1.0, where a higher score is better). Visual analogue scale (VAS) scores (scale of 0–100 expressed in millimetres, where 0 = no pain and 100 = worst pain imaginable) for pain at rest (d) and pain when weight bearing (e). Figures are unadjusted medians. (Tables of means, medians and quartiles are given in Appendix 13.)

With the exception of mental health, recovery was incremental over time and most rapid in the first 4 weeks. However, at 9 months nearly 20% of individuals reported persisting difficulties getting around and 50% of participants scored less than 10–20% of the maximum score range on the FAOS subscales. Overall, mental health declined significantly at 4 weeks in comparison with baseline but had recovered by 12 weeks.

Age was a significant determinant of recovery, which was substantially slower as age increased and less complete by 9 months for all outcomes. Gender was also significant in recovery, particularly at the earlier time points, with men reporting better scores than women on average across all scores. Baseline score was positively related to recovery, with a better score at the outset being associated with better scores later on for all outcomes. No interactions between sex, age or randomisation group were observed.

Self-reported benefit

Participants were asked to rate the benefit that they had received from their treatment at 12 weeks and 9 months on a scale of 0–10. Those in the tubular bandage arm felt that they had received the least benefit of the four groups at both time points, with half of participants reporting a benefit of less than 5 out of 10; the benefit of the other three supports was rated about equally at between 7 and 8 out of 10 (Table 16, Figures 10 and 11).

FIGURE 10.

Self-reported benefit of treatment at 12 weeks. Benefit is measured on a scale of 0–10, where 0 = no benefit and 10 = maximum benefit (density = proportion).

FIGURE 11.

Self-reported benefit of treatment at 9 months. Benefit is measured on a scale of 0–10, where 0 = no benefit and 10 = maximum benefit (density = proportion).

Analysis of treatment uptake

Uptake was measured in terms of the numbers of participants who received the application of the treatment to which they were randomised. The analysis was repeated for groups defined by their uptake: that is those who were randomised to a treatment and received it and those who rejected randomisation group treatment. There were no differences between the two groups with respect to sex, age or baseline scores for primary or secondary outcomes (Table 17). Overall, the results were consistent with the results of the intention to treat analysis.

Sensitivity analysis

Multiple imputation was used to assess the impact of (complete-case) missing data on the analyses. Conclusions from this were not substantially different from those based on the complete-case analysis presented here, and the findings were, as far as can be ascertained, relatively insensitive to missing data.

Introduction

To inform decision-making on the optimum intervention for severe ankle sprain it is necessary to consider not only the measurable benefits or clinical effectiveness of different treatments but also their costs.

Severe ankle sprains can have a range of direct and indirect health-care cost consequences. These include the cost of initial treatment and any subsequent costs incurred in primary and secondary care during the recovery period. Different treatment arms may result in different direct health-care costs and differences in the date at which people return to work, days off work or ‘sick leave’.

The most appropriate economic evaluation technique for a clinical trial will depend on the results, and it is recognised that a well-designed economic study should allow for different eventualities. 80 The simplest eventuality would be when the cheaper intervention is found to be better on at least one outcome measure and no worse on any other, in which case it is preferred. Another eventuality is when two interventions have the same outcomes, in which case the economic evaluation required is a cost-minimisation analysis focusing only on costs. When there is the possibility that better outcomes are achieved at a higher cost, or vice versa, or when more than two interventions are being considered, a full economic evaluation is required.

This chapter presents an economic evaluation of the three interventions (below knee cast, Aircast brace and Bledsoe boot) versus tubular bandage as the comparator group. The evaluation explores costs and cost-effectiveness of these three approaches, and the factors influencing this, and is presented in several sections: (1) analysis of health-care costs alone; (2) impact of including indirect costs (sick leave); (3) cost-effectiveness analysis; (4) cost–utility analyses with and without sick leave costs; and (5) sensitivity analyses.

Methods

The primary analysis adopted an NHS perspective as that is of most interest to NHS decision-makers. A societal perspective, including the impact on productivity costs (i.e. time off work), was included in a secondary analysis. Sensitivity analysis was used to examine the impact of uncertainty on the expected cost-effectiveness of each intervention.

Results

Results given in Chapter 3 showed that at 4 weeks the below knee cast was the most effective treatment with respect to reduction of pain and QoL measures but that by 9 months there were no statistically significant differences observed between groups. These observed early differences appear to be clinically significant.

Cost-utility analyses

Table 23 presents the cost–utility analysis findings of the incremental cost per QALY gained for the three interventions relative to tubular bandage. To produce these ICERs we used bootstrapping techniques to generate 1000 replications of actual cost and effect data. Table 23 presents the point estimates for cost per QALY; more detailed outputs are reported in Appendix 15.

TABLE 23 - Cost–utility analysis: incremental cost-effectiveness ratios (ICERs), direct health-care costs only

Simulation output (1000 trials) ICERs.

	Tubular bandage	Below knee cast	Aircast	Bledsoe
Number of participants	81	78	73	86
ICER (direct health-care costs only)	–	£339a	£301	£2116

The Aircast brace and below knee cast had similar ICER values, whereas the Bledsoe boot had a much higher figure (over £2000 per QALY gained). Although there is debate about the exact amount society should be willing to pay for a QALY, the National Institute for Health and Clinical Excellence (NICE) employs a threshold of approximately £20,000–£30,000 per QALY when considering new technologies. 89 Table 23 shows that, considering direct health-care costs only, all three intervention groups were associated with a cost per QALY gained that is well within this range. The most cost-effective means of achieving a QALY was the Aircast brace, although this intervention was only marginally more cost-effective than the below knee cast.

There are some methodological problems associated with the application of ICERs,91 and summarising clinical trial results in terms of ICERs can be misleading when there are more than two comparators. A particular arm may have the most favourable cost per QALY gained but a lower level of effectiveness than other intervention(s). In these circumstances an alternative intervention that is more effective but also associated with a higher cost may actually represent a more favourable option if the effectiveness gain justifies the additional marginal cost.

In such a situation, and to summarise uncertainty around ICER estimates, the use of cost-effectiveness acceptability curves (CEACs) is becoming increasingly accepted within clinical trials to assess different options. 92 CEACs have the advantage that the curves convey information from which inferences can be made about the statistical significance or otherwise of cost-effectiveness results. 93 For multiple comparators one can attempt to rank the options by cost and then remove the options by simple and extended dominance. 94 However, to do this discernible differences must exist in effectiveness between comparators. In this instance, the differences in average effects between intervention groups (as measured using EQ-5D) are so small that they are not statistically significant. Therefore, although we could assume that the control group was not a cost-effective option (because of the relatively small incremental cost per QALY associated with all of the intervention groups relative to the control group) we lacked a basis to rank interventions using either simple or extended dominance. Thus, we considered the cost-effectiveness of all options relative to the control group.

A CEAC curve (Figure 12) indicates the probability (on the vertical axis) that an intervention is cost-effective relative to the comparator group for a range of possible societal valuations of a QALY (on the horizontal axis). If, for any given valuation of a QALY, the CEAC reaches or exceeds a 95% probability then it is possible to conclude (at the 5% significance level) that this intervention is cost-effective relative to the control group. Several interventions can be displayed on the same graph and thus CEACs can shed light upon cost-effectiveness in more complex decision-making contexts.

FIGURE 12.

Simulation output (1000 trials) cost-effectiveness acceptability curves (CEACs). Graph showing the probability of an intervention being cost-effective relative to tubular bandage at different levels of willingness to pay for an additional quality-adjusted life-year (QALY) (direct health-care costs only; excludes sick leave costs).

The CEAC curves in Figure 12 demonstrate that there is a high probability that all three interventions are more cost-effective than the control (tubular bandage) for most reasonable cost per QALY thresholds. Both the below knee cast and the Aircast brace were virtually the same, indicating that these two interventions had a comparable cost-effectiveness. The Bledsoe boot curve lay below these two, indicative of lower cost-effectiveness.

Appendix 16 provides scattergrams illustrating cost-effectiveness for the 1000 simulations under the different treatment regimes for direct care costs only (Figures 14–16). Similar benefits accrued on use of the Bledsoe boot as a treatment for greater health-care costs than for Aircast or below knee cast in almost all simulations.

Impact of indirect costs on ICERs

To take the costs associated with absence from work into account we could use either a friction cost or a human capital approach. Although the friction cost approach has its advocates,95,96 it lacks a foundation in economic theory. 97,98 Therefore, in the interests of simplicity we used a human capital approach.

When the costs of days off work because of illness were included, a different picture emerged (Table 24). The Aircast brace was now the least cost-effective intervention and the below knee cast remained the best in terms of value for money.

TABLE 24 - Cost–utility analysis: incremental cost-effectiveness ratios (ICERs), direct health-care and indirect sick leave costs

Simulation output (1000 trials) ICERs.

	Tubular bandage	Below knee cast	Aircast	Bledsoe
Number of participants	81	78	73	86
ICER (direct health-care and indirect sick leave costs)	–	£1393a	£3585	£2275

The CEACs shown in Figure 13 suggest that, once costs arising through sick leave were included, the Aircast brace was the least cost-effective option and that, once society values a QALY at around £7500, both the Bledsoe boot and below knee cast appeared to be equally cost-effective. All three interventions were cost-effective compared with the control group (tubular bandage) assuming a cost-effectiveness threshold of £20,000 per QALY.

FIGURE 13.

Simulation output (1000 trials), cost-effectiveness acceptability curves (CEACs). Graph showing the probability of an intervention being cost-effective relative to tubular bandage at different levels of willingness to pay for an additional quality-adjusted life-year (QALY) (direct health-care and indirect sick leave costs).

Appendix 16 provides scattergrams illustrating cost-effectiveness for the different ankle support regimes including indirect costs (Figures 17–19), showing that, if sick leave costs were included, there was little to discriminate between the three interventions in terms of costs or benefits accrued.

Discussion

From a health-care cost perspective the Aircast brace and below knee cast represented the most cost-effective interventions, with the Bledsoe boot ranked third. The cost–utility analyses also demonstrated that the Bledsoe boot was least cost-effective but that all three interventions were highly cost-effective compared with treatments for other types of condition. The cost per QALY figures extended up to a maximum of just above £2000 per QALY for the Bledsoe boot, which was well below the threshold set by bodies such as NICE of £20,000–£30,000.

To capitalise on any reduced prices associated with bulk buy arrangements some centralised purchasing arrangement would need to be established. For Aircast the suppliers offered a bulk buy price of £22.56 (list price £38.19 including VAT), but only for bulk orders of 500 braces on a 3-year contract. The eight trial clinics saw 1522 potentially eligible participants over a period of 27 months, equating to approximately 85 patients annually per centre; therefore, a throughput of 500 braces per annum would only be feasible at a five-centre level. The transaction, storage and transport costs involved in a centralised supply chain would then need to be balanced against any cost savings. The feasibility of such an arrangement is doubtful.

Conclusion

When considered from a health-care perspective alone the below knee cast and Aircast brace are the most cost-effective options for the management of severe ankle sprains. If purchased under bulk buy conditions the Aircast brace is more cost-effective than the below knee cast; however, the feasibility of a bulk buy system is doubtful. If the decision-maker’s main concern is to maximise effectiveness relative to total health-care costs then the Aircast brace may have a marginal cost-effectiveness advantage over the below knee cast. The lack of a clear front-runner in terms of cost-effectiveness reinforces the case for giving patients an informed choice.

When considering societal costs as well, the below knee cast is the most cost-effective treatment for severe ankle sprains. This finding persists throughout analyses using a range of assumptions about sick leave.

The sensitivity of rank order to changes in the assumptions made about sick leave means that results from this wider cost perspective should be treated with some caution.

Introduction

A component of the original brief commissioned by the National Coordinating Centre for Health Technology Assessment (NCCHTA) was to explore the feasibility of a deposit system to promote return of the mechanical supports and reuse. The manufacturers of the Bledsoe boot and Aircast brace do not recommend reuse, although both allow washing of the device.

In the context of a clinical trial it was felt that, given the manufacturer’s recommendations, and the many barriers that already exist to recruitment into trials, implementing a deposit system was impractical. It was agreed with NCCHTA that a small series of interviews would be undertaken as a preliminary exploration of the topic. We undertook a small qualitative study to examine participants’ opinions of the viability of a refundable deposit system for expensive items such as the Bledsoe boot.

Method

A total of 19 CAST trial participants (10 males and 9 females) undertook semistructured interviews. They were conducted by a single researcher (RN). Participants were asked four questions:

1. How would you have felt if we had asked for a refundable deposit on the boot/brace to encourage you to return it?

2. How big a deposit would encourage you to return the boot/brace? £0–5, £5–10, £10–15, £15–20, more than £20?

3. Would you have been willing to pay this amount if asked?

4. If you had been asked to pay a deposit, do you think this would have affected your willingness to take part in the trial? Please explain.

All interviews were audio tape recorded with the participant’s consent and later transcribed verbatim. The key questions were designed to give yes/no responses but a subsequent explanation of the answer was open. Thematic analysis of these responses was undertaken.

Participants

The mean age of participants was 34 (range 16–62). Eight participants were non-drivers. The level of sporting activity of participants ranged from no participation in sport to regular participation in three or more sporting activities. The majority of participants (12, 63%) had injured their ankle between 3 and 6 months before the qualitative study. Four participants were randomised to tubular bandage, four received a below knee cast, six received a Bledsoe boot and five received an Aircast brace. Analysis was performed using the approach of framework analysis. 100

Results

The idea of a refundable deposit to encourage return of the boots was explained to the participants and they were then asked whether they would have been willing to pay (Table 27).

The majority of participants had no problem with the idea of paying a deposit. Some participants had previous experience of shortages of hospital equipment and were willing to participate in a scheme to help prevent this. Two participants were unsure and expressed expectations of ‘free’ health care. Having to pay a deposit (even if refundable) went against these expectations. One participant was flatly against the deposit system, stating that paying a deposit ‘goes against the grain’. Two of the participants suggested that they felt a social obligation to return hospital equipment and that they felt a deposit would not make any difference to them. Most participants (n = 14, 64%) felt that a deposit of between £15 and £20 would be fair considering the cost of the Bledsoe boot. Although most participants would be willing to pay they were then asked whether they would have had sufficient funds on them at the time of their emergency department attendance. About half of the participants felt that they would not have had enough cash to pay such a deposit, although many would have had either a cheque book or credit/debit cards.

Conclusion

This is a small study that aims only to give an initial exploration of the acceptability of a deposit system. Participants were generally happy with the concept of paying a deposit for a device like the Bledsoe boot. However, the mechanical integrity of the devices over prolonged periods of use and different cleaning regimes is unknown and would have to be established before reuse could be recommended. If the device had proven a more clinically effective option, a deposit and reuse system may have reduced the overall cost. No further modelling was undertaken given the results of the trial. The costs of implementing a deposit system for the Aircast brace are likely to outweigh any benefits if the device was reusable. Neither the tubular bandage nor below knee casts are fit for reuse.

Further research is required to determine the effects of deposits in clinical trials.

Main findings of the trial

This is the first large RCT of three types of mechanical support for ankle sprains of sufficient severity to prevent weight bearing. Results showed a clinically and statistically significant advantage for the use of a below knee cast compared with tubular bandage in relation to symptom resolution and return of normal activity and function in the first 3 months. Benefits were maximal in the first 3 months of recovery. The differences between tubular bandage and the Aircast ankle brace and the Bledsoe boot had less clear clinical relevance, although the Aircast brace gave significantly better results for mental well-being in the early stages. The economic evaluation indicates that from a health-care perspective the Aircast brace and below knee cast perform similarly in terms of cost-effectiveness. The Bledsoe boot is the least cost-effective (relative to tubular bandage).

External validity and generalisability of the findings

This multicentred trial recruited from eight centres ranging from teaching hospitals to district hospitals, with a range of sizes of emergency departments, in a variety of metropolitan, urban and semirural environments. The services available in these hospitals reflected the normal services in UK hospitals, with none having specific specialist services applicable to this type of injury. The level of training in application of braces was consistent with what would be expected in routine clinical practice, with some being applied by plaster technicians but most being applied by other clinical staff, as is normal practice for these units. No special instructions regarding the type of below knee casts were given, but all sites used synthetic casts.

The interventions before application of the tubular bandage or mechanical support were standardised to ensure equality between treatment arms. This minimised risk, including compartment syndrome associated with excessive swelling, and ensured that injuries were of sufficient severity to meet the case definition. Injury severity is particularly difficult to ascertain until time has passed and swelling has had a chance to resolve. Audit has demonstrated that there was variable referral into the trial by clinicians, with between 15% and 100% of potential participants being referred to the clinic for consideration of recruitment. It was not possible to collect detailed information about these potential participants. We have no reason to believe that there was a systematic bias in those approached. Of those unable to weight bear at the time of presentation who were given a clinic appointment, 512 (43%) had symptom resolution by the time of clinic attendance and were therefore not eligible for the trial. In practice, it is possible that the tubular bandage, Bledsoe boot or Aircast brace could be applied at the time of emergency department presentation. However, this is highly unlikely to affect the results that we obtained. Larger numbers of less severe self-limiting injuries would have been included. A study of current practice20 showed low usage of mechanical supports at initial presentation, and it is believed that current practice in emergency medicine is for delayed application.

Loss to follow-up was minimal for the first 3 months of the trial and the internal validity of the study was not compromised. Slightly higher losses occurred by the 9-month follow-up; however, the main purpose for continuing follow-up until 9 months was not to detect differences between treatment but to ensure that, overall, no one treatment was associated with a consistently poor profile. Otherwise, internal validity of the study was good (see comments in the section on limitations of the study).

Overall, we believe that the generalisability of this trial is good, with valid representation of severe injury (represented by the inability to weight bear as a proxy for grade II and III injuries), a wide range of hospitals and substantial numbers of participants. It is not known whether the spectrum of patients presenting to emergency departments is similar to those presenting to minor injury units, primary care or sports injury facilities. However, the number of patients with the severity of injury studied here who are treated at these facilities is likely to be small, as all of these patients require radiography. 65 It is believed that the results of this trial should be applicable to all patients conforming to the inclusion/exclusion criteria of the trial.

It is important to note that this trial is related to the subgroup of patients who presented to emergency departments with an ankle sprain of such severity that they were unable to weight bear on that leg at the time of presentation and were still unable to weight bear at a review clinic a few days later. The population in this study had a slightly higher proportion of males than females, similar to previous studies, but average age was slightly higher than in previous studies because of the inclusion criteria (no upper age limit and all types of injuries). All participants received advice to elevate the limb, use ice, rest and undertake gentle non-weight-bearing exercises in the days between emergency department attendance and the clinic appointment. This initial period of 2–3 days may be important for the subsequent outcome, as it results in reduced swelling and also promotes early movement. The reduction in swelling is important both to speed up the healing process and to reduce pain, and also to allow proper fitting of any external support devices. The Aircast brace can be applied in the presence of oedema and gradually adjusted as swelling reduces, but the Bledsoe boot and tubular bandage may be oversized if applied early in the presence of swelling. A cast that can be weight bearing is contraindicated in the first few days because of the risk of compartment syndrome.

Internal validity and limitations of the trial

The trial groups were comparable for age, sex, educational status, baseline symptoms and injury characteristics and we do not believe that there has been any bias in allocation to the four treatment groups. Overall, the trial group was comparable to the English population for sex, height, BMI, employment, educational status and ethnicity, based on 2001 census data and data from the 2004 Health Survey for England. Comparative data for previous ankle injuries and symptoms are not available.

Uptake of treatment varied between groups. The below knee cast is the only treatment that participants cannot remove themselves. The majority of those who declined the trial because they did not want a particular treatment cited plaster as their reason for declining (46 out of 57 participants, 81%) (see Table 4). Insignificant numbers did not receive the allocated intervention for other reasons, were lost to follow-up or discontinued their treatment. The compliance rate over the following 2 weeks is not known; participants in the non-cast group could remove their splints either intermittently or permanently after leaving the clinic.

The follow-up rates in the study were high compared with similar clinical trials, with follow-up rates of 83%, 82% and 76% for the 4-week, 12-week and 9-month questionnaires respectively.

At the time of randomisation (a few days after injury), most participants were abstaining from sport, were not confident in walking, had moderate or frequent pain and had moderately severe symptoms. Nearly all participants reported some difficulty in basic self-care and mobility. In comparison with age- and sex-adjusted population norms, the participants had significantly impaired mobility but similar mental health scores.

We confirm the findings of other investigators in smaller studies that ankle sprains presenting with the inability to weight bear do not recover quickly; many participants still had limited mobility and function resulting from their injuries at 9 months. Patients with ankle sprains who initially present with an inability to weight bear should be given a cautious prognosis and warned of potential long-term effects on their mobility and activities. Recovery is slower and less complete with increasing age. At all follow-up points and across all randomisation groups the scores of men were better than those of women on all primary outcome measures.

Three participants in the study developed DVT and two suffered pulmonary emboli (although one was probably related to pregnancy rather than the injury and the other had a past history of thromboembolic disease and was taking warfarin and was randomised in error). These adverse events were not associated with any particular treatment arm. Thromboembolic events are a recognised but unquantified event after ankle injury. Our study suggested a rate of 1% (4/584) of ankle injuries having a clinically apparent thromboembolic event within 9 months of injury. Numbers were too small to give differential rates by treatment. The risks/benefits of the use of prophylactic anticoagulation in lower limb injury are uncertain101 and could usefully be investigated in future studies.

Clinical findings

The differences observed between treatments were broadly consistent across all physical measures and their subscales.

Mental health status at 4 weeks fell to below the norm for all treatment groups, suggesting that this injury has an adverse effect on mental well-being regardless of the support method used. For mental health the results suggest that the removable supports may give better results than either tubular bandage or the fixed below knee cast. By 9 months only the tubular bandage group had mental health scores below the population norm. Psychological issues will result from a combination of the effects of the injury, the recovery and the treatment. Although the below knee cast resulted in quicker recovery it did not give the best mental health scores early on; however, by 9 months the scores were comparable for all groups except for tubular bandage. It is possible that the inability to remove the boot, for either comfort or hygiene purposes, may be responsible for this effect.

Trials that use disability and HRQoL outcomes face significant challenges in interpreting the clinical meaningfulness of the observations made. There are some benchmarks against which we can assess the clinical importance of the differences between treatments. Changes in scores of between 2.5 and 4.0 points on the SF-12 are generally considered clinically meaningful in terms of the individual. 102 Studies of the use of visual analogue scales for pain severity in the emergency care setting indicate that a change of 13 mm is the accepted level for indicating a clinically significant change. 103,104 There is little evidence to determine clinically significant changes in the FAOS, although the authors have suggested that an 8- to10-point difference, as used in the Knee injury and Osteoarthritis Outcome Score, is clinically significant. 105 These guidelines suggest that differences between the treatments observed on the FAOS scores were small at baseline and of moderate size at 12 weeks. Differences between tubular bandage and the below knee cast were commensurate with a clinically significant change.

An alternative method is to express the differences between groups in relation to the standard deviation. 106 Effect sizes of 0.25 are considered small but may be of clinical relevance, and those of 0.25–0.5 are considered moderate but are likely to be of clinical relevance. Overall, our interpretation of findings of the CAST study are that the benefits afforded by below knee casts in relation to tubular bandage are clinically small at 4 weeks, but clinically more substantial by 12 weeks, affording greater recovery of mobility, comfort and confidence in the ankle.

Present medical opinion has suggested the importance of early mobilisation and the role of early return of proprioceptive stimulation as key determinants of recovery in ankle sprains. Results from our study contradict this theory by demonstrating that the treatments producing most immobilisation resulted in quicker recovery, without any longer-term disadvantage. However, although we used a below knee cast, it was for a relatively short period compared with that in many studies. There are various pathophysiological mechanisms that could explain our findings that a period of immobilisation may be beneficial, including a decreased risk of further reinjury and rebleeding even at the microscopic level, an early decrease in pain encouraging better mobilisation once the cast is removed, short-term abolition of pain reducing plastic changes in the central nervous system that can lead to increased vigilance of the ankle and the development of centrally sensitised pain, decreased swelling because of increased elevation of the cast compared with other patients, a better proprioceptive response when the cast is removed because of decreased swelling and loss of abnormal stimulus that may occur in the early recovery phase, and a decreased inflammatory response with decreased abnormal fibrosis and a recovery period allowing return of tensile strength of ligaments before mobilisation. The initial 2-day period may be important in dispersing initial haematomata that could lead to a fibrotic response.

It is possible that long-term instability (becoming apparent after 9 months) occurs, but this would not be detected in this study. It is well recognised that a period of initial movement can prevent stiffness in joints that are subsequently immobilised. We do not know how much exercise people undertook when in each form of splint or how much/when they removed the device. If a period of relative immobility is the reason for the improved outcome then it is possible that the other groups performed less well because they could remove the splint and therefore moved the ankle more. It would only be possible to determine this by trials of varying regimes of usage of the various splints, to inform any instructions on splint usage. The below knee cast completely immobilises, and therefore gives better analgesia during its use, but this benefit persists at 4 weeks (over 2 weeks after removal). Early diminution of pain may have important consequences for encouraging recovery and could potentially be duplicated by better analgesia. The below knee cast is the only device for which the period of support is guaranteed and which cannot be varied by the patient or by the clinician’s instructions. It may also have other reasons for improved outcome that could be independently manipulated, such as indirectly encouraging more elevation, less dependency and less use of the whole limb, being more likely to be elevated at night and providing better analgesia.

With no benefit in outcome and with worse mental health scores this study does not recommend the use of tubular bandage in severe ankle sprains. It is unlikely that severe injuries would be appropriately treated by the widely used RICE (rest, ice compression, elevation) regime.

Cost and economic analysis

Mean direct health-care costs per participant indicated that the Bledsoe boot was the most expensive form of ankle support, with tubular bandage the least expensive.

Cost–utility analysis, comparing incremental costs with differential impact on HRQoL over 9 months, demonstrated that from a health-care perspective the Aircast brace (£301 per QALY) and below knee cast (£339 per QALY) were more cost-effective than the Bledsoe boot (£2116 per QALY). Simulations generated CEACs, which were indistinguishable for the Aircast brace and below knee cast, indicating that these two interventions had comparable cost-effectiveness; the Bledsoe boot was least cost-effective.

Cost–utility analysis was necessarily based on the sample of patients for whom we had cumulative outcome data over the study period or for whom we had a sound basis for imputation for missing utility data. We used imputation techniques in a small number of cases in which we considered we could reliably make predictions about utility changes, despite some missing data. A conservative approach using average values was adopted as more sophisticated approaches are recognised to lead to varying predictions. 77,78 We assume that attrition of the sample is unlikely to have affected final estimates of relative cost–utility in any systematic manner as analysis of baseline characteristics indicated no significant differences. This was confirmed by analysis of sample characteristics such as baseline utility measures. Other uncertainty around ICER estimates was addressed by the use of CEACs and sensitivity analysis.

Although the Aircast brace and below knee cast could not be differentiated in terms of cost-effectiveness, they differed in terms of participant preference. Differential compliance rates suggested that the below knee cast was the least popular intervention. In a non-trial population, participants’ responses to the offer of this treatment could be even less favourable unless the benefits are explained carefully. Bearing in mind that both types of support were similarly cost-effective, decision-makers may favour the Aircast brace over the below knee cast for routine implementation. We did not formally assess preference or acceptability and so these suggestions should be interpreted with caution. The results of this trial may give participants a greater amount of information with which to make a choice. The cost of a fitted below knee cast was less than half of that for the Aircast brace; however, total health-care costs (including subsequent care costs) did not differ significantly between the two treatment groups and so the below knee cast lost its cost advantage.

If the Aircast brace was acquired at a discount price through bulk purchase it had half the ICER value of the below knee cast. Centralised purchasing arrangements would have to be established to enable this price advantage to be realised; storage and transport costs would need to be added to the intervention cost, although these are unlikely to eliminate purchase cost savings.

From a societal perspective the inclusion of indirect costs associated with sick leave indicated that there was little to discriminate between the three interventions in terms of costs or benefits accrued (CEACs). Because the valuation of indirect costs was contentious, sensitivity analyses were undertaken to include varying the assumed cost of a single day off work and capping reported sick leave attributable to ankle injury at 30 working days. The sensitivity of the economic evaluation conclusions to the assumptions made about sick leave means that the results from this wider cost perspective should be treated with some caution.

Overall, the economic evaluation results of this trial indicated that, if the decision-maker’s main concern is to maximise effectiveness relative to total health-care costs, the Aircast brace and below knee cast perform similarly. The Aircast brace may also be the optimal pragmatic option in terms of universal use unless patients can be offered the two treatment options.

Implications for health care

This study demonstrates that severe ankle sprains, as defined by an inability to weight bear 2–3 days following injury, have long-term effects that can be influenced by treatment. Current treatment could be improved by the use of mechanical supports.

This study recommends the use of either a short-term below knee cast or an Aircast brace, following an initial period of several days to allow swelling reduction and non-weight-bearing exercise, for the treatment of severe ankle sprains. The choice between the two depends on a balance of clinical effectiveness, patient acceptability and cost.

The CAST trial group

We are indebted to the consultants, nurses, physiotherapists and plaster technicians at the collaborating centres for their assistance in participant recruitment and data collection and for application of the ankle supports: Jason Kendall, Sue Kemspon and Caroline Pearce and team, Frenchay Hospital, Bristol; Magdy Sakr, Conor Kelly, Rachel Nakash, Esther Williamson, Mark Williams, Vivien Nichols, Cheryl Ritchie, Gill Doughty, Diane Mooney, Marie Stead, Chris Wootton, Sue Soden, Kim Wilding, Ann McLean and Rona Dixon, University Hospital Coventry and Warwickshire, Coventry; Aidan MacNamara, Raj Patel, Geraldine Saunders, Jo Walker, Saraj Mohammed and Kathy Stretton, Birmingham Heartlands Hospital, Birmingham; David Skinner, Sally Bateman, Jo Sale, Fiona Hooker, Kathryn Lewis, Claire Granville, Sue Norton, Lin Lyzba and Pam Osborne, John Radcliffe Hospital, Oxford; Matthew Cooke and John Hopson, Solihull Hospital, Solihull; Graham O’Byrne, Susan Slater and Tammy Mills-Chester, Alexandra Hospital, Redditch; Jo Williams, Hospital of St Cross, Rugby; Matthew Dunn, Nicky Parker, Esther Noakes, Nic Vaux, Rob Smith and Sheila Poulton, Warwick Hospital, Warwick.

Contribution of authors

Matthew Cooke, Professor of Emergency Medicine, participated in the design of the original experiment, had input into the original grant application, networking to establish clinical sites and supervision of clinical sites, and was the lead author for the final report.

Jennifer Marsh, Lecturer in Medical Statistics and Health Informatics, was involved in the design of the study, discussions on the pilot study, analysis of data from the pilot study, changes in the light of analysis of the pilot study data, discussions on data collection, data cleaning and management for the main study, implementation of data cleaning and management, analysis of data and discussions on analysis and interpretation, discussions on the statistics sections write-up, and undertaking the write-up. Also made a major contribution to the preparation of DMEC reports and attended TSC meetings, DMEC meetings and management and other meetings.

Michael Clark, Senior Research Fellow in Health Economics, undertook the economic components of the study, including having input into the design of data collection tools and the microcosting study, attended regular study meetings, was responsible for economic data analysis and interpretation, and undertook the final economic analysis and economic write-up.

Rachel Nakash, Research Fellow, was trial co-ordinator for the planning and set-up phase and to the end point of recruitment and led on the qualitative study and analysis.

Rose Jarvis was the clinical trials co-ordinator for the final stages of the study and was involved in data cleaning and management, analysis and co-ordination of report writing.

Jane Hutton, Professor of Medical Statistics, was involved in the design of the study, discussions on the pilot study, collection of data for the pilot study, changes in the light of analysis of pilot study data, discussions with statistician on data collection, cleaning and management, discussions with statistician and others on data analysis and interpretation, and writing of the results chapter. Also attended TSC meetings and some management and other meetings.

Ala Szczepura, Professor of Health Services Research, wrote the economics section of the original application, attended TSC and other management meetings and was responsible for the economic analysis and the writing of the economics elements in the report.

Sue Wilson, Professor of Clinical Epidemiology, had input into the original grant application and provided comments on the final draft of the report.

Sallie Lamb, Professor of Rehabilitation, led the design of the original experiment, had responsibility for writing the original grant application, gathered comments from other applicants and integrated them into the grant application, set up the trial, appointed and supervised the trial co-ordinator and administrator, supervised the trial on a day-to-day basis in close liaison with Matthew Cooke, participated in analysis and participated in the writing of reports.

Other roles fulfilled

Bill Gillespie, Chairman of TSC; Sue Wilson, member of TSC; Vicky Staples, member of TSC; Jeremy Dale, member of TSC; Janet Dunn, Chairman of DMEC; Damian Griffin, member of DMEC; Pat Overton-Brown, member of DMEC; Gulnaz Begum and team, randomisation; Emma Withers, trial administrator; Lisa Craven, follow-up; Chris McCabe, advice on economic evaluation; Simon Gates, reviewer for systematic review.

Disclaimers

The views expressed in this publication are those of the authors and not necessarily those of the HTA Programme or the Department of Health.