Article Text

Economic evaluations of diagnostic tests, treatment and prevention for lateral ankle sprains: a systematic review
1. Chung-Wei Christine Lin1,
2. Kimi Uegaki2,
3. Veerle M H Coupé3,
4. Gino M Kerkhoffs4,
5. Maurits W van Tulder2
1. 1Musculoskeletal Division, The George Institute for Global Health and Sydney Medical School, The University of Sydney, Sydney, Australia
2. 2Department of Health Sciences and EMGO Institute for Health and Care Research, VU University, Amsterdam, The Netherlands
3. 3Department of Epidemiology and Biostatistics, VU University Medical Center, Amsterdam, The Netherlands
4. 4Department of Orthopedic surgery, Orthopedic Research Center, Academic Medical Centre, Amsterdam, The Netherlands
1. Correspondence to Maurits W van Tulder, Department of Health Sciences and EMGO Institute for Health and Care Research, VU University, De Boelelaan 1085, U454, 1081 HV Amsterdam, The Netherlands; maurits.van.tulder{at}vu.nl

## Abstract

Objective To assess and summarise the economic evidence regarding diagnostic tests, treatment and prevention for lateral ankle sprains.

Methods Potential studies were identified from electronic databases and trial registries and by scanning reference lists. Risk of bias and methodological quality were evaluated. Two independent reviewers screened, assessed studies and extracted data. Data were synthesised descriptively due to study heterogeneity.

Results A total of 230 records were identified; 10 studies were included. Five studies conducted a full economic evaluation and five studies involved cost analyses. Lack of blinding was the main risk of bias. The methodological quality of the full economic evaluations was fairly good. Valuation of costs, measurement of outcomes and sensitivity analysis were points for improvement. Single studies showed that the Ottawa ankle rules (OAR) was cost effective for diagnosing lateral ankle sprains in the emergency setting compared with existing hospital protocols; acute treatment with anti-inflammatory medication and the plaster cast for severe sprains appeared cost effective; and neuromuscular training was cost effective in preventing ankle re-injury.

Conclusions Results of this current systematic review supplements the evidence provided by reviews of effectiveness. There is evidence to support the implementation of OAR in the emergency setting, the use of anti-inflammatory medication and the plaster cast in the acute phase, and the prescription of neuromuscular exercises to prevent re-injury. Although the evidence is limited due to the low number of studies, shortcomings in methodological quality and small sample sizes, the findings may be used to inform clinical practice and practice guidelines.

## Introduction

Limited healthcare resources for sports injuries mean that cost-effective methods for diagnosis, treatment and prevention are required to ensure a system's sustainability. This need for economic evidence will increase as participation in sport and physical activity is recognised and promoted as an essential component of a healthy lifestyle, whereby the number of injuries will rise. Ankle injuries involving the lateral ankle ligament complex are particularly common (600 to 700 per 100 000 person years)1 and account for 25% of all sports-related injuries.2

Healthcare consumption and socioeconomic costs associated with ankle sprains are considerable. For example, in the USA, an estimated 1.6 million physician office visits and greater than 8000 hospitalisations per year are due to ankle sprains.3 In The Netherlands, half of all ankle injuries require medical attention and the average length of sick leave is 2.5 weeks, with 90% of cases resuming work by 6 weeks.4 The annual costs due to sports-related ankle sprains is estimated at €187 200 000 in The Netherlands, with productivity loss costs due to paid and unpaid work accounting for 80% of the total.3 Furthermore, residual instability can lead to an increased risk of recurrence and interfere with resumption of daily activities including sport, recreation and work.5,,7 Concomitant ankle fractures may be a complication of lateral ankle sprains, occurring in up to 15% of all ankle sprains.8

The current best evidence for the diagnosis, treatment and prevention of acute lateral ankle sprain was evaluated in the recent Dutch clinical guideline.9 The key findings were as follows:

• Diagnosis: fracture is the main red flag. The Ottawa ankle rules (OAR) have been developed and seem accurate to rule out fractures and the need for radiography after acute ankle injuries.8 The gold standard for diagnosing lateral ankle sprain is delayed physical examination (4 to 5 days post trauma), which is more reliable than physical examination within 48 h after trauma.10 Stress radiography, ultrasonography and MRI have no role in the routine examination, but can be useful in diagnosing associated injury (bone, chondral or tendon) when there is a suspicion of such injury.

• Treatment: in the first phase, use ice and compression in combination with rest and elevation. Non-steroidal anti-inflammatory drugs (NSAIDs) have beneficial short-term outcomes for acute ankle sprains. A short period of plaster immobilisation or similar rigid support facilitating a rapid decrease of pain and swelling can also be helpful in the acute phase.9 ,11 Thereafter, functional treatment for 4 to 6 weeks is preferable to immobilisation.

• Prevention: balance or neuromuscular training prevents recurrence of ankle injuries in athletes up to 12 months postinjury,12,,14 is recommended to commence within 12 months after injury and should be included into regular training activities and/or at home. The use of a brace and tape reduces the risk of re-injury in those who are active in sports.

To help make steps toward sustainability, current best practice for acute ankle sprain should be evaluated for its cost effectiveness to supplement clinical evidence and inform clinical guidelines.15 To our knowledge, however, the available economic evidence regarding diagnostic, treatment and prevention has not yet been systematically reviewed. Therefore, the objectives of the present systematic review were to assess and to summarise the evidence from full economic evaluations and cost analyses regarding diagnostic tests, treatment and prevention for lateral ankle sprains.

## Methods

### Search strategy

The primary literature search was conducted in EMBASE.com, a database that combines citations of Medline from 1966 to present and Embase from 1974 to present. The search strategy is presented in table 1. We conducted a supplementary search in the Cochrane National Health Service Economic Evaluation Database, trial registries (ie, Dutch Trial Register, Australian New Zealand Clinical Trials Registry and ClinicalTrials.gov) and scanned the reference lists of included studies to identify additional studies. The literature search was conducted in April 2010 and updated in May 2011.

Table 1

Search strategy in EMBASE.com

The inclusion criteria were as follows: (1) studies which recruited persons 12 years and older, (2) any diagnostic test, prevention and treatment for lateral ankle sprains, (3) any control intervention, (4) a full economic evaluation (ie, cost effectiveness, cost utility or cost-benefit analysis) or a partial economic evaluation (ie, cost analysis). The evaluations could either be conducted alongside randomised controlled trials or economic modelling studies. Exclusion criteria were studies involving children (<12 years), exclusively on fractures or joint fusions.

### Selection phase

To identify relevant studies, we followed a two-step procedure. First, potentially eligible studies were identified based on title and abstract. This screening was conducted by two reviewers independent from each other. The results of the two reviewers were compared and any disagreements were discussed openly. If disagreement remained, then a third reviewer was consulted in order to reach consensus. Second, eligibility was confirmed based on a review of the full text. The full-text screening was also conducted by two reviewers independent of each other and any disagreements were resolved following the same consensus procedure.

### Data extraction and assessment

Study characteristics and key findings of all included studies were extracted by one reviewer and checked for accuracy by a second reviewer. Any discrepancies were discussed openly and corrected as required.

We assessed the risk of bias of included economic evaluations that were based on randomised controlled trials using the Cochrane risk of bias-assessment tool.16 This assessment tool covers six domains (ie, sequence generation, allocation concealment, blinding of participants, personnel and outcome assessors, incomplete outcome data, selective outcome reporting and other sources of bias). Because the blinding and incomplete outcome data domains were broadly defined, we made the following modifications. First, we evaluated the adequacy of blinding separately for the participants, administrators of the intervention or outcome assessors. Second, we assessed the completeness of outcome data in terms of how drop outs were handled and if data were analysed according to the intention-to-treat principle. Studies with positive ratings on more than half the items (ie, five or more) were rated as having a low risk of bias.17 In cases where the nature of the intervention would make blinding of patients or administrators of the intervention impossible, we did not put any weight on these items and excluded them from the denominator when determining overall risk of bias.

We evaluated the risk of bias of the full economic evaluations using the Consensus Health Economic Criteria (CHEC) list. The CHEC-list is a 19-item assessment tool developed through a Delphi procedure involving 23 international experts of health economics. It represents a minimum set of internal and external validity criteria for individual economic evaluation studies.18 For modelling studies, we supplemented the evaluation of methodological quality with a framework proposed by Philips et al.19 This framework is divided into three categories of quality: structure, data and consistency. The risk of bias of the cost analysis studies was not assessed, because there is no tool available.

The risk of bias assessments were conducted by two reviewers independently of each other and any disagreements were again resolved through a consensus procedure. Before the included studies were evaluated by the reviewers, a pilot study with unrelated articles was conducted to test the use of each assessment tool, to check uniform interpretation of the items and to test the consensus procedure.

### Data analysis

A meta-analysis of economic evaluations is not opportune, because of the heterogeneity of study populations (countries and settings), interventions, costs (the way they were measured and valued) and effects. Therefore, we conducted a descriptive synthesis of the currently available economic evidence. The main outcome of interest was a cost-effectiveness ratio for full economic evaluations, or cost data for partial economic evaluations (cost analysis).

## Results

### Search and selection

The search and selection process is summarised in figure 1. The complete search strategy resulted in 230 unique records. Of these, 10 studies met the inclusion criteria. Four of the 10 studies focused on the diagnosis, two on the prevention and four on the treatment of lateral ankle sprains.

Figure 1

Flow diagram of the search and selection process

One ongoing randomised controlled trial20 with a planned economic evaluation was identified via the Dutch Trial Registry (NTR2157), and another21 was identified from ClinicalTrials.gov (NCT01126242). Inclusion of these studies was not possible as data were not yet available.

### Description of the studies

The characteristics of the four diagnostic studies are presented in table S1 (uploaded separately as a supplementary file). Two studies conducted a full economic evaluation involving the OAR for ruling out fractures in the emergency setting.22 ,23 The remaining two studies performed a partial economic evaluation: one compared the costs of adding MRI to radiography24 and the other compared the costs of different strategies for diagnosing ligament rupture.10 Three of the diagnostic studies adopted a societal perspective.22,,24 One study did not specify the economic perspective,10 but collected study-related costs only. Follow-up periods ranged from 1 week23 to 1 year.22

Of the four treatment studies (table S2, uploaded separately as a supplementary file), two conducted a full economic evaluation25 ,26 and two conducted a cost analysis.27 ,28 One study evaluated the use of piroxicam compared to placebo26 and three studies investigated different (im)mobilisation methods25 ,27 ,28 following acute ankle sprains. The follow-up periods ranged from 10 weeks27 to 9 months.25

Of the two prevention studies (table 2), one conducted a full economic evaluation of proprioceptive training for preventing re-injury with a follow-up of 1 year.3 Although the authors used the term ‘proprioceptive training’, the training programme included postural stability, strengthening and functional exercises, so we have replaced it with the term ‘neuromuscular training’ in our review. The other study29 conducted a cost analysis in which a semirigid brace was compared with taping in preventing ankle sprains from the perspective of a high school athletic programme during a 10-week season.

Table 2

Description of prevention studies (n=2)

### Risk of bias of underlying randomised control trials (appendix table 1, web-only file)

There were eight studies based on a randomised controlled trial.3 ,23,,29 The main potential source of bias pertained to the lack of blinding of either the participants, administrators of the intervention or the outcome assessors. Blinding of the participants and administrators of the intervention would be impossible in seven studies due to the nature of the intervention itself.3 ,23,,25 ,27,,29 Overall, five of the eight studies3 ,23,,26 had a low risk of bias. Three studies had a high risk of bias.27,,29

### Risk of bias of full economic evaluations

The risk of bias of the five full economic evaluations assessed by the CHEC list is presented in appendix table 2 (web-only file).3 ,22 ,23 ,25 ,26 The number of criteria fulfilled by the studies ranged from 12 to 18 from a total of 19. The criteria that were least met pertained to cost valuation, not using a validated outcome measure and the lack of a sensitivity analysis. With regard to valuation, shortcomings were that either the source of price weights or index year for the costs was not reported, or tariffs (which contain an element of negotiation and do not correspond with the actual cost for providing a service)30 were used instead of a price weight based on actual resources consumed.

The risk of bias of the decision analysis model by Anis et al22 is presented in appendix table 3 (web-only file). Items pertaining to assess the uncertainty of the model's data via sensitivity analysis were least met. Consistency-related items regarding testing the model's mathematical logic and comparing its results to previous models were also not met.

### Findings of the diagnostic studies (online supplementary table S1)

Two full economic evaluations found that the OAR were cost effective from a societal perspective.22 ,23 First, the base case analysis of the modelling study by Anis et al22 demonstrated that when emergency department physicians implemented the OAR as opposed to existing hospital protocols, there were no missed fractures and costs were saved due to a decrease in radiograph usage (table S1, uploaded separately as a supplementary file). The cost saving per patient by implementing the OAR was CAN$8.89 in Canada, and ranged from US$7.45 to US$32.92 depending on the cost schedule used (hospital charge, Medicare or Medicaid). The sensitivity analysis showed that the cost savings per patient were most influenced by the cost of radiography. Furthermore, the OAR were still associated with cost savings if the sensitivity of the OAR was reduced to 98.5% and the probability of litigation was increased 100 times to that of the observed rate of litigation for missed fractures (ie, CAN$7.16 per patient in Canada and US$4.00 to US$29.29/per patient in USA). Second, the study of Derksen et al23 showed that specially trained emergency department nurses were a cost-effective alternative to house officers in diagnosing and managing acute ankle injuries, using a protocol based on the OAR. The findings of the two cost analyses indicated that (additional) diagnostic imaging tests during the diagnostic process of acute ankle injuries was not cost saving compared to simpler approaches (ie, MRI + radiography vs radiography alone;24 arthrography, ultrasound or radiography +/− physical examination vs physical examination).10

### Findings of treatment studies (online supplementary table S2)

With respect to acute treatment, one full economic evaluation found that piroxicam was dominant to placebo (ie, reduced treatment costs and fewer training days missed).26 Another full economic evaluation indicated that the air cast and the below-knee plaster cast were more cost effective than the Bledsoe boot or bandage in people with severe ankle sprain from the UK healthcare-sector's perspective.25 However, from a societal perspective where the costs of productivity loss were included, the below-knee plaster cast was the most cost-effective option.25 Two cost analyses indicated that lower costs were associated with the use of a semirigid protective device during the acute phase. The use of an air cast compared with compression bandage resulted in cost savings by reducing the amount of sick leave.27 Also, the use of either air -cast or Unna paste plus tape resulted in lower direct costs at 6 months compared to immobilisation with a plaster cast.28

### Findings of prevention studies

Table 2 shows that with regards to preventing (re-)injuries, there was evidence that adding neuromuscular training to usual care was dominant compared with usual care alone among athletes following an ankle sprain from a societal perspective.3 That is, the programme was associated with less treatment costs and a reduction in the risk for recurrence. A cost analysis from an athletic-programme perspective comparing air cast to taping indicated that there was no statistically significant between-group difference in the number of ankle sprains but the use of air cast was associated with cost savings.29

## Discussion

The present systematic review assessed and summarised the currently available economic evidence regarding diagnosis, treatment and prevention for lateral ankle sprains. Ten studies representing five full economic evaluations and five cost analyses were included. We considered the results generated from the full economic evaluations to provide higher level of evidence than cost analyses because they consider costs relative to effectiveness outcomes versus costs alone. The findings from full economic evaluations suggest that implementation of the OAR may lead to wiser allocation of resources in terms of less unnecessary radiography and efficient delegation of work load without loss of health benefits, NSAIDs (piroxicam) compared with placebo or the plaster cast compared with the Bledsoe boot, air cast or bandage for severe sprain were cost-effective treatments and neuromuscular training was dominant over no training at preventing re-injury.

### Interpretation of the evidence

Our findings on cost-effectiveness supplement the latest guideline recommendations of effectiveness, which endorse the OAR and delayed physical examination for diagnosis, NSAIDs and plaster immobilisation for treatment, and neuromuscular training, brace or tape in the athletic population for prevention of lateral ankle sprain.9

The OAR were designed to have 100% sensitivity. The trade-off is lower specificity (ie, specificity for ankle fractures=0.50 (95% CI 0.46 to 0.55)), which means that there will still be ‘unnecessary’ radiographs taken due to false-positive findings.31 However, the number of ‘unnecessary’ radiographs remains lower compared to the practice of ordering radiographs for all cases. Furthermore, while it has been demonstrated that more efficient rules could be devised at the cost of missing a few fractures, it is unlikely that such rules would be accepted in practice by medical and health professionals or patients.32

Despite the high sensitivity of the OAR and evidence that non-medical health professionals can also apply the rules successfully, in practice, there may be hesitation to implement the rules due to fear of litigation for a missed fracture in patients who did not have a radiograph. It should be noted, however, that fractures are also missed on radiographs.22 Also, if the OAR are accepted as good clinical practice, as by an increasing number of international guidelines,9 ,33 ,34 the chance of successful litigation may be lower.35 ,36 Furthermore, Anis et al22 found that the implementation of OAR were still costsaving even when the probability of litigation was 100 times the observed rates (ie, 1% and 5% in Canada and the USA, respectively).

In diagnosis ankle sprain, findings of the two cost analyses pointed to less diagnostic workup. The addition of MRI to radiography should be discouraged as there were no cost savings compared with radiography alone and no significant differences in health-related quality of life or health status. With regard to assessing ligamentous rupture in cases where fracture was already ruled out, attention should be placed on delayed physical examination rather than diagnostic imaging.10 The reduced reliance on diagnostic imaging concurs with evidence in other musculoskeletal conditions, for example, back pain where routine imaging is being discouraged for its lack of effects and risk of radiation exposure.37

The NSAID piroxicam was found to be cost effective compared with placebo.26 However, increased instability was also noted in those taking piroxicam, raising the question of mobilising too early. There are a handful of studies on NSAIDs in ankle sprain, all with short-term results supporting the efficacy of NSAIDs.38 But such benefits in cost effectiveness and efficacy of NSAIDs need to be weighed against the recent evidence on their serious side effects, such as bleeding early, impediment of tissue healing and atrial fibrillation.39 ,40

The findings that plaster casts are more effective11 and cost effective25 compared with other immobilisation strategies appear at odds with other evidence in this area, including a Cochrane review of effectiveness41 and a cost analysis which was also included in the current review.28 The difference might be that Cooke et al25 exclusively recruited patients with severe ankle sprain (unable to weight-bear for at least 3 days). There is no information on whether plaster casts are also cost effective in people with less severe ankle sprain, but given the differences in effectiveness outcomes it is likely that the results in more severe ankle sprain are not transferrable to less severe ankle sprain.

A history of previous lateral ankle sprain is the strongest predictor of an ankle sprain.42 For patients with previous ankle sprains, the results from Hupperets et al3 suggested that neuromuscular training was dominant (ie, better effects and less costly) in preventing re-injury compared with no training. This is in line with subgroup findings of an economic evaluation by Verhagen et al43 which suggested that neuromuscular training would be cost effective if targeted at athletes with a history of ankle sprains. For those without previously injured ankles, the only evidence comes from a cost analysis which suggested that the prophylactic use of an air-cast brace was cost saving compared with taping.29 This finding is in line with the post-hoc cost-benefit analysis of Olmsted et al.44 Effectiveness studies show that in preventing re-injury, a brace or taping are effective. However, it is unclear whether a brace is more effective than tape,45 so the choice of a brace or tape depends on the individual situation, pragmatic considerations and costs..

### Strengths and limitations

To our knowledge, this is the first systematic review of economic evaluations pertaining to the diagnosis, treatment and prevention of lateral ankle sprains. We used standardised assessment tools to evaluate the risk of bias and methodological quality, and both trial-based and modelling studies were included. We tried to minimise this risk of missing studies by scanning reference lists, trial registries and not using any language restrictions. The number of available studies was small and meta-analysis was not possible due to study heterogeneity. Also, the available studies had small sample sizes, risk of bias and certain methodological shortcomings. As such, we cannot draw strong conclusions. Nevertheless, the results provide direction as to which diagnostic tests and treatment or prevention methods warrant further consideration for clinical practice and future research. As more research becomes available, this review can be updated.

### Quality of the evidence

The risk of bias assessment indicated that the main source of potential bias was blinding. The nature of the interventions in the included studies rendered blinding difficult. This argument does not negate the fact that the lack of blinding is a source of potential bias. As such, we kept the criteria in our assessment but excluded them from the denominator when calculating the overall score on the risk of bias.

The quality of the available full economic evaluations was fairly good but lessons can be learnt for future economic evaluations. There were three CHEC-list criteria that were fulfilled less consistently by current studies: cost valuation, outcome measurement and sensitivity analysis. Future economic evaluations could be improved by using quality-assessment tools such as the CHEC list in guiding the study design, data analysis plan and reporting. Two of the included studies did not specify the perspective taken10 ,26 and one used a narrow perspective.29 We recommend that future studies specify the economic perspective undertaken and adopt a wide perspective where possible (ie, societal), so that all relevant economic consequences are captured to provide comprehensive data to inform practice.15

A final recommendation to future studies is the length of follow-up period. Our review identified that current studies had a follow-up period ranging from 7 days23 to 1 year.3 ,22 For diagnostic studies, a follow-up period that covers the relevant costs and effects related to the diagnostic process could be sufficient. For treatment and prevention studies, longer follow-up periods would be desirable given findings that the residual complaints of pain, instability and recurrence of ankle sprain are common even at 3 years follow-up.46 Considering course of recovery for acute ankle sprains and trial resource constraints together, we would recommend at least 6 months of follow-up and the use of modelling to extrapolate findings beyond 1 year.

## Conclusions

Results of this systematic review provide an overview of the currently available economic evidence to supplement evidence available from reviews focusing on effectiveness, and can be used to inform clinical practice and practice guidelines. The current evidence is limited due to the low number of studies, risk of bias, certain methodological shortcomings and the small sample size of the studies. Nevertheless, the findings provide support for the implementation of OAR in the emergency setting, the use of NSAIDs and the plaster cast for severe sprain in the acute phase and the use of neuromuscular training to prevent re-injury.

View Abstract

• ## Supplementary Data

This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Files in this Data Supplement:

## Footnotes

• ▸ Supplementary tables S1 and S2 are published online only. To view the fi les please visit the journal online (http://bjsm.bmj.com/content/early/recent)

• Funding This research was funded by The Netherlands organisation for health research and development (ZonMw) within the framework of the program ‘Management of Knowledge, Quality Curative Care’. CL is funded by a fellowship from the Australian National Health and Medical Research Council.

• Competing interests None.

• Provenance and peer review Not commissioned; externally peer reviewed.

• ▸ References to this paper are available online at http://bjsm.bmj.com

## Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.