What is the prognosis of physical functioning following an ankle injury?

The following electronic databases will be searched: Cochrane Central Register of Controlled Trials via OVID, MEDLINE via PubMed, EMBASE (www.embase.com), CINAHL via EBSCO-host, PEDro (www.pedro.org.au), AMED via OVID and SPORTDiscus via EBSCO-host.

In addition to the electronic searches, we will conduct citation tracking, checking the reference lists of the included studies and relevant systematic reviews.

There will be no language or publication restrictions applied to the search strategy.

All quantitative studies with a longitudinal design will be included, such as inception cohort studies, randomised controlled trials, crossover studies, retrospective studies, case-control and case series studies. Data from all samples will be collected regardless of the intervention received. From case-control studies, data from the cases will be extracted.

Cross sectional studies and case-reports will be excluded.

Studies evaluating a sample of people with a sudden onset injury to musculoskeletal structures of the ankle joint (bony or soft tissue) will be included. Such injuries could include, but need not be limited to, ankle sprain, distal tibial and/or fibular fracture, ankle dislocation, and Achilles tendon rupture.

Studies evaluating musculoskeletal conditions that are non-injury related, for instance rheumatoid arthritis of the ankle and tumour related diseases, will be excluded. Foot injuries (e.g. fracture of the fifth metatarsal or talus fracture) and ankle injuries that present as a gradual onset of symptoms due to overuse (e.g. Achilles tendinitis) will also be excluded. If a study contains a mix of different diagnoses but is possible to extract data for the ankle injury-related conditions separately, the study will be included.

Studies reporting data on female and male samples of all age groups will be included.

Not applicable.

Not applicable.

Primary outcomes

We will assess the prognosis in physical functioning, which will be operationalised as changes in physical activity levels and/or activity limitation. Physical activity levels can be assessed either objectively (e.g. accelerometry) or by self-report (e.g. questionnaires). Activity limitation can be assessed either with questionnaires (e.g. Walking Impairment Questionnaire), scales (e.g. Lower Extremity Functional Scale) or objective tools (e.g. walking tests).

Secondary outcomes

None.

Two reviewers will independently screen titles and abstracts and then, if necessary, the full text of studies identified by the search strategy using an electronic screening form designed to assess eligibility criteria. A reason for exclusion will be provided for the exclusion of all full-text studies screened.

Two reviewers will independently extract data from the included studies using a data extraction form. The following data will be extracted: sample source and size, study design, inception time (time between injury and recruitment), type of injury, tool(s) employed to assess physical activity levels and/or activity limitation, data on physical activity and/or activity limitation at each time point assessed, and the timing of follow up. Data will be extracted for all follow-ups available in each included study. The type of intervention will also be recorded for the relevant included studies.

Both the electronic screening form and the data extraction form will be specially created for the review and will be piloted before use. All disagreements regarding the selection of included studies or data extraction will be solved by discussion and, if necessary, arbitration by a third reviewer.

Authors of the included studies will be contacted if the study reports incomplete or missing data. If the author does not reply within two weeks, a second email will be sent as a reminder. If, at the end of the fourth week, a reply has not been received, we will use the available data.

To assess the risk of bias of included studies, we will use four out of the five measurements suggested by Altman 2001. The scale contains:

1) Defined sample: source of participants and sample selection described;

2) Representative sample: gathered at a common point in course of the disease;

3) Follow-up rate > 80%: outcome data available for at least 80% of participants at one follow-up point; and

4) Analysis: appropriate, including statistical adjustments for important prognostic factors.

The risk of bias assessment will be conducted by two independent reviewers and disagreements will be solved by discussion and, if necessary, arbitration by a third reviewer.

The data will be presented separately according to the type of ankle injury into one of the four groups: (1) ankle fractures, (2) ankle sprains or strains, (3) Achilles tendon ruptures, (4) other injuries. If possible, a meta-analysis will be conducted with the included studies.

For continuous variables, measures of central tendency (mean or median) and dispersion (standard deviation, standard error or 95% confidence intervals) will be converted to or approximated by means and standard deviations. For dichotomous or categorical data, the proportion of subjects in each category will be extracted. Where different tools are employed to assess the outcomes, their measurement properties will be investigated to determine if it is possible to combine the data. For continuous data, pooled estimates of the regression coefficients and the error variance of the course of physical activity or activity limitation across studies will be obtained using linear mixed models using Stata. Dichotomous or categorical data will be pooled using mixed logistic regression. Each study will be assigned a weight equal to the inverse of the mean variance of the estimates from that study. To account for the inconsistency of timing of outcome assessment in the included studies, the relationship between time and the outcome(s) will be investigated. The relationship between the outcome(s) and time (or transformed time) will be modelled by treating time as a continuous variable and the measures of physical activity or activity limitation will be used as the dependent variable in the analysis. For purposes of reporting, outcomes will be categorised in immediate-term (within a week of onset of the condition), short-term (one to six weeks after the onset of the condition), medium-term (six weeks to six months after the onset of the condition), and long-term (over six months after the onset of the condition).

We anticipate that there will be significant heterogeneity. Between-study variability will be quantified using the tau and rho statistics. The former is a model-derived estimate of the standard deviation of outcomes on day one, and the latter is the proportion of the total variance due to between-study variability (due to tau). We will explore the possible causes of heterogeneity from a clinical perspective, examining participant characteristics, interventions and outcomes.

Inception cohort studies are considered to have the ideal design to provide the most robust evidence on systematic reviews of prognosis studies. For this reason, a sensitivity analysis will be conducted including only studies with an inception cohort design if sufficient inception cohort studies are found.

None planned.

The results of this review will be submitted to a peer-reviewed journal for publication as well as presented in international conferences.

Paula R Beckenkamp

PO Box M201

Missenden Rd

NSW 2050

Australia

pbeckenkamp@georgeinstitute.org.au

The George Institute for Global Health and the Sydney School of Medicine, University of Sydney

http://www.georgeinstitute.org/ and http://sydney.edu.au/medicine/

Miss Paula R Beckenkamp, The George Institute for Global Health and the Sydney School of Medicine, University of Sydney
Dr Chung-Wei Christine Lin , The George Institute for Global Health and the Sydney School of Medicine, University of Sydney
Professor Robert D Herbert , Neuroscience Research Australia, NeuRA
Dr Hidde P van der Ploeg , Department of Public and Occupational Health, VU University Medical Center Amsterdam
Dr Anne M Moseley , he George Institute for Global Health and the Sydney School of Medicine, University of Sydney

Mrs Sakina Rashid-Chagpar, The George Institute for Global Health

01 August 2012

30 November 2012

Paula R Beckenkamp is funded by the International Postgraduate Research Scholarship (IPRS) and the Australian Postgraduate Awards (APA) scholarships

Chung-Wei Christine Lin and Robert D Herbert are funded by the NHMRC Research Fellowship

None known

English

Australia

Subject indexing assigned by CRD

Ankle Injuries; Humans; Prognosis; Recovery of Function

20 September 2012

20 September 2012