Methodology

Our Methodology

Systematic Reviews of the evidence to inform the guideline

Aim of the systematic reviews

A systematic review was conducted on each PICO. The aim of each systematic review was to determine the effectiveness of each physiotherapy intervention compared with no intervention, a sham intervention or another physiotherapy intervention on outcomes of impairment, activity limitation or participation.

Methods

Types of studies

Published randomised controlled trials (RCTs) and randomised controlled cross over trials were included. Trials with more than two parallel comparisons were included if two of the comparisons met the inclusion criteria. If trials were reported in more than one publication or interim analyses were published prior to the completion of the trial, then the most recent publication was used. Trials published only in English were included.

Types of participants

Adults (> 16 years) with a traumatic or non-traumatic SCI were included. Trials with a mixture of participants with different neurological conditions were only included if 80% or greater of participants within the trial had an SCI. Congenital condition involving the spinal cord such as spina bifida were excluded.

Types of interventions

All physiotherapy interventions identified in the list of PICOs were included. These were all interventions considered routine clinical practice in Australia and New Zealand.

Types of comparisons

Trials were included if they compared the interventions of interest with no intervention or a sham intervention. Trials that compared interventions with an alternate intervention were also included if they were a PICO of interest. Trials that included a co-intervention or usual care were included if the co-interventions or usual care were administered to both groups (making it possible to determine the added benefit of the intervention of interest).

Types of outcome measures

Trials were included that contained an outcome relevant to each PICO. These typically included measures of impairment, activity limitation and participation restriction. In situations where there was more than one measure of an outcome, we chose the outcome without looking at the results of the trial. A decision rule was used that prioritised measures considered important to clinicians and people with SCI.

Search methods for identification of studies

The following electronic databases were searched to identify reports of relevant studies: Ovid MEDLINE (1946 to August 13th 2020); Ovid EMBASE (1974 to August 13th 2020); EBSCO CINAHL Plus (1937 to August 13th 2020); Physiotherapy Evidence Database (PEDro) (Searched August 13th 2020) and CENTRAL on August 13th 2020.

To search Medline and Embase we used the OVID search strategy for RCTs combined with search terms for SCI. To search CINAHL we used the Cochrane search strategy for RCTs combined with search terms for SCI. To search PEDro we used category Neurotrauma combined with category RCTs. To search Central we used terms for SCI. Full search strategies can be found in the technical section of the guideline (Appendix One).

Searching other resources

In addition, we searched the reference lists of all identified RCTs and systematic reviews.

Selection of studies

Two authors independently screened the identified titles and abstracts using the pre-defined inclusion criteria detailed above. When required, the full text was then assessed to determine whether the trial met the inclusion criteria. If the trial met the inclusion criteria it was included. One author then selected studies from the identified list and matched them to each PICO question. If the trial did not meet the inclusion criteria it was excluded. Disagreements were resolved by discussion.

Data extraction and management

The data were extracted from the studies and recorded on an excel spreadsheet. One author independently extracted descriptive data.

The data extracted included:

trial methodology, type of trial and design of trial
trial participants including age, gender, neurological level of SCI, AIS classification of SCI, type of SCI (traumatic or non-traumatic) and time since injury
the experimental intervention including type, frequency, dosage of exercise or any details of the intervention provided
the comparison intervention
the co-interventions in the experimental and the comparison group
the outcome measure
the trial including authors, year of publication, setting and country

Details of data extraction for synthesis

Two authors independently extracted data for each study to determine mean between-group differences and 95% confidence intervals (95% CI). This included outcome scores and number of participants overall and in each group. Data were estimated from graphs if necessary. The following rules were used (from first to last) when deciding upon which data to extract:

mean between-group difference in post-intervention scores, adjusted for baseline scores.
mean and standard deviation (SD) of change scores provided in the studies (post-intervention scores and change scores were not pooled in meta-analyses in which results were expressed as standardised mean differences (SMD)).
mean (SD) post-intervention scores.

If only medians and inter-quartile ranges (IQR) were provided, medians were extracted and used as means, and SDs were estimated by dividing the interquartile range by 1.35. Cross-over studies were analysed using first period data or combined data if first period were not available. RevMan 5.4.1 software was be used to convert 95% CIs, standard errors, p values and any other appropriate combination of data or statistical results into SDs when necessary. The direction of effect of each outcome was standardised.

Meta-analyses were conducted across studies that made similar comparisons if there were at least two studies without excessive clinical or statistical heterogeneity. Clinical heterogeneity was assessed by examining the type of participants, type and intensity of the intervention, and other issues related to the design and conduct of the studies. Statistical heterogeneity was quantified using the I² statistic where an I² > 75% was considered to indicate excessive heterogeneity and results were not pooled. A fixed-effects model was used to pool data if the I² was less than 50%, and a random-effects model was used if the I² was between 50 and 75%. If studies in a meta-analysis used the same measure and same units, effects were expressed as mean differences (MD) and 95% CI. If different measures or different units were used within a meta-analysis, effects were expressed as SMD and 95% CI. In calculating SMD post-intervention scores were not be pooled with change scores. Data were analysed using RevMan v5.4.1. No sub-group or sensitivity analysis were performed.

Assessment of risk of bias in included studies

The risk of bias in each trial was assessed by one reviewer and checked by one reviewer using the five domains of Version 2 of the Cochrane risk-of bias tool.The domains assessed were potential bias arising from: the randomisation process; deviations from intended interventions; missing outcome data; measurement of the outcome; selection of the reported result.

The level of potential bias was judged as low, high or unclear (due to a lack of information or uncertainty) for each domain. Disagreements were resolved by discussion.

The PEDro score for each study was also extracted from the PEDro database. If scores were not available on the database one author assessed the score for the study.

Measures of treatment effect

Continuous data that used the same units were expressed as mean differences with 95% confidence intervals (CI). Continuous outcomes that use different units were expressed using SMD with 95% CI. Dichotomous outcomes were expressed as risk ratios (RR) with 95% CI. Time to event data were expressed as hazard ratios (HR) with 95% CI. Data were pooled in meta-analyses where appropriate and reasonable.

Unit of analysis issues

Unit of analysis issues was considered in the following three cases:

Cross-over trials

In cross-over trials data were analysed from the first period if available. Data for different periods within the trial were only used if first period data were not provided within the study.

Trials used in meta-analysis in which more than one type of intervention were compared

In trials that compared two or more types of interventions with no training or a sham group, data were analysed for all intervention groups. Double-counting of the control or sham group participants was avoided by using all data from the groups but dividing data from the control or sham groups by the number of groups.

Trials where multiple measures were taken on the same participant

In trials where multiple measures were taken on the same participant data at the end of the intervention period were used.

Dealing with missing data

All feasible available results were included. Authors were only contacted for missing data where clarifications were required. However, no data obtained from authors was used in the guideline. Only published data were extracted to use in analysis. All available data were converted where possible (for example, when data were reported as standard errors) using the calculator incorporated into Review Manager. If results were only presented graphically, we estimated the mean scores and SDs from graphs if it was reasonable to do so.

Assessment of heterogeneity

Data were pooled in a meta-analysis if there were two or more studies, there was clinical homogeneity (studies with similar interventions, participants and outcomes) and not excessive statistical heterogeneity (see the above details of data extraction for synthesis).

The Australian and New Zealand Physiotherapy Guidelines for people with SCI was funded by icare NSW, National Injury Insurance Scheme Queensland, Transport Accident Commission Victoria, and Lifetime Support Authority South Australia.

Acknowledgement of Country – The Australian and New Zealand Clinical Practice Guidelines team acknowledges the Traditional Custodians of country throughout Australia and their connections to land, sea and community. We pay our respect to their Elders past and present and extend that respect to all Aboriginal and Torres Strait Islander peoples today.

Honouring Te Tiriti – Ethical approval, including Māori consultation through University of Otago, Christchurch, Te Komiti Whakarite (Canterbury District Health Board), and through Māori consultation (Counties Manukau District Health Board) was obtained for Phase 1 of the study. In developing these guidelines, we were committed to meeting our obligations under Te Tiriti and reflect the principles of tino rangatiratanga, equity, active protection, options, and partnership.