Methods used in Reviews

All Cochrane Incontinence Reviews, including protocols and updates, are expected to comply with both the latest version of the Cochrane Handbook and the Methodological Expectations of Cochrane Intervention Reviews (MECIR) standards. These supercede the advice given below. If you are a registered author with the Cochrane Incontinence Group and have questions, please contact cochrane.incontinence@newcastle.ac.uk.

Access to specialised register by review authors

Review authors are sent lists of possible studies for inclusion in their review after their protocol has been published at The Cochrane Library. Lists of studies are sent at other times on request of the review authors and, after consultation with review authors, before planned updates for a review. The contents of the specialised register are submitted to The Cochrane Library for inclusion in CENTRAL/CCTR.

Additional search strategies

Review authors are asked to contact the authors of studies included in any systematic review and those from whom clarification of study methodology is sought to ascertain if they know of any other possible RCTs relevant to the review. The citations of each trial report, and any reviews the review author is aware of, should be interrogated to identify further relevant studies.

The editorial team recommends that additional searching is done after consultation with the group's Information Specialist to make sure there is no overlap with past or current searching within the group.

Study selection

Studies that state they are randomised or quasi-randomised should be included in the review. The selection procedure should be done independently by two review authors. Any differences in opinion that cannot be easily resolved will be referred to the editorial team. If it is not clear how participants have been allocated to groups, clarification should be sought from the authors of the study.

Studies which do not state they have used randomisation should remain excluded unless the authors confirm that a randomised or quasi-randomised design was used.

Assessment of methodological quality

Two review authors will evaluate all relevant clinical studies independently for methodological quality. Any disagreements will be resolved by discussion and any differences in opinion, which cannot be easily resolved, will be referred to the editorial team. Assessment of methodological quality will be undertaken by each reviewer using the Cochrane risk of bias tool this includes assessment of: 

  • adequate sequence generation

  • allocation concealment

  • blinding

  • incomplete data addressed

  • freedom from selective reportin

  • freedom from other bias

Blinding can be further assessed as:

  • blinding of participants

  • health care providers or study personnel

  • outcome assessors 

Where appropriate, the description should include verbatim quotes from reports or correspondence. Alternatively, or in addition, it may include a summary of known facts, or a comment from the review authors.

Data collection

Studies which meet the criteria for methodological quality and subject relevance will be passed to the stage of data abstraction. Data relevant to the pre-stated outcome measures, characteristics of the study, interventions and participants will be abstracted independently by two reviewers.

Where data may have been collected, but not reported, further clarification will be sought from the trialists. Included trial data will be processed as described in the Cochrane handbook (Mulrow 1997). Trial data will be analysed according to the treatments compared and grouped by the type of incontinence. Any differences of opinion related to the data extracted will be resolved by discussion.

Analysis

Where appropriate, data will be quantitatively combined using meta-analysis to determine the typical effect of the intervention. Meta-analysis will be undertaken using the Review Manager (RevMan) software. This calculates a relative ratio or Peto's odds ratio for dichotomous data (Yusuf 1994) and a weighted mean difference or a standardised mean difference for continuous data.

The Incontinence Group recommends using a fixed effects model. Evidence of heterogeneity across studies should be discussed in the results. There is currently no consensus about the choice of summary statistics. We recommend using relative risks, as they are generally more intuitive, but recognise that odds ratios may be more appropriate in some circumstances, such as when events are uncommon. Crossover trials can be included in systematic reviews using the analytical methods described by Elbourne et al (Elbourne 2002).

Following the conventions of Cochrane, all comparisons will be framed in terms of unfavourable events, such as adverse symptoms or death, where possible. Relative ratio or odds ratios less than one therefore favour the experimental treatment, while a relative ratio or odds ratio greater than one favour the control treatment. 95% of confidence intervals will be derived for all comparisons.

The Incontinence Group recommend that reviewers always seek to perform intention-to-treat analysis, where participants are analysed in the group to which they were randomised, regardless of the intervention they actually received. For trials with missing data, it is recommended that the primary analysis is based upon the observed data without imputation for missing data. Sensitivity analyses may be carried out using different assumptions about missing data. 

Evidence of statistical heterogeneity across studies will be explored using the Chi-squared test for heterogeneity and the Istatistic. Studies will also be examined for methodological and clinical heterogeneity, particularly if significant statistical heterogeneity is identified (Thompson 1994). Data that cannot be combined quantitatively will be assessed qualitatively.

Studies will be excluded from the analysis if they are not randomised or quasi-randomised trials, or if they make comparisons other than those specified. These studies will be listed in the table of excluded articles.

Reporting of reviews

As of yet, the Group has no written policy on the reporting of results. Attempts to incorporate information on costs are encouraged, as is cross-referencing to other related reviews.