19.03.2015 Views

Guide for Developing a Cochrane Protocol - Cochrane Public Health ...

Guide for Developing a Cochrane Protocol - Cochrane Public Health ...

Guide for Developing a Cochrane Protocol - Cochrane Public Health ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

<strong>Cochrane</strong> Handbook – Chapter 10<br />

Data synthesis<br />

The protocol should outline the procedures you intend to use to analyse and summarise the<br />

study results, including whether or not you intend to carry out meta-analyses. This section should<br />

contain a clear rationale <strong>for</strong> any choices, considering the potential impact of each choice on the<br />

outcomes of the review. Meta-analyses should only be undertaken if participants, interventions<br />

and comparisons are judged to be sufficiently similar to ensure an answer that is meaningful. It<br />

may be useful to organise this section by study type and/or by type of outcome data (e.g.<br />

continuous or binary).<br />

More specifically, if the intention is to carry out a quantitative analysis of results, you should<br />

outline in the protocol:<br />

• the software package that will be used to conduct the analyses (this should<br />

be RevMan and alternatives should be discussed with the CPHG);<br />

• how statistics describing the overall literature will be presented;<br />

• why a particular effect size metric is to be used;<br />

• if adjustments to effect sizes will be made <strong>for</strong> any reason;<br />

• techniques to be used to combine results of separate tests;<br />

• techniques to be used to assess and then analyse variability in findings<br />

across tests.<br />

As a default option, the CPHG recommend using the random-effects model to incorporate<br />

heterogeneity into your meta-analyses (as opposed to the fixed-effect model). The randomeffects<br />

model allows <strong>for</strong> a greater level of natural heterogeneity between studies, assuming that<br />

each study is estimating a unique ‘true’ effect applicable to the time, population and context in<br />

which the study was conducted. Fixed effects meta-analysis may be included in subsequent<br />

sensitivity analysis or when significant justification is provided in the protocol.<br />

For qualitative syntheses (with or without a meta analysis), you need to provide a rationale <strong>for</strong><br />

how you intend to organise studies/findings to arrive at reasonable conclusions and present<br />

in<strong>for</strong>mation in a useful way <strong>for</strong> end-users of the review. For example, you may like to synthesise<br />

studies grouped by the type of intervention, the length of the intervention, the type of outcome,<br />

or by study design. It may be difficult to identify what is most appropriate at protocol stage but<br />

we encourage you to report your anticipated approach or to identify the options <strong>for</strong> analysis and<br />

how the decision on synthesis structure will be reached.<br />

You must include in<strong>for</strong>mation about how you will summarise the findings of the review. Use the<br />

five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness and<br />

publication bias) to assess the quality of the body of evidence <strong>for</strong> each outcome, and to draw<br />

conclusions about the quality of the body of evidence within the text of the review. Note that this<br />

does not mean you have to provide a GRADE rating <strong>for</strong> the quality of the body of evidence (high,<br />

moderate, low, very low), however you must consider the five factors mentioned above in your<br />

review. Be sure to justify and document all assessments so that they can be included in your<br />

review. These factors must be addressed irrespective of whether your review will include a<br />

‘Summary of Findings Table’ (SoFT). If you plan to include a <strong>for</strong>mal SoFT in your review, you<br />

should include this in<strong>for</strong>mation in this section of your protocol. It is good practice to list the<br />

Last updated: 24 November 2011 - 16 -

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!