Publishing results of systematic reviews and metaanalyses

Dissemination of the results of a systematic review should be similar to that of a clinical trial and much of the guidance presented in Chapter 10 concerning publication and presentation of RCTs also applies to systematic reviews.

As discussed in Section 11.1, one of the advantages of systematic review is that objective and pre-specified methodology is used. Transparency of process should ensure that other researchers can easily reproduce the results. It is therefore essential that reports and publications should be consistent with this aim, and so should report analyses and results in sufficient detail to permit reproduction and present results and discussion in an unbiased format. It is important that any potential weaknesses of the review are brought to the attention of the reader. For example, if the review did not consider unpublished data, it should be stated that publication bias might be a problem.

The methods section is particularly important. In the same way that the report of a trial should provide details of the types of patients who were considered for the trial, so the report of a systematic review should report what the inclusion and exclusion criteria for the review were. It is important to give details of all trials that were considered for inclusion (as distinct from trials that were clearly irrelevant) and to provide details of why any particular trials were excluded. It should also be clear which of the eligible trials were able to be included in any analyses, as well as details of and reasons why any could not be included; for example, if publications did not report sufficient data to be included in quantitative analysis, or if trialists would not supply individual patient data. The search strategies used to locate trials should be reported, and it should be stated whether attempts were made to locate unpublished trials and the method for doing so. Methods used to screen trials and extract or obtain data should also be given. Similarly, the quantitative methods should be described fully. The fact that analyses were pre-specified should be stated, as should an outline of planned analyses.

The main characteristics of eligible trials should be reported (it is often easiest to do this in tabular format) and an overall description of patient characteristics given. The results of all analyses should be reported, though some maybe dealt with briefly. Ideally, any planned analyses that were not ultimately possible should also be mentioned. As discussed previously, it is often useful to present results in a variety of formats, reporting for example both relative and absolute effect sizes. The robustness of the meta-analyses in terms of sensitivity analyses should also be reported, and any heterogeneity and its implications noted.

The discussion section of a systematic review often maybe relatively brief. The position that many researchers involved in the field take is that their role is to conduct the review to the highest possible standards and to simply present the results. Many argue that unless the results are overwhelming, interpretation should be left to the reader.

0 0

Post a comment