The Process for Developing Cochrane Systematic Reviews

Cochrane Gynaecological, Neuro-oncology

and Orphan Cancer

The Process for Developing Cochrane Systematic Reviews

Search strategies

Access to specialised register by review authors

Access to the specialised register of gynaecological cancers may be obtained by searching CENTRAL for "SR-GYNAECA".

Additional search strategies

Review authors are strongly advised to:

1) Search for additional trials in the citations of each trial report relevant to their review topics

2) Write to the authors of all trials and/or reviews relevant to their review topic to request information on any similar trials of which they may be aware, and which remain unpublished, or as yet are not in CENTRAL

3) Write to pharmaceutical companies (if appropriate) to request information on pharmaceutical company trials.

Study selection

Review authors are generally advised to include only controlled trials and randomized controlled trials (RCTs).

The CGNOC recommends that the selection of studies for inclusion in a review is carried out independently by at least two review authors and that any differences are resolved by discussion. Differences that are not resolved by discussion between the review authors of a review should be brought to the attention of a third party and a majority decision taken.

Data collection

For each included study, data extraction should be carried out by more than one person, with differences resolved by discussion, or if that is not possible, by intervention of the editor responsible for the review. Where appropriate, kappa scores should be calculated for agreement between data extractors. Review authors are strongly advised to contact the author(s) of the primary research to verify the data and to provide further data where the report is incomplete. Where this is not possible, it should be stated in the text of the review.

Where the results of an RCT have not been published, but the author has satisfied himself/herself as to the quality of the data, that data should be collected. It should be stated in the discussion section of the review which of the data are from unpublished trials.

Review authors should gather information on rare adverse events and long-term complications. RCTs may not always be the best source of such information.

Analysis

Meta-analytic techniques, where possible, should be used in reviews. The statistical methods employed should be decided on a discussion with a statistician experienced in meta-analysis. When statistical methods are used to combine studies the methods should be stated in the protocol for the review. Where heterogeneity exists, the advice of a statistician should be sought, and where appropriate, sensitivity and other analyses should be performed.

Analyses are not specified a priority in the protocol but should be identified in the review. Where confidence intervals are used these are usually 95% CI for individual results and 99% for pooled estimates.

Help in finding statistical advice may be sought from the editorial base.

Reporting of reviews

Reviews will normally comment on the strength and applicability of the results. Where possible, review authors will include data on the potential harms and benefits of the intervention as well as costs if they are available. Such data may include non-randomised study derived data (e.g. rare adverse events).

Review authors should declare any conflict of interest in the appropriate section of the review.

Editorial process

Titles

All title proposals should be forwarded to the editorial base using the form provided giving information on the background, objectives, criteria for included studies and a target date for completion of the review. The Managing Editor will inform the editors of the group, and once accepted the title will be registered. Where two review authors’ teams notify the editorial base at the same time, it will be the responsibility of the Co-ordinating Editor to find a mutually agreed/agreeable solution.

All potential review authors of a review should notify the editorial base of their interest in the review to avoid conflict over review authorship. Where there are disputes about review authorship these should be brought to the Managing Editor in the first instance. If these cannot be resolved the matter should be referred to the Publication Ombudsman of Cochrane.

If a fortnight after the title has been distributed via the Titles Management System, the Group has not been alerted to any potential overlap with Cochrane reviews already being carried out, the author(s) will be advised that he/she may proceed, and a contact editor assigned to the review.

Review authors preparing protocols are strongly encouraged to attend a protocol development workshop organised by their reference Cochrane Centre.

It is expected that a first draft of the protocol will be submitted to the editorial base within six months, but no more than one year of the acceptance of the title. After one year, if no protocol has been received, a title may be deregistered and be offered to an alternative author team. Under special circumstances, extra time may be given to an author team. All communication regarding the review will be through the named contact person.

Protocols

The contact editor will decide if a protocol is ready to be sent to referees for peer review. Referees will, where possible and appropriate, include internal and external referees, methods/statistical referee, a consumer and a representative from a low-income country.

Referees are asked to return their comments (on the group's feedback form if desired) within three weeks. No attempt is made to conceal the identity of the review authors of the review from the referees, or vice versa.

The contact editor will summarise the comments of the referees and these will be passed on to the contact author. Following any necessary revisions, and after approval by the contact editor, protocols are submitted for inclusion in the Cochrane Library.

If an author requires statistical support they may be put in touch with the group's statistical advisor (via the Managing Editor). The review author(s) may attend the editorial base for consultation and/or assistance in using RevMan and Archie. The editorial base is willing to provide support in developing protocols and in finding expert help where needed.

Review authors who disagree on any aspect relating to the content of the review or to its method of preparation are encouraged to settle their dispute between themselves. If this is not possible, then the contact or co-ordinating editor will mediate, and if necessary involve the reference Cochrane Centre.

Where there is a disagreement between the editor and the author(s), the co-ordinating editor will mediate, and if necessary involve the reference Cochrane Centre.

Copy-editing, according to Cochrane Style Guide, is carried out on the protocol before submission for publication.

Protocols will normally be removed from the Cochrane Library after two years if no review has been received.

Reviews

For continuity, wherever possible the views of the referees who commented on the protocol will be sought again for the review once the contact editor has given his/her approval.

The contact editor will collate the comments and pass them to the author within three weeks. Following revision and approval by the contact editor, and further comment by the referees and others if this is thought appropriate the review will be submitted to the Cochrane Library. If the contact editor and the review authors do not agree on the submission further views from external referees will be sought. Where agreement cannot be reached, the CGNOC editorial team will make a final decision on whether the review will be submitted.

In addition to the support that is offered to the review author(s) at the protocol stage, review authors can expect to receive citations of newly-found trial reports on a regular basis. The trials search co-ordinator will search the Group's specialised register for each review up to the publication of the review on the Cochrane Library. Results of searches will be sent to the contact author whose responsibility it is to share the information with the co-review authors. Help with obtaining full papers or translations may be sought from the editorial base.

The full review will be subject to meet Methodological Expectations for Cochrane Intervention Reviews (MECIR) standards and copy-edited according to Cochrane guidelines before submission for inclusion in the Cochrane Library.

Updating

Review authors will be expected to update their reviews regularly, or as soon as new evidence becomes available. Where there is no new evidence, this should be stated together with the search strategy used to look for new data. Newly-found trial reports should be placed in the 'Studies awaiting assessment' or 'Classification pending' sections of RevMan during the updating process so they may be visible to readers of the review. Review authors will be expected to respond to criticisms as quickly as possible.

Once a review has been published on the Cochrane Library, searches of the specialised register will be carried out three months before a review needs to be updated and sent to the contact author.

People maintaining a review are afforded the same level of support as those preparing a review, and substantive updates are submitted to the same procedures as the initial review.

Similarly, the policy for dealing with disagreements between co-review authors and between review authors and editors is the same as that for those preparing a review.

If a review which has been published on the Cochrane Library, to the knowledge of the editorial team becomes out of date, the review authors will be called upon to rectify this. If this is unlikely to happen, the editors may select alternative reviewers to maintain the review, or if this is not possible, remove the review from the Cochrane Library.