Ressurser for oversiktsforfattere

 

• Cochrane Handbook for Systematic Reviews of Interventions - offisiell håndbok for utarbeiding av Cochrane-oversikter
• RevMan homepage - dokumentasjon og veiledning for programvaren som benyttes for å lage og oppdatere Cochrane-oversikter.
• Cochrane Style Resource – sjekk din Cochrane-oversikt opp mot de offisielle språklige retningslinjene
• Using Individual Patient Data - Power Point-presentasjon
• Browse the list of Cochrane Review Groups
• Re-publishing of reviews - forklaring av prosedyrer og søknadsskjema for å få tillatelse til å publisere din Cochrane-oversikt I et annet vitenskapelig tidsskrift QUOROM statement checklist (PDF-dokument). Sjekkliste over ting som oversiktsforfattere kan framheve I oversikten for å forsikre leserne om påliteligheten av funnene.

Opplæring ansikt til ansikt

Kontakt Cochrane-senter eller oversiktsgrupper for informasjon om lokale workshops og kurs I hvordan lage systematiske oversikter. Noen av disse er omtalt på ”the Cochrane workshops”-siden.

Nettbasert opplæring

Open Learning Materials – lær framgangsmåten I passe store, nettbaserte moduler Dette er et supplement til Cochrane Reviewers' Handbook – til hjelp I prosessen med å ferdigstille en oversikt.

Opplæringstilbud fra andre organisasjoner:

Undertaking Systematic Reviews of Research on Effectiveness – omfattende veileder fra NHS Centre for Reviews & Dissemination  

 

Methods used in reviews

Search strategies

Once the title of the review has been registered, the Trials Search Co-ordinator (TSC) of the Skin Group will contact the lead author and offer assistance in the development of the search strategy section of the protocol. 
On publication of the protocol the TSC will run searches of the listed databases and will cut-and-paste all the search strategies into the Revman file of the review.  The citations will be sent to the author.
Once the review is published the TSC will send a reminder to the author 18 months after publication about updating the review. Once the author has responded, the TSC will start the searches for the update.

When searching for trials the TSC will search the databases in the following order:

1. The Cochrane Skin Group Specialised Skin Register.

Access to specialised register by reviewers

Authors may search the Skin Group's Specialised Register themselves, by typing the search term SR-SKIN into The Cochrane Library. This is the tag given to the records that are submitted in the Skin Group's Specialised Register. However the version of the Specialised Register on CENTRAL is always out of date because of the lag time between submission and publication, it will also lack many of the abstracts that are part of the citations. This is because abstracts cannot be submitted to CENTRAL due to copyright restrictions. You will get the most up-to-date information from the Skin Group handsearching project through searches the TSC does for you.

2. Cochrane Central Register of Controlled Trials (CENTRAL).

3. MEDLINE: The TSC will initially work with the author to develop a 'draft' search strategy in MEDLINE which is then pasted into the protocol and used as the basis from which to develop search strategies in other databases.

MEDLINE has been searched by the UK Cochrane Centre to 2004 using the 'highly sensitive search strategy for identifying reports of randomized controlled trials' as given in the Cochrane Handbook. Reports of trials have been transferred to CENTRAL. Therefore MEDLINE only needs to be searched from 2005 to the present. See www.thecochranelibrary.org and go to Product Descriptions > CENTRAL creation details.

4. EMBASE: This is a biomedical database that has a particularly comprehensive coverage of pharmacology, drug research and toxicology. It includes references from some European journals not included in other resources. It has been searched from 1974-2006 by the UK Cochrane Centre for the following free-text terms: random$; factorial$; crossover$; placebo$;doubl$ adj blind$; singl$ adj blind$; assign$; allocat$; volunteer$; and for the following index terms, known as EMTREE terms: crossover-procedure; double-blind procedure; randomized controlled trial; single-blind procedure.

All of these reports are now published in CENTRAL. Therefore EMBASE only needs to be searched from 2007 onwards. See www.thecochranelibrary.org and go to Product Descriptions > CENTRAL creation details.

5. LILACS (Latin American and Caribbean Health Science Information database): This database is searched on http://bases.bireme.br using the clinical trials filter on the site.

Additional search strategies

Other databases that may be of use to your review  and which the TSC will search for you include:

6. PsycINFO (psychology database).

7. AMED (Allied and Complementary Medicine database).

There may be other databases that are relevant to your particular review. If you search these then the search strategy used for each database must be saved and pasted into the full review. The date the search was conducted must also be noted in the review.

The Skin Group also recommends that the author does the following searches:

a) References from published studies
The bibliographies should be scanned for possible reference to RCTs. The authors must report in the review what they have done i.e. whether they have looked at the bibliographies of included studies, excluded studies or all papers referenced in the review.

b) Unpublished literature
Unpublished and grey literature should be obtained via correspondence with authors and a record made of what sources were searched.

With regard to ongoing trials the following websites should be searched:
www.controlled-trials.com;
www.clinicaltrials.gov;
www.anzctr.org.au;
www.who.int/trialsearch; and 
www.nottingham.ac.uk/ongoingskintrials/.

When searching the above sites remember to record the date of the search and the terms used.

c) Conference proceedings
The abstracts from conferences that have been attended should be scanned for reference to RCTs but before doing this check the Conference Proceedings (in the Handsearching section) that have already been handsearched by the Skin Group to save you unnecessary work.

d) Adverse effects
The Cochrane Skin Group encourages authors to investigate the adverse effects of the interventions used in their reviews. For common side effects summarise what is reported in the included randomised controlled studies. However RCTs are unlikely to report serious, rare or long term adverse effects. For serious side effects a separate search for non-RCTs may need to be done and the information summarised qualitatively. See chapter 14 of the Cochrane Handbook for advice on including adverse effects in Cochrane reviews: The Cochrane Handbook by Higgins JPT, Green S (editors). Cochrane Handbook for Systematic Reviews of Interventions Version 5.0.1 [updated September 2008] The Cochrane Collaboration, 2008. This is also available from www.cochrane-handbook.org.

The TSC will offer assistance, when the protocol is being written, so that the author considers the most practical approach to deal with this issue. Later when the protocol is published and the author is writing the review, the TSC will help with running a search strategy to identify the adverse effects if that was planned in the protocol.

Study selection

The Cochrane Skin Group recommends that the most reliable evidence comes from well designed randomised controlled trials, and that these should form the basis of its systematic reviews of interventions. Therefore, the reviews from the Cochrane Skin Group are normally based on randomised trials only. It is however important to also look at all papers that are classed as controlled clinical trials as the method of allocation may be random but may not have been clearly stated. Other forms of evidence such as case control studies are occasionally used when discussing issues such as adverse events. "Quasi-randomised" trials, where allocation is by non-random methods such as alternation, or is based on characteristics such as date of birth, name or case number, should not be included.

Authors assess each article to determine if it has met the predefined inclusion criteria for the review. These criteria include both the study design and the characteristics of the main outcomes under study. We recommend that the selection of studies for inclusion in reviews is independently carried out by more than one author. Differences between selections are then resolved through discussion and by consensus. Pilot testing of study selection on a subset of the trial reports should normally be carried out.

We do not consider blinding of authors to authorship of the trial reports to be necessary.

For more information please see the Cochrane Handbook www.cochrane-handbook.org.

Assessment of methodological quality

Within RevMan 5 you will see the Characteristics of included studies table has five entries for each study: Methods, Participants, Interventions, Outcomes and Notes. Up to three further columns may be specified for items not covered by these categories, for example, to provide information on length of follow-up or funding source.

See also: Detailed guidance on Characteristics of included studies tables is provided in Chapter 11 (Section 11.2) of the Cochrane Handbook www.cochrane-handbook.org.

Important change

Characteristics of studies that have the potential for bias should now be entered into the new Risk of bias table, which is an extension of the Characteristics of included studies table. Its use is not yet mandatory but is strongly recommended. The new RevMan 5 program will automatically show the item for allocation concealment but other characteristics must be selected from the risk of bias list. These categories are: assessments for sequence generation, allocation sequence concealment, blinding, incomplete outcome data, selective outcome reporting and other issues. For each item, the table provides a description of what was reported to have happened in the study and a subjective judgement regarding protection from bias (Yes for a low risk of bias, No for a high risk of bias; Unclear otherwise).

See also: Risk of bias tables which are discussed in Chapter 8 (Section 8.6) of the Cochrane Handbook www.cochrane-handbook.org.

Additional guidance from the CSG on quality criteria:
For more information please see the Cochrane Handbook www.cochrane-handbook.org

Data collection

It is highly desirable that data extraction from trial reports be independently carried out by at least two authors. A data extraction form is available from the Editorial Base. Any differences in data interpretation should be resolved through discussion among the authors or through a third person if a clear compromise can not be achieved. When necessary, authors are encouraged to contact trialists to ask them to clarify and/or verify their published (and unpublished) data.

Unpublished RCT data are eligible for assessment and possible inclusion in systematic reviews.

For more information please see the Cochrane Handbook www.cochrane-handbook.org.

Analysis

Pooling of results across studies should be done when appropriate, to derive more precise estimates of treatment effect than is available from individual studies. When statistical methods are used to combine studies or test for heterogeneity among included studies, the methods used are stated in the protocol for the review. Analyses not specified, a priori, in the protocol are identified as such in the review.

If possible a weighted pooled treatment effect should be calculated across studies using a random effects model (DerSimonian and Laird model). For dichotomous outcomes, the results should be expressed as risk ratios (RR) with 95% confidence intervals (CI), and as number needed to treat (NNT), where appropriate, with 95% CI and the baseline risk to which it applies. For continuous outcomes, the results should be expressed as difference in means (MD) with 95% CI, or as standardised mean differences (SMD) with 95% CI, where different outcome scales are pooled.

Sensitivity analyses:
Heterogeneity in studies should be explored using I. Where substantial heterogeneity exists between studies for the primary outcome (I > 50%), sources for such heterogeneity should be sought and explored in sensitivity analyses. Thus, if the study estimates appears to differ from others, reasons such as participant factors (e.g. age, diagnosis, sex race, co-morbidity), treatment factors (e.g. dosage, formulation), and study factors (e.g. concordance rates, quality of reporting) should be investigated further to explain such differences.

Non-randomised controlled studies should be listed but not discussed further. Studies relating to side-effects should be described qualitatively.

For more information please see the Cochrane Handbook www.cochrane-handbook.org.

Additional guidance from the CSG on analysis issues:

1. Using RR and not OR
Because risk and odds are different when events are common, the risk ratio RR and the odds ratio OR also differ when events are common. The Skin Group recommends that because many of the outcomes of trials of skin conditions are common events, then RR should be used.

2. Are there multiple lesions for example per individual?
If so these need to be treated as cluster trials which statistically adjust for the lack of complete independence of outcomes, say, of lesions in the same individual.

3. Has the trial compared cure for example between one side of the body treated with A and the other side treated with B?
This analysis requires the assistance of a statistician.

Additional guidance from the CSG on outcome measures:

1. Your protocol should include an outcome as assessed by the participant/patient.

2. Outcomes and timing: you need to pre-specify a clinically meaningful time to assess cure; this may often be weeks or months after treatment since many treatments for skin diseases may only have a transient effect.

3. Studies that do not include any of your outcomes - are they in or out?
The problem:
Having pre-specified your primary and other outcomes in your protocol, it is not uncommon to find randomised controlled trial that are otherwise eligible for inclusion in your review which do not contain any of the outcome measures that you have pre-specified. Alternatively, you may encounter trials that do include one of your outcomes, but they have not described it in the way that you wanted them to (e.g. they describe proportion of participants with 20% skin re-pigmentation as opposed to proportion with 50% skin re-pigmentation).

What do you do?  Do you include such studies (and change your outcomes) or do you exclude them because they have not measured what you think is important?

There is no clear guidance in the Cochrane Handbook at the moment about what to do in such situations, so what follows is guidance developed by the Cochrane Skin Group which will be replaced as new Cochrane guidance becomes available. The guidance is moulded by the guiding principles of the need to remain clinically relevant whilst striving to reduce bias.

The risks of the inclusion and exclusion approach are described below:

1. Include the studies. You have to be careful here, because if all the studies measure some meaningless short-term outcome, then you don't want to collude with that unsatisfactory state of affairs and magnify such irrelevant data in your review.

2. Exclude the studies. This seems the most obvious choice if the studies don't contain any of the data you want, but you are effectively excluding studies that otherwise fit your inclusion critera (unless you have pre-specified excluding studies that do not contain any of your outcome data in your protocol). In a field that is not bursting with RCTs, it seems a pity to exclude an RCT (or at least not to provide some details about them in the main body of the review), just because the trial did not measure what you wanted them to or in the way you wanted them to. In other words "throwing the baby out with the bathwater".

Possible solutions:
We recommend that you do the following:

1. Contact the authors to get the outcomes specified in your review. Include the new data if you can and acknowledge the authors' efforts, or consider adding them to 'Studies awaiting assessment' if getting the additional data is likely to take a long time.

2. If a trial identified in your main review does contain the outcomes specified in your protocol, but not in the way you need eg 20% repigmentation as opposed to 50% repigmentation, then in one sense that trial does contain your pre-specified outcome, but not measured in the right way. If you can't get the desired data from the author, then you should include the RCT in your review. If you do this, then you can describe the study briefly under the appropriate intervention category in the results section, pointing out that it did not contain any of your pre-specified outcomes. If readers are interested in what outcomes they did measure, then these will be available to them as it will be recorded in your table of included studies. By all means describe what results they found briefly in the results section, but they should be used in your summary results or forest plots.

3. If a trial identified in your review does not contain any of the outcome measures specified in your review at all, then you should also include that study, in your results section, pointing out that the study did not record any of your pre-specified outcomes. If your team feels that the other outcomes reported in the study are clinically important, then you can mention the results briefly in the text of your results section, but they should not be used in your summary results or forest plots. Details of what outcomes they did measure will of course be stated in your table of included studies. When it comes to your discussion section, you have an opportunity to comment on the fact that the study(ies) did not contain any of the outcomes that your review team considered to be the most important.

4. If you do decide that after reading the studies that you want to modify the outcomes specified in your protocol because those described in the studies seemed very relevant eg if you pre-specified improvement by 6 weeks as short-term response in acne, and found that none of the studies had recorded response at 6 weeks, but that they all did at 8 weeks (ie not much difference in what was a very arbitrary decision in the first place), then it may be possible to change your time point of assessment to 8 weeks, providing you state this clearly in your final review in the section 'Differences between protocol and review' and why you chose to depart from your protocol. Be very careful with this approach as it violates the whole principle of pre-specifying your outcomes and could possibly introduce bias. Only consider it if the differences are clinically trivial and where the over-zealous application of strict protocol criteria would exclude nearly all your studies.

Hywel Williams, Jo Leonardi Bee, Sarah Garner and Dedee Murrell
Cochrane Skin Group, November 2007

Reporting of reviews

Authors are encouraged to refer to other systematic reviews in The Cochrane Library where this is relevant especially to the Background text of the review or in the Discussion. This also has important implications for improving the Impact Factor of the Collaboration.

We encourage authors to publish their reviews in print journals or books, however we ask that authors contact the Cochrane Skin Group when considering publishing in another journal. Many other journals allow dual publication of systematic reviews in The Cochrane Library and their journal. We ask all authors to publish their review on The Cochrane Library before publishing in another journal or to arrange "simultaneous" publication.

Please refer to section 2.4 Publication of Cochrane reviews in print journals and books of the Cochrane Handbook www.cochrane-handbook.org for further information on this subject.

 

Editorial process

Titles

When a suggestion for a new title is received the Co-ordinating Editor decides whether to accept it as relevant to the scope of the Skin Group. The title is then held with others until we periodically ask all the membership to vote and list the available titles in their order of priority.  We then publicise the winning titles to the membership and on the website. We openly invite interested groups to apply to do the reviews. These titles are not only open to current Skin Group members, but to all who feel they would be interested in becoming part of a Cochrane Skin Group review team and would have something to offer the review team. In order for selection of the review teams to be as fair as possible we ask that those who wish to declare their interest to answer the following questions:

1. List the members of your team and their roles.
2. State why your team is well positioned to prepare and maintain this review (250 words maximum).
3. List up to 4 references that you or your team have published relevant to your application.
4. List the Cochrane reviews that you have completed so far. We require anyone leading a review team to have completed a Cochrane review before starting another review (to clarify: as leading a Cochrane review takes much time and effort, we advise you not to take on more than one Cochrane review at a time). 
5. It is also a Cochrane and a Skin Group requirement that your review team has a methodologist such as a statistician as part of the team. Please provide the name and email of this person.

If anyone is interested in becoming involved in writing a review for one of the selected titles, but is unable to build a team, or cannot fulfil all of the above 5 points, we still welcome expressions of interest. We ask that they tell us how they would like to be involved and how they could contribute. We then endeavour to inform the membership and help them to become part of a review team.

When other author(s) have previously expressed an interest in the same or a similar title, the individuals concerned are encouraged to collaborate on the project. In addition, the editorial base may be able to suggest to authors the names of potential co-authors and consumers that can participate in the review if that help is required. Occasionally, additional co-authors from specific backgrounds such as primary care or child health are solicited by contacting the appropriate Cochrane field.

We will then forward all applications to our panel of 3 independent editors who will decide which teams seem most able to complete a Cochrane systematic review.

Once the successful teams are chosen the lead author is asked to submit a title registration form to the editorial base csg@nottingham.ac.uk. This form includes a plan of work and clarifies the roles played by members of the review team.

1. We require each new review team to include a methodologist.
2. We require each new review team to include a team member with previous experience leading a Cochrane review.
3. We strongly recommend that all new review teams include a consumer: their participation ensures that the review is written in language accessible to lay people and that the outcomes chosen are consumer focussed.
4. Ideally, there should be a minimum of three people in addition to the consumer.
5. It is important that authors agree between themselves at an early stage how each will contribute to the process of delivering a final review.
6. It is the responsibility of the lead author to confirm that co-authors agree to their assigned tasks before the title registration form is submitted. Completion of the form is a prerequisite for registration. With the Title Registration form we also ask each author to read and agree to the terms of the "Author agreement of responsibilities and commitments" form and to sign a "Declaration of Interests" form. If title registration has not been completed within 3 months of acceptance of the title, the title may be released for other authors to pick up. 
7. We also recommend that at least one member of the review team is a native speaker of English.

Completion of the form is a prerequisite for registration.

If title registration has not been completed within 3 months of acceptance of the title, the title may be released for other authors to pick up.

After the title registration form has been accepted, the title is entered into the Information Management System for two weeks and checked for possible overlaps with other Cochrane Groups. At the end of that time period the authors are sent all the appropriate electronic paperwork to enable them to start writing their protocol. New titles are displayed on the Group's web site.

Protocols and reviews are expected to follow the format recommended in the Cochrane Handbook www.cochrane-handbook.org which can be accessed through RevMan 5. Also authors can make use of the Skin Group's help documents which can be found on the Resources page of the Cochrane Skin Group website. www.csg.cochrane.org.

The editorial base may be able to offer office space and a computer to authors during the development of their protocols and reviews. Those authors outside the UK may wish to contact their nearest Cochrane Centre for support http://www.cochrane.org/contact/country.htm.

Protocols

The Skin Group encourages UK authors to attend a Cochrane workshop on 'Developing a protocol for a review' before submission of the finished protocol. See http://www.cochrane.org/news/workshops.htm.

The editorial base expects that a title will be developed into a protocol within 6 months of its registration. Titles that have been registered for more than 6 months, for which protocols have not been forthcoming, may be 'de-registered' and become generally available once more for review by other Group members.

When the protocol is ready for peer review it is submitted to the editorial base. The protocol is sent to our Statistical Editor, Methods Editor, one designated 'Key' Editor, an external content referee and a consumer.

All referees are asked to give their response as free text divided into major and minor comments. The consumer referee is given a guide and checklist that focuses on readability, clarity, ease of understanding, and relevance. There are also comments from the editorial base and the Trials Search Co-ordinator. The editorial base collates the comments from all the referees and sends them to the lead author.

Authors show how their revision has addressed each of the referees' comments. When the necessary corrections and revisions have been made the revised protocol is re-submitted to the editorial base. The Co-ordinating Editor decides whether to accept the protocol for publication in The Cochrane Library, or if it will need further revision. If there is disagreement by authors with comments from referees, the Co-ordinating Editor will resolve the issue in consultation with the authors. Accepted protocols undergo copy editing, and any major copy-editing changes are sent to the authors for approval.

Before a protocol can be accepted for publication a Licence for Publication form must be signed by each of the authors and returned to the Skin Group. All authors must complete a Licence for Publication form which is available from within the RevMan5 file before the protocol can be submitted for publication. The Skin Group then sends these forms on to the permissions office of Wiley and sons.

The Skin Group aims to process protocols from their acceptance at the editorial base to submission to The Cochrane Library within three months, however this is greatly influenced by the replies from referees and whether the authors provide prompt responses to the referees' comments.

Reviews

UK authors are encouraged to attend the Cochrane workshop on "Introduction to analysis" when beginning the analysis in their review http://www.cochrane.org/news/workshops.htm.

In general the Skin Group expects reviews to be published within two years of registration.

The same referees that were used for the protocol usually assess the completed review, and the procedure for assessment is the same as that for protocols, except that one additional external content referee is used at this stage.

Disagreements between the editorial team and authors about the content of reviews are resolved by discussion with the Co-ordinating Editor. Disagreements between authors may be resolved through mediation by the Co-ordinating Editor.

Before a review can be accepted for publication a Licence for Publication form must be signed by each of the authors and returned to the Skin Group. All authors must complete a Licence for Publication form which is available from within the RevMan5 file before the review can be submitted for publication. The Skin Group then sends these forms on to the permissions office of Wiley and sons.

The editorial base expects to receive the review for peer review within 15 months of the protocol being accepted for publication in The Cochrane Library.

The Skin Group aims to process reviews from their acceptance at the editorial base to submission to The Cochrane Library within six months, however this is greatly influenced by the time it takes to receive replies from all the referees and the authors prompt responses to the referees' comments.

Updating

Authors are required to update their reviews every two years, and as quickly as possible in response to feedback. However, authors are free to update more frequently if they so wish.

The Trials Search Co-ordinator writes to the author 18 months after publication of their review to remind them that the review needs to be updated within six months and to offer help with running searches for trials. Revised reviews are commented on by both our Statistical Editor and Methods Editor, as a minimum, prior to re-publication. If the conclusions have changed and /or many new trials have been included then the update will also be sent to an external content referee and a consumer. Copy-editing of updated reviews follows the same pattern as for protocols and other reviews.

Updating reviews in response to the Cochrane feedback system will be negotiated between the authors and the Group's Feedback Editor.

If a review becomes out-of-date in an important way and its authors do not update it within fifteen months of being advised that this is necessary, the review may be removed from The Cochrane Library and published on the York Database of Reviews of Effectiveness (DARE).

 

Inclusion criteria

Clinical trials included in the Skin Group's Specialised Register must be:
a) on topics that fall within the scope of the Cochrane Skin Group
b) on a single population of living human beings
c) planned prospectively.

There must be a comparison between two or more interventions administered concurrently. These interventions must be assigned to the participants either by random allocation or by some quasi-random method of allocation (such as by alternation, date of birth, medical record number, social security number).

Suitable trials are identified from the full article.