Minutes of the RevMan Advisory Group special meeting held at the UK Cochrane Centre Thursday 3 February 2000, 10.30-14.30

Present: Clive Adams, Mike Clarke (Convenor and minutes), Mandy Collingwood (as stand-in for Sonja Henderson), Monica Fischer, Julian Higgins, Heather Maxwell, Rasmus Moustgaard, Helen Smith, Mark Starr, Lindsay Stead, Paul Whatling, Phil Wiffen, Sue Wilson.

Apologies: Phil Alderson, Chris Cates, Sonja Henderson, Peter Langhorne

Contents:
Welcome and introductions
Timetable for RevMan/MetaView 4.1
Future development of Information Management System
Date of next meeting

Welcome and introductions

Mike thanked everyone for coming to the meeting and explained that it was a special meeting of RAG with only two items on the agenda: an update on RevMan/MetaView 4.1 and the proposal for the future development of the Information Management System. It was agreed that the meeting should be restricted to these issues and that the routine business of the RAG would take place at the next regular meeting in April. Helen Smith and Julian Higgins (as the new statistical representative) were welcomed to their first meeting of the RAG.

Timetable for RevMan/MetaView 4.1

The initial beta-testing of MetaView 4.1 should be completed by the Nordic Cochrane Centre in mid February. The findings will be passed to Update Software for revision. They are unable to give a delivery date for the revised version before knowing the extent of the programming changes needed. About 20 people have agreed to beta test the combined RevMan/MetaView 4.1 for at least one full day each. It is expected that this would take one month but it will be extended if necessary to ensure that full, adequate testing is done. When the beta version is released to the registered testers, it will also be available on the Internet allowing interested, non-registered users to have a look at the new software. When the final version of the software is ready, it will be distributed on CD-ROM by the Nordic Cochrane Centre to all contact reviewers. Co-reviewers will be invited to register on a web site if they would also like to receive a CD-ROM copy. It is expected that 2000 CD-ROMs will be distributed and that this will take about two weeks.

A major difference between MetaView 3.1 and 4.1 is that the latter allows reviewers to specify the statistical settings used for each outcome, while the former has fixed defaults. Since MetaView 4.1 is a 32-bit program, Windows 3.1 users (of which there is still a substantial proportion) will not be able to run it. This is not a problem for the Internet version of The Cochrane Library where the graphs are viewed as graphic images generated from the reviewers' settings (and it might also be possible to use MetaView when viewing Cochrane reviews on the Internet). For the CD-ROM version, however, where MetaView is the default way of viewing graphs there may be a problem if MetaView 3.1 has to be used by Windows 3.1 users, since they would not necessarily see the graphs intended by the reviewers. Several solutions were suggested: 1) Include graphic images on the CD-ROM as for the Internet version (but this may not be possible because these may not fit on the CDs); 2) Modify MetaView 3.1 to allow reviewers to specify the statistical settings used for each outcome (Mark reported that this would be too difficult to do); 3) Display a warning message in The Cochrane Library if Windows 3.1 users run MetaView (3.1) for a review where some outcomes are set to anything other than the Peto Odds Ratio or WMD (which are the defaults in MetaView 3.1); and 4) Place warning messages in all reviews for which some outcomes are set to anything other than the Peto Odds Ratio or WMD. This issue will be discussed at the next meeting of the RevMan Advisory Group (April 2000) when there should be a better idea of when RevMan 4.1 will be released and Update Software will have a better idea of how much space the relevant images would occupy on the CD-ROMs.

Future development of Information Management System

Monica described the flow of information in the current system. There was some discussion about the representation of this as a circular flow. It was suggested that the complexity of the interaction between reviewers and CRGs should be made more explicit. This might be done by including an additional flow diagram, perhaps one that a CRG (eg PaPaS) already has. The flow through users and criticisms editors needs to be modified as reviews are updated for reasons other than this type of feedback. Mark pointed out that the content of the Parent Database is not reproduced exactly when published.

Monica showed a grid describing various information elements within the collaboration and the separate systems in use now. The purpose of this grid is to show that the same information is held independently in different places. In particular, this produces lots of duplication between what is held in the software used for review preparation and what is held in one or more of the other systems. Several changes were suggested to the format of this grid and an additional row should be added for study data. MeerKat should be noted as possible software for specialised registers. The RAG agreed that the key problem to be resolved is that information held in RevMan is also held independently in several other places and it would be better if information was held once only and linked to as necessary.

Monica clarified that although a needs assessment will be done to guide the development of the Information Management System, applications for funding need to be initiated now. This is because the current funding for the support and development of RevMan (to the Nordic Cochrane Centre) finishes at the end of 2000. The RAG agreed that the current proposal should be one of the options for the development of the Information Management System but that it is not the only one. It would be part of a vision for the future that would retain the idea of a standard format for Cochrane reviews but would make it possible for different pieces of software to produce the review through interaction with information held in the Internet system (or its offline equivalent). For example, if a RevMan 5 was developed, the proposal should make it possible for some reviewers to chose to continue with RevMan 4.1, or to use separate word-processing or analysis software, if they did not need all the features of RevMan 5.

The proposal is likely to be sent to a variety of funders - for example, in healthcare or software development. This will be done after consultation with organisations (initially in the US, in late February) who will be asked for advice on how best to refine it for particular funders and after changes are made to reflect the issues raised by the RAG.

Monica presented the outline for a fuller document which could go to funders who expressed an interest following the initial proposal. It was agreed that this should be circulated to RAG for comment by email so that Monica would be able to use it as her personal background paper at her meetings in the US. It was suggested that sections should be added to stress the benefits of the investment to the funder and to explain why the Collaboration is not able to fund the proposal itself. This should include an explanation of how the Collaboration has been funded until now and how it is expected that it will be funded in the next few years. It was agreed that ways of reducing the overheads' component of the projected budget (currently 40%) should be explored.

The needs assessment should be done through a survey and, possibly, focus groups at the Colloquium. It was agreed that it would not be possible or sensible to conduct any of this in time to influence the proposal that Monica will take to her meetings at the end of February.

In conclusion, the RAG agreed that the current proposal should be pursued with possible funders, subject to changes that reflect the discussions by the RAG and advice from those with fund-raising expertise. This decision was not an endorsement of this as the single vision for the future as that will need to be guided by the needs assessment. It was anticipated that the process of securing funding would allow adequate opportunity for further scrutiny, discussion and guidance from the RAG and Software Development Group. In addition it was agreed that the option of seeking funding to help with the early stages of the development process, before any particular model was adopted, should not be ruled out.

Date of next meeting

The date for the next meeting of the RAG will be determined by email but it might be adjacent to the UK CochraneContributors meeting to minimise the travel demands on members.

RAG minutes | MinMan home