The Cochrane Schizophrenia Group

Technology; making the review writing process more efficient and accurate

on August 2, 2013


This is a follow up to the post 
Ferreting in CRS and RevMan…it’s technical

Health Info Mgmt students from Ulm

The plan:

Six months.

Two Information Management and Healthcare students from the University of Applied Sciences in Ulm, Germany.

Two great projects that enhance the systematic review writing process within the Cochrane Schizophrenia Group.

The result:

A program called ‘Ferret’ was created.  Its purpose: To find both hardcopy and electronic sources of studies relating to a particular subject e.g. Cannabis for schizophrenia (due to be updated if anyone’s interested, loud hint), and produce a report that sorts the results by the nearest library for our dear Drew to recover for our review teams.  This is an essential part of review writing that most people aren’t aware of.  You’d think we would just scoured Google, but in order to create high quality systematic reviews a more rigorous and reliable approach is needed…cue previous post Is Google a reliable tool for trial searches? and exciting news from Drew Davey, Research Assistant for CSzG and the Official Keeper of the Ferrets:

ferret-drawing-268x300

The new search program for the CSzG which will be invaluable in our post- CRS era.  The new Ferret F2 will integrate the CRS with our own elderly Ferret’s records of library holdings.

This evolution will enhance the various functions of the invaluable search program. For example, it

will be possible to teach the critter that specific issues of a journal are missing from a library’s holdings, and thus prevent false positives in library specific search lists.  While I could often remember many of the anomalies and realize that a trip to the British Library would be futile given how many errors the “old” Ferret had generated, my memory was rarely perfect. It is obvious that this improvement in accuracy will limit my frustration and trial searching expense.  There is also a new level of automation with cross referencing by both cities and journals.

Given that Ferret F2 will interact with the CRS in much the same way as its predecessor checked Meerkat for “Copy Not Obtained” status, it is crucial that the CRS be brought up to speed: the currently missing thousands of “Copy obtained” notes must be attached before the F2 can be taken for a meaningful run in the park. This will allow a well deserved retirement of the original CSzG Ferret who has been struggling with imports and exports since the arrival of the CRS.  The two students working on this project have certainly earned our gratitude with commitment and application.

A program called ‘RevMan HAL’ (yes, as in 2001: A Space Odyssey (1968) “”hello Dave”…we have a Co-Ed with a sense of humour) was created.  Its purpose:  To automatically generate accurate text (a first draft) for the “Effects of intervention” section of a review using data saved in the review by the review team in RevMan.  Now, I saw a demonstration of this and it was actually amazing.  There were a few buttons to press which will take some getting used to but the amount of text and the level of accuracy it adhered to was amazing.  Amazing still to think that this is stuff review authors have been made to write each time and often make mistakes in as it’s quite a boring regurgitating kind-of laborious task.

RevMan HAL

As a side-line it also helps us at the Editorial Base correct those common editorial mistakes such as ‘teh’ (‘the’), lower-case at the beginning of sentences and random spacing around punctuation; kind of like Find and Replace in Microsoft Word.  For people like me who commonly make such mistakes, programs and people that can ‘clean up after me’, as my mum and partner would say, are the unappreciated but highly necessary pieces in the review game.  In addition to this genius it also has some additional functionality that helps with copy editing.  I can’t wait until the copy editors get an eye full of this.

What does Steph Sampson, our in-house full-time systematic reviewer think of the program?

All Cochrane multi-review authors know the feeling; you’ve got your nicely-extracted data into nicely-formatted forest plot graphs, ready for a nice result-writing session. This feeling starts with enthusiasm, seeing as you’re ‘almost there’ in completing your review – however, things are only getting started. Repetition, double-triple-checking and uncertainty are afoot; and for reviews with more than 50 included studies, multiple-comparisons and outcomes with loads of data can serve to prolong the writing-up process with complexity.

At the click of a button, HAL identifies all comparisons, outcomes and included studies within your review, and automatically constructs your written results section. This includes your risk ratios/ odds ratios/ mean differences, confidence intervals and other meta-analysed data – formatted-and-all, it also takes into account other important factors such as heterogeneity and whether the results are statistically significant, picking-up on P= value numbers.

Hal 9000 A - No Chrome

Hal 9000 A – No Chrome (Photo credit: K!T)

This software is not only well-received news for systematic review authors, but on the grander-scale of things, the software has the potential to re-sculpt the way in which reviews are processed and received. Cutting-down writing time will excel the amount of reviews written, allowing authors to preserve energy for interpreting the data instead of getting lost in how it is to be presented in RevMan.

Even though it is still incredibly important that review authors know how to use the RevMan software alone, and know where to look to find their data, HAL is a programmers-send and is in existence to assist in pace of our research, not to render our interpretation skills redundant.

Both of these programs were shown to the top dog of Cochrane, David Tovey, Editor in Chief at The Cochrane Library, and other influential members: Toby Lasserson, Senior Editor at The Cochrane Library; Chris Mavergames, Web Development Director for the Cochrane Collaboration; and, Jacob Riis.  They were shocked at the progress made by two techie students in such a short space of time.  This could be revolutionary in the review writing world (it could also be the beginning of Terminator, but we won’t go there).  With a few more developments and a lengthy sign-off process commonly associated with the public and third sector we could have initiated change in the speed at which reviews are produced and the accuracy of the information recorded from included studies.  Watch this space.

If you use RevMan (which is free to download from the IMS Cochrane website by the way) you should definitely also download RevMan HAL (also free to download from our website) along with the guidance of how to use it.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: