Data Extraction - Questions about using EPPI-Reviewer

Forum (Archive)

This forum is kept largely for historic reasons and for our latest changes announcements. (It was focused around the older EPPI Reviewer version 4.)

There are many informative posts and answers to common questions, but you may find our videos and other resources more informative if you are an EPPI Reviewer WEB user.

Click here to search the forum. If you do have questions or require support, please email eppisupport@ucl.ac.uk.

<< Back to main Help page

02/09/2011 23:58

Monali Malvankar

Joined: 17/10/2011

Posts: 16

Data Extraction

Hi,

Following are the papers we like to extract data from:

1) Arshinoff, S.A., Y.N.J. Strube, and R. Yagev, Simultaneous bilateral cataract surgery. Journal of Cataract & Refractive Surgery, 2003. 29(7): p. 1281-1291.

2) Acosta-Rojas, E.R., et al., Association Between Visual Impairment and Patient-Reported Visual Disability at Different Stages of Cataract Surgery. Ophthalmic Epidemiology, 2006. 13(5): p. 299-307.

3) Alfonso, J.F., et al., Visual acuity comparison of 2 models of bifocal aspheric intraocular lenses. Journal of Cataract & Refractive Surgery, 2009. 35(4): p. 672-676.

the last paper compares 2 lens. my question is: in the paper they have mentioned baseline (before surgery) characteristics of patients and after surgery (outcomes). In EPPI, they have option of outcome but how to input the baseline characteristics?

05/09/2011 12:30

Jeff Brunton

Joined: 17/10/2011

Posts: 594

Re: Data Extraction Modified By Jeff Brunton on 05/09/2011 11:33:44

Hello Monali,

EPPI-Reviewer supports the meta-analysis of both 'd' and 'r' families of continuous effect size (as well as binary outcomes). Calculating effect sizes are one of the most difficult and involved components of meta-analysis, but useful texts are:

The Handbook of Research Synthesis by Harris M. Cooper (Editor), Larry V. Hedges (Editor), Jeffrey C. Valentine (Editor);

and Practical Meta-Analysis (Applied Social Research Methods) by Mark W. Lipsey and David Wilson.

A useful online calculator for converting published statistics to effect sizes is at:

http://www.campbellcollaboration.org/resources/effect_size_input.php

(Michael Borenstein and colleagues also have a book in preparation: 'Computing Effect Sizes for Meta-analysis' that is currently due for publication next April.)

EPPI-Reviewer has built-in functionality for calculating effect sizes from a range of statistics that are often presented in publications, though it cannot be exhaustive, and cover every way that a study may present its results. You may need to do some prior calculation before using the 'manual entry' outcome option, to enter the effect size and standard error directly. (SMD and SE are the most likely statistics you'll use). It looks as though one of the challenges you face is dealing with studies with a before/after design. It's quite possible to calculate effect sizes from these studies - and the Lipsey & Wilson book and online calculator above may help here - though they are not as strong a design for attributing causality as a controlled trial.

(EPPI-Centre researchers may be able to help with specific calculations, but we'd probably need to charge for their time.)

Best regards,

Jeff

06/09/2011 15:21

Monali Malvankar

Joined: 17/10/2011

Posts: 16

Re: Data Extraction

My question is: how to input baseline charateristics since they cannot be classifed as outcome or intervention......or should I just have a checkbox in front of charateristics so that the reviewer can enter them manually in the info tab?

Since if it is classified as outcome then only "manual entry" comes into play.

Monali

07/09/2011 10:04

Jeff Brunton

Joined: 17/10/2011

Posts: 594

Re: Data Extraction

Hello Monali,

This depends whether you are wanting to carry out ‘simple’ meta-analysis in EPPI-Reviewer, or export the data and carry out more complex analyses in another package such as stata. If you want to use your baseline characteristics as covariates in a meta-regression (in stata / spss etc), then you should enter them as ‘outcome classification codes’ (please see the section on Meta-analysis starting on page 60 in the user manual). If you are capturing information according to which you want to subdivide studies in EPPI-Reviewer, you should use standard checkbox codes.

Best regards,

Jeff

Page 1 of 1