Testing the form

Now you have designed your form, you need to test it to see if it works. The important features of a data collection form are that it is easy to use, comprehensive (it is very frustrating to get further along your review and find you need a piece of data you didn't record) and consistent across reviewers. Questions or items need to be phrased in an unambiguous way and you need to check they mean the same thing to all your reviewers.

Section 7.7 to 7.10 of the Reviewers' Handbook discusses the testing of data collection forms and you should read it now.

In summary the main components to testing your data collection form are

• Pilot testing the form with a sample of studies to ensure in is understandable, easy to complete and comprehensive
• Reliability testing the form in a more formal way by comparing a sample of completed forms by two or more reviewers to ensure they are in agreement.

Using the form

Two issues you may need to consider when starting to use your form to extract and record data are whether or not you need to blind reviewers and what to do about missing data. These issues are covered in Section 7.9 and 7.10 of the Reviewers' Handbook. As with the decision about trial inclusion and quality assessment, bias and error are likely to be reduced if more than one reviewer extracts the data, independent of each other. Whether or not bias is further reduced by blinding the reviewer to the source and authors of the study is yet to be determined, but some reviewers have gone to considerable lengths to mask studies of all identifying information.

By the completion of this module you should have designed and tested your data extraction form and will have also completed Modules 8, 9 and 10 as you worked your way through designing your data extraction form. Progress now to Module 11 for some more detailed information about the analysis components of your review.