Collecting data you want, in a format you want it

Now that we have identified studies we are going to consider for inclusion in the review, we need to work out how we are going to use them and what information we need to extract from the studies to include in the review. Like all other components of the review we try to do this in a systematic way. At first it may seem odd to have a whole module devoted to the issue of collecting data from the studies you've found. But you've somehow got to manage the process of getting information about a study from a paper publication (or more than one paper, or a letter or a phone call….), organise it and then present it in your review. If you don't have a system for doing this, it's easy to forget where information came from, just collect what is easiest to get out of papers, or make mistakes.

In addition to collecting information from the studies, you will need a system for collecting information about the process you follow, for example the number of studies you find, how many of those you include in your review and so on.

What information do we need to collect?

There are four main types or categories of information we need to collect from studies considered for inclusion in our review:

• how the study measures up against our eligibility criteria and whether it should be included in the review (and if not, why not)
• descriptive information about the study to complete the 'Table of Included Studies' or 'Table of Excluded Studies in RevMan' in RevMan
• information about the quality of the study
• information about the results of the study in the form of data to be used in your review

Why bother with a specific form?

You may be thinking that taking the time and effort to design and test a structured form to extract data from studies is not worth it when you could just sit down with your included studies and enter the information you need straight into RevMan. Experience suggests that it is worth taking time over this step and Section 7.1 of the Reviewers' Handbook gives you some good reasons why we bother with data collection forms. You should read that now.

In summary we need data collection forms as they:

• specifically mirror your review question and so allow you to reassess all important components of your question and ensure you extract the relevant data relevant
• document the history of the process of taking the information from the study into the review and so allow you to backtrack and identify any errors or assumptions you have made in your review
• record the information you need to generate the results of your review, without any additional, unnecessary information. This will allow more efficient data entry into RevMan and highlight any gaps in the data
• allow more than one reviewer to extract data from a study and compare the results of their work, increasing the reliability of the data collected for the review. This is an important step in reducing bias.

Format of data collection forms

There are several different approaches to the format of your data collection form and no single style is necessarily the best. Some reviewers prefer an open ended form with a few headings for structure and lots of space to record information in flexible formats. Others prefer to list every item they wish to extract in a highly structured, checklist type approach.

The following two examples will give you some idea of the variation in data extraction forms. You should design your form to suit your preferred style of working.

Example_from_a_review_ in_schizophrenia.pdf

Example_from_EPOC.pdf

Paper versus electronic forms

Section 7.2 of the Reviewers' Handbook devotes a little time to discussion of the benefits of paper versus electronic data collection forms. On the surface it may appear that an electronic form where data can be transferred directly from the form into RevMan would save time, however, at this stage that can only occur for transferring numerical data in spreadsheet form.

If however you prefer working with electronic media you may prefer to extract data into an electronic form and enter into RevMan from there. To date, most reviewers have used paper data collection forms.

Some review groups may have a template for a data collection form you can modify to your particular review but in many cases individual reviews are so specific that a form needs to be designed to match that review.