Research Guides: Systematic Reviews: Data Extraction, Study Characteristics, Results

Data Extraction: PRISMA Items 9-10

The next step is for the researchers to read the full text of each article identified for inclusion in the review and extract the pertinent data using a standardized data extraction/coding form. The data extraction form should be as long or as short as necessary and can be coded for computer analysis if desired, particularly for a meta-analysis.

GW School of Medicine, School of Public Health, and School of Nursing faculty, staff, and students can use the various statistical analytical software in the Himmelfarb Library such as SPSS, Stata, and SAS.

Software to help you create coded data extraction forms from templates include: Covidence (free; our recommended tool), DistillerSR (needs subscription), EPPI Reviewer (subscription, free trial), or AHRQ's SRDR tool (free) which is web-based and has a training environment, tutorials, and example templates of systematic review data extraction forms.

If you prefer to design your own coded data extraction form, you can use AHRQ's Systematic Review Data Repository SRDR tool, or online survey forms such as Qualtrics, RedCAP, or Survey Monkey, or design and create your own coded fillable forms using Adobe Acrobat Pro or Microsoft Access.

Examples of data extraction forms:

Data Extraction Form Example (suitable for small-scale literature review of a few dozen studies)
This example was used to gather data for a poster reporting a literature review of studies of interventions to increase Emergency Department throughput. The poster can be downloaded from http://hsrc.himmelfarb.gwu.edu/libfacpres/62/
Data Extraction Form for the Cochrane Review Group (uncoded & used to extract fine-detail/many variables)
This is one example of a form, illustrating the thoroughness of the Cochrane research methodology. You could devise a simpler one page data extraction form for a more simple literature review.
Coded data extraction form (fillable form fields that can be computerized for data analysis) See Table 1 of Brown, Upchurch & Acton (2013)

Study characteristics: PRISMA Item 17

The data extraction forms can be used to produce a summary table of study characteristics that were considered important for inclusion.

A bibliography of the included studies should always be created, particularly if you are intending to publish your review. Read the advice for authors page on the journal website, or ask the journal editor to advise you on what citation format the journal requires you to use. Himmelfarb Library recommends using RefWorks to manage your references.

Results: PRISMA Items 18-22

In the final report the results from individual studies should be reported for PRISMA Item 19 as follows:

"For all outcomes, present, for each study:

(a) summary statistics for each group (where appropriate) and
(b) an effect estimate and its precision (e.g. confidence/credible interval), ideally using structured tables or plots."

In a review where you are reporting a binary outcome e.g. intervention vs placebo or control, and you are able to combine/pool results from several experimental studies done using the same methods on like populations in like settings, then in the results section you should report the relative strength of treatment effects from each study in your review and the combined effect outcome from your meta-analysis. For a meta-analysis of randomized trials you should represent the meta-analysis visually on a “forest plot”. If your review included heterogenous study types (e.g. some combination of experimental trials and observational studies) you won't be able to do a meta-analysis, then instead your analysis could follow the Synthesis Without Meta-analysis (SWiM) guideline.

The PRISMA required reporting items include: risk of bias, results of individual studies, results of syntheses, reporting biases, and certainty of evidence.