Table 4 Suggestions for planning and performing data extraction in systematic reviews

Consider the specific review requirements early in the process including review complexity, size, resources, and experience and expertise of data collectors.

[6, 34]

Develop a thoughtful data collection form with clear instructions.

[6, 13]

Consider using existing and proven forms and adapt them as required, if available.

[23, 35]

Consider specific data collection requirements for more complex methods such as individual patient data or network meta-analyses and make use of available guidance for such situations.

[36]

Pilot data collection forms using a purposive sample of studies in light of the review specifics. This could, for example, include a mix of well and less well reported studies and different study designs or outcomes.

[13, 23]

Consider the merits and downsides of different extraction methods in the light of resources requirements, risk of errors and the severity of possible errors.

[3, 4, 30]

Be cognisant and reflective of the intricacies of coding such as stability, accuracy, reproducibility, and effects of framing, learning and fatigue.

[34]

When resolving disagreements between reviewers, make sure that a fair procedure is in place to avoid decision making simply based on seniority, experience, or power.

[30, 37]

Take advantage of software that can help to support workflows, keep a paper trail, and reduce risk of extraction errors.

[3, 6, 31, 38]