Cochrane Collaboration > colloquia > abstracts > dublin 2006 Extracting data for a systematic review of non-randomized studies (NRS) of a public health intervention

Barney Reeves, George Wells, Beverley Shea

Objectives: The workshop aims to improve awareness among editors and review authors of the key issues when including NRS in systematic reviews of effectiveness. The majority of the time during the workshop will be taken up by participants working in small groups to extract data from a primary NRS of a public health intervention. Summary: This workshop is aimed at review authors and editors who are considering whether or not to include non-randomized studies (NRS) in Cochrane reviews when there are no randomized controlled trials (RCTs), only poor RCTs or very few small RCTs but where an answer to the question addressed by the review is a priority. Evaluations of public health interventions often have these limitations. Decision makers typically view some evidence as better than none, although Cochrane reviews usually exclude NRS because of concern that the data are biased. The workshop will follow the format used in the last two years but differ in that it will focus on evaluation of public health interventions. It will be based on draft guidance produced by the Cochrane NRS Methods Group (NRSMG):

• participants will consider the kinds of information that the NRSMG recommends extracting from primary studies;
• participants will investigate the availability and quality of the information required in a sample primary study (previously included in a systematic review of NRS).

Three instruments will be used for data extraction about:

• study design (a checklist developed by the NRSMG);
• the Newcastle-Ottawa Scale (for quality assessment of NRS);
• a form for extracting data on study findings (adapted from a form used previously for a systematic review of NRS).

The implications of varying amounts and quality of information from primary NRS for systematic reviews of NRS will be discussed. Varying amounts and quality of information are also the norm for systematic reviews of RCTs, so the discussion will contrast the implications for systematic reviews of NRS and RCTs. Level of knowledge required to attend: intermediate.

Dublin 2006 -- W09