Using qualitative research to explore heterogeneity in a Cochrane Review

Article type
Authors
Glenton C1, Carlsen B2, Flottorp S3, Jamtvedt G3
1Norwegian branch of the Nordic Cochrane Centre, Norway
2Uni Rokkan Centre, Norway
3Norwegian Knowledge Centre for the Health Services, Norway
Abstract
Background: The Cochrane Handbook suggests several ways in which qualitative research can contribute to Cochrane Reviews. Supplemental guidance to the Handbook suggests that one way of doing this is by providing information, for instance on diversity in population characteristics, that can contribute to decisionmaking about subgroup analyses, and thereby offer explanations for heterogeneity on study findings.

Objectives: To describe how a systematic review of qualitative research was used to explore heterogeneity in a Cochrane Review.

Methods: The Cochrane Review on the effects of audit and feedback on professional practice and healthcare outcomes offers a large body of evidence. However, the included studies showed large variation in effectiveness. When updating the review, the authors therefore explored factors that might explain this heterogeneity and provide a basis for exploratory analysis. A systematic review of qualitative research exploring GPs’ attitudes to clinical practice guidelines concluded that GPs’ reasons for not following guidelines differed according to whether the guideline in question was prescriptive, in that it encouraged a certain type of behaviour or treatment, or proscriptive, in that it discouraged certain treatments or behaviours. These findings were used as a basis for exploratory analysis in the Cochrane Review. The authors examined the direction of change required in each trial, i.e. whether feedback aimed to increase or decrease current behaviour. In addition, four other intervention characteristics (format, source, frequency and instruction for improvement/action plan) were identified through an additional meta-analysis and qualitative research, from theories for change (Control Theory, Feedback Intervention Theory) and common sense.

Results: The direction of change required, along with the other four characteristics of the intervention, helped explain variation in effects (expected difference in adjusted RD = 6%).

Conclusions: This experience supports the suggestion that qualitative data may provide a useful source of information when preparing systematic reviews of effectiveness.