10.3.2  Including unpublished studies in systematic reviews

Publication bias clearly is a major threat to the validity of any type of review, but particularly of unsystematic, narrative reviews. Obtaining and including data from unpublished trials appears to be one obvious way of avoiding this problem.  Hopewell and colleagues conducted a review of studies comparing the effect of the inclusion or exclusion of ‘grey’ literature (defined here as reports that are produced by all levels of government, academics, business and industry in print and electronic formats but that are not controlled by commercial publishers) in meta-analyses of randomized trials (Hopewell 2007b).  They included five studies (Fergusson 2000, McAuley 2000, Burdett 2003, Hopewell 2004), all of which showed that published trials had an overall greater intervention effect than grey trials. A meta-analysis of three of these studies suggested that, on average, published trials showed a 9% larger intervention effect than grey trials (Hopewell 2007b).


The inclusion of data from unpublished studies can itself introduce bias. The studies that can be located may be an unrepresentative sample of all unpublished studies. Unpublished studies may be of lower methodological quality than published studies: a study of 60 meta-analyses that included published and unpublished trials found that unpublished trials were less likely to conceal intervention allocation adequately and to blind outcome assessments (Egger 2003). In contrast, Hopewell and colleagues found no difference in the quality of reporting of this information (Hopewell 2004).


A further problem relates to the willingness of investigators of located unpublished studies to provide data. This may depend upon the findings of the study, more favourable results being provided more readily. This could again bias the findings of a systematic review. Interestingly, when Hetherington et al., in a massive effort to obtain information about unpublished trials in perinatal medicine, approached 42,000 obstetricians and paediatricians in 18 countries they identified only 18 unpublished trials that had been completed for more than two years (Hetherington 1989).


A questionnaire assessing the attitudes toward inclusion of unpublished data was sent to the authors of 150 meta-analyses and to the editors of the journals that published them (Cook 1993). Researchers and editors differed in their views about including unpublished data in meta-analyses. Support for the use of unpublished material was evident among a clear majority (78%) of meta-analysts while journal editors were less convinced (47%) (Cook 1993).  This study was recently repeated, with a focus on the inclusion of grey literature in systematic reviews, and it was found that acceptance of inclusion of grey literature has increased and, although differences between groups remain (systematic review authors: 86%, editors: 69%), they may have decreased compared with the data presented by Cook et al. (Tetzlaff 2006).


Reasons for reluctance to include grey literature included the absence of peer-review of unpublished literature. It should be kept in mind, however, that the refereeing process has not always been a successful way of ensuring that published results are valid (Godlee 1999). The team involved in preparing a Cochrane review should have at least a similar level of expertise with which to appraise unpublished studies as a peer reviewer for a journal. On the other hand, meta-analyses of unpublished data from interested sources are clearly a cause for concern.