Scaling the Tower of Babel; the impact of including foreign language articles in a large-scale systematic review

Article type
Authors
Fayter D1, Heirs M1, Corbett M1, Eastwood A1, Fox D1
1CRD, University of York, York, UK
Abstract
Background: It has been argued that the inclusion of research in languages other than English (LOE) in systematic reviews is costly, time-consuming and unnecessary as it is often of poor quality and is unlikely to change the results of meta-analyses. Current advice from Cochrane suggests that reviewers make decisions about the inclusion of articles in LOE on a case-by-case basis. Objective: To explore the impact of a recent decision to include studies in LOE in a large-scale systematic review and scoping review of photodynamic therapy in cancer. Methods: We devised a comprehensive search strategy which included searching databases in LOE and writing to those conducting trials in photodynamic therapy worldwide. We engaged native speakers or experienced translators to help identify and data extract papers in LOE. Results: In addition to indirect costs involved in database searching and recruitment of native speakers / translators, the direct cost of LOE papers was 117. Thirty-three studies in LOE were identified as possibly relevant. Of these, one trial, published in Chinese, was included in the systematic review, fifteen in the scoping review and the remainder excluded. The included Chinese trial on palliative photodynamic therapy for oesophageal cancer reported longer follow-up than the other seven trials in the group and considered adverse effects. Conclusion: Although the inclusion of papers in LOE in this review incurred extra time and cost, there was a gain in the completeness of results. Furthermore, in a controversial topic area, we demonstrated a commitment to thoroughly identifying and assessing all the relevant literature, enabling us to state with increased confidence the need for better quality research. We believe these factors helped to increase the credibility of our review and suggest that studies in LOE should be considered for inclusion in systematic reviews unless there is a clear justification for their exclusion.