Topics and issues reported as 'other bias' in randomised clinical trials included in systematic reviews by Cochrane authors during 2017

Article type
Authors
Perez-Bracchiglione J1, Madrid E1, Franco J2, Rada G3, Bravo G4, Meza-Concha N5, Olguín P6, Garnham R6, Vergara L5, Urrea G5, Verdejo C5, Loézar C1, Papuzinski C1, Vargas M1, Arancibia M1, Vargas I7
1Interdisciplinary Centre for Health Studies Universidad de Valparaiso - Cochrane Chile
2Instituto Universitario Hospital Italiano - Cochrane Argentina
3Epistemonikos Foundation
4Centro Evidencia UC - Cochrane Chile
5School of Medicine Universidad de Valparaíso Chile
6School of Medicine Universidad de Valparaíso Chile
7Midwifery School Universidad de Valparaiso
Abstract
Background: Cochrane recommends a specific tool for assessing risk of bias, where the domain called 'other bias' (OB) refers to any potential systematic error not included within the classical bias which may lead to deviations from the truth. Nevertheless, the interpretation of other bias is often vague, no guidelines have been provided. Moreover, there is no clear appraisal about what are the issues Cochrane authors currently assess as OB.

Objectives: To assess and describe the issues or topics described as OB by Cochrane authors.

Methods: We retrieved all new systematic reviews (SRs) of interventions published by Cochrane during 2017 and reviewed the domain OB from the tables of characteristics of studies of all the primary studies as a way to assess rationale for judgement of this domain. We mapped each of the judgements into topics and calculated their frequency.

Results: We retrieved a total of 371 SRs, of which 77 (20.8%) did not assess risk of OB. The remaining 294 made a total of 6359 assessments of risk of OB of primary studies. Of these 6359 assessments of OB, 3394 (53.4%) were classified as 'low', 970 (15.3%) as 'high', and 1995 (31.4%) as 'unclear'. The most common topics for judgement of this domain were: "issues related to the funding sources” (24.8%), “baseline imbalance” (10.9%), “small sample size” (2.6%) and “issues related to the involvement of the sponsor during design, protocol, conduction, analysis or publication” (1.9%). The criteria for assessing and rating a certain issue widely varied among authors of of systematic reviews.

Conclusions: Cochrane authors considered and assessed the OB domain in most SRs conducted in 2017. Although we found great heterogeneity for the assessment of this domain, when it was evaluated, OB generally referred to issues related to the source of funding and the role of the sponsor during the study. It is desirable that appropriate instruments or guidelines are available to consider when a certain issue must be rated as high, low or uncertain for the most common topics considered as OB.

Patient or healthcare consumer involvement: This research was conducted by a multidisciplinary team composed by health professionals and researchers. The Institutional Review Board and the local Healthcare Patients Association contributed with external revision of the draft.