Do intellectual relationships create conflicts of interest?




Poster session 4 Saturday: Evidence implementation and evaluation


Saturday 16 September 2017 - 12:30 to 14:00


All authors in correct order:

Hakoum MB1, Noureldine H2, Guyatt G3, Akl EA4
1 Clinical Research Institute, American University of Beirut Medical Center, Lebanon
2 Faculty of Arts and Sciences, Lebanese American University, Lebanon
3 Department of Health Research Methods, Evidence, and Impact, McMaster University, Canada
4 Department of Internal Medicine, American University of Beirut Medical Center, Lebanon
Presenting author and contact person

Presenting author:

Maram Hakoum

Contact person:

Abstract text
Background: Studies that investigated whether conflicts of interest (COI) introduce bias have focused on financial COIs. Whether intellectual COIs influence judgment is unknown.

Objectives: We hypothesise that an intellectual relationship is associated with inferences from bodies of evidence. To test this hypothesis, we will study the association between trial authors’ best guess of the true effect size of a specific intervention, and the effect size to which they might be attached, i.e. the one reported in their trial.

Methods: A cross-sectional study of a random sample of Cochrane and non-Cochrane clinical systematic reviews published in 2015. Eligible reviews included a meta-analysis of a dichotomous outcome, including at least three randomised-controlled trials, and showing a substantial level of heterogeneity. We included the first three and last two trial authors of the trials contributing to the primary meta-analysis of the ‘outcome of interest’. We will survey these authors to ask about their best guess for the true effect size. We will run a regression analysis to measure whether the divergence (∆ BestGuess (i)) of trial authors’ best guess (EE BestGuess (i)) from the pooled effect estimate (EE Meta-Analysis (i)) is independently associated with the difference (∆ Trial (i)) between their trial’s effect size (EE Trial (i)) and the pooled effect estimate (please see Figure 1 below). We will adjust for trial authors’ financial COI in the analysis.

Results: We included 36 eligible meta-analyses (included 23 Cochrane reviews and 13 non-Cochrane reviews) that included a total of 236 trials and 885 trial authors. A significant independent association will support the hypothesis that intellectual COI affects judgment. The data-collection phase is under way and we will present results at the Summit.

Conclusions: While there is increasing interest in intellectual COI (e.g, in the setting of clinical practice guidelines development), there is little empirical evidence of its impact on judgment. The findings of our study will contribute to filling this important gap in the literature.