Verification of conflicts of interest disclosures in health research: a conceptual analysis

Article type
Authors
Khamis AM1, Hakoum MB2, El-Rayess HM3, Haddad S3, Akl EA3
1Clinical Research Institute, American University of Beirut
2Department of Family Medicine
3Department of Internal Medicine, American University of Beirut
Abstract
Background:
Systematic reviewers are expected to collect and report the conflicts of interests (COI) of authors of included primary studies, and to declare their own COIs when publishing their reviews. Since the pharmaceutical and medical device companies have started publishing their payments to physicians online, investigators have started verification of conflicts of interest declarations in health research reports.

Objective:
To provide a conceptual analysis of the processes of conflicts of interest (COI) disclosure, and of verification of COI disclosure in health research.

Methods:
We based our analysis on studies that verified COI disclosures in health research reports. We systematically searched the following electronic databases: MEDLINE, Embase, the World Health Organization (WHO) Global Health Library, and Scopus. We analyzed the eligible studies for the steps used to conduct the verification of COI, and the reported limitations and challenges. We used an iterative process to draft and refine models for two processes:
1) COI disclosure; and
2) verification of COI disclosure.
The model for each process includes the steps, and ‘what could go wrong’ for each step.

Results:
We included 35 eligible studies. The model for the process of COI disclosure by the researcher consists of the following three steps:
1) defining relevant COI according to the topic of the research;
2) considering the organization’s COI disclosure policy requirements; and
3) the researcher choosing what to include in the declaration.
The model for the process of verification of COI disclosure consists of the following five steps (with the first four to collect ‘identified COI’):
1) identify the appropriate verification source;
2) identify which COI are appropriately captured by the verification source;
3) determine which of these are relevant to the to the topic of the research;
4) determine which of these are required by the organization’s COI disclosure policy; and
5) compare the declared COI to identified COI.

Conclusion:
We provide two models respectively for the process of COI disclosure and the process of verification of COI disclosure. Editors, systematic reviewers, guideline organizations, and consumers of health research publications can use these models to understand better the risk of bias associated with COI, and to verify COI declarations when relevant.

Patient or healthcare consumer involvement:
No involvement.