Extended definition
Conflict of interest (COI) is the situation in which secondary interests — financial (paid consulting, equity, royalties, travel support), personal (close relationships with assessed competitors or collaborators), professional (pursuit of promotion, prestige, future funding), institutional (affiliation loyalties) — may unduly influence judgment about the primary interest, which in scientific research is methodological rigor and integrity of findings. Lo and Field (2009, Conflict of Interest in Medical Research, Education, and Practice, National Academies Press) consolidated the definition and governance framework that guides contemporary institutional policies. Drazen et al. (2010, NEJM) described the uniform ICMJE COI disclosure form, today adopted by virtually all top-tier health journals. Reporting a conflict does not eliminate it: transparency is the institutional defense, allowing readers to assess the work with full context. Absence of declaration when substantive COI exists is an ethical violation documented in COPE flowcharts.
When it applies
COI declaration applies to every manuscript submitted to journals adopting ICMJE — effectively all Q1/Q2 in health, and growing in other fields. It applies to grant proposals, peer review reports (reviewers declare COI to the editor before agreeing to review), participation on editorial boards, advisory committees, clinical guideline panels. It applies to top-tier conference presentations (declaration at the start of the talk). It applies to editorials and invited commentary — where editorial position must be free of financial bias. Typical window of relevance: three to five years before submission (varies by journal). Covers financial relations with sectors whose activity has non-trivial overlap with the work’s topic.
When it does not apply
It does not apply as an optional exception: absence of COI must be declared explicitly (“The authors declare no conflict of interest”), not merely omitted. It does not apply as a shield against criticism: declaring COI does not exempt the author from substantively addressing methodological critiques. It does not apply retroactively as justification for error: undeclared COI is a failure; subsequently declared in correction is partial mitigation. It does not cover all possible influences: cognitive, ideological, and theoretical biases are not typically classified as COI in the editorial sense. It does not replace independent statistical analysis: declaring industry support does not automatically validate the results of a clinical trial.
Applications by field
— Health and biomedical sciences: ICMJE form is standard; top-tier journals (NEJM, Lancet, BMJ, JAMA) enforce strictly; editorials and clinical guidelines have additional rules. — Social sciences: growing declaration in studies with specific corporate or governmental funding; APA and ASA have guidelines. — Engineering and computer science: strong growth in ML and AI with Big Tech funding; NeurIPS and similar introduced mandatory declarations. — Public policy research: declaration of think tank affiliations, government ties, advocacy group funding.
Common pitfalls
The first pitfall is interpreting “no industry funding” as absence of COI: personal, professional, and institutional ties are also declarable COIs. The second is declaring COI with vague language (“I received support from several companies”): ICMJE requires specification by entity, approximate value, and relation to the work’s topic. The third is confusing COI with misconduct: COI is a situation, not a failure — it can exist without affecting research, and transparent declaration is the correct practice. The fourth is assuming co-authorship with someone in COI contaminates the work: each author declares separately; integrity emerges from the set of declarations + methodological rigor. The fifth is failing to update declarations between submission and publication: changes (new employment, new consultancy) during the editorial process must be communicated to the editor.