Reporting of commercial involvement and the direction of stated conclusions in the neuromodulation-for-pain literature: a meta-research study of 896 publications

Background Efficacy in neurostimulation for chronic pain rests on patient-reported outcomes, which admit no external verification, so procedural transparency substitutes for it. The sponsorship effect is established for drugs and barely characterized for devices. Methods Meta-research study of 896 publications (1995–2025, 155 journals) on neurostimulation for pain, 712 of them primary studies, coded under a prespecified schema with a verbatim quotation per field. The primary exposure was industry sponsorship, analyzed where the funding source was determinable; authors9 declared ties were entered both separately and in the same model; a four-level hierarchy of commercial involvement was secondary. Undeterminable sources were never scored. Firth penalized logistic regression related exposure to a favorable stated conclusion, whose coding was re-applied independently and by two human coders. Results A disclosure statement was locatable for 80.6% of publications; 38.5% stated no funding source, and of 239 declaring industry funding, 69.5% described the sponsor9s role nowhere. Sponsorship was associated with a favorable conclusion (OR 4.03 (95% CI 2.42 to 6.85); n=333), persisting after adjustment for reporting quality (3.90) and among randomized trials (5.59 (95% CI 2.25 to 14.71)). With both exposures in one model, sponsorship gave 5.19 (95% CI 2.81 to 9.85) and authors’ ties 0.95 (95% CI 0.48 to 1.87). The hierarchy reached 7.36 (95% CI 2.68 to 23.44) where the sponsor had an active role. Sponsored studies were better reported, not worse (ρ=+0.23); outcome misclassification was non-differential. Conclusions The association lies with funding and the sponsor9s described role, not with the authors9 declared ties the disclosure system captures. This design does not establish causation.