SOS: BIO: improving the integrity of harms reporting in biomedical research: a science-of-science framework for measurement, causality, and intervention

The National Institute of General Medical Sciences awarded Oklahoma State University Center for Health Sciences $219,557 on August 14, 2026, under the Biomedical Research and Research Training program (CFDA 93.859) to develop and validate metrics for assessing the integrity and rhetorical framing of adverse event reporting in biomedical research.

The project will create two novel indices: the Harms Reporting Integrity Index, measuring completeness, concordance, transparency, and timeliness of adverse event disclosures across trial registries and publications, and the Framing Risk Index, capturing rhetorical strategies that minimize reported harms. The recipient will link these indices to downstream regulatory and scientific outcomes—FDA labeling changes, guideline incorporation, citation dynamics, and workforce trajectories—using causal inference methods. The project will pilot pragmatic editorial interventions including discrepancy flagging tools, harms checklists, and framing prompts to improve reporting practices. Work is performed in Tulsa, Oklahoma, with a period of performance through April 30, 2030. The investigative team includes expertise in meta-research, machine learning and natural language processing, causal modeling, and clinical specialties, supported by preliminary data from 16 registry-publication concordance studies and qualitative analyses of framing practices.

Abstract

(see instructions): the biomedical evidence ecosystem depends on accurate and transparent reporting of harms from clinical trials. Yet adverse events are often incompletely reported, inconsistently framed, or contradicted between trial registries and publications. These deficiencies compromise systematic reviews, misinform regulatory and clinical decision-making, and erode trust in science. This project will create the first validated, enterprise-level metrics to assess both the integrity and rhetorical framing of harms reporting. Using these metrics, we will map trends across specialties and test their consequences for regulatory action, guideline uptake, and scientific influence. Finally, we will pilot pragmatic editorial interventions to improve reporting practices. Intellectual merit. The project advances the science of science by moving from descriptive audits to reproducible, quantitative, and testable frameworks. Aim 1 develops two novel indices: the harms reporting integrity (HRI) index, measuring completeness, concordance, transparency, and timeliness, and the framing risk index (FRI), capturing rhetorical strategies that minimize harms. Aim 2 links these indices to downstream enterprise outcomes (FDA labeling, guideline incorporation, citation dynamics, and workforce trajectories) using causal inference. Aim 3 pilots editorial tools such as discrepancy flagging, harms checklists, and framing prompts. The project benefits from a multi-investigator team with expertise in meta-research (vassar, alicia ford), machine learning and NLP (gurupur), causal modeling (timothy ford), and clinical specialties. Feasibility is demonstrated by preliminary data from 16 registry-publication concordance studies and qualitative analyses of framing practices.