Public Health Research Preprint Withdrawn

A Withdrawn Preprint Highlights a Persistent Problem in Public Health Research

A 2026 preprint on the link between diet, oxidative stress, and osteoarthritis was formally withdrawn before peer review. Authors Tao Liu, Qian Ma, and their colleagues pulled the manuscript from medRxiv due to a fundamental flaw: their osteoarthritis diagnosis was based solely on self-reported survey data. The retraction notice is an unusually detailed public lesson in scientific rigor. The team’s planned study on the Oxidative Balance Score and osteoarthritis could not proceed with the available data.

The Oxidative Balance Score Was the Planned Exposure

The central concept of the intended research was the Oxidative Balance Score. This is not a single supplement but a composite dietary and lifestyle score. Researchers calculate it by aggregating factors that influence the body’s oxidative state. Pro-oxidant elements, like saturated fat intake, receive negative scores. Antioxidant elements, such as intake of nutrients like beta-carotene, vitamins C and E, selenium, and flavonoids, plus lifestyle factors like physical activity, receive positive scores. A higher total OBS theoretically indicates a stronger antioxidant balance. The study aimed to see if people with a higher score had lower odds of osteoarthritis, proposing a direct link between redox status and joint health.

Self-Reported Data Introduced Unfixable Bias

The authors’ withdrawal explanation details why self-reported osteoarthritis is inadequate for this type of study. Participant confusion between osteoarthritis and other conditions like rheumatoid arthritis or gout was a primary concern. They noted this misreporting was more common among individuals with lower health literacy. In their dataset, 15.94% of the osteoarthritis group had a junior high school education or less, compared to 9.82% in the control group. This created a differential, non-random bias. The “osteoarthritis” group was likely a heterogeneous mix of true cases and people with other or no conditions, making any association with OBS unreliable. Furthermore, without imaging data, the study could not differentiate between knee, hip, or hand osteoarthritis, which have different biological mechanisms.

Withdrawal Stops a Flawed Chain of Analysis

The retraction was necessary because the flawed outcome definition corrupted the entire analytical pipeline. The initial epidemiological analysis of OBS and osteoarthritis in the NHANES dataset served as the foundation for subsequent steps. Planned transcriptomic analysis of RNA-seq data and Mendelian Randomization studies intending to assess causality were all based on the same poorly defined patient group. The authors concluded that “routine revision cannot address this fundamental flaw.” This decision prevents potentially misleading conclusions from entering the scientific literature and underscores a commitment to methodological integrity over publishing speed.

The incident emphasizes a critical point for longevity science: high-quality mechanistic research requires precise clinical phenotyping. For a field interested in biomarkers and multi-omics signatures, using noisy, real-world data without rigorous validation can lead to false leads. The authors’ commitment to restarting the project with a validated osteoarthritis definition is the correct, albeit more difficult, path forward.

Source:
WITHDRAWN: Oxidative Balance Score and Osteoarthritis: A Multi-Omics Study Based on NHANES, RNA-seq, and Mendelian Randomization (medRxiv preprint, 2026-05-13)