Hemoglobin-to-red blood cell distribution width ratio is inversely associated with prevalent gastrointestinal cancers in NHANES and serves as an informative feature in machine-learning models.

Authors

Shiming Chen, Kexin Zhao, Jingxia Zhang, Yingxue Niu, Ran Zou, Min Wang, Yifang Li

Published in

BMC cancer. Volume 26. Issue 1. Jul 21, 2026. Epub Jul 21, 2026.

Abstract

The hemoglobin-to-red blood cell distribution width ratio (HRR) is an inexpensive index derived from routine blood tests, but its relationship with gastrointestinal (GI) cancers in the general population remains unclear.
We pooled NHANES 2001-2018 data (adults ≥ 20 years). GI cancer history was screened using MCQ220 and site-confirmed using site-specific follow-up items for colon, rectal, stomach, esophageal, liver, and pancreatic cancers. HRR was calculated as hemoglobin (g/dL) divided by RDW (%). Survey-weighted logistic regression estimated odds ratios (ORs) and 95% CIs for prevalent GI cancer across HRR (per 0.1-unit increase and tertiles), with restricted cubic splines and subgroup interaction tests. In parallel, eight machine-learning (ML) algorithms were trained and evaluated (70/30 split) to assess predictive performance (AUC, calibration, and decision-curve analysis), with SHAP for interpretability.
Among 39,663 participants, 344 had site-confirmed GI cancer. Higher HRR was inversely associated with GI cancers in the fully adjusted model (per 0.1-unit increase: OR 0.86, 95% CI 0.77-0.95; P = 0.003). Participants in the highest HRR tertile had lower odds than the lowest tertile (OR 0.57, 95% CI 0.35-0.93; P = 0.026; P for trend = 0.028). Sex modified the association (P for interaction = 0.002), with an inverse association in men (OR 0.74) but not women. ML models showed similar discrimination (top AUC ≈ 0.82), and SHAP highlighted HRR as an important contributor alongside age.
Higher HRR was inversely associated with prevalent, site-confirmed GI cancers in NHANES. In machine-learning analyses, HRR functioned as an informative classification feature.

PMID:
42920116
Bibliographic data and abstract were imported from PubMed on 24 Oct 2026.

Read full publication at:
Please sign in to see all details.

Stats

1-terrible, 9-excellent. How would you rate this publication? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.