Hemoglobin-to-Red Blood Cell Distribution Width Ratio and All-Cause Mortality in Older Adults: The Mediating Role of Cognitive Impairment — A Prospective Cohort Study from NHANES 2011–2014

Background: The hemoglobin-to-red blood cell distribution width ratio (HRR) is a novel integrative hematological parameter reflecting both oxygen-carrying capacity and erythrocyte volume heterogeneity. While HRR has been associated with adverse health outcomes, the causal pathways linking HRR to cognitive impairment and all-cause mortality remain poorly understood, and no study has quantified the mediating role of cognitive function in this pathway. We aimed to examine the association between HRR and all-cause mortality, test the mediating role of cognitive impairment, and identify the dominant cognitive domain pathway, thereby providing evidence for early risk stratification and targeted public health interventions in aging populations. Methods: We included 2,433 community-dwelling older adults (≥ 60 years) from the National Health and Nutrition Examination Survey (NHANES) 2011–2014, with mortality follow-up through the National Death Index (NDI) until December 31, 2019. HRR was calculated as hemoglobin (g/dL) divided by red blood cell distribution width (%) and categorized into quartiles (Q1: <0.93; Q2: 0.94–1.02; Q3: 1.03–1.16; Q4: >1.17). Cognitive function was assessed using the Digit Symbol Substitution Test (DSST), Animal Fluency Test (AFT), CERAD Immediate Recall and Delayed Recall Tests, with a global composite Z-score calculated. Cognitive impairment was defined as Z-score < − 1 standard deviation (SD). Multivariable linear regression was used for cognitive associations, Cox proportional hazards models for mortality, and causal mediation analysis under the counterfactual framework was employed to quantify mediation effects. Domain-specific mediation, competing risk models (Fine-Gray), and extensive sensitivity analyses (excluding early deaths, inverse probability of treatment weighting [IPTW], E-value assessment) were conducted. All analyses accounted for the NHANES complex multistage sampling design by incorporating survey weights, stratification, and clustering variables. Results: Compared with the lowest quartile (Q1), the highest HRR quartile (Q4) was significantly associated with lower all-cause mortality risk (fully adjusted Model 4: HR = 0.47, 95% CI: 0.36–0.63, P for trend < 0.001), with a significant dose-response relationship. HRR was positively associated with DSST (β = 8.62, 95% CI: 4.91–12.33, P < 0.001), AFT (β = 0.34, 95% CI: 0.07–0.62, P = 0.015), and global Z-score (β = 0.26, 95% CI: 0.01–0.52, P = 0.042) in fully adjusted models. Causal mediation analysis revealed that cognitive impairment significantly mediated the HRR-mortality association, with the indirect effect accounting for approximately 19%–21% (binary: 18.78%, 95% CI: 11.93%–27.46%; continuous Z-score: 21.25%, 95% CI: 15.14%–28.06%). Domain-specific analysis identified processing speed (DSST) as the dominant cognitive pathway (mediation proportion: 25.74%, P < 0.001), followed by global cognition (17.43%, P < 0.001) and executive function (5.84%, P = 0.036), while episodic memory showed negligible mediation (0.46%, P = 0.844). Competing risk models and sensitivity analyses confirmed the robustness of all findings (E-value = 3.42, lower CI bound = 2.76). Conclusions: Higher HRR was independently associated with reduced all-cause mortality risk and better cognitive performance in older adults. Cognitive impairment significantly mediated this association, with processing speed identified as the dominant cognitive pathway, suggesting that HRR influences mortality primarily through cerebrovascular-white matter mechanisms rather than hippocampal degeneration pathways. Given that HRR is routinely available from complete blood counts, it represents a cost-effective biomarker for community-based risk stratification and targeted public health interventions aimed at healthy aging.