Red cell distribution width and 5-year mortality across older adult care settings: a pooled European cohort analysis with US validation

Background Red cell distribution width (RDW), generated automatically with every complete blood count, predicts mortality across multiple populations, but its independent prognostic value at a fixed cut-off across older care settings is unresolved. Methods We pooled individual-level data from four prospective European cohorts of older adults: community-dwelling (ECHA-PC-MUSA, n = 812), nursing home residents (SADEL, n = 813), long-lived families with area-matched controls (GEHA, n = 579), and hospitalized patients (REPORTAGE, n = 4,400); total n = 6,604, followed for up to 5 years. RDW coefficient of variation (CV) ≥ 15% was the primary exposure and all-cause mortality the primary outcome. Multilevel Cox regression with cohort as a random effect adjusted for age, sex, anaemia, albumin, neutrophil-to-lymphocyte ratio (NLR), number of chronic diseases and medications, and impaired activities of daily living (ADLs), with pre-specified stratification at age 85 years. External replication drew on 2,471 participants of the 2016 wave of the Health and Retirement Study (HRS) in the United States. Results In the European cohorts, mean age was 84.8 years (SD 9.2); 41.4% had RDW-CV ≥ 15%. Mortality rate at 5 years was 69.9%. The pooled adjusted HR for RDW-CV ≥ 15% was 1.58 (95% CI 1.48–1.68; p < 0.001), with low between-cohort heterogeneity (θ = 0.06); the association was concordant across three of four cohorts; it was externally replicated in HRS (HR .64, 95% CI 1.23–2.19; p < 0.001) and strongest in the oldest-old. Conclusions RDW-CV ≥ 15% independently predicts all-cause mortality during follow-up of up to five years across heterogeneous older-adult populations after adjustment for major geriatric prognostic factors, including anaemia, ADL, NLR, and albumin, suggesting that RDW captures a dimension of systemic dysregulation.