Dietary iron intake predicts all-cause and cardiovascular mortality in patients with diabetes

Diabetes mellitus is a growing global health concern, associated with elevated risks of all-cause and cardiovascular mortality. Iron is essential for energy metabolism, oxygen transport, and enzymatic processes, but both deficiency and excess are linked to disease. Prior research has connected higher iron intake—particularly heme iron from red meat—with increased type 2 diabetes risk, while non-heme iron may be inversely related in certain groups. Iron status may also affect risks of diabetic complications (retinopathy, kidney injury, neuropathy). However, the relationship between dietary iron intake and mortality in people with diabetes is not well established. This study investigates whether dietary iron intake is associated with all-cause and cause-specific mortality (CVD, cancer) among adults with diabetes, and whether the association is nonlinear or differs by sex.

The paper summarizes evidence indicating that elevated heme iron intake is positively associated with diabetes risk in multiple cohorts, whereas non-heme iron may be inversely associated in some populations (e.g., postmenopausal women). Iron status has been implicated in diabetic complications (retinopathy, kidney injury, neuropathy). Studies in general and high-risk populations have linked iron biomarkers (serum iron, ferritin, transferrin saturation) to mortality, including in prediabetes and in patients with diabetes and CVD. Evidence on iron intake and CVD or cancer mortality is mixed: some meta-analyses suggest heme iron relates to higher CHD incidence, and transferrin saturation inversely to CHD mortality, while other studies report contrary findings. Differences between dietary intake measures and serological markers may reflect absorption, transport, and storage variability. Potential sex differences in iron stores and intake patterns have been noted, with men generally having higher ferritin and intake than women, and possible changes across menopausal status.

Design and data source: Observational cohort analysis using NHANES cycles 1999–2014 with mortality follow-up through the National Death Index to December 31, 2015. Population: Adults with diabetes identified by self-report, use of hypoglycemic medications, or biochemical criteria (HbA1c > 6.5% or fasting plasma glucose > 7.0 mmol/L). Exclusions: age <18 years (n=109) and missing iron intake (n=26). Final sample: 5,970 participants. Dietary assessment: One 24-hour dietary recall using a standardized multi-pass method administered by trained interviewers via Computer-Assisted Personal Interview; nutrient totals (including iron and total energy) calculated. Dietary supplement data were cross-checked against the NCHS Product Label Database; total dietary iron intake quantified. Outcomes: Primary outcome was all-cause mortality; secondary outcomes were CVD mortality (ICD-10: I00–I09, I11, I13, I20–I51) and malignant neoplasm mortality (ICD-10: C00–C97), ascertained by linkage to the National Death Index. Covariates: Demographics (sex, age, race/ethnicity, education, poverty income ratio), lifestyle (physical activity: vigorous/moderate/inactive; alcohol; smoking: current/past/never), medical history (hypertension, CVD), medications (antihypertensives, hypoglycemics, lipid-lowering), and biomarkers (fasting glucose, HbA1c, triglycerides, LDL-C, serum iron, eGFR via CKD-EPI). Definitions: hypertension defined by history, BP ≥140/90 mm Hg, or antihypertensive use; CVD by self-reported CHF, CHD, angina, MI, and stroke. Missing data in Cox models were imputed using predictive mean matching. Statistical analysis: Iron intake categorized into quintiles: F1 <8.3 mg, F2 8.3–11.1 mg, F3 11.1–14.4 mg, F4 14.4–18.8 mg, F5 >18.8 mg. Descriptive comparisons across quintiles via ANOVA (continuous) and chi-square (categorical). Survival analyses included Kaplan–Meier and log-rank tests. Cox proportional hazards models estimated HRs and 95% CIs for all-cause, CVD, and cancer mortality with three adjustment models: Model 1 (age, gender); Model 2 adds race, education, PIR, BMI, smoking, alcohol, physical activity; Model 3 adds hypertension, CVD, eGFR, serum iron, and energy intake. Nonlinear associations were examined using restricted cubic splines, including sex-stratified analyses. Subgroup analysis assessed effect modification by baseline serum iron categories (<57, 57–74, 74–96, >96 mg/dL). Statistical significance set at P<0.05; analyses conducted in R 3.6. Ethics: NHANES IRB approval (Protocol #98-12); written informed consent obtained.

- Cohort characteristics: n=5,970 adults with diabetes; mean age 61.3 years; 51.3% male (n=3,059); mean dietary iron intake 14.1 ± 7.4 mg/day. - Follow-up and events: 41,425 person-years; 1,497 deaths. - All-cause mortality: After multivariable adjustment, iron intake 11.1–14.4 mg/day (F3) had the lowest all-cause mortality risk vs reference <8.3 mg/day (F1): HR 0.83 (95% CI 0.70–0.99), P<0.05. - Nonlinearity and sex differences: Restricted cubic spline analyses showed a nonlinear association overall (P=0.005) with an L-shaped pattern in men (P≈0.052) and a J-shaped pattern in women (P=0.016). Higher iron intake appeared protective in women at moderate levels but potentially harmful in men at higher intakes. - Cause-specific mortality: An apparent lower CVD mortality risk in the 14.4–18.8 mg/day group (HR 0.68, 95% CI 0.47–1.00) in less-adjusted models attenuated and lost significance after full multivariable adjustment. No significant associations were observed between iron intake and cancer mortality. - Subgroup by serum iron: No significant interaction between baseline serum iron and dietary iron intake on all-cause mortality (P for interaction = 0.296).

The study demonstrates a nonlinear relationship between dietary iron intake and all-cause mortality among adults with diabetes, with increased risk at both low and high intakes. Sex-specific patterns emerged: in men, the curve suggested increasing risk at higher intakes (L-shaped), whereas in women a J-shaped relationship indicated potential benefit at moderate intake but harm at both extremes. These findings align with literature indicating complex roles of iron in metabolic health. Mechanistically, iron excess may promote oxidative stress, impair insulin secretion, and contribute to cardiometabolic disease, while iron deficiency can compromise oxygen transport and cellular function. Mixed evidence regarding CVD and cancer mortality likely reflects differences between dietary measures and circulating iron biomarkers, and potential confounding by dietary sources (e.g., heme iron from red meat) and comorbidities. The absence of consistent associations with CVD and cancer mortality after full adjustment suggests that overall mortality risk may be more sensitive to total iron intake balance and underlying health status than to specific causes, or that residual confounding remains. Clinically, the findings suggest there may be an optimal intake range for individuals with diabetes—potentially around 11–14 mg/day—with consideration for sex-specific differences.

Dietary iron intake displays a nonlinear association with all-cause mortality in adults with diabetes, with an L-shaped pattern in men and a J-shaped pattern in women. A threshold near 12 mg/day appears to minimize risk. No robust associations were found with CVD or cancer mortality after full adjustment. These results highlight the need for cautious, potentially sex-specific recommendations for iron intake in diabetes and call for further research to elucidate mechanisms and to refine personalized dietary guidance.

- Iron status biomarkers: Key indicators of body iron stores (e.g., ferritin, transferrin, transferrin saturation) were not included, limiting inference about true iron stores versus intake. - Measurement error: Disease history and lifestyle factors were self-reported, introducing potential recall bias; 24-hour dietary recalls may not reflect usual intake. - Residual confounding and observational design: As a non-randomized observational analysis, unmeasured confounders may persist; causality cannot be established. - Dietary iron bioavailability: Differences in absorption and source (heme vs non-heme) were not fully accounted for, which may attenuate associations between intake and outcomes.