Albuminuria and the triglyceride–glucose (TyG) index represent different dimensions of cardiometabolic and kidney health. Albuminuria, measured by urinary albumin-to-creatinine ratio (UACR), signals kidney endothelial and filtration abnormalities, while TyG is an index derived from fasting triglyceride and glucose values used as a metabolic surrogate. The authors tested whether a comparatively lower TyG would attenuate the elevated mortality risk associated with albuminuria in a nationally representative US cohort.
The analysis used data from the National Health and Nutrition Examination Survey (NHANES) cycles 1999–2018. The analytic sample comprised 21,694 adults from the NHANES fasting subsample with survey-weighted analyses and linked mortality follow-up through 2019. UACR was classified at 30 mg/g (albuminuria-positive versus negative). TyG was dichotomized at its survey-weighted median value of 8.575 to create high versus low TyG groups. Cross-classification of UACR and TyG produced four phenotypes: concordant-low (UACR−/TyG−), metabolic-predominant (UACR−/TyG+), albuminuria-predominant (UACR+/TyG−), and concordant-high (UACR+/TyG+).
The primary outcome was cardiovascular mortality. The primary modeling approach was survey-weighted cause-specific Cox proportional hazards regression. Models included progressive covariate adjustment; the fully adjusted model is reported as Model 3. Supplementary analyses included competing-risk approaches, formal testing for UACR-by-TyG interaction, time-varying analyses, and multiple imputation for missing covariates. The authors applied fasting-subsample survey weights to produce population-representative estimates.
During a median follow-up of 9.25 years there were 997 cardiovascular deaths. Compared with the concordant-low phenotype, Model 3 hazard ratios for cardiovascular death were:
A formal UACR-by-TyG interaction test did not demonstrate statistical interaction (P = 0.468). The pattern indicates that excess cardiovascular mortality was concentrated in phenotypes with albuminuria, whereas isolated elevation of TyG (metabolic-predominant) was not independently associated with higher cardiovascular death after adjustment.
In a pre-specified lower-risk subgroup—participants without prior cardiovascular disease or diabetes and with estimated glomerular filtration rate (eGFR) ≥60 mL/min/1.73 m2—the albuminuria-predominant phenotype remained associated with higher cardiovascular mortality (hazard ratio 2.27; 95% CI, 1.24–4.15). The authors also provided a standardized 10-year absolute risk difference in that subgroup: 1.71 percentage points, comparing albuminuria-predominant with the concordant-low phenotype. Sensitivity analyses including competing-risk methods and multiple imputation were performed; the main pattern of findings persisted across these approaches.
Beyond cardiovascular death, albuminuria-positive phenotypes were associated with increased noncardiovascular mortality (hazard ratio 2.36; 95% CI, 1.77–3.16). This reinforces that albuminuria marked broader excess mortality in this population-representative cohort.
The study’s central observation is that albuminuria-positive phenotypes carried substantially higher cardiovascular and noncardiovascular mortality, regardless of a concurrent lower TyG index. In contrast, isolated TyG elevation (without albuminuria) was not associated with excess cardiovascular mortality after multivariable adjustment. The authors emphasize that the cross-classified, population-relative phenotype may help characterize heterogeneity in risk but does not constitute a fixed clinical threshold or a treatment rule.
Limitations inherent to the design include reliance on single-time UACR and TyG measures for phenotype assignment, potential residual confounding despite multivariable adjustment, and the observational, secondary-analysis nature of NHANES data. The report is a preprint and has not been peer-reviewed; the authors state it should not be used to guide clinical practice until validated.
NHANES protocols were reviewed and approved by the National Center for Health Statistics Ethics Review Board, and participants provided written informed consent. The present work is a secondary analysis of publicly available, deidentified data and linked mortality files; no additional IRB approval was required. The processed analysis dataset, code, data dictionary, survey-design specifications, and reproducibility materials are archived in Zenodo (doi:10.5281/zenodo.21858614) as reported by the authors.