The central question was methodological. But eGFR captures only one dimension of kidney damage. The researchers fitted four statistical models on the same complete-case sample, adding adjustment for eGFR, log UACR, or both, and estimated prevalence ratios by marginal standardization.

The results were striking. Adjusting for log UACR attenuated the association between serum albumin and documented retinopathy by 73.3 percent, compared with 34.3 percent under eGFR adjustment—a difference of 38.9 percentage points, with a 95 percent confidence interval of 30.7 to 49.0. In other words, much of the apparent link between lower blood albumin and retinopathy was statistically absorbed by albuminuria rather than by filtration rate.

Not every modeling choice produced the same picture. When the researchers replaced continuous log UACR with a binary indicator of albuminuria, attenuation was substantially weaker, indicating that collapsing the ratio into a yes-or-no variable discards information that matters statistically. Sensitivity analyses using multiple imputation to handle missing data supported the comparison between the two renal adjustments, although after mutual adjustment for both eGFR and UACR, the confidence interval for the albumin association included the null, leaving some uncertainty about its independent contribution.

One additional analysis sharpened the distinction between the two kidney measures. When the sample was restricted to patients whose eGFR was 90 mL/min/1.73 m² or higher—people with apparently preserved filtration—serum albumin remained associated with documented retinopathy, with a prevalence ratio of 0.762 (95 percent CI 0.649 to 0.873). Within that same stratum, eGFR itself showed no clear association with retinopathy (prevalence ratio 0.981, P = 0.823).

This suggests that albuminuria captures kidney-related risk information that a normal filtration reading can miss. The authors are careful about what the attenuation figures mean. The percentages describe the proportional reduction in the absolute log prevalence-ratio estimate after adjustment, not a causal proportion of the association explained by kidney measures.

As a cross-sectional study of hospitalized patients, the analysis also cannot establish temporal ordering between the markers and retinopathy, and the outcome definition classified unassessed patients without a retinopathy diagnosis as having no documented disease, a limitation the researchers acknowledge in their design. Still, the practical implication for future research is concrete. Studies probing routine blood markers for associations with diabetic retinopathy may reach different conclusions depending on whether they adjust for eGFR, UACR, or both, and dichotomizing UACR can mask much of its statistical contribution.

The work, published in Frontiers in Endocrinology in the journal's clinical diabetes section, does not claim that any of the four blood markers causes or predicts retinopathy in a clinical screening sense. Rather, it documents how a single covariate decision reshapes the evidence base in a large single-center inpatient cohort.

Subject of Research: UACR adjustment attenuates associations of selected routine blood markers with documented diabetic retinopathy more than eGFR adjustment: a cross-sectional study in type 2 diabetes

Article Title: UACR adjustment attenuates associations of selected routine blood markers with documented diabetic retinopathy more than eGFR adjustment: a cross-sectional study in type 2 diabetes

Article References: Zhang, J., Xu, M., Xiong, B., Luo, B., & Pang, L. (2026). UACR adjustment attenuates associations of selected routine blood markers with documented diabetic retinopathy more than eGFR adjustment: a cross-sectional study in type 2 diabetes. Frontiers in Endocrinology, 17, 1983060. https://doi.org/10.3389/fendo.2026.1983060

Image Credits: AI Generated

DOI: 10.3389/fendo.2026.1983060

Keywords: UACR, adjustment, attenuates, associations, selected, routine, blood, markers, documented, diabetic, retinopathy, than