# A clinically actionable nomogram integrating HbA1c, renal function, and blood pressure for early prediction of diabetic macular edema in working-age patients with type 2 diabetes

## TL;DR

This study creates a tool to predict diabetic macular edema in working-age type 2 diabetes patients using common blood tests and blood pressure.

## Contribution

A new nomogram integrating HbA1c, eGFR, and SBP for early DME prediction in working-age T2DM patients.

## Key findings

- DME prevalence was 15.71% in the study cohort of 490 patients.
- The nomogram showed strong predictive accuracy with an AUC of 0.905 in training and 0.884 in validation.
- Four key predictors of DME were identified: diabetes duration, SBP, eGFR, and HbA1c.

## Abstract

This study aims to identify factors associated with diabetic macular edema (DME) presence in working-age (18–60 years) patients with type 2 diabetes mellitus (T2DM) by developing a model that integrates HbA1c, renal function (eGFR), and hemodynamic parameters (SBP). The model addresses critical gaps in current screening strategies by using routinely available biomarkers, thereby enabling non-ophthalmologists to efficiently identify high-risk individuals.

This cross-sectional study prospectively collected data from 490 patients with type 2 diabetes mellitus (T2DM), aged 18–60 years, who were consecutively enrolled at a single medical center between January 2020 and March 2025. The participants were randomly allocated into two groups: a training cohort (n=343) and a validation cohort (n=147). Predictors were selected via LASSO regression with 10-fold cross-validation from an initial set of 19 variables, encompassing renal function, metabolic parameters, and hemodynamic indices. Subsequently, a multivariate logistic regression model was developed and illustrated through a nomogram. The model’s predictive accuracy was evaluated through receiver operating characteristic (ROC) curves (AUC), calibration curves, and decision curve analysis (DCA).

The overall prevalence of DME in the study cohort was 15.71% (77 of 490). Four predictors independently associated with DME were identified using LASSO regression, namely diabetes duration (OR = 1.460, 95% CI: 1.212–1.457), SBP (OR = 1.066, 95% CI: 1.037–1.095), eGFR (OR = 0.938, 95% CI: 0.916–0.961), and HbA1c (OR = 1.484, 95%CI: 1.189–1.852). The resulting nomogram exhibited robust discriminatory ability (training AUC = 0.905, 95% CI: 0.858–0.951; validation AUC = 0.884, 95% CI: 0.820–0.949) and strong calibration performance (Hosmer-Lemeshow test, P = 0.878). DCA further confirmed substantial clinical applicability within a threshold probability range from 2% to 100%, achieving a maximum net benefit of 0.14, thereby potentially preventing unnecessary intervention in 14 out of every 100 patients.

This nomogram effectively integrates HbA1c, renal function, and hemodynamic parameters to identify key factors associated with diabetic macular edema (DME) risk in working-age patients with type 2 diabetes mellitus (T2DM), demonstrating high accuracy. By utilizing routine clinical measures, it facilitates implementation in primary care settings, offering the potential to reduce vision loss through timely referrals. Future multicenter studies are warranted to verify its generalizability and explore its integration with emerging biomarkers.