Development of a Practical Nomogram for Depression Risk Stratification in Older Adults With Hypertension and Diabetes: Retrospective Analysis of Data From the China Health and Retirement Longitudinal Study

TL;DR

A new tool helps doctors quickly assess depression risk in older adults with hypertension and diabetes using everyday health and lifestyle data.

Contribution

A practical nomogram was developed using readily available data to predict depression risk in older adults with hypertension and diabetes.

Findings

The nomogram includes nine predictors like daily living activity and social engagement, achieving strong accuracy in depression risk prediction.

Retirement status significantly lowers depression risk, likely due to increased social participation.

The model offers excellent discrimination (AUC=0.825) and is suitable for use in primary care settings.

Abstract

Depression affects over 40% of middle-aged and older Chinese adults living with both hypertension and diabetes, amplifying cardiovascular risk, functional decline, and mortality. Existing screening instruments—such as the 10-item Center for Epidemiologic Studies Depression Scale—focus narrowly on mood symptoms and are rarely feasible in busy primary care consultations. They also omit routine functional, cognitive, and social data that may jointly drive depressive states in cardiometabolic populations.

This study aimed to develop and validate a concise, clinically actionable nomogram that quantifies individual depression risk using readily available information in Chinese adults aged ≥45 years who have diagnosed hypertension and type 2 diabetes.

We analyzed anonymized wave 5 China Health and Retirement Longitudinal Study data collected between July 2020 and August 2020. Of 1504 eligible participants, 635 (42.2%) met the Center for Epidemiologic Studies Depression Scale cutoff score of >10 for probable depression. A total of 42 candidate predictors spanning demographics, laboratory values, comorbidities, functional status, and socioenvironmental factors were screened. Least absolute shrinkage and selection operator regression with 10-fold cross-validation identified the most parsimonious set. A multivariable logistic model was built on a 70% training set (n=1052) and evaluated on a 30% testing set (n=452). Performance was assessed using the area under the receiver operating characteristic curve (AUC), calibration plots, decision curve analysis, and Shapley additive explanations for interpretability. Multiple imputation was used to handle <20% missingness.

Nine nonredundant predictors entered the final nomogram: activity of daily living score, memory impairment, number of pain sites, sleep duration, life satisfaction score, self-rated health score, social activity engagement score, retirement status, and memory test score. The model achieved excellent discrimination (training AUC=0.819; testing AUC=0.825) and calibration (mean absolute error ≤0.018). Decision curves demonstrated positive net clinical benefit across clinically relevant threshold probabilities. Shapley additive explanations analysis revealed a 3-fold increase in depression odds per 1-point increase in activity of daily living score, whereas retirement conferred substantial protection (prevalence of depression: 103/635, 16.2% in the retired group vs 269/869, 31.0% in the nonretired group; P<.001), mediated by greater social participation.

The 9-item nomogram enables <3-minute depression risk stratification in resource-limited primary care settings for adults with hypertension and diabetes. Functional decline, affective-cognitive burden, and socioeconomic disengagement constitute the dominant causal pathway. Prospective trials should examine whether interventions targeting postretirement social engagement and functional rehabilitation can reduce incident depression in this high-risk population.

Figures