Blood pressure phenotypes and kidney impairment in adults with diabetes: a machine learning study

Q: What is the clinical evidence from this study?

Study design: Cross-Sectional. Population: Diabetes mellitus and concurrent diabetic kidney disease (DKD) (n=1240). Intervention: Systolic-diastolic hypertension (SDH, SBP ≥140 and DBP ≥90 mmHg) and isolated systolic hypertension (ISH, SBP ≥140 and DBP <90 mmHg) vs. Normal blood pressure (NBP, SBP <140 and DBP <90 mmHg). Primary outcome: Concurrent diabetic kidney disease (DKD) phenotype (UACR ≥30 mg/g or eGFR <60 mL/min/1.73 m2) (OR 2.465, 95% CI 1.634-3.717, p=<0.001).

Q: What does this research mean for the field?

Isolated systolic hypertension and systolic-diastolic hypertension are significantly associated with increased odds of concurrent diabetic kidney disease in adults with diabetes. Novelty: ClaimNovelty.INCREMENTAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

Q: What question did this study set out to answer?

To assess associations between specific blood pressure phenotypes and concurrent diabetic kidney disease and evaluate machine learning models for risk discrimination.

Jialin Song; Juexuan Li; Huating Zhang; Junqiang Ding; Kai Wei; Xiangyi Ma; Bishan He; Rongyang Zhou; Hengbing Zhou; Qingyu Zhou; Tian Cao; Peiyan Dong; Biying Wu; Wencai Ke; Ding Yang; Jianming Zhou; Qinyu Deng; Yiming Sun; Yao Yao; Hongmei Zi; Lu Zheng; Jianhua Zhang; Qiudan Chen; Xiaotong Chen; Yong Lin; Shuying Chen

doi:10.3389/fragi.2026.1964420

Key result

Systolic-diastolic hypertension is linked to ~2.5-fold higher odds of concurrent diabetic kidney disease.

OR 2.465
95% CI 1.634-3.717
P<0.001
n=1,240

Why the study?

Although hypertension is a key modifiable risk factor for diabetic kidney disease (DKD), the associations between specific blood pressure phenotypes and concurrent DKD risk in adults with diabetes remain poorly understood.

Are specific blood pressure phenotypes (ISH, IDH, SDH) associated with concurrent diabetic kidney disease in adults with diabetes, and can machine learning models using routine clinical variables discriminate this concurrent risk pattern?

Key Points

To assess associations between specific blood pressure phenotypes and concurrent diabetic kidney disease and evaluate machine learning models for risk discrimination.
Cross-sectional analysis of 1,240 adults with diabetes from community health records in Shanghai, defining concurrent diabetic kidney disease as UACR ≥30 mg/g or eGFR <60 mL/min/1.73 m².
Blood pressure phenotypes (140/90 mmHg threshold) were categorized as normal, isolated systolic hypertension (ISH), isolated diastolic hypertension (IDH), and systolic-diastolic hypertension (SDH).
The dataset was split into training (n = 992) and independent test (n = 248) sets to train multivariable logistic regression and evaluate five machine learning models.
Concurrent diabetic kidney disease was identified in 331 of 1,240 participants (26.69%), occurring in 18.78% with normal blood pressure, 31.78% with ISH, 10.00% with IDH, and 32.20% with SDH.
Multivariable logistic regression demonstrated significantly higher odds of concurrent diabetic kidney disease for ISH (OR 1.80, 95% CI 1.31–2.48, P < 0.001) and SDH (OR 2.46, 95% CI 1.63–3.72, P < 0.001) compared with normal blood pressure.
On the test set, discrimination did not differ significantly between XGBoost (ROC-AUC 0.776, 95% CI 0.704–0.838) and logistic regression (ROC-AUC 0.771, 95% CI 0.699–0.832; paired difference 0.005, 95% CI -0.038 to 0.048, P = 0.812).

Study Design

Type

Cross-Sectional (n=1,240)

Multicenter

No

Structured PICO

Are specific blood pressure phenotypes (ISH, IDH, SDH) associated with concurrent diabetic kidney disease in adults with diabetes, and can machine learning models using routine clinical variables discriminate this concurrent risk pattern?

P

Population

1,240 older adults (mean age 71.17 years, 53.23% female) with documented diabetes from community health checkups in Shanghai, evaluated cross-sectionally from 2018 to 2022.

E

Exposure

Hypertension phenotypes (140/90 mmHg threshold): isolated systolic hypertension (ISH: SBP ≥140 and DBP <90 mmHg), isolated diastolic hypertension (IDH: SBP <140 and DBP ≥90 mmHg), and systolic-diastolic hypertension (SDH: SBP ≥140 and DBP ≥90 mmHg); and classification using five machine learning algorithms (L2-penalized logistic regression, random forest, XGBoost, radial-kernel support vector machine, and Gaussian naïve Bayes).

C

Comparator

Normal blood pressure (NBP: SBP <140 and DBP <90 mmHg) as the reference category in regression models; comparison among the five machine learning algorithms.

O

Outcome

Concurrent diabetic kidney disease (DKD) phenotype defined by single-visit UACR ≥30 mg/g or eGFR <60 mL/min/1.73 m² (calculated using the CKD-EPI 2009 creatinine equation).surrogate

Main Result

Odds Ratio: 2.465 (95% CI 1.634–3.717)

Absolute Event Rate: 32.2% vs 18.78%

p-value: p=<0.001

Isolated systolic hypertension and systolic-diastolic hypertension are strongly associated with concurrent diabetic kidney disease in community-dwelling adults with diabetes, and integrating these phenotypes with routine clinical variables offers moderate discrimination for opportunistic risk pattern recognition.

Limitations

Cross-sectional design cannot determine temporality or whether blood pressure abnormalities preceded kidney damage
Single-visit UACR and eGFR measurements do not establish chronicity or exclude non-diabetic kidney disease
Lack of medication data (e.g., antihypertensives, ACEi/ARBs, SGLT2 inhibitors) introduces unmeasured confounding and potential misclassification of blood pressure phenotypes
Unavailable clinical variables such as diabetes duration, smoking status, and cardiovascular disease history
Small sample size in the isolated diastolic hypertension (IDH) subgroup (n=20, 2 positive cases) resulting in wide confidence intervals
Low sensitivity of machine learning models (0.106-0.303) at the 0.50 probability threshold limits their utility as clinical screening tools
Lack of external validation in an independent population
Cross-sectional design using single-visit UACR and eGFR measurements, which does not establish chronicity or causal attribution of kidney abnormality to diabetes.
Source dataset lacked detailed laboratory logs to reconstruct formal diabetes diagnostic pathways or differentiate diabetes subtypes.
Lack of methodological details on blood pressure measurement (number of readings, device, cuff specification) and detailed antihypertensive-treatment status, meaning treated hypertension could be classified as a lower measured blood-pressure phenotype.
Small sample size in the isolated diastolic hypertension (IDH) subgroup (n = 20 total, 2 positive cases), resulting in extreme statistical imprecision.
Hold-out test set contained only 66 positive observations, contributing to wide confidence intervals for calibration and threshold-based metrics.

Ask about this paper

Answers are grounded in this paper and cite their sources.

Implication

May support opportunistic DKD recognition via BP phenotypes in diabetes; extends cross-sectional associations but leaves causal inference open.

Abstract

Background Diabetic kidney disease (DKD) is a leading cause of end-stage renal disease, with hypertension being a key modifiable risk factor. However, associations between specific blood pressure (BP) phenotypes and concurrent DKD risk remain poorly understood. Methods This cross-sectional study analyzed 1,240 adults with diabetes from community health records in Shanghai. Concurrent DKD was defined by UACR ≥30 mg/g or eGFR <60 mL/min/1.73 m 2 . BP phenotypes (140/90 mmHg threshold) were classified as normal (NBP), isolated systolic (ISH), isolated diastolic (IDH), and systolic-diastolic hypertension (SDH). Multivariable logistic regression (LR) estimated adjusted odds ratios (aORs). The cohort was split 80:20 into training (n = 992) and test (n = 248) sets. Five machine learning algorithms were evaluated using strict within-training-fold preprocessing. Results Concurrent DKD was present in 331/1,240 participants (26.69%). After adjustment, ISH and SDH were significantly associated with higher DKD odds versus NBP. On the independent test set, ROC-AUCs were 0.776 for XGBoost and 0.771 for LR, with no statistically significant discrimination difference. XGBoost showed a slightly lower Brier score, while LR offered superior calibration stability and nomogram interpretability. Conclusion ISH and SDH are associated with a higher prevalence of concurrent DKD in adults with diabetes. Routine clinical features and BP phenotypes provide moderate discrimination. These models function for opportunistic concurrent risk pattern recognition rather than predictive screening, supporting further external validation prior to clinical implementation.

Discussion

No takes yet. Share an insight, caveat, or question.

Cite This Study

Song et al. (2026) conducted a cross-sectional in Diabetes mellitus and concurrent diabetic kidney disease (DKD) (n=1,240). Systolic-diastolic hypertension (SDH, SBP ≥140 and DBP ≥90 mmHg) and isolated systolic hypertension (ISH, SBP ≥140 and DBP <90 mmHg) vs. Normal blood pressure (NBP, SBP <140 and DBP <90 mmHg) was evaluated on Concurrent diabetic kidney disease (DKD) phenotype (UACR ≥30 mg/g or eGFR <60 mL/min/1.73 m2) (OR 2.465, 95% CI 1.634-3.717, p=<0.001).

synapsesocial.com/papers/6af3463ebdf2424ebd04f85b https://doi.org/10.3389/fragi.2026.1964420

View Full Paper

Ask AI

Bookmark

Share

Blood pressure phenotypes and kidney impairment in adults with diabetes: a machine learning study

Key Points

Study Design

Structured PICO

Main Result

Limitations

Ask about this paper

Implication

Abstract

Discussion

Cite This Study

Blood pressure phenotypes and kidney impairment in adults with diabetes: a machine learning study

Key Points

Study Design

Structured PICO

Main Result

Limitations

Ask about this paper

Implication

Abstract

Discussion

Cite This Study

Keep reading

Keep reading