Related Experiment Video
Updated: Jan 7, 2026

Hydra, a Computer-Based Platform for Aiding Clinicians in Cardiovascular Analysis and Diagnosis
Published on: September 26, 2018
Hypertension Phenotypes and Mortality Risk in the United States of America: A Data-Driven Cluster Analysis
Rodrigo M Carrillo-Larco1,2, Jithin Sam Varghese1,2, Arshed Quyyumi3
1Hubert Department of Global Health, Rollins School of Public Health, Emory University, Atlanta, USA, emory.edu.
Insights
This study identified four distinct hypertension phenotypes in adults. Glucose-related hypertension showed the highest mortality risk, highlighting the need for personalized hypertension management.
Area of Science:
- Cardiovascular Medicine
- Clinical Epidemiology
- Biostatistics
Background:
- Hypertension is a major global cause of mortality, with current treatments lacking personalization.
- Existing blood pressure classifications do not capture the full biological heterogeneity of hypertension.
- Variations in treatment outcomes suggest underlying differences in hypertension subtypes.
Purpose of the Study:
- To identify data-driven clinical phenotypes of primary hypertension.
- To examine the associations between these identified phenotypes and mortality risk.
Main Methods:
- Pooled analysis of 10 cross-sectional surveys (NHANES 1999-2020) including 4084 adults (≥30 years) with newly diagnosed or undiagnosed hypertension.
- K-means clustering of eight variables (age, BMI, blood pressure, cholesterol, HbA1c, eGFR) to define phenotypes.
- Mortality risk (all-cause and cardiovascular) assessed using data linked to the National Death Index.
Main Results:
- Four distinct phenotypes were identified: Early-onset hypertension (EOH), Late-onset hypertension (LOH), Glucose-related hypertension (GRH), and Lipid-related hypertension (LRH).
- GRH (4.5% of cohort) exhibited the highest all-cause and cardiovascular mortality risk compared to EOH.
- LOH and LRH did not show significantly different mortality risks compared to EOH.
Conclusions:
- Data-driven cluster analysis successfully identified four hypertension phenotypes with differential mortality risks.
- These findings underscore the biological heterogeneity of primary hypertension and its impact on patient outcomes.
- The identified phenotypes may inform more personalized approaches to hypertension diagnosis and management.
Background:
Hypertension is a leading, yet modifiable, cause of mortality worldwide. While current treatment guidelines apply uniformly, variation in outcomes suggests unrecognized biological heterogeneity. Existing classifications based solely on systolic and diastolic blood pressure fail to capture this complexity. We identified data-driven clinical phenotypes of primary hypertension and examined their associations with mortality.
Methods:
Pooled analysis of 10 cross-sectional surveys (NHANES 1999-2020). Data from 4084 adults (≥ 30 years) with newly diagnosed or undiagnosed hypertension were collected. Hypertension was defined by self-report (in the last 2 years) or those with undiagnosed high systolic or diastolic blood pressure (≥ 140/90 mmHg). Predictors: age, body mass index, systolic blood pressure, diastolic blood pressure, total cholesterol, high-density lipoprotein cholesterol (HDL-c), hemoglobin A1c, and estimated glomerular filtration rate (eGFR). We used these variables because they are readily available in primary care settings, enabling clinical translation of these findings. We used k-means clustering of eight variables to identify phenotypes. Using mortality data linked to the National Death Index, we estimated the risk of all-cause and cardiovascular mortality.
Results:
Four phenotypes: Early-onset hypertension (EOH), late-onset hypertension (LOH), glucose-related hypertension (GRH), and lipid-related hypertension (LRH). EOH (37.3%) consisted of younger adults with high BMI and diastolic blood pressure, and low systolic blood pressure and HDL-c. LOH (32.6%) consisted of older adults with low diastolic blood pressure, total cholesterol, and eGFR. GRH (4.5%) consisted of adults with high BMI and HbA1c. LRH (25.6%) consisted of adults with high systolic blood pressure, total cholesterol, and HDL-c and low BMI and HbA1c. Compared to EOH, mortality was the highest in GRH (all-cause: 3.45 [1.80-6.61]; cardiovascular: 5.40 [2.18-13.37]), yet not significant for LOH (1.18 [0.74-1.87]; 1.04 [0.49-2.21]) and LRH (1.01 [0.62-1.63]; 0.93 [0.46-1.87]).
Conclusions:
This data-driven cluster analysis identified four phenotypes with different mortality risks in people with newly diagnosed hypertension.
Related Concept Videos
Hypertension III: Clinical Manifestations and Diagnostic Studies
Hypertension I: Introduction
Hypertension II: Pathophysiology
Hypertension and Regulation of Blood Pressure
Analysis of Population Pharmacokinetic Data
Hypertension V: Nursing Management
