Establishment and Evaluation of a Risk Prediction Model for Abnormal Circadian Rhythm of Blood Pressure in Young Hypertensive Patients

Introduction

Hypertension is a significant independent risk factor for cardiovascular diseases (CVD), including cardiac death, coronary heart disease, heart failure, and stroke.¹ Blood pressure in the human body follows a circadian rhythm regulated by the nervous and endocrine systems. Typically, blood pressure exhibits two peaks, occurring between 6:00 to 10:00 a.m. and 4:00 to 6:00 p.m., with the lowest levels observed between 2:00 and 3:00 a.m., often decreasing by 10–20%.² These circadian variations can be monitored using 24-hour ambulatory blood pressure monitoring (ABPM). Based on nocturnal blood pressure decline, blood pressure rhythms are classified into dipper, non-dipper, anti-dipper, and super-dipper patterns. Dipper blood pressure represents the normal physiological pattern, whereas non-dipper, super-dipper, and anti-dipper patterns are associated with target organ damage and an increased risk of cardiovascular and cerebrovascular mortality.³

Previous studies have highlighted the critical role of inflammatory factors in the development of hypertension. Peripheral blood cell counts and derived indicators that assess inflammatory status have been widely employed in clinical research on chronic inflammation. Examples include the systemic immune-inflammatory index (SII), systemic inflammatory response index (SIRI), and lymphocyte-monocyte ratio (LMR). These indicators encompass different components of blood cells, reflecting various inflammatory pathways and mechanisms of immune regulation. As reliable, cost-effective, and easily obtainable markers of inflammation, their levels are closely linked to a range of cardiovascular and cerebrovascular diseases.⁴ Inflammatory state plays a key role in the pathogenesis of non-dipper blood pressure, and higher levels of inflammation may lead to impaired nighttime blood pressure drop by promoting endothelial dysfunction and sympathetic nervous system overactivity.⁵ Akyüz et al’s study showed that white blood cell counts and neutrophil levels were significantly higher in patients with non-dipper hypertension, highlighting the critical role of inflammation in mediating adverse outcomes associated with non-dipper hypertension.⁶

In recent years, there has been a concerning trend of hypertension manifesting at a younger age. A national survey conducted between 2012 and 2015 revealed a prevalence of 10.1% among individuals aged 18–35, a figure that continues to rise.⁷ Young hypertensive patients have a significantly increased lifetime cardiovascular risk due to prolonged exposure to hypertension.⁸ Early identification and treatment are key to reducing future cardiovascular risk, and strengthening blood pressure reduction early can reduce lifelong cardiovascular mortality.⁹

Consequently, early detection, prevention, and effective management of hypertension are crucial for safeguarding cardiovascular health. Predictive models are becoming increasingly pivotal in clinical decision-making, particularly for assessing early disease risks. Unfortunately, there are currently limited predictive tools available in clinical practice for assessing the risk of young hypertensive patients, especially for nomogram that have not been validated specifically for young hypertensive patients. Therefore, this study aimed to identify independent risk factors for abnormal circadian blood pressure rhythms in young patients with hypertension and to develop a predictive model. Such a model could facilitate the early identification of high-risk individuals and provide essential clinical guidance for the intervention and treatment of young hypertensive patients.¹⁰

Materials and Methods

Study Participants

This study included 327 young hypertensive patients who received treatment and evaluation at the 901st Hospital of the Chinese People’s Liberation Army Joint Logistics Support Force from January 2023 to June 2024 through consecutive sampling. After excluding patients with incomplete personal information, a total of 211 study subjects were ultimately included. Additionally, 203 hypertensive patients admitted to the First Affiliated Hospital of the University of Science and Technology of China during the same period were recruited for external validation. Based on the model area under the curve (AUC) value and predictor variables, and following the improved Events Per (EPV) criterion (EPV ≥ 7). To accurately estimate the confidence interval of AUC (target with ≤ 0.10) and ensure statistical performance (detecting AUC decrease ≥ 0.05, α=0.05, β=0.2). Hypertension in this study was defined according to the Guidelines for the Prevention and Treatment of Hypertension in China (Revised Edition 2018). Inclusion criteria were as follows: (1) Patients with a history of hypertension or currently using antihypertensive medications, with blood pressure consistently below 140/90 mmHg; (2) Diagnosis of hypertension based on ambulatory blood pressure monitoring (ABPM): average systolic/diastolic blood pressure ≥ 130/80 mmHg; daytime ≥ 135/85 mmHg; nighttime ≥ 120/70 mmHg; (3) Young hypertensive patients, Age range 15–35 years. Exclusion criteria were: (1) Secondary hypertension; (2) Systemic inflammatory diseases; (3) Malignant tumors; (4) Obstructive sleep apnea syndrome; (5) Severe impaired consciousness and/or mental symptoms precluding cooperation; (6) Recent use of glucocorticoids or immunosuppressants; (7) Active tumors, acute and chronic inflammation, blood system, and immune system diseases; (8) Severe liver and kidney failure.

All participants underwent 24-hour ABPM. Individuals showing a nighttime blood pressure decrease of 10%–20% compared to daytime levels were classified into the dipper group, while those without such a decrease were classified into the non-dipper group.

Research Methods and Grouping

General Information

Clinical demographic information was collected via a survey questionnaire, which included data on gender, age, height, weight, nocturnal sleep duration, and history of smoking and drinking.

Blood Pressure Measurement

Blood pressure was assessed according to WHO standards. Patients rested in a seated position for 30 minutes, avoiding any disruption to the measurement site and refraining from alcohol or stimulating beverages. Blood pressure was measured using the Korotkoff auscultatory method, with three measurements taken at 2-minute intervals. The average of the three readings was recorded as the final result. For the detection of dynamic blood pressure, the same brand and model of equipment are used.

24-Hour Ambulatory Blood Pressure Monitoring

ABPM was programmed to automatically measure blood pressure every 30 minutes during the day (8:00–22:00) and every hour at night (22:00–8:00). Patients were instructed to remain in a quiet state during measurements. Data not meeting predefined criteria—such as systolic blood pressure > 260 mmHg or < 60 mmHg, and diastolic blood pressure > 160 mmHg or < 40 mmHg—were excluded from the analysis. Participants wore the device for 24 hours, with data retention required to exceed 90% for inclusion; otherwise, measurements were repeated. The following blood pressure parameters were collected: 24-hour average systolic blood pressure (24h SBP), 24-hour average diastolic blood pressure (24h DBP), daytime average systolic blood pressure (DSBP), daytime average diastolic blood pressure (DDBP), nighttime average systolic blood pressure (NSBP), nighttime average diastolic blood pressure (NDBP), 24-hour systolic blood pressure load, 24-hour diastolic blood pressure load, 24-hour systolic blood pressure variability, 24-hour diastolic blood pressure variability, daytime systolic blood pressure load, daytime diastolic blood pressure load, daytime systolic blood pressure variability, daytime diastolic blood pressure variability, nighttime systolic blood pressure load, nighttime diastolic blood pressure load, nighttime systolic blood pressure variability, nighttime diastolic blood pressure variability, morning peak systolic blood pressure, and morning peak diastolic blood pressure. The nighttime blood pressure dip rate was calculated as follows: Nighttime blood pressure dip rate % = (average daytime blood pressure - average nighttime blood pressure) / average daytime blood pressure × 100%. This calculation typically refers to systolic blood pressure.

Laboratory Testing

For the detection of peripheral blood indicators, we used Ethylene Diamine Tetraacetic Acid (EDTA) anticoagulant tubes of the same specifications from the same manufacturer, using the same testing reagents and standard range. Fasting overnight for 6–8 hours, peripheral venous blood was collected in the morning. A fully automated biochemical analyzer was used to measure blood routine parameters, including fasting blood glucose (FBG), total cholesterol (TC), triglycerides (TG), high-density lipoprotein (HDL), low-density lipoprotein (LDL), creatinine (Cr), glomerular filtration rate (GFR), urea (UREA), uric acid (UA), aspartate aminotransferase (AST), and alanine aminotransferase (ALT).

Inflammatory indices were calculated as follows: Systemic Inflammatory Response Index (SIRI) = absolute neutrophil count (N) × absolute monocyte count (M) / absolute lymphocyte count (L); Systemic Immune-Inflammatory Index (SII) = peripheral platelet count (PLT) × absolute neutrophil count (N) / absolute lymphocyte count (L); Lymphocyte-to-Monocyte Ratio (LMR) = L / M; Neutrophil-to-Lymphocyte Ratio (NLR) = N / L; Platelet-to-Lymphocyte Ratio (PLR) = P / L; Neutrophil-to-Monocyte Ratio (NMR) = N / M.

Statistical Analyses

Statistical analyses were conducted using SPSS 25.0 and R software (version 4.2.1), incorporating the rms, Hmisc, pROC, and ggplot2 packages. Data conforming to normal distribution were presented as mean ± standard deviation (SD), and comparisons between groups were assessed using independent sample t-tests. Non-normally distributed data were expressed as median (interquartile range, IQR), and non-parametric tests were applied. Categorical data were presented as frequencies and analyzed using the Chi-square test. The binary logistic model was constructed using the glm function using R software, and the Nomogram correlation model was constructed and visualized using the rms package. For the inclusion criteria of variables in multivariate regression analysis, we included factors that were meaningful in univariate regression analysis, and combined with previous studies, included variables with strong clinical significance. Statistical significance was set at P < 0.05.

Results

Comparison of General Data Between the Two Groups of Hypertensive Patients

The study included a total of 211 patients, with 93 in the dipper group and 118 in the non-dipper group. A comparison between the groups revealed significant differences in TG, HDL, LDL, Cr, UA, AST, and ALT levels (P < 0.05). Specifically, the non-dipper group showed higher levels of TG, LDL, Cr, and UA, and lower levels of HDL, AST, and ALT compared to the dipper group. No significant differences were observed in gender, age, smoking history, drinking history, nighttime sleep duration, BMI, FBG, TC, GFR, or UREA between the groups (Table 1).

Table 1 Comparison of General Data Between the Two Groups

Comparison of Inflammatory Indicators Between the Two Groups of Hypertensive Patients

A comparison between the two groups revealed significant differences in LMR, SIRI, and SII (P < 0.05). The non-dipper group exhibited higher SIRI and SII levels and lower LMR levels compared to the dipper group. However, no significant differences were found in WBC, N, NLR, PLR, or NMR between the groups (Table 2).

Table 2 Comparison of Inflammatory Indicators Between the Two Groups

Comparison of Blood Pressure Levels and Rhythms Between the Two Groups of Hypertensive Patients

The nocturnal drop rate of blood pressure differed significantly between the dipper and non-dipper groups (P < 0.05). The non-dipper group exhibited a significantly lower nocturnal drop rate compared to the dipper group. No significant differences were observed in other blood pressure indicators (Table 3).

Table 3 Comparison of Blood Pressure Levels and Rhythms Between the Two Groups

Univariate and Multivariate Logistic Regression Analysis of Abnormal Circadian Rhythm of Blood Pressure in Young Hypertensive Patients

Univariate Logistic regression analysis revealed significant differences in LMR, SIRI, SII, TG, HDL, LDL, and UA between the dipper and non-dipper groups (P < 0.05). Abnormal circadian rhythm of blood pressure in young hypertensive patients was set as the dependent variable (non-dipper group = 1, dipper group = 0), while significant variables from the univariate analysis were included as independent variables in the multivariate logistic regression model (Table 4). The results identified decreased LMR, increased SIRI, elevated TG, and elevated UA as independent risk factors for abnormal circadian rhythm of blood pressure in hypertensive patients.

Table 4 Logistic Regression Analysis of Influencing Factors of Abnormal Circadian Rhythm of Blood Pressure

Nomogram Prediction Model for the Risk of Abnormal Circadian Rhythm of Blood Pressure in Hypertensive Patients

A nomogram was developed based on the four independent risk factors identified through multivariate logistic regression analysis, as shown in Figure 1. Each factor corresponds to a score (Points) on the scale. The total score (Total Points) is calculated by summing these individual scores, corresponding to a specific value on the risk axis. This value represents the risk of abnormal circadian rhythm of blood pressure in hypertensive patients.

Figure 1 Risk nomogram for abnormal circadian rhythm of blood pressure.

Evaluation of Risk Nomogram

The discrimination ability of the model using Harrell concordance index analysis method. Similarly, the ROC curve analysis showed an AUC of 0.883, indicating high discrimination ability and accuracy. Calibration of the nomogram was evaluated using calibration curves, where the X-axis represents the predicted risk of abnormal circadian rhythm, and the Y-axis represents the actual occurrence. The Apparent curve reflects prediction performance, the Bias-corrected curve shows calibration accuracy, and the Ideal curve represents a perfect prediction. The model closely approximated the ideal prediction, demonstrating strong consistency with theoretical expectations (Figures 2 and 3).

Figure 2 Area under the ROC curve of the risk nomogram.

Figure 3 Calibration curve of risk nomogram.

External Validation of the Risk Prediction Model for Abnormal Circadian Rhythm of Blood Pressure in Hypertensive Patients

For external validation, 203 patients were included, with 99 in the dipper group and 104 in the non-dipper group. The ROC curve analysis yielded an AUC of 0.844, indicating strong discriminatory ability. Additionally, the calibration curve showed close alignment between predicted and actual values, highlighting the model’s high accuracy in predicting abnormal circadian rhythms of blood pressure in hypertensive patients (Figures 4 and 5).

Figure 4 Area under the ROC curve of the external validation risk nomogram.

Figure 5 Calibration curve of the external validation risk nomogram.

Discussion

Hypertension, a prevalent chronic condition, is strongly associated with an increased risks of heart attack, stroke, and damage to vital organs. Its incidence has been steadily rising, particularly among younger individuals. Effective management of hypertension is essential to prevent target organ damage and reduce the likelihood of adverse cardiovascular events. The circadian rhythm of blood pressure is recognized as an independent risk factor in the progression of hypertension. Abnormal blood pressure patterns throughout the day significantly heighten the risk of target organ damage, as well as cardiovascular and cerebrovascular complications.¹¹ According to the 2011 guidelines from the British Hypertension Society, except for patients with stage 3 hypertension, antihypertensive treatment decisions should be guided by dynamic blood pressure monitoring results.

Prior studies have demonstrated that hypertensive patients with abnormal circadian blood pressure rhythms experience imbalances between sympathetic and vagal nerve activity.¹² In particular, nocturnal non-dipping blood pressure pattern is closely associated with an increased risk of cardiovascular, cerebrovascular, and renal complications.^13,14 While the precise mechanisms underlying non-dipping blood pressure patterns remain unclear, research suggests links to impaired autonomic nervous system function, heightened sympathetic nerve activity, and elevated levels of inflammatory markers.

The pathophysiology of hypertension involves endothelial damage, inflammatory activation, insulin resistance, and platelet activation. Hypertension is not merely characterized by elevated blood pressure but is a complex cardiovascular disorder. Increased blood pressure correlates with higher levels of systemic inflammation, and prolonged inflammatory responses play a critical role in the progression of hypertension. Circulating inflammatory factors are strongly associated with the severity of hypertension and the extent of target organ damage.

Neutrophils, the most abundant white blood cells, are pivotal in responding to inflammation. They release numerous cytokines in response to inflammatory stimuli, such as bacterial infection, cancer, or environmental exposure, thereby initiating and amplifying the inflammatory response. Neutrophils also promote the secretion of inflammatory mediators, including arachidonic acid, myeloperoxidase, superoxide radicals, interleukin-1β, and matrix metalloproteinases.¹⁵ Additionally, they activate the renin-angiotensin-aldosterone system and reduce levels of coenzyme II,¹⁶ contributing to endothelial dysfunction, arteriosclerosis, and an increased risk of hypertension and arrhythmias. Monocytes play a central role in vascular inflammation and tissue remodeling by secreting chemokines and cytokines, generating reactive oxygen species (ROS), expressing coagulation factors, and transforming into macrophages.¹⁷ Lymphocytopenia, a condition linked to blood pressure disorders, is associated with the activation of adaptive immunity and stress responses. Inflammatory responses may induce apoptosis of regulatory lymphocytes, inhibit the secretion of anti-inflammatory factors, and disrupt the balance between anti-inflammatory and pro-inflammatory systems.¹⁸ These processes affect vascular homeostasis and promote the development of hypertension. The inflammatory response progressively disrupts the circadian rhythm of blood pressure, increasing sympathetic nervous system activity during nighttime. This disruption exacerbates the release of inflammatory factors and causes further vascular endothelial damage. The interplay of these factors ultimately contributes to the progression of hypertension and increases the risk of target organ damage.

SIRI and LMR are emerging inflammatory markers derived from multiple blood cell components. In this study, SIRI and LMR were demonstrated to be independent risk factors for circadian rhythm abnormalities in young hypertensive patients. Evidence increasingly highlights the critical role of innate and adaptive immune responses in the inflammatory pathways of hypertension. Activated immune cells infiltrate target tissues, causing organ damage by excessively secreting cytokines and chemokines. Monocytes, which are integral to both SIRI and LMR, play a key role in hypertension pathogenesis. LMR, as a novel inflammatory parameter, has been widely studied in cardiovascular diseases. For example, in a small sample single center retrospective study, it was found that LMR is a novel marker of blood pressure variability, connected to target-organ damage, in children with primary and secondary renal hypertension.¹⁹ In addition, a retrospective study by Aky ü z, A et al found that SII was an independent predictor of non-dipper hypertension.⁶ The high SII value in hypertension patients can be used as an early warning parameter to identify non-dipper hypertension patients. Unlike our study, our research indicates that SII has certain predictive value in the single factor analysis stage, but further multivariate analysis has no significant significance. We speculate that besides queue differences, its collinearity with SIRI may also be a potential reason. Monocyte levels are positively correlated with all-cause mortality and cardiovascular disease risk, underscoring their clinical significance.²⁰ Animal studies further support this, demonstrating the influence of inflammatory markers on hypertension outcomes.^21,22 For instance, Wenzel et al showed that monocyte depletion reduced blood pressure and improved vascular function in mice.²³ Similarly, mice deficient in tumor necrosis factor α (TNFα) or interleukin-6 (IL-6) exhibited lower blood pressure.²⁴ These findings suggest that SIRI and LMR, as comprehensive inflammatory markers, have predictive value comparable to traditional markers.

We also observed significantly higher triglyceride levels in the non-dipper group compared to the dipper group, identifying triglycerides as an independent risk factor for abnormal circadian blood pressure rhythm. This finding suggests that disrupted circadian rhythms and nocturnal sympathetic nerve excitation contribute to elevated nighttime blood pressure and abnormal lipid metabolism. Previous studies have similarly linked elevated LDL levels to the progression of non-dipper hypertension. Hyperlipidemia can stimulate angiotensin-2 receptors, promoting hypertension development.²⁵ Additionally, hyperlipidemia impairs vascular endothelial function and disrupts pressure sensory reflex pathways, further contributing to elevated blood pressure.²⁶ Thus, in addition to managing blood pressure actively, it is crucial to address blood lipid metabolism in hypertensive patients. Prospective studies have established a strong relationship between blood lipids and the development of hypertension. Elevated blood lipid levels independently predict the future risk of hypertension, even in apparently healthy individuals, preceding its onset.^27,28

This study found that UA levels were significantly higher in the non-dipper group than in the dipper group, identifying UA as an independent predictor of abnormal circadian blood pressure rhythms in hypertensive patients. Supporting this, a study involving adolescents with hypertension reported that nearly 90% of patients had elevated serum UA levels, a condition rare in healthy individuals.²⁹ Animal studies have shown that elevated UA induces hypertension through mechanisms such as stimulating oxidative stress, impairing endothelial function, and activating the renin-angiotensin system.³⁰ Elevated UA can also lead to renal microvascular disease (arteriosclerosis), contributing to salt-sensitive hypertension that may persist even after UA reduction.³¹

The subjects of this study were young hypertensive patients aged 18–35, who were significantly characterized by a low incidence of age-related comorbidities. In addition, this population has a single type of antihypertensive medication, a low rate of combination therapy, and minimal interference from non-cardiovascular drugs. The young cohort is not a limitation, but a scientific choice to enhance the intrinsic effectiveness of research. In addition, we found that gender, age, BMI, smoking history, and drinking history had no significant effect on the circadian rhythm of blood pressure. This may be attributed to the narrow age range of the participants and the fact that most subjects were young individuals without underlying diseases.

Nomograms are widely used in predictive models as visual tools to represent the relationships between variables. They graphically display the results of logistic regression analysis, enabling intuitive assessment of disease risk and prognosis based on statistically significant factors. In this retrospective study of 211 hypertensive patients from our hospital, a nomogram was developed to predict abnormal circadian blood pressure rhythms. Scores were assigned to LMR, SIRI, TG, and UA, with the total score reflecting the overall risk level. The predictive ability, consistency, and clinical utility of the model were evaluated and found to be favorable, offering valuable clinical insights. External validation, conducted using data from 203 hypertensive patients from other hospitals, confirmed the model’s reliability and robustness.

Previous studies have developed predictive models for spoon type hypertension based on inflammatory indicators, such as pan-immune-inflammation value (PIV)(AUC=0.725), uric acid/albumin ratio (UAR) (AUC=0.783 95% CI: 0.743–0.822).^32,33 The model established in this study has higher predictive accuracy compared to previous studies (AUC=0.883 (95% CI: 0.835–0.931)), has been validated in an independent third-party cohort (AUC=0.844 (95% CI: 0.791–0.896)), making it more generalizable.

The inflammation integration model developed in this study provides a new tool for precise management of hypertension in young people. In the future, it is necessary to verify its generalization in multi center queues and explore the impact of genetic environmental interactions on predictive performance. More importantly, prospective intervention trials will determine whether individualized treatment based on this model can improve prognosis. If the verification is successful, the guideline committee may consider incorporating SIRI and other factors into the risk assessment process for non-dipper hypertension, to supplement the potential application value of inflammation assessment.

It cannot be denied that although we have reduced the impact of some diseases and comorbidities on the study through exclusion criteria, the factors that may cause inflammation and cardiovascular changes are complex and varied, and there may still be potential factors that could lead to model bias. In this study, our univariate analysis results showed that factors such as LDL, HDL, AST, ALT were significant in univariate analysis but not in multivariate analysis, and there may be a risk of overfitting or collinearity.

Conclusions

In conclusion, decreased LMR, increased SIRI, elevated TG, and higher UA levels were identified as independent risk factors for abnormal circadian blood pressure rhythms in young hypertensive patients. This study successfully developed a nomogram-based risk prediction model incorporating LMR, SIRI, TG, and UA as key indicators. The model is simple, rapid to apply, clinically accurate, and holds significant relevance for practical applications in hypertension management.