Association of Blood Pressure with Suicide Attempt in First-Episode Drug-Naive MDD: Anxiety Mediates

Summary

This study found that blood pressure parameters were U‑/J‑shaped associations with suicide attempt risk in first-episode, drug-naive Major Depressive Disorder (MDD) patients, and anxiety played a partial mediating role.

Abstract

This study investigated the relationships between blood pressure parameters and suicide attempt risk in first-episode, drug-naive major depressive disorder (MDD) patients, and examined anxiety's mediating role. Blood pressure parameters were measured in 1,718 first-episode, drug-naive patients with MDD. Relationships with suicide attempt risk were analyzed using logistic regression with curve fitting and threshold effect analysis. Mediation analysis assessed anxiety's role. A U-shaped relationship existed between pulse pressure (PP) and suicide attempt risk (inflection point: 43 mmHg). Below this threshold, each 1 mmHg PP increase was associated with 6.5% lower risk (OR = 0.935, 95% CI: 0.897–0.974); above it, each 1 mmHg increase was associated with 7.1% higher risk (OR = 1.071, 95% CI: 1.036–1.107). Systolic (SBP) and diastolic blood pressure (DBP) showed J-shaped relationships with suicide risk (inflection points: 132 mmHg and 77 mmHg, respectively). No significant associations existed below these thresholds, while above them, each 1 mmHg increase in SBP and DBP was associated with 21.9% (OR = 1.219, 95% CI: 1.144–1.299) and 9.0% (OR = 1.090, 95% CI: 1.049–1.132) higher suicide attempt risk, respectively. Anxiety partially mediated the effects of PP (29.34%), SBP (31.89%), and DBP (35.75%) on suicide risk. Blood pressure parameters, particularly elevated SBP and DBP, significantly predict suicide attempt risk in MDD patients, with anxiety playing a partial mediating role. These findings highlight the importance of blood pressure monitoring and anxiety management in suicide prevention strategies for MDD patients.

Introduction

Major depressive disorder (MDD) is a prevalent and debilitating mental illness, affecting approximately 4.4% of the global population. As projected by the World Health Organization (WHO, 2008), MDD is anticipated to become a principal cause of disease burden by 2030. Individuals diagnosed with MDD frequently present with a constellation of symptoms, including persistent depressed mood, loss of interest in activities, sleep disturbances, appetite changes, and recurrent suicidal thoughts¹. The severity of MDD markedly elevates the risk of suicide, with studies indicating that individuals with MDD are markedly more likely to attempt suicide compared to the general population. In China, the prevalence of suicide attempts among MDD patients is strikingly high, with estimates ranging from 16% to 33.7%². The relationship between MDD and suicide is multifaceted, involving a complex interplay of biological, psychological, and social factors^3,4,5,6. While psychosocial factors⁷, such as hopelessness, anxiety, and substance abuse, have been extensively studied, emerging evidence suggests that physiological factors, particularly cardiovascular health, may also play a critical role in suicide risk.

Recent literature has increasingly focused on biological markers^8,9 including hypothalamic‑pituitary‑adrenal (HPA) axis activity, immune/inflammatory markers, and composite inflammatory indices, as potential predictors of suicidality in depression. For instance, elevated adrenocorticotropic hormone (ACTH) levels have been associated with lower suicidal ideation, while systemic inflammatory markers such as SII (systemic immune‑inflammation index) and SIRI (systemic inflammation response index) show predictive value for suicide risk in MDD patients^10,11,12.

Blood pressure (BP) parameters, including systolic blood pressure (SBP), diastolic blood pressure (DBP), and pulse pressure (PP, defined as SBP minus DBP), have also been recognized as potential biomarkers of suicidal ideation and behavior^13,14. However, the precise nature of these associations remains poorly understood.

The present study design—focusing on first‑episode, drug‑naive (FENT) MDD patients—offers distinct advantages over studies including medicated or chronically ill populations. Antidepressants, particularly tricyclics, can alter BP, and chronic disease progression may induce cumulative physiological changes that obscure true BP‑suicide relationships. By excluding these confounders, the analytical approach (including non‑linear threshold modeling and mediation analysis) provides clearer insight into the intrinsic associations between BP parameters and suicide risk.

Protocol

This study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of the First Hospital of Shanxi Medical University (approval number: 2016-Y27). All participants signed an informed consent form prior to participation in the study.

Methods
Study design and population
This retrospective cross‑sectional study was conducted at the psychiatric outpatient department of the First Hospital of Shanxi Medical University in China between 2015 and 2017. A total of 1,718 patients with first‑episode drug‑naive MDD were enrolled.
Inclusion criteria: (1) diagnosis of FENT‑MDD according to DSM‑IV criteria; (2) age 18‑60 years.
Exclusion criteria: (1) pregnancy or lactation; (2) other severe DSM‑IV Axis I disorders or serious physical illnesses; (3) substance use disorders (except nicotine); (4) inability or unwillingness to provide written informed consent.

Diagnostic confirmation
Diagnoses were established by board‑certified psychiatrists using the Structured Clinical Interview for DSM‑IV (SCID‑I). First‑episode status was defined as the first psychiatric contact for a depressive episode with no prior antidepressant or psychotropic medication exposure, verified by medical records and patient/family interviews.

Measurements
Blood pressure assessment
Blood pressure was measured following standardized procedures. Participants rested quietly for 5 min in a seated position before measurement. Three readings were taken at 2 min intervals, and the average was calculated. PP was computed as the difference between SBP and DBP.

Suicide attempt assessment
Lifetime suicide attempts were evaluated through structured face-to-face interviews using the standardized WHO/EURO Multicenter Study item asking whether participants had ever attempted suicide. For participants who reported attempts, detailed information on frequency, methods, and timing was collected. When necessary, family members were contacted to verify the information. A suicide attempt was defined as any self-directed injurious behavior with the intent to die.

Covariates
Covariates included age, sex, marital status, education level, body mass index (BMI), age of onset, illness duration, Hamilton Depression Rating Scale (HAMD‑17) score, Hamilton Anxiety Rating Scale (HAMA) score, Positive and Negative Syndrome Scale (PANSS‑P), Clinical Global Impressions‑Severity (CGI‑S) score, thyroid function (FT3, FT4, TSH, TgAb, TPOAb), lipid profile, and blood glucose.

Statistical analysis
Continuous variables were expressed as mean ± SD (normal) or median (IQR) (non‑normal). Categorical variables as frequencies (%). Normality was tested with the Shapiro-Wilk test. For three‑group comparisons (low, middle, high PP): a one-way ANOVA was used for normally distributed continuous variables, a Kruskal-Wallis test for non-normally distributed continuous variables, and a chi‑square test for categorical variables. Post hoc pairwise comparisons were corrected using Bonferroni adjustment.

The association between blood pressure parameters and suicide attempts was analyzed in three steps. (1) Logistic Regression Models. Three logistic regression models were constructed to evaluate the association between PP and suicide attempts: Model 1 was unadjusted univariate analysis; Model 2 was adjusted for demographic characteristics (age, sex, marital status, education level, and BMI); Model 3 was further adjusted for clinical features (e.g., age of onset, illness duration, HAMD-17 score, HAMA score) and laboratory indicators (e.g., thyroid function, lipid profile, blood glucose) listed in Table 1. This stepwise adjustment strategy was used to assess the robustness of the results. Odds ratios (ORs) with 95% confidence intervals (CIs) were reported. (2) Non-linear Relationship Assessment. Generalized additive models (GAM) with penalized splines were used to explore potential non-linear relationships between PP and suicide attempts. If a non-linear relationship was detected (P for non-linearity < 0.05), the following steps were performed: inflection points were identified using a recursive algorithm; piecewise linear regression models were constructed on either side of the inflection point; the log-likelihood ratio test was employed to ascertain the goodness-of-fit between the standard linear regression model and the piecewise linear model, identifying the most appropriate model for elucidating the association. (3) Categorical Analysis. To validate the findings from the continuous variable analysis and assess potential non-linear patterns, pulse pressure was classified into three categories: ≤39 mmHg, 40–45 mmHg, and ≥46 mmHg. The 40–45 mmHg group was employed as the reference group. The cutoff values for this classification were derived from previous literature¹⁴, and logistic regression was performed to evaluate the association between PP categories and suicide attempts.

Mediation analysis was conducted to investigate whether anxiety (HAMA score) mediated the relationship between PP and suicide attempts. The product of coefficients method was employed, which involved: 1. Estimating the direct effect of PP on suicide attempts. 2. Calculating the indirect effect of PP on suicide attempts mediated through anxiety. 3. Computing the total effect of PP on suicide attempts. 4. Determining the proportion of the total effect mediated by anxiety.

Bootstrap resampling (5000 iterations) was used to calculate 95% CIs for the indirect effects.
All statistical analyses were conducted using R software (R Foundation for Statistical Computing). Two-sided P-values < 0.05 were considered statistically significant.

Results

Demographic and clinical parameters
A total of 1,718 patients were stratified into three groups based on pulse pressure (PP): low PP group (≤39 mmHg, n = 456), middle PP group (40–45 mmHg, n = 598), and high PP group (≥46 mmHg, n = 664) (Table 1). Significant differences in demographic characteristics were observed across PP tertiles. The high PP group demonstrated significantly higher mean age (43.2 years ± 10.6 years vs. 31.7 years ± 10.9 years vs. 26.8 years ± 9.2 years, P

Discussion

In this large-scale study of 1,718 first-episode, drug-naive MDD patients, we investigated the association between blood pressure parameters and suicide attempts, with anxiety as a potential mediator. The analysis yielded a distinctive U-shaped relationship between PP and suicide risk, with an inflection point at 43 mmHg. Below this threshold, each 1 mmHg increase in pulse pressure was associated with a 6.5% decrease in suicide risk¹⁵, while above the threshold, each 1 mmHg increase corresponded t...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
caret R package	CRAN / Max Kuhn	Version 6.0-94	Used for stratified 7:3 development/hold-out validation split with the createDataPartition function. Official package page: https://cran.r-project.org/package=caret
glmnet R package	CRAN / glmnet authors	Version 4.1-8	Used for least absolute shrinkage and selection operator penalized logistic regression with binomial family and alpha = 1. Official package page: https://cran.r-project.org/package=glmnet
mice R package	CRAN / amices project	Version 3.16.0	Used for multiple imputation by chained equations with 20 imputed datasets and 20 iterations. Official package page: https://cran.r-project.org/package=mice
pROC R package	CRAN / pROC authors	Version 1.18.5	Used for receiver operating characteristic curve analysis, area under the curve estimation, confidence intervals, and DeLong comparisons. Official package page: https://cran.r-project.org/package=pROC
cobas 8000 modular analyzer series	Roche Diagnostics	System catalog no. 05641446001 / SYS_128	Automated clinical chemistry and immunochemistry analyzer series for biochemical assays and ion-selective measurements. Use only if this platform matches the hospital laboratory record; otherwise replace with the actual local analyzer. Official product page: https://diagnostics.roche.com/global/en/products/systems/cobas-8000-analyzer-series-sys-128.html
XN-1000 Automated Hematology Analyzer	Sysmex Corporation	Catalog number not publicly specified; product model XN-1000	Automated hematology analyzer for complete blood count-related variables such as white blood cell count and platelet count. Use only if this platform matches the hospital laboratory record; otherwise replace with the actual local analyzer. Official product page: https://www.sysmex.com/en-us/lab-solutions/hematology/xn-series/xn-1000
CS-5100 Automated Blood Coagulation Analyzer	Sysmex Corporation	Catalog number not publicly specified; product model CS-5100	Automated coagulation analyzer for coagulation variables such as prothrombin time, activated partial thromboplastin time, fibrinogen, and D-dimer. Use only if this platform matches the hospital laboratory record; otherwise replace with the actual local analyzer. Official product page: https://www.sysmex.com/en-us/lab-solutions/hemostasis/sysmex-cs-5100
ABL90 FLEX PLUS blood gas analyzer	Radiometer Medical ApS	Catalog number not publicly specified; product model ABL90 FLEX PLUS	Blood gas and acute-care analyzer; included for lactate measurement only if lactate was obtained through this or an equivalent blood gas platform in the hospital record. Otherwise replace with the actual local analyzer. Official product page: https://www.radiometer.my/en-my/products/blood-gas-testing/abl90-flex-plus-blood-gas-analyzer
Electronic medical record system	Participating hospitals	Institution-specific system; catalog number not applicable	Used to extract admission notes, progress notes, intensive care unit records, procedure records, vital-sign records, discharge records, and endpoint-adjudication source documents. Replace with the local vendor/system name if the manuscript names a commercial electronic medical record system.
Laboratory information system	Participating hospitals	Institution-specific system; catalog number not applicable	Used to extract laboratory timestamps, raw values, raw units, reporting units, and harmonized laboratory data. Replace with the local vendor/system name if the manuscript names a commercial laboratory information system.
Endpoint adjudication form	Study team	Custom study document; not applicable	Standardized form used by two trained reviewers and a third senior reviewer to adjudicate MODS status, organ system involved, onset time, source record, and supporting value.
Data dictionary and unit-conversion table	Study team	Custom study document; not applicable	Defines variable names, units, conversion rules, missingness flags, center code, cohort indicator, and analysis-ready variable formats.