The association between high-sensitivity C-reactive protein and depression symptom severity among U.S. adults: A multivariable analysis of NHANES data

Abstract

Background

Systemic inflammation may play a role in depression. This study examined the association between high-sensitivity C-reactive protein (hs-CRP), a well-established biomarker of systemic inflammation, and the severity of depression symptoms as measured by the Patient Health Questionnaire (PHQ-9).

Methods

Data from the National Health and Nutrition Examination Survey (NHANES) 2015-2023 were utilized. Depression was classified as low (PHQ-9 < 10) and high (PHQ-9 ≥ 10). hs-CRP levels were dichotomized into low (≤3 mg/L) and high (>3 mg/L). Logistic regression models were used to analyze the association between hs-CRP and depression symptom severity, adjusting for demographic, socioeconomic, health-related covariates, while considering the NHANES complex sampling design.

Results

Among 15,525 U.S. adults, elevated hs-CRP levels were significantly associated with high depression symptom severity. hs-CRP > 3 mg/L was associated with 26% greater odds of high depression symptom severity (adjusted OR = 1.26; 95% CI: 1.04-1.51; p-value = 0.018). Other factors associated with greater odds of high depression symptom severity included female sex, younger age (20-39 years), low-income status, unmarried status, tobacco use, multimorbidity, and timing of survey responses in relation to the onset of the COVID-19 pandemic. Exploratory depression symptom-domain analysis showed that hs-CRP was significantly associated with somatic symptom-domain (p-value <0.001) while not significantly associated with cognitive/affective symptom-domain (p-value = 0.056).

Conclusion

Depression symptom severity is influenced by many demographic, socioeconomic, behavioral, and health-related factors. The study findings support an independent association between elevated hs-CRP and depression symptom severity, and the significance of the association could depend on the specific depression symptom-domain.

Highlights

• •Systemic inflammation may influence the presentation of depression symptoms.
• •Elevated High-Sensitivity C-Reactive Protein (hs-CRP) levels are significantly associated with moderate to severe depression symptoms in U.S. adults, even after adjusting for confounders.
• •Findings support the inflammation-depression hypothesis.

1. Introduction

Depression is a prevalent and debilitating mental health disorder associated with impaired cognitive and physical functioning, chronic recurrence, medical comorbidities, and increased risk of self-harm and suicide (Hasin et al., 2018; CDC, 2023; Lee and Giuliani, 2019; Bains and Abdijadid, 2023; Wang et al., 2025). Although treatments for depression exist, including psychotherapy, brain stimulation, and pharmacological treatments, 20-50% of patients do not respond adequately (Akil et al., 2017; Qiao et al., 2018). These overlapping comorbidities and limited treatment response have highlighted the need to identify the biological mechanisms contributing to the development and persistence of depression symptoms.

Growing evidence suggests that systemic inflammation may play a role in the pathophysiology of depression (Maes et al., 2009; Peruga et al., 2011; Malhi and Mann, 2018; Lee and Giuliani, 2019; Lin et al., 2025). Animal and human studies suggest that elevated inflammatory markers, such as C-reactive protein (CRP) and proinflammatory cytokines, may influence depression symptoms by altering neurotransmitter signaling, neuroendocrine activity, and brain structure (Dantzer et al., 2008; Capuron and Miller, 2011; Raison and Miller, 2011; Kiecolt-Glaser et al., 2015; Thomas and Apovian, 2017; Varatharaj and Galea, 2017; Tabatabaeizadeh et al., 2018; Osimo et al., 2020; Salvador et al., 2021; Suneson et al., 2023). However, the directionality of this relationship remains debated, with ongoing research seeking to clarify whether inflammation is a cause, consequence, or merely a correlate of depression symptoms.

Several studies using the National Health and Nutrition Examination Survey (NHANES) data have examined the association between inflammatory markers and depression. Ford and Erlinger (2004) found that elevated levels of CRP was associated with major depression in men, but not women. More recent studies using high-sensitivity CRP (hs-CRP) and depression symptom scales such as the Patient Health Questionnaire-9 (PHQ-9), have similarly demonstrated positive associations between inflammation and depressive symptoms (Case and Stewart, 2014; Jung and Kang, 2019). However, findings remain variable across different subgroups. For instance, Ji et al. (2024) reported that the association between hs-CRP and depressive symptoms weakened in subgroup analyses, suggesting variability among demographic and clinical populations. Similarly, Zhang et al. (2025) highlighted the role of metabolic factors, demonstrating interactions between body mass index (BMI), hs-CRP, and depression symptoms. In contrast, Rengasamy et al. (2022) examined mechanistic pathways and demonstrated that physical illness and in turn, peripheral inflammatory markers (CRP and white blood cell count), may mediate the relationship between inflammation and depression symptoms.

Furthermore, depression is increasingly recognized as a multifaceted disorder characterized by different clusters of symptoms, including somatic symptoms (e.g., fatigue, sleep disturbance, appetite changes, psychomotor disturbances) and cognitive-affective symptoms (e.g., hopelessness, guilt, depressed mood, and suicidal ideation). Emerging evidence suggests inflammatory processes may be more strongly associated with somatic manifestations of depression than with cognitive-affective symptoms, supporting the hypothesis that inflammation contributes to specific depression symptoms (Fried et al., 2020; Milaneschi et al., 2020; Frank et al., 2021).

Building on this literature, the present study seeks to further assess the association between hs-CRP and depression symptom severity among U.S. adults by using recent NHANES data (2015-2023). Besides overall depression symptom, as measured by PHQ-9, depression symptom-domain level associations will also be investigated. By incorporating more recent data spanning the COVID-19 period and multimorbidity, this study aims to clarify whether systemic inflammation independently increases the likelihood of increased depression symptom severity and better characterize populations at elevated risk.

2. Methods

This study utilized data from NHANES across three recent data cycles: 2015-2016, 2017-March 2020, and 2021-2023. NHANES is a research initiative conducted by the National Center for Health Statistics (NCHS) that is designed to assess the health and nutritional status of adults and children across the U.S. by collecting data through interviews, physical examinations, and laboratory data. For more detailed information on the NHANES data collection process, refer to the NHANES Survey Methods and Analytic Guidelines (CDC, n.d.-c). To ensure that the data accurately represents the entire U.S. population, NHANES purposefully oversamples individuals from underrepresented groups (e.g., older adults, African Americans, low-income individuals) to ensure that key subpopulations are accurately represented. Sample weights are then assigned to each study participant and indicate the number of people in the U.S. population represented by each sample person. See NHANES Sample Design Module for more details (CDC, n.d.-b).

Due to the suspension of NHANES field operations in March 2020 caused by the COVID-19 pandemic, the data collection for the 2019-2020 cycle was truncated. As a result, data collected from 2019 to March 2020 was merged with the 2017-2018 data cycle to form a dataset covering a 3-year period, while ensuring national representativeness. See NHANES All Continuous NHANES for more details (CDC, n.d.-a).

2.1. Study sample and participants

Mobile Examination Center (MEC) examination sample weights were used for all analyses as the analytic sample included examination and laboratory measures. Following NHANES analytic guidelines, the sample weight corresponding to the most restrictive analytic component was selected. Specifically, WTMEC2YR were used for the 2015-2016 and 2021-2023 cycles, and WTMECPRP were used for the 2017-March 2020 cycle. To account for the unequal lengths of the NHANES cycles, pooled MEC weights were constructed following NHANES analytic guidelines. The 2015-2016 and 2021-2023 two-year weights were multiplied by 2/7.2, while the 2017-March 2020 wt were multiplied by 3.2/7.2, reflecting the total 7.2-year analytic period. The weighted cycles were then merged into a single dataset. All analyses incorporated the pooled MEC weights.

NHANES does not collect certain health variables (e.g., data on arthritis, stroke, and chronic respiratory illnesses) from participants under 20 years of age. As chronic health conditions were relevant to the present analysis, this younger age group was excluded from the study. Individuals missing responses to any of the nine questions obtained from the Mental Health – Depression Screener Questionnaire, or missing hs-CRP values were also excluded. To avoid confounding from acute infection, participants with hs-CRP values > 10 mg/L were removed, as this threshold typically reflects acute rather than chronic, low-grade systemic inflammation (Pearson et al., 2003; Shanahan et al., 2014). After applying all exclusion criteria, the final analytic sample included 15,525 participants, reduced from an initial sample of 37,464 (Fig. 1).

Fig. 1.

Study flowchart.

2.2. Key variables

The variables analyzed in this study include responses to the Mental Health – Depression Screener Questionnaire, laboratory results (hs-CRP levels), demographic characteristics, health characteristics, and timing of survey responses in relation to the onset of the COVID-19 pandemic.

2.2.1. Depression symptoms severity

The primary outcome was depression symptom severity (DSS), which was assessed using data from the NHANES Mental Health – Depression Screener Questionnaire. This questionnaire directly corresponds to the PHQ-9, a validated and widely used self-report tool that aligns with the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) diagnostic criteria for depression. The NHANES variables (DPQ010-DPQ090) represent the nine PHQ-9 items and are identical in content and wording, each assessing the frequency of depressive symptoms over the past two weeks on a 4-point scale ranging from 0 (“not at all”) to 3 (“nearly every day”). The responses are summed to yield a total score ranging from 0 to 27, with higher scores indicating greater depression symptom severity. For this study, the total scores were dichotomized. Scores <10 represented minimal to mild symptoms, while scores ≥10 represented moderate to severe symptoms. This threshold has been found to offer 88% sensitivity and 88% specificity in identifying depression requiring treatment, with a Cronbach's alpha ranging from 0.86 to 0.89 (Kroenke et al., 2001).

Depression symptoms were further divided into somatic and cognitive/affective symptom-domains (Rathore et al., 2014; Grapp et al., 2019). Somatic symptoms included sleep disturbance, fatigue, appetite disturbances, and psychomotor changes, while cognitive/affective symptoms include anhedonia, depressed mood, low self-esteem, difficulties in concentration, and suicidal ideation. Composite scores for each symptom-domain were calculated by summing the corresponding PHQ-9 item responses. Somatic symptom scores were dichotomized as low (<5) versus high (≥5), while cognitive/affective symptom scores were dichotomized as low (<6) versus high (≥6).

2.2.2. hs-CRP levels

hs-CRP measurements were collected from laboratory records and reported as continuous variables; however, for the purposes of this study, these values were classified into two groups: low (≤3 mg/L) and high (>3 mg/L). hs-CRP >3 mg/L has been the established cut-off for low-grade inflammation according to the CDC and American Heart Association (AHA) (Pearson et al., 2003) and has been used as the cut-off in several related studies (Jung and Kang, 2019; Osimo et al., 2020).

2.2.3. Demographic characteristics

Demographic characteristics included sex, age at the time of the survey, race/ethnicity, highest level of education completed, marital status, and income status. Age was grouped as follows: young adulthood (20-39 years), middle adulthood (40-64 years), and older adulthood (65 years and above). Highest level of educational attainment was categorized into high school graduate or less, some college or Associate of Arts (AA) degree, and college graduate or above. Marital status was categorized into married/living with partner, divorced/separated/widowed, and never married. Income status was based on the family income-to-poverty ratio (FIPR) and was grouped into low-income (FIPR ≤1.85), middle-income (FIPR 1.86-4.99), high-income (FIPR ≥5.00), and unreported.

2.2.4. Health-related variables

The health-related variables assessed in this study included BMI, tobacco use, and the number of reported chronic health conditions: angina, arthritis, asthma, coronary artery disease (CAD), congestive heart failure (CHF), chronic obstructive pulmonary disease (COPD), emphysema, bronchitis, diabetes mellitus, hyperlipidemia, hypertension, kidney disease, liver disease, cancer, myocardial infarction, stroke, and thyroid disease.

BMI was classified into healthy weight (<25.0), overweight (25.0-29.9), and obese (≥30.0). Tobacco use was recorded as a binary variable (yes/no), with participants considered as current users if they reported using cigarettes or any form of nicotine within the preceding five days.

Prior research has demonstrated that higher multimorbidity is associated with increased depressive symptoms and reduced quality of life (Birk et al., 2019; Triolo et al., 2020). Therefore, chronic health conditions were consolidated into a single variable and grouped as follows: none, one to two, or three or more chronic health conditions. This approach reflects overall disease burden and aligns with our focus on multimorbidity and its potential association with DSS, rather than examining the individual effects of comorbidities.

2.2.5. Timing of survey response

Lastly, a variable representing survey timing relative to the onset of the COVID-19 pandemic was included to account for potential long-term physical and psychological effects of the COVID-19 pandemic (Mayo Clinic, 2024). Participants were categorized as either pre-pandemic or post-pandemic onset based on whether they completed the survey before or after the onset of the COVID-19 pandemic in the U.S.

2.3. Statistical analysis

All data management and analyses were conducted using R (v4.3.2; R Core Team, 2023). Analyses proceeded in five stages. First, descriptive statistics were generated to summarize the overall study population, with unweighted sample counts (n) and weighted proportions (%) reported for all variables. Second, bivariate analyses were conducted to examine unadjusted differences in each predictor variable with respect to DSS. Chi-squared tests with Rao and Scott's second-order correction were used to account for the NHANES complex survey design. Third, univariable logistic regression models were employed to estimate the unadjusted association between each predictor variable and DSS. Results were reported as crude odds ratios (ORs) with 95% confidence intervals (CIs). Fourth, multivariable logistic regression models were implemented to evaluate the association between depression symptoms severity and hs-CRP levels, while adjusting for potential confounders (i.e., sex, age, race/ethnicity, education level, marital status, family income-to-poverty ratio, tobacco use, BMI, number of chronic health conditions, and timing of survey response in relation to the onset of the COVID-19 pandemic). Finally, separate survey-weighted multivariable logistic regression models were conducted for somatic and cognitive/affective symptom-domain severity to explore whether hs-CRP levels demonstrated different associations with different depression symptom-domains. As BMI and multimorbidity may function as mediators as well as confounders, we also conducted sensitivity analyses by excluding either BMI or the chronic disease variable from the multivariable model to assess their impact on the association between hs-CRP and DSS.

Given that the percentage of missing data was small (<1%) for all variables, an all-completer analysis was conducted by removing participants with missing values. Multicollinearity was assessed using variance inflation factors (VIFs). Influential observations were evaluated using Cook's distance, which measures the extent to which individual cases disproportionately affect model estimates. Overall model fit was evaluated using the design-adjusted Rao-Scott Likelihood Ratio Test. The full model's ability to distinguish between individuals with high depressive symptom severity from those with low depressive symptom severity was assessed by the area under the curve (AUC) using a Receiver Operating Characteristic curve. All multivariable logistic regression models adjusted for sex, age, race/ethnicity, education level, marital status, family income-to-poverty ratio, tobacco use, BMI, and timing of survey response in relation to the onset of the COVID-19 pandemic. Results were reported as adjusted ORs with 95% CIs and statistical significance defined as p-value<0.05.

3. Results

3.1. Univariate analysis

The study included 15,525 participants, of whom 8.6% reported high DSS and 28.1% presented elevated hs-CRP levels (>3 mg/L). The sample was distributed evenly by sex (50.0% male, 50.0% female). Middle-aged adults (40-64 years) comprised the largest age group (43.3%), followed by young adults aged 20-39 years (35.3%), and older adults aged ≥65 years (21.3%). Most participants identified as Non-Hispanic White (64.7%), while the rest of the sample identified as Hispanic origin (15.5%), Non-Hispanic Black (10.0%), Non-Hispanic Asian (5.4%), and Other/Multiracial (4.4%). 35.4% of participants reported an educational attainment of high school graduate or less, followed by 33.6% reporting college graduate or above, and 31.0% reporting some college/AA degree. Most participants were married or living with a partner (62.7%). 37.9% of participants were classified as middle-income and 25.7% as low-income. 23.2% of participants reported current tobacco use. 39.4% were classified as obese and 33.0% as overweight. 34.7% reported no chronic health conditions, 38.5% reported having 1-2, and 26.8% had three or more chronic health conditions. Most responses were obtained before the onset of the COVID-19 pandemic (72.7%) (Table 1).

Table 1.

Characteristics comparison between participants with low vs. high DSS.

Variables	Totala (n)	Depression Symptom Severityb		p-valuec
		Low (PHQ-9 Score <10) (n)	High (PHQ-9 Score ≥10) (n)
hs-CRP Levels (mg/L)
Low (≤3 mg/L)	10,921 (71.9%)	10,031 (92.4%)	890 (7.6%)	<0.001
High (>3 mg/L)	4604 (28.1%)	4059 (88.8%)	545 (11.2%)
Sex
Male	7636 (50.0%)	7058 (93.1%)	578 (6.9%)	<0.001
Female	7889 (50.0%)	7032 (89.7%)	857 (10.3%)
Age
Young Adulthood (20-39 yrs)	4607 (35.3%)	4131 (89.9%)	476 (10.1%)	<0.001
Middle Adulthood (40-64 yrs)	6702 (43.3%)	6059 (91.7%)	643 (8.3%)
Late Adulthood (≥65 yrs)	4216 (21.3%)	3900 (93.3%)	316 (6.7%)
Race/Ethnicity
Non-Hispanic White	6770 (64.7%)	6113 (91.7%)	657 (8.3%)	<0.001
Non-Hispanic Black	2952 (10.0%)	2692 (90.8%)	260 (9.2%)
Hispanic Origin	3575 (15.5%)	3234 (90.9%)	341 (9.1%)
Non-Hispanic Asian	1467 (5.4%)	1403 (95.6%)	64 (4.4%)
Other Race/Multi-Racial	761 (4.4%)	648 (85.3%)	113 (14.7%)
Highest Level of Education
High School Graduate or Less	6059 (35.4%)	5378 (89.5%)	681 (10.5%)	<0.001
Some College or AAd degree	4863 (31.0%)	4354 (90.2%)	509 (9.8%)
College Graduate or Above	4597 (33.6%)	4354 (94.6%)	243 (5.4%)
Marital Status
Married or Living w/Partner	9032 (62.7%)	8432 (93.7%)	600 (6.3%)	<0.001
Divorced/Separated/Widowed	3508 (18.4%)	3072 (88.0%)	436 (12.0%)
Never Married	2974 (19.0%)	2577 (87.4%)	397 (12.6%)
Income Status based on FIPRe
Low (FIPR ≤1.85)	5373 (25.7%)	4654 (85.9%)	719 (14.1%)	<0.001
Middle (FIPR 1.86-4.99)	5398 (37.9%)	4992 (92.4%)	406 (7.6%)
High (FIPR ≥5)	2942 (26.1%)	2811 (95.5%)	131 (4.5%)
Unreported	1812 (10.3%)	1633 (91.3%)	179 (8.7%)
Tobacco Use
No	11,944 (76.8%)	11,084 (93.4%)	860 (6.6%)	<0.001
Yes	3560 (23.2%)	2987 (84.8%)	573 (15.2%)
Body Mass Index
Healthy Weight (<25.0)	4126 (27.6%)	3786 (91.3%)	340 (8.7%)	<0.001
Overweight (25.0-29.9)	5133 (33.0%)	4724 (93.1%)	409 (6.9%)
Obese (≥30.0)	6139 (39.4%)	5480 (90.2%)	659 (9.8%)
Number of Chronic Health Conditions
0	4828 (34.7%)	4529 (93.6%)	299 (6.4%)	<0.001
1-2	5804 (38.5%)	5316 (92.1%)	488 (7.9%)
≥3	4893 (26.8%)	4245 (87.7%)	648 (12.3%)
Timing of Survey Response
Pre-Pandemic	11,028 (72.7%)	10,119 (92.5%)	909 (7.5%)	<0.001
Post-Pandemic	4497 (27.3%)	3971 (88.6%)	526 (11.4%)

^aColumn unweighted count (weighted %). Due to missing values, not all counts sum up to a total of 15,525. Percentages were calculated based on available data for each variable.

^bRow unweighted count (weighted %).

^cp-value based on chi-squared test with Rao & Scott's second-order correction.

^dAA = Associate of Arts.

^eFIPR = Family Income to Poverty Ratio.

3.2. Bivariate analysis

Bivariate analyses revealed variations in DSS across demographic, socioeconomic, and health-related factors (all p-values<0.001) (Table 1). Individuals with high hs-CRP levels were more likely to report greater DSS compared to individuals with low hs-CRP levels (11.2% vs. 7.6%).

Women had a higher prevalence of increased DSS than men (10.3% vs. 6.9%). Young adults exhibited the highest prevalence (10.1%) of greater DSS compared to middle-adults (8.3%) and late-adults (6.7%). Racial/ethnic differences were also evident, with Other/Multiracial reporting the highest prevalence (14.7%) among all racial/ethnic groups and Non-Hispanic Asian demonstrating the lowest prevalence (4.4%). Greater DSS was more common among those with high school degree or less (10.5%), and less common among those with a college degree or higher (5.4%). Participants who reported low-income had a higher prevalence of elevated DSS (14.1%) compared to those who reported high-income (4.5%).

Health and lifestyle characteristics also demonstrated significant associations. Tobacco users had a significantly higher prevalence of increased DSS (15.2%) compared to non-users (6.6%). Prevalence of increased DSS was slightly lower among individuals classified as overweight compared to those classified as healthy weight (6.9% vs. 8.7), while individuals classified as obese demonstrated a modest increase compared to individuals classified as healthy weight (9.8% vs. 8.7%). This finding suggests that the relationship between BMI and DSS may not be linear, with overweight and obesity demonstrating distinct associations. Participants with three or more chronic health conditions had nearly double the prevalence of high DSS compared to those with no health conditions (12.3% vs. 6.4%). Finally, high DSS was more common among respondents after the pandemic onset (11.4%) compared to those surveyed prior to the pandemic onset (7.5%).

3.3. Univariable logistic regression models

These associations were further analyzed using univariable logistic regression models, as presented in Table 2 and Fig. 2. Participants with elevated hs-CRP levels (>3 mg/L) had 54% higher odds of reporting high DSS compared to those with low levels of hs-CRP (crude OR = 1.54; 95% CI: 1.30-1.83; p-value<0.001).

Table 2.

Summary of univariable and multivariable logistic regression analyses.

Variable	Crude OR (95% CI)	Crude p-value	Adjusted OR (95% CI)a	Adjusted p-valuea
hs-CRP Levels (mg/L)
Low (≤3 mg/L)
High (>3 mg/L)	1.54 (1.30, 1.83)	<0.001	1.26 (1.04, 1.51)	0.018
Sex
Male
Female	1.54 (1.32, 1.80)	<0.001	1.61 (1.36, 1.91)	<0.001
Age
Young Adulthood (20-39 yrs)
Middle Adulthood (40-64 yrs)	0.80 (0.67, 0.96)	0.018	0.62 (0.49, 0.79)	<0.001
Late Adulthood (≥65 yrs)	0.64 (0.50, 0.80)	<0.001	0.37 (0.26, 0.53)	<0.001
Race/Ethnicity
Non-Hispanic White
Non-Hispanic Black	1.12 (0.92, 1.35)	0.256	0.76 (0.62, 0.93)	0.008
Hispanic Origin	1.10 (0.88, 1.37)	0.388	1.05 (0.84, 1.32)	0.643
Non-Hispanic Asian	0.51 (0.37, 0.70)	<0.001	0.69 (0.48, 0.99)	0.042
Other Race/Multi-Racial	1.91 (1.42, 2.57)	<0.001	1.28 (0.95, 1.71)	0.100
Highest Level of Education
High School Graduate or Less
Some College or AAb Degree	0.92 (0.78, 1.08)	0.308	1.02 (0.86, 1.22)	0.792
College Graduate or Above	0.49 (0.37, 0.65)	<0.001	0.85 (0.61, 1.17)	0.300
Marital Status
Married or Living w/Partner
Divorced/Separated/Widowed	2.00 (1.68, 2.39)	<0.001	1.45 (1.16, 1.82)	0.002
Never Married	2.14 (1.71, 2.67)	<0.001	1.77 (1.36, 2.30)	<0.001
Income Status based on FIPRc
Low (FIPR ≤1.85)
Middle (FIPR 1.86-4.99)	0.50 (0.42, 0.60)	<0.001	0.64 (0.54, 0.76)	<0.001
High (FIPR ≥5)	0.29 (0.23, 0.36)	<0.001	0.46 (0.36, 0.60)	<0.001
Unreported	0.58 (0.47, 0.72)	<0.001	0.73 (0.57, 0.92)	0.011
Tobacco Use
No
Yes	2.55 (2.22, 2.93)	<0.001	2.06 (1.75, 2.44)	<0.001
Body Mass Index
Healthy Weight (<25)
Overweight (25-29)	0.78 (0.66, 0.93)	0.007	0.81 (0.67, 0.98)	0.031
Obese (≥30)	1.14 (0.94, 1.39)	0.188	0.94 (0.73, 1.21)	0.600
Number of Chronic Health Conditions
0
1-2	1.25 (1.05, 1.50)	0.016	1.71 (1.43, 2.06)	<0.001
≥3	2.04 (1.67, 2.48)	<0.001	3.49 (2.69, 4.54)	<0.001
Timing of Survey Response
Before COVID-19 Onset
After COVID-19 Onset	1.58 (1.33, 1.87)	<0.001	1.66 (1.42, 1.93)	<0.001

^aThe adjusted ORs (95% CIs) and p-values were based on the multivariable logistic regression model which included all listed variables.

^bAA = Associate of Arts.

^cFIPR = Family Income to Poverty Ratio.

Fig. 2.

Forest plot of adjusted odds ratios (ORs) and their 95% confidence intervals (Cis) of the multivariable logistic regression analysis.

Female participants also had significantly higher odds of reporting elevated DSS compared to males (crude OR = 1.54; 95% CI: 1.32-1.80; p-value<0.001). Compared to young adults (20-39 years old), middle-aged adults (40-64 years old) had 20% lower odds (crude OR = 0.80; 95% CI: 0.67-0.96; p-value = 0.018) and older adults (≥65 years old) had 36% lower odds (crude OR = 0.64; 95% CI: 0.50-0.80; p-value<0.001) of elevated DSS. DSS varied significantly across race/ethnicity. Non-Hispanic Asian adults had about half the odds of high DSS compared with non-Hispanic Whites (crude OR = 0.51; 95% CI: 0.37-0.70, p-value<0.001), while Other/Multi-Racial participants had nearly twice the odds (crude OR = 1.91; 95% CI: 1.42-2.57, p-value<0.001) when compared to non-Hispanic White adults. Adults with a college degree or higher had significantly lower odds of elevated DSS compared to those with a high school education or less (crude OR = 0.49; 95% CI: 0.37-0.65; p-value<0.001). Compared to married/cohabitating adults, those who were divorced/separated/widowed had twice the odds of elevated DSS (crude OR = 2.00; 95% CI: 1.68-2.39; p-value<0.001), while adults who were never married had more than double the odds (crude OR = 2.14; 95% CI: 1.71-2.67; p-value<0.001). Income status showed strong inverse associations with DSS. Compared to low-income participants, those with middle-income (crude OR = 0.50; 95% CI: 0.42-0.60; p-value<0.001), high-income (crude OR 0.29; 95% CI: 0.23-0.36; p-value<0.001), and even unreported income (crude OR = 0.58; 95% CI: 0.47-0.72; p-value<0.001) all had significantly lower odds of DSS.

Tobacco users had more than double the odds of high DSS compared to non-users (crude OR = 2.55; 95% CI: 2.22-2.93; p-value<0.001). Compared to adults classified as healthy BMI, those classified as overweight had significantly lower odds of DSS (crude OR = 0.78; 95% CI: 0.66-0.93; p-value = 0.007), while those classified as obese were not significantly different (crude OR = 1.14; 95% CI: 0.94-1.39; p-value = 0.188). Multimorbidity was associated with higher odds of high DSS. Compared with adults reporting no chronic health conditions, those with 1-2 conditions had 25% higher odds of high DSS (crude OR = 1.25; 95% CI: 1.05-1.50; p-value = 0.016), while those with ≥3 chronic conditions had more than double the odds (crude OR = 2.04; 95% CI: 1.67-2.48; p-value<0.001). Finally, participants surveyed after the COVID-19 pandemic onset had significantly higher odds of reporting elevated DSS compared to those surveyed before the COVID-19 pandemic (crude OR = 1.58; 95% CI: 1.33-1.87; p-value<0.001).

3.4. Multivariable logistic regression model

Multivariable logistic regression was used to evaluate all predictors simultaneously, adjusting for demographic, socioeconomic, and health-related covariates (Table 2). Several associations remained significant, though some were attenuated after adjustment, with an value of 0.733 for the multivariable adjusted model.

Those with elevated hs-CRP levels had 26% higher odds of experiencing greater DSS compared to those with lower hs-CRP levels (adjusted OR = 1.26; 95% CI: 1.04-1.51; p-value = 0.018). While statistically significant, this effect size was modest, and the reported AUC reflects the performance of the fully adjusted model rather than hs-CRP alone.

Sex, age, race/ethnicity, education, marital status, and income level also remained significant predictors of DSS. Female participants remained more likely than male participants to report elevated DSS, with 61% higher odds (adjusted OR = 1.61; 95% CI: 1.36-1.91, p-value<0.001). Age continued to show an inverse association with DSS. Compared to young adults, middle-aged adults (40-64 years) had 38% lower odds (adjusted OR = 0.62; 95% CI: 0.49-0.79; p-value<0.001), and those ≥65 years old had 63% lower odds (adjusted OR = 0.37; 95% CI: 0.26-0.53; p-value<0.001) of elevated DSS. Non-Hispanic Asian adults continued to produce significantly lower odds of DSS compared to Non-Hispanic White adults (adjusted OR = 0.69; 95% CI 0.48, 0.999; p-value = 0.042), as well as Non-Hispanic Black adults (adjusted OR = 0.76; 95% CI: 0.62-0.93; p-value = 0.008). Compared to married/cohabitating adults, divorced/separated/widowed adults had 45% higher odds of higher DSS (adjusted OR = 1.45; 95% CI: 1.16-1.82; p-value = 0.002) and adults who were never married had 77% higher odds (adjusted OR = 1.77; 95% CI: 1.36-2.30; p-value<0.001). Compared to low-income participants, individuals classified as middle-income had 36% lower odds of high DSS (adjusted OR = 0.64; 95% CI: 0.54-0.76; p-value<0.001) and those who reported high-income had 54% lower odds (adjusted OR = 0.46; 95% CI: 0.36-0.60; p-value<0.001). The crude association of higher educational attainment was no longer statistically significant after adjustment for covariates.

Tobacco use remained significantly associated with high DSS (adjusted OR = 2.06; 95% CI: 1.75-2.44; p-value<0.001). Associations with BMI became weaker after adjustment. Compared to adults classified as healthy weight, overweight adults had 19% lower odds of elevated DSS (adjusted OR = 0.81; 95% CI: 0.67-0.98; p-value = 0.031).

Multimorbidity continued to have a significant association with DSS. Compared to participants without chronic health conditions, participants with 1-2 chronic health conditions were 71% more likely to have elevated DSS (adjusted OR = 1.71; 95% CI: 1.43-2.06; p-value<0.001), and those with 3 or more chronic health conditions were 249% more likely to report high DSS (adjusted OR = 3.49; 95% CI: 2.69-4.54; p-value<0.001).

Lastly, responses in relation to the COVID-19 pandemic continued to have a significant association with DSS. Those who were surveyed after the onset of the COVID-19 pandemic were 66% more likely to report high DSS compared to those surveyed before the COVID-19 pandemic (adjusted OR = 1.66; 95% CI: 1.42-1.93; p-value<0.001), suggesting a potential influence of broader psychological and social stressors during of the pandemic.

3.5. Exploratory symptom-domain analyses

To further explore whether the association between systemic inflammation and DSS differed across symptom domains, exploratory symptom-domain analyses were conducted examining somatic and cognitive/affective depression symptoms separately while still adjusting for all covariates (Table 3). In adjusted analyses, elevated hs-CRP levels remained significantly associated with greater odds of somatic DSS (adjusted OR = 1.26; 95% CI: 1.10-1.44; p-value<0.001). In contrast, the association between elevated hs-CRP levels and cognitive/affective DSS did not reach statistical significance (adjusted OR = 1.25; 95% CI: 0.99-1.58; p-value = 0.056).

Table 3.

Summary of symptom-domain multivariable logistic regression analyses.

Variable	Somatic Symptoms		Cognitive/Affective Symptoms
	Adjusted OR (95% CI)a	Adjusted p-valuea	Adjusted OR (95% CI)a	Adjusted p-valuea
hs-CRP Levels (mg/L)
Low (≤3 mg/L)
High (>3 mg/L)	1.26 (1.10, 1.44)	<0.001	1.25 (0.99, 1.58)	0.056
Sex
Male
Female	1.84 (1.60, 2.10)	<0.001	1.34 (1.16, 1.55)	<0.001
Age
Young Adulthood (20-39 yrs)
Middle Adulthood (40-64 yrs)	0.70 (0.57, 0.87)	0.001	0.68 (0.52, 0.87)	0.003
Late Adulthood (≥65 yrs)	0.44 (0.43, 0.57)	<0.001	0.33 (0.23, 0.48)	<0.001
Race/Ethnicity
Non-Hispanic White
Non-Hispanic Black	0.74 (0.62, 0.88)	<0.001	0.90 (0.72, 1.13)	0.356
Hispanic Origin	1.10 (0.91, 1.34)	0.320	1.08 (0.84, 1.39)	0.532
Non-Hispanic Asian	0.63 (0.47, 0.85)	0.002	0.81 (0.58, 1.12)	0.198
Other Race/Multi-Racial	1.06 (0.80, 1.42)	0.674	1.59 (1.06, 2.38)	0.025
Highest Level of Education
High School Graduate or Less
Some College or AAb Degree	0.97 (0.87, 1.08)	0.591	0.93 (0.77, 1.12)	0.437
College Graduate or Above	0.72 (0.59, 0.88)	0.001	0.79 (0.60, 1.04)	0.095
Marital Status
Married or Living w/Partner
Divorced/Separated/Widowed	1.19 (0.99, 1.43)	0.058	1.49 (1.13, 1.97)	0.005
Never Married	1.58 (1.34, 1.86)	<0.001	1.67 (1.30, 2.14)	<0.001
Income Status based on FIPRc
Low (FIPR ≤1.85)
Middle (FIPR 1.86-4.99)	0.84 (0.71, 0.99)	0.039	0.63 (0.50, 0.79)	<0.001
High (FIPR ≥5)	0.51 (0.43, 0.61)	<0.001	0.44 (0.31, 0.60)	<0.001
Unreported	0.80 (0.63, 1.01)	0.064	0.75 (0.56, 0.99)	0.041
Tobacco Use
No
Yes	2.02 (1.74, 2.35)	<0.001	1.74 (1.39, 2.19)	<0.001
Body Mass Index
Healthy Weight (<25)
Overweight (25-29)	0.88 (0.73, 1.07)	0.200	0.96 (0.73, 1.26)	0.781
Obese (≥30)	1.11 (0.92, 1.35)	0.283	0.92 (0.69, 1.23)	0.586
Number of Chronic Health Conditions
0
1-2	1.84 (1.56, 2.17)	<0.001	1.55 (1.22, 1.98)	<0.001
≥3	3.75 (3.02, 4.64)	<0.001	3.66 (2.66, 5.04)	<0.001
Timing of Survey Response
Before COVID-19 Onset
After COVID-19 Onset	1.29 (1.14, 1.46)	<0.001	1.63 (1.34, 2.00)	<0.001

^aThe adjusted ORs (95% CIs) and p-values were based on the multivariable logistic regression model which included all listed variables.

^bAA = Associate of Arts.

^cFIPR = Family Income to Poverty Ratio.

Several covariates, including female sex, age, multimorbidity, tobacco use, and survey participation after the onset of the COVID-19 pandemic, remained significantly associated with increased odds of DSS across both symptom domains. Notably, participants with three or more chronic health conditions demonstrated substantially greater odds of both somatic DSS (adjusted OR = 3.75; 95% CI: 3.02-4.64; p < 0.001) and cognitive/affective DSS (adjusted OR = 3.66; 95% CI: 2.66-5.04; p < 0.001).

Other covariates demonstrated more variable associations across symptom domains. Although education level was not significantly associated with DSS in the primary fully adjusted model, having a college education or higher was significantly associated with lower odds of elevated somatic DSS in symptom-domain analyses (adjusted OR = 0.72; 95% CI: 0.59-0.88; p-value = 0.001). Similarly, divorced/separated/widowed marital status was not significantly associated with elevated somatic DSS (adjusted OR = 1.19; 95% CI: 0.99-1.43; p-value 0.58) or elevated cognitive/affective DSS (adjusted OR = 1.49; 95% CI: 1.13-1.97; p-value = 0.005). In contrast, never married status remained significantly associated with increased odds of DSS across both symptom domains. Additionally, while overweight status demonstrated a significant association with lower DSS in the primary fully adjusted model, this relationship was no longer significant in either symptom-domain model.

4. Discussion

This study examined the association between DSS and a wide range of demographic, socioeconomic, and health-related characteristics in a nationally representative sample of civilian, non-institutionalized U.S. adults. The findings provide support for the role of systemic inflammation in the disease process of depression, while also highlighting the role of demographic, socioeconomic, and health-related factors.

4.1. Depression symptom severity (DSS) and hs-CRP

Elevated hs-CRP levels remained significantly associated with high DSS after adjusting for covariates, consistent with prior literature proposing a potential role of inflammation in depression. Although the present study cannot establish causality due to its cross-sectional design, experimental studies provide insight into potential biological mechanisms underlying this relationship.

Animal models have demonstrated that inflammatory activity may induce depression-like behaviors through neuroinflammation, altered neurotransmitter activity, impaired neurogenesis, and increased blood-brain barrier permeability (Anisman and Merali, 2002; Peruga et al., 2011; Kraus et al., 2017; Lee and Giuliani, 2019). Human studies similarly reported elevated CRP and proinflammatory cytokine levels among individuals with depression, even after adjustment for confounding variables (Dantzer et al., 2008; Raison and Miller, 2011; Tabatabaeizadeh et al., 2018; Osimo et al., 2020; Suneson et al., 2023). Proposed mechanisms include inflammatory-mediated alterations in immune-brain signaling, neurotransmitter pathways, and emotional regulation (Capuron and Miller, 2011; Thomas and Apovian, 2017; Varatharaj and Galea, 2017; Salvador et al., 2021). However, the directionality of this association remains uncertain, as depression itself may also contribute to the inflammatory responses, suggesting a potentially bidirectional relationship (Kiecolt-Glaser et al., 2015).

In a study conducted using data from the 2016 Korean NHANES, it was found that elevated hs-CRP levels were significantly associated with an increased risk of depression (Jung and Kang, 2019). However, the association only remained significant for younger adults after controlling for potentially confounding factors (Jung and Kang, 2019). In addition to hs-CRP levels, other inflammatory biomarkers, such as interleukin-6 and interleukin-8, have also be found to be elevated in depressed patients (Suneson et al., 2023). Although effect sizes were modest, the persistence of this relationship after adjustment suggests that systemic inflammation may play a crucial role in the pathophysiology of depression. Additional studies also reported positive associations between low-grade inflammation and depressive symptoms, further supporting this link (Orsolini et al., 2022; Ji et al., 2024).

Although the observed association in this study was modest, these findings contribute to a growing body of literature indicating the role of inflammation in depression. Further longitudinal and interventional studies are necessary to determine whether inflammatory biomarkers such as hs-CRP have a role in the screening, diagnosing, classifying, or treatment selection for depression.

4.2. Depression symptom severity (DSS) and covariates

Female participants in the current study had higher odds of elevated DSS compared to male participants. This observed sex difference is consistent with prior studies demonstrating higher rates of depression among women (Siu and the US Preventive Services Task Force, 2016; Hasin et al., 2018; Azarmanesh et al., 2022). This pattern has been attributed to a combination of biological and societal/cultural factors, including hormonal fluctuations (e.g., premenstrual dysphoric disorder, postpartum depression, and postmenopausal depression), gendered social roles and expectations (e.g., temperament, coping styles), unequal power and advancement opportunities, and different exposure to stressors (Albert, 2015; Kuehner, 2017). The current analysis showed that young adults had higher odds of high DSS compared to middle- and late-adulthood populations. This protective association between older age and DSS aligns with prior research suggesting that depression symptoms are more common in early-adulthood, potentially due to cultural shifts (e.g., rising use of electronic communication and social media), financial instability, and the aftermath of the COVID-19 pandemic (Twenge et al., 2019; Villaume et al., 2023). The prevalence of elevated DSS among females and young adults is supported by a 2025 public health report by the NCHS, which showed depression prevalence decreasing with increasing age (Brody and Hughes, 2025). When participants were stratified by sex, the trend continued, but with depression prevalence being higher in females than males in every age group.

The strong association between low-income and high DSS in this study aligns with previous findings on the impact of socioeconomic status as a social determinant of health. Financial strain, increased exposure to environmental irritants, limited access to healthcare and other essential resources, and disproportionate exposure to trauma, violence, and crime, have been well-known pathways linking income to poor mental health (Hasin et al., 2018; Ridley et al., 2020; Zare et al., 2022). Similarly, unmarried adults, particularly those who are divorced or widowed, had greater odds of increased DSS, which is consistent with literature highlighting the role of social support in reducing depression risk, especially later in life (Adamczyk and Segrin, 2014). While being single may be seen as freedom by some, divorce and widowhood expose individuals to vulnerabilities such as lower socioeconomic conditions and poor health status, which may lead to or compound already existing depression (Srivastava et al., 2021).

Behavioral habits and health status predictors identified are also in alignment with prior studies. Some researchers believe that depressed individuals are more likely to smoke in order to improve their mental health status (self-medication hypothesis), while others argue that the biological effects of nicotine lead to psychological damage that increases the risk of depression (Wu et al., 2023).

The relationship between BMI and DSS in this study differed from prior literature that has reported a positive association between obesity and depression symptoms (Jokela and Laakasuo, 2023). Our sensitivity analysis suggests that the magnitude and significance of the association between hs-CRP and DSS was very stable whether BMI was included as covariate or not in the multivariable model (results not shown). In the present analysis, individuals classified as overweight demonstrated a lower prevalence and odds of elevated DSS, while individuals classified as obese demonstrated a higher prevalence but lower odds of elevated DSS compared to participants classified as healthy weight. These findings should be interpreted cautiously, as the apparent protective association observed among overweight participants may depend on how BMI categories were defined and may not fully reflect underlying metabolic or inflammatory health status. Additionally, BMI is a screening tool rather than a direct measure of total fat, fat distribution, or metabolic health. BMI does not account for body composition or fat distribution, which are more strongly associated with metabolic and inflammatory risk (National Academies of Sciences, Engineering, and Medicine, 2023).

The presence of certain chronic health conditions (e.g., chronic respiratory illnesses, heart disease) pose as a major risk factor for depression, as these conditions cause physical burdens, financial strains, barriers to relationships, and lead to a decline in quality of life (Xiang et al., 2025). Multimorbidity was strongly associated with higher DSS in the fully adjusted model, emphasizing the relationship between physical and mental health. Even though chronic disease burden is closely linked to systemic inflammation, our sensitivity analysis by leaving out chronic disease burden from the multivariable model does not significantly impact the association between hs-CRP and DSS (results not shown).

The increased odds of depression among respondents surveyed after the COVID-19 pandemic began found in the current study are consistent with global studies showing increases in DSS during and after the onset of the pandemic. Social isolation, uncertainty, financial instability, and disruptions in daily routines likely contributed to this effect (Varma et al., 2021; WHO, 2022).

The current study's findings underscore the importance of addressing depression risk among vulnerable populations. Routine depression screening in primary care and community settings may be particularly important for women, low-income populations, unmarried adults, and individuals with multiple chronic health conditions. Integrating behavioral health into chronic disease management could also improve both physical and mental health outcomes. Additionally, public health strategies aimed at reducing socioeconomic burdens, such as expanding access to healthcare, improving job security, creating spaces that encourage physical activity, and strengthening social safety nets, may reduce factors associated with increased DSS.

4.3. Exploratory symptom-domain analyses

A notable finding of the present study was the differential association observed between hs-CRP levels and depression symptom domains. In the exploratory symptom-domain analyses, elevated hs-CRP levels remained independently associated with elevated somatic DSS, whereas the association with cognitive/affective DSS did not reach statistical significance. These findings suggest that systemic inflammation may demonstrate stronger associations with somatic manifestations of depression, including fatigue, appetite disturbance, sleep disruption, and psychomotor-related symptoms, compared to cognitive/affective manifestations. These findings are supported by previous animal studies demonstrating that systemic inflammation can induce depression-like behaviors, while suppression of the immune response may reduce these symptoms (Anderson et al., 2015; Lee and Giuliani, 2019).

The differential findings observed across symptom domains may help explain some of the variability reported in prior studies examining inflammation and depression. While inflammatory processes may contribute more strongly to physical and behavioral symptoms of depression, cognitive/affective symptoms may be influenced more by psychosocial, environmental, or other biological factors not fully captured by peripheral inflammatory markers such as hs-CRP, further supporting the heterogenous nature of depression. These findings should be interpreted cautiously and warrant further investigation in longitudinal studies using broader inflammatory biomarker panels.

Additionally, multimorbidity remained strongly associated with both elevated somatic and cognitive/affective DSS, suggesting that cumulative chronic disease burden may contribute broadly to DSS regardless of symptom domain. These findings further support the importance of considering chronic disease burden when evaluating the relationship between systemic inflammation and depression.

4.4. Strengths & limitations

This study has several strengths. It uses a large, nationally representative dataset spanning multiple NHANES cycles, models that adjust for complex survey design, utilizes hs-CRP as a more sensitive biomarker of low-grade systemic inflammation, and incorporates a broad range of demographic, socioeconomic, lifestyle, and health-related covariates. These features enhance generalizability and reduce biases. Additionally, the inclusion of survey cycles before and after the onset of the COVID-19 pandemic provides insight into how major societal disruptions may influence chronic systemic inflammation and DSS.

Several limitations should also be acknowledged. First, the cross-sectional design does not allow for causal inferences regarding the directionality of the associations. It is important to consider the possibility of reverse causation. While elevated hs-CRP may contribute to increased DSS, it is also possible that depression itself promotes inflammatory responses through behavioral and physiological pathways. Second, the removal of subjects with missing values from analysis could lead to potential selection bias. Third, depression symptoms were self-reported via the Mental Health - Depression Screener Questionnaire, which poses the risk of recall or reporting biases. Fourth, hs-CRP was the only inflammatory marker consistently collected among the three data cycles. While hs-CRP is a sensitive marker of systemic inflammation, it represents only one mediator in the inflammatory response. More comprehensive biomarker panels may provide a better picture of biological mechanisms underlying depression. Fifth, despite adjustment for a wide range of demographic, socioeconomic, and health-related covariates, residual confounding cannot be excluded. Depression is influenced by a complex interplay of biological, psychological, and social factors (e.g., genetics, prior trauma, current stressors, medication use) that could not be included in the current analysis due to limitations in the variables available within the NHANES dataset.

Future research should employ longitudinal designs to clarify causal relationships, incorporate a broader panel of inflammatory markers, and evaluate the sensitivity and specificity of hs-CRP in identifying depression.

5. Conclusion

This study demonstrates that depression symptom severity among U.S. adults is influenced by a complex interplay between demographic, socioeconomic, behavioral, and health-related factors. Elevated hs-CRP levels remained independently associated with higher DSS after adjustment for multiple covariates, supporting the potential relationship between systemic inflammation and depression. Exploratory symptom-domain analyses further suggested that inflammatory processes may be more associated with somatic depression symptoms than cognitive/affective symptoms, supporting the heterogenous nature of depression. Beyond hs-CRP levels, significant associations were also observed with female sex, young-adulthood, low-income status, unmarried status, tobacco use, multimorbidity, and post-pandemic status. Together, these findings emphasize the importance of considering both biological and social factors when evaluating depression risk and support the need for future longitudinal studies investigating inflammation-related depression symptom profiles.

Funding

EL, CS, and JJC were partially supported by the Ola HAWAII grant (2U54MD007601-36) from the National Institutes on Minority Health and Health Disparities, and EL was also partially supported by the PIKO grant (U54GM138062) from the National Institute of General Medical Sciences, National Institutes of Health.

CRediT authorship contribution statement

Jediah Grace Arellano: Conceptualization, Data curation, Formal analysis, Methodology, Writing – original draft. Eunjung Lim: Data curation, Funding acquisition, Methodology, Writing – review & editing. Chathura Siriwardhana: Methodology, Writing – review & editing. John J. Chen: Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Supervision, Validation, Writing – original draft, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Public Domain NHANES