Explore the U-shape between the stress hyperglycemia ratio and diabetes or prediabetes with depressive symptoms in United States adults: insights in NHANES 2005–2018

Abstract

Background

Diabetes is a prevalent health problem worldwide, depression is a common psychological complication in diabetics, and studies have now confirmed that diabetes with depressive symptoms is associated with blood glucose levels. Stress hyperglycemic ratio (SHR) can reflect the collective transient elevation of blood glucose during acute stress. Meanwhile, the Triglyceride-Glucose (TyG) Index, a surrogate marker of insulin resistance, is associated with long-term metabolic dysfunction. This study explores and compares the predictive value of SHR and TyG in diabetes with depressive symptoms among adult Americans, aiming to provide insights into their clinical and metabolic implications.

Methods

This is a cross-sectional study of adult participants from the National Health and Nutrition Examination Survey (NHANES) from 2005 to 2018. SHR values were calculated from fasting glucose (FPG) and glycosylated hemoglobin (HbA1c) measurements using a specific formula. The Triglyceride-Glucose (TyG) index is calculated as the natural logarithm of the product of fasting plasma glucose and triglyceride levels, divided by two, and is used as a marker of insulin resistance. Depressive symptoms were assessed using the Patient Health Questionnaire-9 (PHQ-9). Multivariate logistic regression modeling was used to explore the relationship between SHR and depressive symptoms, and the nonlinear relationship was explored by smoothed curve fitting. Comparing the predictive value of SHR and TyG for diabetes with depressive symptoms using ROC curves.

Results

This study included 13,905 U.S. participants who were at least 20 years old. Out of them, 358 individuals with diabetes or prediabetes had a diagnosis of depression. After controlling for variables, multifactorial logistic regression revealed an OR of 1.06, meaning that every unit rise in SHR was linked to a 6% increase in the risks of depressive symptoms. Quartile analysis showed that the risk of depressive symptoms was significantly higher in the Q4 of SHR compared to the Q1. When we adjusted for TYG in Model 4, the association between SHR and depressive symptoms became negatively significant (OR = 0.92, 95% CI: 0.87 0.97). Nonlinear analysis showed a “U”-shaped relationship, with a significant inflection point at an SHR value of 16.07. Above this point, the chance of having depressive symptoms is dramatically higher; otherwise, the risk is lower. In the prediction of depressive symptoms in patients with diabetes, the area under the curve (AUC) of the TyG index versus the SHR reached 0.663 and 0.534, respectively, suggesting that the TyG index may have superior predictive efficacy compared to the SHR.

Conclusion

This study reveals a significant U-shaped relationship between SHR and depressive symptoms in individuals with diabetes. Likewise, a distinct non-linear association was identified between TyG and depressive symptoms. Both TyG and SHR show potential as simple and effective strategies for identifying diabetes-related depressive symptoms, with TyG demonstrating slightly greater predictive value compared to SHR. Nonetheless, the overall predictive performance of these markers remains limited.

Introduction

Diabetes poses a significantly high risk, with its global prevalence and widespread occurrence of prediabetes presenting a major public health challenge [1, 2]. The International Diabetes Federation (IDF) reported in 2021 that over 500 million persons worldwide suffer from diabetes, accounting for approximately 10.5% of all adult populations globally [3]. Beyond directly impacting quality of life and health status, diabetes frequently coexists with serious consequences such as neuropathy, retinopathy, and cardiovascular illnesses, which substantially increase medical burdens and treatment complexities for patients [4]. Notably, depression, as a prevalent psychological complication among diabetes patients, further compromises both their physical health and social functioning, significantly affecting overall well-being [5].

The incidence of depressive symptoms is significantly higher in people with diabetes. According to research, the prevalence of depression in individuals with diabetes ranges from 10% to 30%, while in the general population, it normally falls between 5% and 10% [6]. Depression can lead to a decline in self-care behaviors and reduced adherence to treatment [7, 8]. The intricate relationship between depression and diabetes makes therapy more challenging and places a heavier burden on patients’ health [9].

Stress hyperglycemia, characterized by transient elevations in blood sugar concentrations during acute stress or illness, is a well-documented phenomenon [10]. Stress hormones like cortisol and catecholamines are released, promoting gluconeogenesis and decreasing insulin sensitivity. This is a physiological reaction to stress [11]. The SHR, which quantifies this response, has been extensively studied in various clinical contexts, particularly in critically ill patients, where it is associated with outcomes such as mortality and morbidity [11–13]. However, recent evidence suggests that the relevance of SHR extends beyond acute care settings. SHR has been linked to an increased risk of cardiovascular events and long-term mortality in patients with chronic diseases [14]. For example, a recent study demonstrated that SHR is associated with both all-cause and cardiovascular mortality in individuals with diabetes or prediabetes, highlighting its potential predictive value in identifying patients at higher risk of adverse outcomes [15, 16].

In addition to SHR, the Triglyceride-Glucose (TyG) index has gained significant attention as a marker of insulin resistance and metabolic dysfunction [17]. Calculated using fasting plasma glucose and triglyceride levels, TyG has been strongly correlated with metabolic disorders, including type 2 diabetes and cardiovascular diseases [18]. Unlike SHR, which reflects acute metabolic stress, TyG is considered a more chronic marker of insulin resistance. Given the established connection between insulin resistance and depressive symptoms, particularly in individuals with diabetes, TyG may provide valuable insight into the underlying mechanisms linking metabolic dysregulation with mental health [19]. Investigating how chronic markers of metabolic dysfunction, such as TyG, differ from acute markers like the Stress Hyperglycemia Ratio (SHR) in predicting depressive symptoms among individuals with diabetes or prediabetes could provide valuable insights. This distinction may enhance our understanding of how metabolic disturbances contribute to the development and progression of depression.

This study, therefore, aims to investigate the relationship between SHR and depressive symptoms in individuals with diabetes or prediabetes, with a specific focus on American adults over the age of 20. Additionally, by including the TyG index in our analysis, we seek to explore its potential role in predicting depressive symptoms and compare its effectiveness to SHR. This dual approach will offer a more comprehensive perspective on the complex relationship between metabolic dysfunction and depression in diabetic patients, and may inform strategies to mitigate the negative impact of these conditions on overall health.

Methods

Study population

This study is a cross-sectional study based on seven cycles of data. We analyzed data on U.S. adults collected by National Health and Nutrition Examination Survey (NHANES) between 2005 and 2018 to characterize the distribution of depressive symptoms among individuals with diabetes and population health status and to explore the relationship between certain factors and disease. NHANES is a nationally representative sample survey of noninstitutionalized adult Americans. The survey is conducted jointly by the National Center for Health Statistics (NCHS) and the Centers for Disease Control and Prevention (CDC), with implementation protocols approved by the NCHS Institutional Review Board (NCHS IRB/ERB). NHANES regularly assesses the health and nutritional status of adults and children in the U.S. through laboratory tests, physical examinations, and interviews. All participants provided informed consent.

From the 2005–2018 NHANES dataset, which included 70,190 subjects, 30,441 self-reported their age as 20 or older. We excluded individuals without diabetes or prediabetes symptoms (n = 78), as well as those with missing values for outcome variables and exposure data: depressive symptoms data (n = 5,419); fasting glucose (n = 18,362); HbA1c (n = 34); and the exclusion of missing data for other covariates is shown in Fig. 1 (total n = 1,690). Additionally, SHR outliers were excluded (n = 261). Outliers were identified based on the standard deviation (SD) method for data following a normal or approximately normal distribution. Specifically, values falling outside the range of Mean ± SD, which encompasses 99.7% of the data, were considered outliers and excluded from the analysis. Ultimately, the total number of adult participants was 13,905. The screening process for the study population is showed in Fig. 1.

Fig. 1.

Flowchart of the sample selection from NHANES 2005–2018

Determination of Stress-hyperglycemia ratio and triglyceride-glucose (TyG) index

SHR values were calculated according to the equation: [FPG (mmol/L)]/[1.59 * HbA1c (%) − 2.59] [20]. The TyG index was computed according to the triglyceride (TG) and fasting blood glucose (FBG) concentrations according to the equation: Ln [TG (mg/dl) × FBG (mg/dl)/2] [21]. In this case, FPG was collected and measured by a standardized procedure and detected by the enzyme hexokinase method. HbA1c was determined by high-pressure liquid-phase ion-exchange chromatography and measured by the Diabetes Laboratory at the University of Missouri-Columbia campus using a Tosoh G8 Glycohemoglobin Analyzer.

Assessment of depressive symptoms

NHANES assesses symptoms of depression using the PHQ-9. This questionnaire contains nine questions assessing whether the patient has experienced symptoms related to depression in the past two weeks. Each question is rated from 0 (none) to 3 (almost every day), and the total score ranges from 0 to 27 [22]. In this research, a PHQ-9 score of ≥ 10 was deemed indicative of a depressed disposition [23]. Due to its utility and applicability to different populations, PHQ-9 is widely used in primary care, the mental health field, and epidemiologic studies as a tool to screen for depressive symptoms [23].

The criteria to diagnose diabetes and prediabetes

A diagnosis of diabetes and prediabetes is defined according to the 2013 American Diabetes Association guidelines [24] as:

Diabetes (DM) was defined as either treatment or medical diagnosis of hyperglycemia with hemoglobin A1c ≥ 6.5%, fasting blood glucose ≥ 126 mg/dl, or a 2-h blood glucose ≥ 200 mg/dl.

Prediabetes was defined as any one of the following: 5.7% ≤ hemoglobin A1c (HbA1c) < 6.5%, fasting plasma glucose (FPG) between 5.6 mmol/L and 7.0 mmol/L, and a 2 h FPG value between 7.8 mmol/L and 11.1 mmol/L during an oral glucose tolerance test (OGTT).

Covariates and definition

Demographic variables include:

Age (20 ~ 59 years, ≥ 60 years); sex (male female); race (categorized as non-Hispanic White; non-Hispanic Black; Mexican America; other Race includes multiracial and other Hispanic races); Education levels (categorized as < 9th grade, 9th − 11th grade, high school graduation, or General Educational Development (GED), a college or associate’s degree (AA), Bachelor’s Degree or higher); Marital Status (categorized as Never Married; Married or Living with a Partner; Widowed, Divorced, or Separated); BMI: [(18.5 kg/m2 ≤ BMI < 25 kg/m2), superheavy (25 kg/m2 ≤ BMI < 30 kg/m2) sum fattening (BMI ≥ 30 kg/m2)]; Smokers were defined as those who smoked at least 100 cigarettes in their lifetime (dichotomous variable, yes/no), alcohol consumption (Individuals with at least 12 drinks/1 year (dichotomous variable, yes/no).

Complications/comorbidities include:

Ever told you had hypertension(yes/no);

Ever told you had coronary heart disease(yes/no);

Ever told you had heart failure(yes/no);

Ever told you had weak/failing kidneys(yes/no);

Ever told you had stroke(yes/no);

Diabetes affected by eye with retinopathy(yes/no);

Laboratory parameters include:

Serum total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), fasting blood glucose (FBG), hemoglobin, and HbA1, which were determined by enzymatic, direct immunoassay, rate method, latex enhanced scattering turbidimetry, and high-pressure liquid chromatography, respectively.

Statistical analysis

R Studio (version 4.2.3) and EmpowerStats (version 4.2) software were used for the statistical analysis of this study. Continuous variables with a regularly distributed distribution were expressed as mean SD, while continuous variables with an irregular distribution were expressed as median (interquartile spacing) and categorical variables as frequency of occurrence (%). T-tests or chi-square tests were used to compare differences when patients were categorized by SHR (continuous variable) quartiles. For multivariate logistic regression analyses, we constructed three models: model 1 (unadjusted for covariates), model 2 (adjusted for age, sex, and race), and model 3 (adjusted for age, sex, race, education level, marital status, smoking, alcohol consumption, HDL-C, TC, triglycerides, LDL-C, use of insulin, BMI and comorbidities). Model 4 further removes the covariate triglycerides after the addition of TyG to Model 3. While adjusting for covariates, we also performed smoothed curve fitting to deal with nonlinear relationships between variables. Smooth curve fitting was performed using the MGCV package, which creates thin plate regression splines. The default maximum degrees of freedom were set to 10, and the smoothness penalty was optimized using generalized cross-validation (GCV). To evaluate the discriminative ability of SHR and the TyG index in predicting depressive symptoms among diabetic patients, we conducted a Receiver Operating Characteristic (ROC) curve analysis using R Studio. The Area Under the Curve (AUC) was calculated to quantify the overall performance of each predictive model. AUC values range from 0.5 (no discriminative ability, equivalent to random chance) to 1.0 (perfect discrimination). Threshold effect analyses were subsequently conducted using segmented regression models to estimate the breakpoints, Likelihood Ratio Tests (LRT) to assess model fit improvement, and Bootstrap resampling methods to evaluate the stability and confidence intervals of the identified thresholds. In addition, we performed subgroup analyses and interaction term analyses. Subgroup analyses were conducted using variables selected for their clinical relevance and hypothesized interaction with the association between SHR/TyG and depressive symptoms. These included demographic factors (age, sex), BMI categories, education levels, complications (e.g., congestive heart failure, coronary heart disease, and chronic kidney disease), and TyG quartiles to assess the dose-response relationship. P-values that were equivalent to or less than 0.05 were regarded as statistically significant.

Results

Baseline characteristics

This study included 14,460 participants, including 358 patients with diabetes comorbid with depression. The mean age of diabetes without depressive symptoms formers was 49.33 years, and that of diabetes comorbid with depressive symptoms formers was 56.79 years (P < 0.001). Table 1 demonstrates the characteristics of participants. Compared to diabetes without depressive symptoms, those diabetes comorbid with depression were older, more female, non-Hispanic black, widowed/divorced/separated, as well as having higher BMI, SHR levels and TyG levels. In addition, participants who have depressive symptoms associated with diabetes had a significantly higher prevalence of complications such as coronary heart disease, stroke, CKD, and hypertension, diabetes affected by eye with retinopathy compared with diabetes without depressive symptoms (all P < 0.001).

Table 1.

Characteristics of study sample

Characteristic	Non- Diabetes with depressive symptoms formers (N = 13547)	Diabetes with depressive symptoms formers (N = 358)	P-value
Age, years (mean ± SD)	49.30 ± 17.75	56.79 ± 13.40	< 0.001
Age, Strata(%)			< 0.001
20–59	9062 (66.89%)	190 (53.07%)
≥ 60	4485 (33.19%)	168 (47.31%)
BMI, kg/m2 (mean ± SD)	28.89 ± 6.69	33.47 ± 7.88	< 0.001
BMI (kg/m2) (%)			< 0.001
< 25	4075 (30.08%)	41 (11.45%)
25–29.9.9	4558 (33.65%)	82 (22.91%)
≥ 30	4914 (36.27%)	235 (65.64%)
Sex (%)			< 0.001
Male	6683 (49.33%)	129 (36.03%)
Female	6864 (50.67%)	229 (63.97%)
Race (%)			0.002
Mexican American	2119 (15.64%)	57 (15.92%)
Other Hispanic	1347 (9.94%)	50 (13.97%)
Non-Hispanic White	5951 (43.93%)	149 (41.62%)
Non-Hispanic Black	2714 (20.03%)	83 (23.18%)
Other race	1416 (10.45%)	19 (5.31%)
Education levels (%)			< 0.001
< 9th grade	1284 (9.48%)	70 (19.55%)
9-11th grade (includes 12th grade with no diploma)	1925 (14.21%)	75 (20.95%)
High school graduate or GED or equivalent	3112 (22.97%)	78 (21.79%)
College graduate or above	3952 (29.17%)	96 (26.82%)
Some college or AA degree	3274 (24.17%)	39 (10.89%)
Marital status			< 0.001
Married or living with partner	8317 (61.39%)	164 (45.81%)
Widowed or divorced or separated	2843 (20.99%)	141 (39.39%)
Never married	2387 (17.62%)	53 (14.80%)
Comorbidities
Congestive heart failure (%)			< 0.001
Yes	369 (2.72%)	38 (10.61%)
No	13,178 (97.28%)	320 (89.39%)
Coronary heart disease (%)			< 0.001
Yes	526 (3.73%)	37 (9.95%)
No	13,562 (96.27%)	335 (90.05%)
Diabetes affected by eye with retinopathy			< 0.001
yes	251 (1.85%)	46 (12.85%)
no	1123 (8.29%)	149 (41.62%)
NA	12,173 (89.86%)	163 (45.53%)
Use of insulin			< 0.001
yes	369 (2.72%)	51 (14.25%)
no	13,178 (97.28%)	307 (85.75%)
Stroke (%)			< 0.001
Yes	453 (3.34%)	43 (12.01%)
No	13,094 (96.66%)	315 (87.99%)
CKD (%)			< 0.001
Yes	374 (2.76%)	36 (10.06%)
No	13,173 (97.24%)	322 (89.94%)
Hypertension (%)			< 0.001
Yes	4713 (34.79%)	250 (69.83%)
No	8834 (65.21%)	108 (30.17%)
Laboratory results
Fasting glucose (mg/dL)	106.70 ± 29.97	140.87 ± 56.43	< 0.001
SHR	16.56 ± 2.10	16.85 ± 2.71	0.029
HDL-C (mmol/L)	1.41 ± 0.42	1.28 ± 0.39	< 0.001
TC (mmol/L)	4.97 ± 1.06	4.88 ± 1.10	0.039
Triglycerides (mmol/L)	1.33 ± 0.75	1.65 ± 0.83	< 0.001
LDL-C (mmol/L)	2.96 ± 0.92	2.84 ± 0.99	0.007
TC(mg/dL)	191.80 ± 41.10	187.13 ± 43.06	0.022
TyG	9.19 ± 0.29	9.40 ± 0.40	< 0.001
HbA1c	5.69 ± 0.99	6.87 ± 1.76	< 0.001

BMI, body mass index; CKD, chronic kidney diseases; HbA1c, hemoglobin a1c; HDL-C, high-density lipoprotein cholesterol; LDL-C, Low-density lipoprotein cholesterol; TC, total cholesterol; SHR Stress–hyperglycemia ratio; TyG Triglyceride-Glucose index

Association between SHR and depressive symptoms with diabetes and prediabetes

Results are shown in Table 2. SHR was positively correlated with depressive symptoms in multiple models. In the fully adjusted model (Model 3), a 1-unit increase in SHR was associated with a 5% increase in the likelihood of depression (OR = 1.05, 95% CI: 1.00–1.10). Quartile analysis of SHR revealed that, in Model 3, the Q4 group had a significantly higher likelihood of depressive symptoms compared to Q1 (OR = 1.34, 95% CI: 1.01 1.79). In Model 4, after adjusting for TyG, the likelihood of depression decreased significantly with each 1-unit increase in SHR (OR = 0.92, 95% CI: 0.87 0.97,). Quartile analysis showed that, in Model 4, the Q4 group had a lower likelihood of depressive symptoms compared to Q1 (OR = 0.71, 95% CI: 0.51 0.98), indicating that adjusting for TyG in the model reversed the SHR-depression relationship. Due to a high proportion of missing data for diabetes-related eye disease (88.7%), this variable was excluded from Models 3 and 4 to ensure result stability.

Table 2.

The associations of the SHR with depressive symptoms with diabetes and prediabetes

Exposure SHR	Depressive symptoms OR (95% CI)
	Crude model 1	Model 2	Model 3	Model 4
Continuous	1.06 (1.01, 1.12) **	1.09 (1.04, 1.14) ***	1.05 (1.00, 1.10) *	0.92 (0.87, 0.97) **
quartile
Q1	1.0	1.0	1.0	1.0
Q2	0.66 (0.48, 0.91) *	0.74 (0.54, 1.02)	0.80 (0.58, 1.11)	0.66 (0.48, 0.92) *
Q3	0.69 (0.50, 0.94) *	0.82 (0.60, 1.13)	0.81 (0.58, 1.12)	0.60 (0.43, 0.84) **
Q4	1.32 (1.01, 1.73) *	1.59 (1.21, 2.09) **	1.34 (1.01, 1.79) *	0.71 (0.51, 0.98) *

SHR Stress Hyperglycemia Ratio

Crude model was adjusted for none

Model 2 was adjusted for categorical age, sex, race

Model 3 was adjusted for categorical age, sex, race, Education levels, marital status, smoking, drinking, Congestive heart failure (yes or no), Coronary heart disease (yes or no), CKD (yes or no), Hypertension (yes or no), HDL-C, TC, Triglycerides, LDL-C, BMI, use of insulin (yes or no)

Model 4 was adjusted for categorical age, sex, race, Education levels, marital status, smoking, drinking, Congestive heart failure (yes or no), Coronary heart disease (yes or no), CKD (yes or no), Hypertension (yes or no), HDL-C, TC, TyG, LDL-C, BMI, use of insulin (yes or no)

*P < 0.05

**P < 0.01

***P < 0.001

We performed smoothed curve-fitting analysis, which revealed a U-shaped relationship between SHR and diabetes-related depressive symptoms, with a significant inflection point at SHR = 16.07 (Table 4; Fig. 2). Below this point, each 1-unit increase in SHR was associated with a reduced risk of depression (OR = 0.88, 95% CI: 0.79–0.97, p = 0.0134). However, above the inflection point, SHR increases were associated with a significantly higher risk of depression (OR = 1.14, 95% CI: 1.07 1.22).

Table 4.

The associations of the TyG with depressive symptoms with diabetes and prediabetes

Exposure TyG	Depressive symptoms OR (95% CI)
	Crude model 1	Model 2	Model 3	Model 4
Continuous	6.00 (4.59, 7.83) ***	5.55 (4.21, 7.32) ***	5.37 (3.80, 7.59) ***	7.46 (5.00, 11.11) ***
quartile
Q1	1.0	1.0	1.0	1.0
Q2	1.32 (0.90, 1.91)	1.28 (0.88, 1.86)	1.44 (0.97, 2.13)	1.48 (0.99, 2.20)
Q3	1.39 (0.96, 2.02)	1.32 (0.91, 1.92)	1.63 (1.08, 2.46) *	1.70 (1.12, 2.59) *
Q4	3.75 (2.72, 5.16) ***	3.38 (2.43, 4.68) ***	3.56 (2.38, 5.33) ***	3.89 (2.52, 6.00) ***

TyG Triglyceride-Glucose (TyG) index

Crude model was adjusted for none

Model 2 was adjusted for categorical age, sex, race

Model 3 was adjusted for categorical age, sex, race, Education levels, marital status, smoking, drinking, Congestive heart failure (yes or no), Coronary heart disease (yes or no), CKD (yes or no), Hypertension (yes or no), HDL-C, TC, Triglycerides, LDL-C, BMI, use of insulin (yes or no)

Model 4 was adjusted for categorical age, sex, race, Education levels, marital status, smoking, drinking, Congestive heart failure (yes or no), Coronary heart disease (yes or no), CKD (yes or no), Hypertension (yes or no), HDL-C, TC, SHR, LDL-C, BMI, use of insulin (yes or no)

*P < 0.05

**P < 0.01

***P < 0.001

Fig. 2.

Nonlinear correlation between the SHR and TyG with depressive symptoms in individuals with diabetes and prediabetes.(a) Nonlinear relationship between SHR and the risk of depressive symptoms in individuals with diabetes and prediabetes.(b) Nonlinear relationship between TyG and the risk of depressive symptoms in individuals with diabetes and prediabetes.The solid red line represents the smooth curve fit between variables. Blue bands represent the 95% confidence interval from the fit.

Association between TyG and depressive symptoms with diabetes and prediabetes

Results are shown in Table 3. The TyG index was positively correlated with depressive symptoms in multiple models. In the fully adjusted model (Model 3), each 1-unit increase in TyG significantly increased the likelihood of depressive symptoms (OR = 5.37, 95% CI: 3.80 7.59). Quartile analysis revealed that the Q4 group had a substantially higher likelihood of depressive symptoms compared to Q1 (OR = 3.56, 95% CI: 2.38 5.33). In Model 4, after adjusting for SHR, the likelihood of depressive symptoms further increased with each 1-unit increase in TyG (OR = 7.46, 95% CI: 5.00 11.11). Quartile analysis showed that, after full adjustment, the Q4 group had the highest likelihood of depressive symptoms compared to Q1 (OR = 3.89, 95% CI: 2.52 6.00). Smoothed curve analysis revealed a nonlinear positive association between TyG and depressive symptoms in diabetes and prediabetes. The risk of depressive symptoms increased gradually at lower TyG levels but steepened significantly when TyG exceeded 9.5. The widening confidence intervals at higher TyG levels indicated increased variability, highlighting a stronger link between elevated TyG and depressive symptoms risk.

Table 3.

Saturation effect analysis of SHR on diabetes with depressive symptoms

Diabetes with depressive symptoms	Model: saturation effect analysis OR (95%CI) P value
SHR turning point (K)	16.07
< K, effect1	0.78 (0.71, 0.87) < 0.0001
>K, effect2	1.08 (1.01, 1.16) 0.0283
Difference in effect of 2 and 1	1.38 (1.20, 1.59) < 0.0001
Log-likelihood ratio	< 0.0001

age, sex, race, education, marriage status, BMI, smoking, drinking, HBP, HDL, LDL, TC, TyG, Coronary heart disease, Congestive heart failure, CKD, use of insulin were adjusted

LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol; TC, Total cholesterol; HBP hypertension

Predictive value of SHR and TyG for depressive symptoms in diabetes and prediabetes: ROC curve analysis

We also applied ROC to explore the predictive value of TyG and SHR for diabetes mellitus comorbid depressive symptoms. The AUC for TyG was 0.663, while SHR had an AUC of 0.534, suggesting TyG provides a stronger prediction for depressive symptoms in individuals with insulin resistance (as shown in Fig. 3).

Fig. 3.

Receiver operating characteristic curve (ROC) analyses for the prediction of depressive symptoms with diabetes and prediabetes

Subgroup analysis

The consequences of the subgroup analyses and interaction tests are shown in Fig. 4. In patients who have diabetes with depressive symptoms, SHR levels did not differ in most of the pre-specified subgroups except for insulin use and TyG quartile. In patients without insulin (OR = 1.09, 95% CI: 1.03–1.14), high SHR levels were independently associated with depressive symptoms, whereas in patients with insulin, high SHR levels were not associated with depressive symptoms. Additionally, the association between SHR and diabetic depression strengthened as TyG levels increased from Q1 to Q4, with the strongest effect observed in the highest TyG group (Q4) (OR = 1.42, 95% CI: 1.21 1.66).

Fig. 4.

Subgroup analysis for the association between SHR and depressive symptoms with diabetes and prediabetes

Discussion

Our preliminary cross-sectional study, focusing on a representative group of US adults, revealed a significant non-linear association between stress hyperglycemic response (SHR), triglyceride-glucose product index (TyG), and the prevalence of depressive symptoms in individuals with diabetes. Fully adjusted models revealed a significant association between SHR and depressive symptoms in individuals with diabetes, with those in the highest SHR quartile showing a higher likelihood of depressive symptoms compared to those in the lowest quartile (OR = 1.34, 95% CI: 1.01, 1.79). Similarly, TyG exhibited a strong association with depressive symptoms, with individuals in the highest TyG quartile having a greater likelihood of depressive symptoms than those in the lowest quartile (OR = 3.56, 95% CI: 2.38, 5.33).

Notably, the association between SHR and depressive symptoms shifted after adjusting for TyG in Model 4 of Table 3, becoming negatively significant (OR = 0.92, 95% CI: 0.87–0.97). In contrast, the effect of TyG was amplified after adjusting for SHR in Model 4 of Table 4. In Model 4, after adjusting for TyG, the association between SHR and depressive symptoms shifted from positive to negative. This likely reflects the distinct roles of SHR and TyG in metabolic and mental health. TyG, a well-established marker of insulin resistance, represents chronic metabolic dysfunction, which is closely linked to depression and other mood disorders [25]. The negative correlation observed after adjusting for TyG suggests that insulin resistance might be a critical factor influencing the relationship between SHR and depressive symptoms. Specifically, TyG may mediate or confound the relationship between SHR and depressive symptoms, underscoring the complex interplay between stress hyperglycemia, insulin resistance, and mental health [26]. These findings suggest that the role of SHR may differ based on an individual’s level of insulin resistance. In individuals with higher TyG (indicating greater insulin resistance), the stress hyperglycemic response may not exacerbate depressive symptoms, but instead may reflect a more adaptive metabolic response to stress.

Subgroup analysis showed that SHR levels did not differ significantly in most pre-specified subgroups, except for insulin use and TyG quartile. Additionally, a negative correlation was observed between SHR values below 16.07 and depressive symptoms in diabetes. This suggests that within a certain range, elevated SHR may have a protective effect, with better stress regulation and less impact on mental health when SHR is below the threshold (e.g., 16.07). This threshold effect indicates that excessive stress hyperglycemia could exacerbate depressive symptoms [27], while moderate blood glucose levels may help maintain better mental health [28].

In addition, research has revealed a complex relationship between metabolic health and mental health. SHR levels below a certain threshold may reflect an individual’s better metabolic status, including good glycaemic control and lower insulin resistance, factors that contribute to a lower risk of depression associated with metabolic disorders [29]. Metabolically healthy individuals typically exhibit greater brain function and mood regulation, which is closely related to insulin action, neurotransmitter homeostasis, and immune system regulation [5, 30].

Previous studies have confirmed that high blood sugar can indeed have a negative impact on brain function in a variety of areas, including cognitive function, neurological health, and mood states [31, 32]. Acute hyperglycemia was linked, in one study, to modest cognitive impairment in individuals with diabetes [33]. Another study found that people with type 2 diabetes experience severe mood swings as a result of acute hyperglycemia, and that these patients experience higher levels of anxiety and despair during hyperglycemic episodes [27]. Research has also shown that sadness in individuals with diabetes might worsen depressive symptoms by increasing the risk of stress hyperglycemia, poor glycemic control, and poor treatment adherence [34]. This bidirectional relationship allows the two to influence each other in a vicious cycle. Our findings complement, and together with previous findings, reveal a strong link between stress hyperglycemia and diabetes with depressive symptoms.

The present study provides evidence of an association between SHR and depressive symptoms in individuals with diabetes. However, as this is a cross-sectional analysis, the exact mechanisms underlying this association remain unclear. Previous clinical and epidemiological studies suggest that depression and metabolic disturbances in diabetes are linked to hyperactivation of the hypothalamic–pituitary–adrenal (HPA) axis and elevated cortisol secretion. Alvarez et al. (2013) reported higher urinary free cortisol levels in diabetic patients with comorbid depression, consistent with subclinical hypercortisolism due to HPA axis overactivation [35]. Prolonged exposure to elevated cortisol levels can lead to hippocampal atrophy and impaired negative feedback regulation [36]. In addition, insulin resistance may exacerbate glucocorticoid receptor resistance, thereby further impairing the regulatory feedback mechanisms of the HPA axis [37]. Elevated cortisol levels may lead to an imbalance in neurotransmitters, particularly 5-hydroxytryptamine (5-HT) and norepinephrine (NE), which play a key role in mood regulation [38]. Research has demonstrated a correlation between insulin resistance and the onset of depression [26]. Insulin resistance alters insulin signaling pathways in the brain, which may have an impact on depressive symptoms in addition to glucose metabolism [39]. Inflammatory responses may also be a potential mechanism by which SHR affects depressive symptoms in diabetes [40]. Levels of inflammatory markers (e.g., CRP, TNF-α, IL-6) are typically higher in depressed patients and may affect the brain’s neurotransmitter system, leading to depression [41–43]. It has been found that chronic hyperglycemia and stress may affect the brain’s neuroplasticity, such as causing atrophy of the hippocampus, which is strongly associated with the development of depressive symptoms [44–46].

In this case, SHR was found to be significantly associated with the occurrence of depressive symptoms among diabetic patients. This result remains significant in specific populations. For example, the correlation between the two remains significant in the age group < 60 years and obese group (BMI ≥ 30). Studies in diabetic patients have shown that SHR is essentially associated with all-cause and cardiovascular mortality. This result was similarly confirmed in a retrospective cohort study of US adults, and the association was consistent across age, sex, and racial subgroups [47]. In addition, another study noted a U-shaped relationship between SHR and diabetes mortality, implying that within a certain range, either too high or too low SHR increases the risk of death [16]. This shows that SHR has an equally important role in predicting cardiovascular disease as it does in promoting the mental health of diabetes individuals.

In summary, the relationship between SHR and mental health is not a simple linear relationship, but is influenced by multiple factors including individual differences, metabolic health, immune response and neurotransmitters. SHR levels below 16.07 may play a protective role in improving mental health through multiple pathways. However, future studies are needed to further explore the relationship and its underlying mechanisms. We hope this explanation provides clarity on the unexpected negative association observed in this group.

Limitation

This research possesses several strengths. Firstly, it encompasses a sizable sample from a multiethnic population, representative of the national demographic, thus enhancing the generalizability of our findings. Secondly, this study expands the application of SHR as a prognostic indicator beyond cardiovascular disease to include mental health in patients with diabetes. Our data suggest that clinicians should consider both glycemic control and mental health in managing diabetic patients, taking SHR levels into comprehensive consideration.

However, some limitations must be acknowledged. The PHQ-9 is a self-reported scale, meaning its accuracy depends on patients’ self-assessment and honesty. Patients may underreport or overreport their symptoms, potentially affecting diagnostic accuracy. Moreover, although we accounted for some potential covariates during data analysis, other potential confounders—such as the use of medications like anxiolytics/antidepressants, hormonal metabolism abnormalities, and other social and environmental variables—could not be entirely excluded. In addition, because the NHANES database does not always allow for a clear differentiation between type 1 diabetes (T1D) and type 2 diabetes (T2D), we were unable to distinguish between these subtypes in our analyses. This limitation should be acknowledged, as the inflammatory mechanisms underlying T1D (autoimmune-related) and T2D (innate immunity-related) differ, and this could have influenced the interpretation of our findings. Furthermore, while SHR and TyG were both significantly associated with depressive symptoms, their discriminative performance—as indicated by the relatively low AUC values (e.g., SHR: 0.534; TyG: 0.663)—was limited. This suggests that although these markers may reflect a link between metabolic dysfunction and depressive symptoms at the population level, they lack sufficient predictive power to serve as standalone clinical tools for detecting depressive symptoms in individual patients. Last but not least, the study’s cross-sectional design makes it difficult to pinpoint precise causal relationships between exposure parameters and outcome variables. Future studies could explore the role of TyG and SHR in diabetes-related depressive symptoms more comprehensively by investigating underlying mechanisms, designing prospective cohort studies, developing targeted interventions, and constructing predictive models. These efforts could deepen our understanding of their clinical implications, promote practical applications, and ultimately improve both the metabolic and psychological health of diabetic patients.

Conclusion

This study reveals a U-shaped relationship between SHR and depressive symptoms in individuals with diabetes, with an inflection point at 16.07. A similar non-linear association was observed between TyG and depression. Both markers show promise for identifying depressive symptoms in diabetes and prediabetes, with TyG slightly outperforming SHR. However, their overall predictive value remains limited.

Abbreviations

NHANES

National health and nutritional examination survey

NCHS

National center for health statistics

CDC

Centers for disease control and prevention

IDF

International diabetes federation

SHR

Stress hyperglycemic ratio

TyG

Triglyceride-Glucose index

BMI

Body mass index

PHQ-9

Patient health questionnaire-9

FPG

Fasting blood glucose

HbA1c

Glycosylated hemoglobin

TC

Total cholesterol

HDL-C

High-density lipoprotein cholesterol

LDL-C

Low-density lipoprotein cholesterol

5-HT

Particularly 5-hydroxytryptamine

NE

Norepinephrine

CKD

Chronic kidney diseases

Q1

Quartile 1

Q2

Quartile 2

Q3

Quartile 3

Q4

Quartile 4

Author contributions

Mingzhu Wang contributed to the design of the study and wrote the manuscript. Mingzhu Wang, Chengchao Peng conceptualized, collected and analyzed data and carried out the initial analyses, Tingting Jiang and Chong Lu collected, analyzed and interpreted the data. Danping Li reviewed and revised the manuscript. Min Lu critically reviewed, edited and approved the manuscript.

Funding

Not applicable.

Data availability

More information about the NHANES could be obtained at: http://www.cdc.gov/nhanes.

Declarations

Ethics approval and consent to participate

The ethics committee of the National Center for Health Statistics gave its permission. Every procedure followed the applicable rules and regulations (the Helsinki Declaration). Prior to their participation in the study, every subject gave written, informed consent.

Consent for publication

All authors have read and agreed to the published version of the manuscript.

Competing interests

The authors declare no competing interests.