Triglyceride–glucose index and psychotic symptoms in major depressive disorder: a cross-sectional study of first-episode, drug-naïve outpatients

Psychotic symptoms (SP) in major depressive disorder (MDD) portend greater severity and adverse outcomes. The triglyceride–glucose (TyG) index, a pragmatic surrogate of insulin resistance, may be informative for SP risk, but evidence within MDD remains limited. In a cross-sectional cohort of first-episode, drug-naïve MDD outpatients ( N = 1,718), the study outcome was defined as PANSS-defined psychotic symptom burden, operationalized as a PANSS positive subscale score ≥ 15, rather than as a DSM-IV diagnosis of MDD with psychotic features. Continuous predictors were z-standardized. Multivariable logistic regression estimated adjusted associations between TyG and SP while controlling for sex, age, illness duration, BMI, SBP, DBP, HAMD, and HAMA. Stratified models examined sex and onset-age differences. Exploratory indirect-effect analysis was performed using a product-of-coefficients approach with 5,000 nonparametric bootstrap resamples to estimate confidence intervals for indirect associations; because all variables were measured at baseline, these models were used to quantify statistical indirect associations rather than to infer temporal or causal pathways. SP prevalence was 9.9% (171/1,718). Higher TyG was independently linked to increased odds of SP (per + 1 SD: OR = 1.35, 95% CI 1.07–1.70; p = 0.011). Sensitivity analyses excluding participants with diabetic-range fasting glucose or BMI-defined obesity yielded generally consistent estimates. The associations were significant in females (OR = 1.35, 95% CI 1.02–1.79; p = 0.034) but not in males (OR = 1.38, 95% CI 0.91–2.09; p = 0.134), with a larger point estimate in late-onset cases (OR = 1.52, 95% CI 0.97–2.38; p = 0.067). Exploratory bootstrap indirect-effect analyses showed a very small statistically detectable TyG-indexed indirect association for HAMA (OR = 1.038, 95% CI 1.014–1.075) and a very small borderline indirect association for HAMD (OR = 1.042, 95% CI 0.999–1.093), accounting for only 1.7% and 2.5% of the corresponding total associations, respectively; both estimates were interpreted as clinically limited exploratory signals. TyG is independently linked to SP in MDD, with the association statistically evident in women and showing a suggestively larger point estimate in late-onset depression. TyG may serve as an adjunctive psychometabolic marker for SP risk characterization in MDD; however, the very small indirect-effect estimates should be considered hypothesis-generating and not clinically actionable on their own. Further longitudinal studies are needed to assess its prognostic and interventional value.