Disease duration-stratified associations of thyroid hormone levels with psychopathology in hospitalized patients with schizophrenia: A cross-sectional study

Abstract

BACKGROUND

Neuroendocrine dysfunction, especially involving the hypothalamic-pituitary-thyroid axis, plays a critical role in the onset and progression of schizophrenia. Alterations in thyroid stimulating hormone (TSH), triiodothyronine (T3), free T3 (FT3), thyroxine (T4), and free T4 have been implicated in this process. Although previous studies have established an association between thyroid function and psychiatric symptoms, how thyroid hormone levels vary with disease duration remains underexplored.

AIM

To investigate duration stage-specific associations between thyroid hormones and psychotic symptoms among inpatients with stable schizophrenia.

METHODS

This cross-sectional study was conducted at Zigong Mental Health Center, China, and included 237 hospitalized patients with stable schizophrenia. Participants were stratified into three groups based on disease duration: 0-10 years, 10.1-20 years, and over 20 years. Peripheral blood samples were collected to measure serum thyroid hormone levels. Psychotic symptoms were assessed using the Positive and Negative Syndrome Scale. Covariate-adjusted linear regression analyses were performed to assess the relationships between thyroid hormone levels and Positive and Negative Syndrome Scale sub-scale scores.

RESULTS

The relationship between thyroid hormones and psychotic symptoms varied by disease duration. In patients with a disease course of 0-10 years, T4 [β = -0.848; 95% confidence interval (CI): -1.564 to -0.133; P = 0.021] and FT3 (β = -2.483; 95%CI: -4.693 to -0.273; P = 0.028) levels were significantly inversely associated with general psychopathology scores. Among those with 10.1-20 years of disease, only TSH showed a significant negative correlation with general psychopathology (β = -1.429; 95%CI: -2.348 to -0.509; P = 0.003). No significant correlations were found in the > 20 years group.

CONCLUSION

The associations between thyroid hormones and psychotic symptoms vary according to the duration of schizophrenia (T4/FT3 early; TSH mid), enabling the development of stage-adapted models and management.

Key Words: Thyroid hormones; Schizophrenia; Symptom; Disease duration; Positive and Negative Syndrome Scale

Core Tip: Although associations between thyroid function and psychopathology have been demonstrated, the link with disease stage remains unclear. This study explored the relationships between thyroid function and psychiatric symptoms in different stages of the disease to elucidate stage-specific interactions, aiming to expand the understanding of neuroendocrine pathways and enable personalized management for schizophrenia.

INTRODUCTION

Schizophrenia is a persistent and disabling mental disease characterized by enduring psychotic symptoms, including hallucinations, delusions, and disorganized thought processes[1]. Its chronic course often requires ongoing clinical care, frequently within inpatient facilities, placing significant strain on patients, their families, and healthcare systems[2-4]. Although pharmacological treatments have advanced, achieving personalized care remains challenging due to the disorder’s complex underlying mechanisms and the lack of dependable biomarkers.

Emerging evidence increasingly points to neuroendocrine dysfunction, particularly disturbances in the hypothalamic-pituitary-thyroid (HPT) axis, as a contributing factor in the pathogenesis of schizophrenia[5-7]. Beyond their role in metabolic homeostasis, thyroid hormones play essential neuromodulatory roles, influencing central nervous system development, neurotransmitter regulation, and synaptic activity[8,9]. Clinical findings indicate a high prevalence of thyroid dysfunctions such as hypothyroidism and elevated thyroid autoantibodies among patients with both first-episode and chronic schizophrenia[10,11]. Moreover, deviations in thyroid hormone levels have been associated with more severe psychiatric symptoms, cognitive impairments, and an elevated risk of suicide in this population[12-14].

Recent evidence has emphasized the potential relevance of thyroid abnormalities as biomarkers associated with schizophrenia spectrum disorders. Jiang et al[15] reported that patients with schizophrenia exhibit significantly lower levels of triiodothyronine (T3), thyroxine (T4), free T3 (FT3), free T4 (FT4), and thyroid stimulating hormone (TSH) compared to healthy controls. T3, T4, and TSH levels were associated with the severity of clinical symptoms, supporting the relevance of thyroid function in the clinical profile of schizophrenia. However, schizophrenia is a progressive condition characterized by ongoing neurobiological changes such as chronic inflammation, oxidative stress, and metabolic dysregulation that may lead to compensatory shifts in endocrine activity over time[16,17]. Existing studies have not explored changes in thyroid hormone levels in terms of disease duration and their specific associations with psychiatric symptoms.

To address gaps in existing research, this study used cross-sectional data to explore the relationship between thyroid hormone levels and psychotic symptoms in individuals with schizophrenia, with a particular focus on potential stage-specific differences related to disease durations. For this exploration, stable inpatients with schizophrenia were categorized into three groups based on disease duration (0-10 years, 10.1-20 years, and > 20 years). These groups were analyzed to elucidate how the thyroid hormone-psychopathology relationship evolves over the course of the disease. These insights could potentially inform the development of stage-specific pathophysiological models and tailored management approaches.

MATERIALS AND METHODS

Study design and participants

This cross-sectional study analyzed data collected in 2023 as part of an ongoing multi-cohort longitudinal investigation initiated by the Department of Psychiatry at Zigong Mental Health Center in Zigong, China. The study adhered to the ethical principles outlined in the Declaration of Helsinki and received approval from the institutional ethics committee, approval No. 2023024. Diagnoses of schizophrenia were made by board-certified psychiatrists in accordance with the criteria set forth in the International Classification of Diseases, 10^th Revision. Stable schizophrenia was defined as the absence of symptom exacerbation for over one month and a stable medication regimen for two or more months prior to study enrollment.

Sample size calculations were not conducted as this study employed a census sampling approach. The target sample comprised 325 individuals, recruited and enrolled by consecutively including all schizophrenia inpatients admitted to the center between August 1 and 31, 2023. Inclusion criteria for this study were: (1) Age ≥ 18 years; (2) A psychiatrist-confirmed diagnosis of stable schizophrenia; and (3) Provision of written informed consent to participate. Exclusion criteria were: (1) Absence of a stable schizophrenia diagnosis; (2) Inability or unwillingness to provide informed consent; (3) Evidence of significant hepatic or renal dysfunction; (4) History of autoimmune disorders or active cancer treatment; and (5) Missing data on thyroid hormone measurements.

After applying these criteria, 237 participants were retained for final analysis. Comprehensive clinical data including demographic information, physical measurements, cognitive and psychiatric assessments, and biochemical profiles were collected within 2-3 days following routine blood testing.

Psychiatric and cognitive function evaluation

Psychiatric symptom was evaluated using the Positive and Negative Syndrome Scale (PANSS)[18], a widely recognized instrument for clinical assessment in schizophrenia. The PANSS comprises 30 items distributed across three sub-scales: Positive symptoms, negative symptoms, and general psychopathology[19]. Each item is rated on a 7-point scale, ranging from 1 (absent) to 7 (extreme), with higher scores indicating more severe psychopathology. Anxiety symptoms and their functional impact over the preceding two weeks were measured using the Generalized Anxiety Disorder 7-item (GAD-7) scale[20], while depressive symptoms were assessed via the Patient Health Questionnaire-9 (PHQ-9)[21].

Cognitive performance was assessed using the Montreal Cognitive Assessment-Chinese version (MoCA-C)[22], which has demonstrated validity in populations with cognitive impairments, including individuals with schizophrenia[23,24]. The MoCA-C is a 30-point instrument designed for rapid screening, with administration typically requiring no more than 15 minutes. Lower scores reflect greater cognitive impairment. All evaluations were conducted by trained and experienced psychiatrists.

Thyroid hormone measurements

Serum concentrations of thyroid hormones (T3, T4, FT3, FT4, TSH) were extracted from the hospital’s electronic patient record system. Thyroid hormone quantification was performed as follows: Fasting venous blood samples (2-3 mL) were collected using sterile vacuum tubes. The samples were left to clot at room temperature for 30 minutes, after which they were centrifuged to isolate serum. The resulting serum was used for biochemical analysis of thyroid hormone levels.

Covariates

Covariate data were gathered through patient self-reports, electronic patient records, and standardized assessments. These variables included demographic, clinical, and psychopathological characteristics of patients with stable schizophrenia. Demographic parameters encompassed age, sex, educational attainment, body mass index (BMI), and marital status (married, single, widowed, or divorced). Clinical information included age at onset (years), disease duration (in years), chlorpromazine equivalent dose (mg/day), length of hospitalization (in months), family psychiatric history, whether the case was a first episode, antipsychotics types (typical, atypical, or combined use), substance use history (smoking and alcohol consumption), presence of chronic diseases, vision impairment, hearing impairment, falls history, coronavirus disease 2019 (COVID-19) infection, GAD-7, PHQ-9, and MoCA-C scores.

Statistical analysis

Participants were categorized into three groups based on illness duration: 0-10 years (n = 58), 10.1-20 years (n = 74), and over 20 years (n = 105). Given that antipsychotic treatment can affect thyroid hormone levels[25,26], subgroup analysis by specific medication type was considered. However, over 10 different antipsychotics, either used alone or in combination, were prescribed within each group, resulting in insufficient sample sizes for valid statistical comparisons between medication-based subgroups. Therefore, stratification by antipsychotic class was not performed.

All statistical analyses were conducted using SPSS version 25.0 (IBM Corp., Armonk, NY, United States). Two-tailed tests were applied, with a significance threshold set at P < 0.05. Categorical variables were expressed as frequencies and percentages, while non-normally distributed continuous variables were reported as medians and interquartile ranges (IQRs).

Comparisons of thyroid hormone levels across disease-duration groups were conducted using Mann-Whitney U tests with Bonferroni correction (adjusted α = 0.017, based on three pairwise tests). Spearman correlation coefficients were calculated to explore associations between thyroid hormone levels and age, disease duration, antipsychotic dosage, and PANSS sub-scale scores. Based on these correlations, significant relationships were identified: In the 0-10 years group, T3, T4, and FT3 were significantly associated with general psychopathology scores; in the 10.1-20 years group, significant associations emerged between TSH and general psychopathology scores and between FT4 and total PANSS scores (all P < 0.05).

To further investigate these associations, fully adjusted linear regression models were constructed. For the 0-10 years group, models assessing T3 and T4 as primary predictors were adjusted for antipsychotic medication type and COVID-19 history, while the FT3 model, in addition, controlled for age due to a significant FT3-age correlation. In the 10.1-20 years group, models with FT4 and TSH as predictors were adjusted for sex, length of hospitalization, chlorpromazine-equivalent dosage, PHQ-9, GAD-7, and MoCA-C scores. Multicollinearity was evaluated using the variance inflation factor (VIF), with values ≥ 5 indicating problematic collinearity[27]; all included variables showed variance inflation factors < 5, supporting model stability. Data visualization was performed using Python’s matplotlib (v3.10.0) for box plots and SPSS 25.0 for scatter plots.

RESULTS

Characteristics of schizophrenia inpatients

Of the 325 patients initially screened, 88 were excluded for the following reasons: Age under 18 years (n = 12), presence of severe hepatic or renal dysfunction or active cancer treatment (n = 20), absence of thyroid hormone data (n = 18), failure to meet criteria for clinical stability (n = 8), and refusal to participate (n = 30). Therefore, the final analysis included 237 patients diagnosed with stable schizophrenia.

Table 1 presents the demographic and clinical characteristics of the enrolled participants. Of the 237 patients, 58.2% (n = 138) were male and 41.8% (n = 99) were female. The median (IQR) values were as follows: Age: 50 (40-56) years, BMI: 24.49 (21.88-27.53) kg/m², disease duration 30 (24-34) years, length of hospitalization 26 (8-51.5) months, age at onset 26 (20-35) years, chlorpromazine-equivalent dose 325 (200-550) mg/day, PHQ-9 score 3 (1-5), GAD-7 score 1 (0-3), MoCA-C score 18 (11.5-23), total PANSS score 61 (51-72), positive symptom score 13 (9-17), negative symptom score 18 (14-23), and general psychopathology score 29 (25-35).

Table 1 General information of enrolled in patients with stable schizophrenia, n (%).

General information		Total participants (n = 237)
Age (years), median (IQR)		50 (40-56)
Male sex		138 (58.2)
BMI (kg/m²), median (IQR)		24.49 (21.88-27.53)
Marital status (married)		49 (20.7)
Education attainment	Illiterate	15 (6.3)
	High school and below	201 (84.8)
	University and above	21 (8.9)
Age at onset(years), median (IQR)		26 (20-35)
Disease duration (years), median (IQR)		30 (24-34)
Hospitalization duration (months), median (IQR)		26 (8-51.5)
Family psychiatric history (yes)		54 (22.8)
First episode (yes)		4 (1.7)
Substance use history (yes)	Smoking	97 (40.9)
	Alcohol use	59 (24.9)
Antipsychotics type	Typical	4 (1.7)
	Atypical	216 (91.1)
	Combined	17 (7.2)
Chlorpromazine equivalent dose (mg/day), median (IQR)		325 (200-550)
Chronic diseases (≥ 1)		79 (33.3)
Vision impairment		32 (13.5)
Hearing impairment		16 (6.8)
Falls history (yes)		15 (6.3)
COVID-19 infection history (yes)		94 (39.7)
PHQ-9 (depression), median (IQR)		3 (1-5)
GAD-7 (anxiety), median (IQR)		1 (0-3)
MoCA-C (cognitive function), median (IQR)		18 (11.5-23)
PANSS profile-total score, median (IQR)		61 (51-72)
PANSS profile-positive symptoms, median (IQR)		13 (9-17)
PANSS profile-negative symptoms, median (IQR)		18 (14-23)
PANSS profile-general psychopathology, median (IQR)		29 (25-35)
TSH (μIU/mL), median (IQR)		2.36 (1.78-3.47)
FT3 (pmol/L), median (IQR)		4.89 (4.44-5.27)
FT4 (pmol/L), median (IQR)		15.73 (14.14-17.68)
T3 (ng/mL), median (IQR)		1.08 (0.97-1.22)
T4 (μg/dL), median (IQR)		7.8 (6.91-8.8)

BMI: Body mass index; COVID-19: Coronavirus disease 2019; IQR: Interquartile range; PHQ-9: Patient Health Questionnaire-9; GAD-7: Generalized Anxiety Disorder 7-item; MoCA-C: Montreal Cognitive Assessment-Chinese version; PANSS: Positive and Negative Syndrome Scale; TSH: Thyroid stimulating hormone; T3: Triiodothyronine; FT3: Free triiodothyronine; T4: Thyroxine; FT4: Free thyroxine.

Regarding sociodemographic and clinical variables, 20.7% of participants were married, and 91.1% had educational attainment ranging from illiterate to high school level. A family history of psychiatric illness was reported by 22.8% of patients, while only 1.7% were experiencing a first episode. Smoking and alcohol use were reported by 40.9% and 24.9% of patients, respectively. The vast majority (92.8%) were being treated with either typical or atypical antipsychotic medications. Furthermore, 33.3% of the cohort had chronic physical comorbidities, 13.5% had vision impairment, 6.8% had hearing impairment, 6.3% reported a history of falls, and 39.7% had previously contracted COVID-19.

Median (IQR) serum concentrations of thyroid hormones were: TSH: 2.36 (1.78-3.47) μIU/mL, FT3: 4.89 (4.44-5.27) pmol/L, FT4: 15.73 (14.14-17.68) pmol/L, T3: 1.08 (0.97-1.22) ng/mL, and T4: 7.8 (6.91-8.8) μg/dL. Moreover, as shown in Figure 1, analysis of thyroid hormone concentrations across disease duration groups revealed a significant difference for FT4 Levels. FT4 Levels differed significantly between the 10.1-20 years and > 20 years groups (P = 0.007).

Figure 1 Box plots of intergroup comparisons of thyroid hormone concentrations between groups with different disease duration. ^aP < 0.017. A: Triiodothyronine concentrations; B: Free triiodothyronine concentrations; C: Thyroxine concentrations; D: Free thyroxine concentrations; E: Thyroid stimulating hormone concentrations, each compared across the three groups. TSH: Thyroid stimulating hormone; T3: Triiodothyronine; FT3: Free triiodothyronine; T4: Thyroxine; FT4: Free thyroxine.

Correlations between thyroid hormone levels and age, disease duration, and antipsychotic dose

Table 2 summarizes the results of the Spearman correlation analysis examining associations between thyroid hormone levels and age, disease duration, and antipsychotic dosage in patients with stable schizophrenia. In the overall sample, age was negatively correlated with FT3 (r = -0.174), while disease duration was positively associated with FT4 (r = 0.155) and T4 (r = 0.177). Moreover, the chlorpromazine-equivalent dose was inversely correlated with FT4 (r = -0.130), T3 (r = -0.273), and T4 (r = -0.192), with all correlations reaching statistical significance (P < 0.05).

Table 2 Correlations between thyroid hormone levels and age, disease duration, and antipsychotic dose in stable schizophrenia patients, stratified by disease duration.

Variable	Total participants			0-10 years group			10.1-20 years group			> 20 years group
	Age	Disease duration	Chlorpromazine equivalent dose	Age	Disease duration	Chlorpromazine equivalent dose	Age	Disease duration	Chlorpromazine equivalent dose	Age	Disease duration	Chlorpromazine equivalent dose
TSH	0.021	0.013	0.02	0.024	-0.21	0.058	0.033	0.103	-0.076	0.036	0.031	0.068
FT3	-0.174b	0.001	-0.08	-0.263a	-0.106	-0.013	-0.194	0.089	-0.022	-0.206a	-0.012	-0.157
FT4	0.079	0.155a	-0.130a	-0.131	-0.06	-0.107	-0.059	0.024	-0.092	0.055	-0.058	-0.134
T3	0.021	0.118	-0.273b	-0.055	0.061	-0.135	0.026	0.289a	-0.230a	-0.084	0.019	-0.365b
T4	0.04	0.177b	-0.192b	-0.023	0.182	-0.166	-0.105	0.246a	-0.074	-0.005	0.043	-0.246a

^aP < 0.05.

^bP < 0.001.

The P value obtained by comparing the three groups stratified by disease duration. TSH: Thyroid stimulating hormone; T3: Triiodothyronine; FT3: Free triiodothyronine; T4: Thyroxine; FT4: Free thyroxine.

Stratified analyses by disease duration revealed distinct patterns. Age was significantly associated with FT3 in both the 0-10 years (r = -0.263) and > 20 years (r = -0.206) subgroups. Disease duration showed positive correlations with T3 (r = 0.289) and T4 (r = 0.246) in the 10.1-20 years group. Furthermore, the chlorpromazine-equivalent dose was negatively associated with T3 in both the 10.1-20 years (r = -0.230) and > 20 years (r = -0.365) groups, as well as with T4 in the > 20 years group (r = -0.246), all with P < 0.05. No significant correlations were observed between TSH, FT4, T3, or T4 and age across the sample. Further, TSH and FT3 were not significantly associated with either disease duration or antipsychotic dosage in any subgroup.

Correlations between thyroid hormones and PANSS profiles

Table 3 displays the Spearman correlation results between thyroid hormone levels and PANSS scores across different disease duration groups. In the 0-10 years group, FT3 (r = -0.385), T3 (r = -0.291), and T4 (r = -0.298) were significantly negatively correlated with general psychopathology scores (all P < 0.05). In the 10.1-20 years group, TSH (r = -0.310) was significantly associated with general psychopathology scores, and FT4 (r = -0.248) was negatively correlated with total PANSS scores (both P < 0.05). No significant associations were found between thyroid hormone levels and any PANSS sub-scale scores in the > 20 years group.

Table 3 Correlational analysis of thyroid hormone levels and Positive and Negative Syndrome Scale profile scores in patients with stable schizophrenia.

Variable	0-10 years group				10.1-20 years group				> 20 years group
	Total score	Positive symptoms	Negative symptoms	General psychopathology	Total score	Positive symptoms	Negative symptoms	General psychopathology	Total score	Positive symptoms	Negative symptoms	General psychopathology
TSH	-0.027	-0.109	0.084	-0.057	-0.219	-0.185	-0.096	-0.310b	-0.114	-0.014	-0.123	-0.13
FT3	-0.207	-0.041	-0.116	-0.385b	0.07	-0.055	0.121	0.091	-0.001	0.081	-0.058	-0.065
FT4	-0.044	-0.003	-0.106	-0.089	-0.248a	-0.213	-0.169	-0.165	0.001	0.02	-0.009	0.033
T3	-0.102	-0.145	0.091	-0.291a	0.006	0.089	0.139	0.06	0.081	0.053	0.077	0.015
T4	-0.224	-0.158	-0.177	-0.298a	-0.115	-0.195	0.11	-0.101	-0.014	-0.09	0.015	0.007

^aP < 0.05.

^bP < 0.001.

TSH: Thyroid stimulating hormone; T3: Triiodothyronine; FT3: Free triiodothyronine; T4: Thyroxine; FT4: Free thyroxine.

Potential covariates for regression modeling were identified through univariate analyses stratified by disease duration (Table 4). In the 0-10 years group, antipsychotic type (β = -0.416) and COVID-19 history (β = -0.258) were both negatively associated with general psychopathology scores (P < 0.05). In the 10.1-20 years group, several variables, including sex, hospitalization duration, chlorpromazine-equivalent dose, PHQ-9, GAD-7, and MoCA-C scores, were significantly associated with both total PANSS and general psychopathology scores (P < 0.05 for all). Hospitalization duration showed a negative relationship with both scores (β = -0.027 for total; β = -0.248 for general), whereas sex, antipsychotic dose, PHQ-9, and GAD-7 were positively associated with both outcomes (β range: 0.247-0.336). MoCA-C scores were inversely associated with total (β = -0.434) and general psychopathology scores (β = -0.410) (P < 0.05 for both).

Table 4 Univariate analysis of general information and Positive and Negative Syndrome Scale profile scores in patients with stable schizophrenia.

Variable	PANSS profile
	Standardized β (disease duration: 0-10 years, n = 58)				Standardized β (disease duration: 10.1-20 years, n = 74)				Standardized β (disease duration: > 20 years, n = 105)
	Total score	Positive symptoms	Negative symptoms	General psychopathology	Total score	Positive symptoms	Negative symptoms	General psychopathology	Total score	Positive symptoms	Negative symptoms	General psychopathology
Age	0.136	0.04	0.173	0.095	-0.125	-0.128	-0.033	-0.152	0.16	0.24a	0.019	0.136
Sex	0.09	0.027	0.155	0.022	0.294a	0.136	0.253a	0.336a	0.167	0.153	0.129	0.133
BMI (kg/m²)	0.03	-0.055	0.156	-0.037	-0.026	-0.038	-0.022	-0.024	-0.1	0.007	-0.037	-0.176
Marital status	-0.135	-0.099	-0.212	0.001	-0.036	0.051	-0.056	-0.113	0.059	0.011	0.111	0.006
Education attainment	-0.109	-0.145	-0.167	0.059	0.081	0.157	-0.025	0.055	-0.247a	-0.127	-0.181	-0.263a
Hospitalization duration	-0.126	-0.091	-0.058	-0.143	-0.027a	-0.25a	-0.166	-0.248a	-0.253a	-0.102	-0.218a	-0.279a
Family psychiatric history	0.071	0.004	0.088	-0.067	-0.183	-0.112	-0.133	-0.202	0.087	0.038	0.036	0.133
First episode	-0.033	0.205	-0.058	-0.053	0.038	-0.138	0.169	0.067	-0.117	-0.008	-0.154	-0.094
Smoking	0.043	0.161	-0.138	0.089	0.164	0.216	0.062	0.091	-0.331b	-0.242a	-0.276a	-0.258a
Alcohol use	0.003	0.111	-0.138	0.043	0.062	0.066	0.061	0.012	-0.139	-0.094	-0.133	-0.092
Age at onset	0.138	0.049	0.169	0.100	-0.116	-0.079	-0.064	-0.151	0.056	0.176	-0.102	0.075
Antipsychotics type	-0.4b	-0.423b	-0.097	-0.416b	0.183	0.169	0.054	0.204	0.062	0.117	0.019	0.032
Chlorpromazine equivalent dose	-0.012	0.038	-0.071	0.009	0.264a	0.397b	-0.021	0.255a	-0.139	0.063	-0.193	-0.150
Chronic diseases	0.143	0.04	0.157	0.128	0.06	-0.093	0.18	0.057	0.088	0.078	0.04	0.081
Vision impairment	-0.032	0.057	-0.047	-0.079	0.029	0.008	-0.041	0.092	0.029	0.053	-0.025	0.054
Hearing impairment	0.116	0.008	0.111	0.143	0.168	0.216	0.291a	0.194	-0.025	-0.013	-0.062	0.022
Falls history	-0.035	-0.118	0.007	0.024	0.104	-0.079	0.13	0.199	0.116	0.148	0.064	0.085
COVID-19 infection history	-0.261a	-0.305a	-0.051	-0.258a	0.125	-0.021	0.203	0.102	-0.008	-0.054	0.023	0.009
PHQ-9 (depression)	0.007	-0.002	-0.1	0.119	0.262a	0.067	0.277a	0.302a	0.165	0.051	0.118	0.219a
GAD-7 (anxiety)	-0.036	-0.051	-0.194	0.163	0.247a	0.257a	0.08	0.285a	0.138	0.112	0.029	0.198a
MoCA-C (cognitive function)	-0.112	0.015	-0.299a	0.037	-0.434b	-0.123	-0.515b	-0.41b	-0.494b	-0.041	-0.589b	-0.481b

^aP < 0.05.

^bP < 0.01.

BMI: Body mass index; COVID-19: Coronavirus disease 2019; PHQ-9: Patient Health Questionnaire-9; GAD-7: Generalized Anxiety Disorder 7-item; MoCA-C: Montreal Cognitive Assessment-Chinese version.

Multiple linear regression results are summarized in Table 5. After adjusting for relevant covariates, T4 (β = 0.848; 95% confidence interval (CI): -1.564 to -0.133; P = 0.021) and FT3 (β = -2.483; 95%CI: -4.693 to -0.273; P = 0.028) remained significantly associated with lower general psychopathology scores in the 0-10 years group. In the 10.1-20 years group, only TSH was significantly associated with reduced general psychopathology scores (β = -1.429; 95%CI: -2.348 to -0.509; P = 0.003). Figure 2 illustrates scatter plots of thyroid hormone levels vs PANSS scores, including both raw values and standardized residuals from the regression models to assess model fit.

Figure 2 Scatter plots of thyroid hormone levels vs Positive and Negative Syndrome Scale subscale scores, showing unadjusted data and standardized residuals. A: Triiodothyronine (T3): Unadjusted; B: T3: Standardized residuals; C: Thyroxine (T4): Unadjusted; D: T4: Standardized residuals; E: Free T3: Unadjusted; F: Free T3: Standardized residuals; G: Free T4: Unadjusted; H: Free T4: Standardized residuals; I: Thyroid stimulating hormone: Unadjusted; J: Thyroid stimulating hormone: Standardized residuals. TSH: Thyroid stimulating hormone; T3: Triiodothyronine; FT3: Free triiodothyronine; T4: Thyroxine; FT4: Free thyroxine; PANSS: Positive and Negative Syndrome Scale.

Table 5 Correlations between thyroid hormone levels and Positive and Negative Syndrome Scale profile scores in patients with stable schizophrenia.

Variable		Total score		General psychopathology
		β (95%CI)	P value	β (95%CI)	P value
Disease duration: 0-10 years	T3	-	-	-5.457 (-13.899 to 2.985)	0.2
	T4	-	-	-0.848 (-1.564 to -0.133)	0.021
	FT3	-	-	-2.483 (-4.693 to -0.273)	0.028
Disease duration: 10.1-20 years	FT4	-0.981 (-2.139 to 0.177)	0.095	-	-
	TSH	-	-	-1.429 (-2.348 to -0.509)	0.003

Linear regression models assessing the association between thyroid hormone level and Positive and Negative Syndrome Scale profile scores. General psychopathology: For the subgroup with disease duration 0-10 years: When triiodothyronine or thyroxine were primary exposures, adjusted for antipsychotic type and coronavirus disease 2019 history; when free triiodothyronine was the primary exposure, adjusted for antipsychotic type, coronavirus disease 2019 history, and age. For the subgroup with disease duration 10.1-20 years: When thyroid stimulating hormone was the primary exposure, adjusted for sex, hospitalization duration, chlorpromazine equivalent dose, Patient Health Questionnaire-9 scores, Generalized Anxiety Disorder 7-item scores, and Montreal Cognitive Assessment-Chinese version scores. CI: Confidence interval; TSH: Thyroid stimulating hormone; T3: Triiodothyronine; FT3: Free triiodothyronine; T4: Thyroxine; FT4: Free thyroxine.

DISCUSSION

This study elucidates disease duration stage-specific associations between thyroid hormone levels and psychotic symptomatology in patients with stable schizophrenia. Distinct correlation patterns emerged across different disease duration. In 0-10-year group, T4 and FT3 Levels were inversely correlated the general psychopathology scores. Among those with a disease duration of 10.1-20 years, TSH was negatively associated with general psychopathology scores. No statistically significant relationships between thyroid hormone indices and clinical symptoms were identified in patients with over 20 years of disease duration. These findings highlight distinct patterns in thyroid hormone-psychiatric symptom associations across different disease stages, suggesting that stage-specific factors may play a role and indicating the need for further investigation into tailored models and management strategies for schizophrenia.

Previous research has identified several mechanisms underlying thyroid dysfunction in schizophrenia, including dysregulation of the HPT axis, disrupted secretory rhythms, inflammatory-immune imbalances, and age-related metabolic changes[5,7,28]. Reductions in the activity of deiodinase, known to be associated with aging[28], lead to impaired conversion of T4 to active T3, resulting in lower levels of FT3. This age-related physiological change likely explains the inverse correlation between age and FT3 observed in our study. Beyond aging, antipsychotic-induced interference with thyroid function represents another key factor. First-generation antipsychotics such as phenothiazines have been shown to reduce both TSH[29] and T4 Levels[30]. Second-generation agents have been linked to subclinical hypothyroidism, typically presenting with decreased FT3, FT4, T3, and T4 Levels, accompanied by elevated TSH levels that correlate with overall antipsychotic burden[26,31]. In line with these findings, our study demonstrated significant associations between antipsychotic dosage (chlorpromazine equivalent) and levels of FT4, T3, and T4 in the overall patient sample. However, TSH and FT3 Levels showed no significant correlation with dosage. This inconsistency may be attributable to the use of chlorpromazine equivalents as proxy measures rather than actual plasma drug concentrations, which are subject to individual variability in hepatic and renal metabolism[32,33].

Stratified analyses further revealed duration-dependent patterns. Age-related declines in FT3 were evident in both early-stage (0-10 years) and late-stage (> 20 years) patients but not in the mid-stage group (10.1-20 years). Antipsychotic dosage was associated with T3 or T4 Levels only in patients with over 10 years of disease duration. These observations collectively emphasize that the impact of aging and antipsychotic medications on thyroid function in schizophrenia does not remain constant over the course of the disease. The observed variability across different disease durations lends support to the concept of stage-specific endocrine regulation and suggests that future therapeutic and management paradigms should be adapted according to the patient’s disease stage.

Various studies have demonstrated the essential regulatory role of thyroid hormones in the neurobiology of psychiatric disorders, including schizophrenia. These hormones influence key processes such as neural development, synaptic plasticity, and the modulation of major neurotransmitter systems[34,35]. Mechanistically, thyroid hormones are known to affect dopaminergic, serotonergic, glutamatergic, and gamma-aminobutyric acidergic pathways, neurochemical circuits that are central to the pathophysiology of schizophrenia[36,37]. Thyroid dysfunction is associated with distinct psychiatric manifestations: Hyperthyroidism can precipitate agitated psychotic symptoms[38], whereas hypothyroidism in schizophrenia has been linked to affective disturbances and cognitive impairments[10,39]. Epidemiological data further support this association, showing an increased prevalence of hypothyroidism among individuals with schizophrenia[40].

Previous studies have examined the relationship between thyroid hormone levels and psychiatric symptoms in individuals with schizophrenia[14,41,42]. For example, Zhu et al[42] reported significantly lower T3 and T4 Levels in hospitalized patients compared to healthy controls, with these reductions correlating with symptom severity. Similarly, Jiang et al[15] identified inverse associations between serum T3/T4 Levels and psychotic symptomatology. The current findings partially corroborate these observations. In particular, a significant inverse association between T4 Levels and psychotic symptoms was observed in the 0-10 years disease duration group. However, no consistent correlation between T3 Levels and symptom severity emerged across the entire cohort. In contrast to Jiang et al[15], who reported a positive association between TSH levels and PANSS scores, our analysis revealed a significant inverse association between TSH and general psychopathology scores within the 10.1-20 years subgroup, with no significant associations detected in other subgroups. Furthermore, stratified analysis based on disease duration demonstrated a significant inverse association between FT3 and general psychopathology scores in the 0-10 years group. These discrepancies may be attributable to differences in sample characteristics, including variations in clinical heterogeneity, pharmacological treatment, metabolic status, and illness trajectory. Stratifying by disease duration likely enabled the identification of subtle, stage-dependent relationships between thyroid function and psychiatric symptoms that may be masked in unstratified analyses.

The stage-specific associations between thyroid function and psychopathology observed in this study likely reflect distinct underlying pathophysiological processes across the different phases of schizophrenia. Mechanistically, T4, as a prohormone for the bioactive T3, plays a critical role in T3-mediated functions such as oligodendrocyte maturation and myelination[43]; thus, reduced T4 Levels may disrupt these processes and contribute to worsening psychiatric symptoms. This mechanism supports our finding of a significant inverse association between T4 Levels and general psychopathology scores in patients within the 0-10 years disease duration group. In the 10.1-20 years group, dysregulation of the HPT axis, potentially involving impaired negative feedback mechanisms[44], and dopamine-induced suppression of TSH secretion[45], may account for the emergence of TSH as the primary thyroid marker associated with psychopathology. On the other hand, the lack of significant associations in patients with disease duration exceeding 20 years may reflect cumulative neuroprogressive alterations that obscure earlier endocrine-psychiatric interactions. These findings underscore the dynamic, stage-dependent nature of thyroid function in schizophrenia and highlight the importance of exploring phase-specific endocrine profiles. Given the potential involvement of multiple endocrine systems in schizophrenia pathophysiology[46], comprehensive hormonal assessments are warranted to place thyroid abnormalities in the broader neuroendocrine context of the disorder.

Several limitations of this study warrant consideration. First, participants were recruited from a single psychiatric hospital in China, which may restrict the generalizability of the findings to populations with different ethnic, environmental, or treatment backgrounds. Second, the cross-sectional nature of the study precludes any inference of causality between alterations in thyroid hormones and the progression of psychiatric symptoms. Third, although key covariates were adjusted for, the potential influence of unmeasured confounders such as cumulative antipsychotic exposure, specific medication classes, other endocrine disorders, and nutritional status was not fully accounted for. Fourth, antipsychotic exposure was assessed using chlorpromazine equivalent doses rather than plasma drug concentrations, which may have overlooked interindividual pharmacokinetic variability. Fifth, the stratification of disease duration into three broad categories (0-10, 10.1-20, and > 20 years) is relatively coarse and may not capture the nuanced temporal dynamics of thyroid function in schizophrenia. More granular or biologically informed stratification may be necessary in future research. Sixth, thyroid function was evaluated solely through peripheral hormone levels, without inclusion of central nervous system-specific measures or thyroid autoantibodies, limiting the depth of mechanistic interpretation. Lastly, the study population comprised long-term hospitalized and clinically stable patients, whose living conditions, nutritional intake, and social environments likely differ from those of community-dwelling individuals with schizophrenia. This may further limit the external validity of the findings. Future studies should adopt longitudinal, multicenter designs with appropriate control groups to distinguish disease-specific effects from treatment-related influences. Incorporating detailed endocrine profiling, neuroimaging, and age- or stage-stratified analyses will be critical to elucidating the role of HPT axis dysregulation across the clinical course of schizophrenia and to validating the associations identified in this study.

CONCLUSION

In summary, this cross-sectional study revealed disease-duration-specific associations between thyroid hormone levels and psychiatric symptom severity among inpatients with schizophrenia. In the 0-10 years group, T4 and T3 Levels were significantly linked to general psychopathology scores, while in the 10.1-20 years group, TSH levels emerged as the primary correlate. These findings underscore distinct, stage-dependent relationships between thyroid function and psychiatric symptoms, suggesting that different phases of the illness may involve unique neuroendocrine dynamics. Further research is warranted to explore these stage-specific mechanisms and to inform the development of individualized, endocrine-informed approaches to schizophrenia management.

ACKNOWLEDGEMENTS

We thank all participants and staffs involved in this research.