Predictive value of serum prealbumin and ferritin combined with Positive and Negative Syndrome Scale on nutritional risk in hospitalized patients with schizophrenia: A cross-sectional study

Abstract

Objective

To investigate the predictive value of serum prealbumin and ferritin combined with the Positive and Negative Syndrome Scale score for concurrent nutritional risk in hospitalized patients with schizophrenia.

Methods

This retrospective study enrolled 90 hospitalized patients with schizophrenia from the Department of Psychiatry in a psychiatric hospital between May 2023 and March 2024. Nutritional risk was screened and categorized using the Malnutrition Universal Screening Tool. Psychiatric symptoms were assessed using the Positive and Negative Syndrome Scale. General demographic data and serum biochemical indicators were collected. Univariate analysis and binary logistic regression analysis were used to identify independent associated factors for nutritional risk. The predictive efficacy of the combination of prealbumin, ferritin, and Positive and Negative Syndrome Scale total score for nutritional risk was evaluated using receiver operating characteristic curve analysis.

Results

Among the 90 patients, 66 (73.3%) had no nutritional risk (Malnutrition Universal Screening Tool = 0), and 24 (26.7%) had nutritional risk (Malnutrition Universal Screening Tool ≥ 1). Binary logistic regression analysis identified the Positive and Negative Syndrome Scale total score (odds ratio = 1.156, 95% confidence interval: 1.046–1.278), prealbumin (odds ratio = 0.902, 95% confidence interval: 0.866–0.958), and ferritin (odds ratio = 0.890, 95% confidence interval: 0.802–0.981) as independent associated factors for nutritional risk in patients with schizophrenia (p < 0.05). Receiver operating characteristic curve analysis showed that the combined model of prealbumin, ferritin, and Positive and Negative Syndrome Scale total score had an area under the curve of 0.831 (95% confidence interval: 0.730–0.931) for predicting nutritional risk. At the optimal cutoff value of ≥0.311, the sensitivity was 0.750 and the specificity was 0.803.

Conclusion

Hospitalized patients with schizophrenia have a high prevalence of nutritional risk. The combination of serum prealbumin, ferritin, and Positive and Negative Syndrome Scale total score demonstrates good predictive value for concurrent nutritional risk and shows promise as an effective indicator for the early identification of nutritional risk in this patient population in clinical settings.

Keywords

Schizophrenia nutritional risk ferritin prealbumin Malnutrition Universal Screening Tool Positive and Negative Syndrome Scale

Introduction

Schizophrenia is a severe mental disorder with high heterogeneity, exhibiting significant individual differences in disease severity and clinical prognosis.^1,2 Beyond core psychiatric symptoms, patients often experience abnormal eating behaviors³ and nutritional metabolic disorders due to cognitive impairment, adverse drug reactions, and social withdrawal,⁴ significantly increasing the nutritional risk. Among patients with schizophrenia, nutritional risk is prevalent yet easily overlooked. Previous studies have confirmed that patients with schizophrenia are at increased risk of malnutrition, with a higher prevalence of underweight and nutritional risk in inpatients compared with outpatients and the general population.^5,6 Severe nutritional complications, such as Wernicke encephalopathy, have also been reported in this population.⁷ Nutritional risk can not only exacerbate the severity of psychiatric symptoms but also affect medication adherence and efficacy, creating a vicious cycle in which psychiatric symptoms and nutritional disorders mutually reinforce each other, ultimately negatively affecting long-term prognosis and quality of life.^8,9 Currently, commonly used nutritional screening tools and comprehensive nutritional indicators often fail to fully consider the specific characteristics of individuals with mental disorders. Limited by patients’ cognitive impairments and communication difficulties, conventional assessment methods cannot accurately reflect their true nutritional status.¹⁰ Importantly, conventional nutritional screening methods may underestimate the true nutritional risk in patients with psychiatric disorders,¹¹ as nearly all patients with mental illness report at least one nutritional risk when comprehensively assessed.¹² This underestimation underscores the need for more comprehensive predictive models that integrate both biochemical markers and psychiatric symptom assessments. Furthermore, existing research on the nutritional status of this population often focuses on describing single nutritional indicators^13–15 and lacks comprehensive analysis of factors influencing nutritional risk; moreover, a targeted predictive model for nutritional risk has not yet been established.

Considering the social realities and clinical operability for patients with schizophrenia, a comprehensive assessment combining routine laboratory indicators with commonly used clinical scale results to evaluate concurrent nutritional risk represents a more practical and feasible strategy. Therefore, this study used the Malnutrition Universal Screening Tool (MUST), a routine nutritional management tool for psychiatric inpatients at our institution that primarily relies on objective indicators, as the starting point for nutritional management in patients with mental disorders. This approach was integrated with general demographic data, peripheral blood biomarkers, and Positive and Negative Syndrome Scale (PANSS) scores, which are widely used in psychiatry, to systematically screen for potential associated factors of nutritional risk in this patient group. Furthermore, a predictive model was constructed to provide a theoretical basis and practical reference for the early identification, targeted intervention, and prognostic improvement of nutritional risk in patients with schizophrenia.

Materials and methods

This study was reported in accordance with the Standards for Reporting Diagnostic Accuracy Studies (STARD) guidelines.¹⁶

Study design and participants

This was a cross-sectional study. Patients were consecutively identified from psychiatric inpatients between May 2023 and March 2024. The inclusion criteria were as follows: (a) admission diagnosis meeting the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) criteria for schizophrenia spectrum disorders; (b) age 18–60 years; and (c) complete clinical data, including PANSS scores, biochemical indicators, and nutritional screening results. The exclusion criteria were as follows: (a) discharge diagnosis revised to a nonschizophrenia spectrum mental disorder or physical disease; (b) comorbid severe physical illness (e.g. heart, lung, liver, or renal failure) or cognitive impairment preventing cooperation with assessments; (c) presence of signs of acute infection (e.g. obvious respiratory symptoms, abnormal lung auscultation, elevated white blood cell count, or body temperature >38.0°C); and (d) pregnancy or lactation. Initially, 106 patients were identified. After applying the exclusion criteria, 16 patients were excluded (9 due to severe physical illness and 7 due to incomplete data), resulting in a final sample of 90 patients who completed all assessments and were included in the analysis.

This study was conducted in accordance with the principles of the Declaration of Helsinki (2024 revision). The study involved retrospective analysis of de-identified clinical data collected during routine patient care. According to institutional guidelines, formal ethical approval and individual informed consent were waived for this retrospective cross-sectional study. All patient data were de-identified to ensure privacy.

Data collection and measurements

Demographic and clinical data were collected using a self-designed questionnaire, including sex, age, educational level, body mass index (BMI), and calf circumference.

Nutritional risk assessment was conducted using the MUST, typically within 48 h of admission. MUST scores were routinely recorded by the ward-assigned specialist as part of the psychiatric admission assessment protocol and were extracted from the medical records. The MUST evaluates three components: BMI, unintentional weight loss during the past 3–6 months, and reduced food intake during acute illness. Based on the total score, patients were divided into a no nutritional risk group (MUST = 0) and a nutritional risk group (MUST ≥ 1). The MUST has been validated in hospitalized populations, with a reported sensitivity of 80.0% and specificity of 74.7%,¹⁷ and is recognized as one of the most accurate screening tools.¹⁸

Psychiatric symptom assessment was conducted using the PANSS, typically within 1 week of admission. PANSS scores were routinely recorded by the attending psychiatrists as part of the admission assessment protocol and were extracted from the medical records. The PANSS comprises three dimensions: positive symptoms (7 items; score range: 7–49), negative symptoms (7 items; score range: 7–49), and general psychopathology (16 items; score range: 16–112).¹⁹

Laboratory measurements were extracted from the hospital laboratory information system. Fasting venous blood samples were collected on the morning after admission and analyzed using automated biochemical analyzers with standard reagent kits. The tested indicators and methods were as follows:

Blood routine. White blood cell count (WBC), neutrophil fraction (NEUT), and hemoglobin (HGB) were measured.

Lipid profile. Triglycerides (TG) were measured using the free glycerol removal method; total cholesterol (TC) was measured using the cholesterol oxidase method; high-density lipoprotein cholesterol (HDL-C) and low-density lipoprotein cholesterol (LDL-C) were measured using the clearance method.

Liver and kidney function. Total protein (TP) was measured using the biuret method; albumin (ALB) was measured using the bromocresol green method; prealbumin (PAB) and retinol-binding protein (RBP) were measured via immunoturbidimetry; creatinine (Cre) and blood urea nitrogen (BUN) were measured using the urease ultraviolet (UV) rate method; and aspartate aminotransferase (AST) and alanine aminotransferase (ALT) were measured using the rate method.

Thyroid function. Thyroid-stimulating hormone (TSH), triiodothyronine (T3), and thyroxine (T4) were measured via electrochemiluminescence.

Tumor markers. Alpha-fetoprotein (AFP) and carcinoembryonic antigen (CEA) were measured via electrochemiluminescence.

Coagulation function. Prothrombin time (PT), activated partial thromboplastin time (APTT), and thrombin time (TT) were measured using the optical method; D-dimer (DD) was measured using enzyme-linked immunosorbent assay.

Anemia-related indicators. Serum ferritin (Fer), vitamin B12 (Vit. B12), folate, and homocysteine (Hcy) were measured via electrochemiluminescence.

Statistical analysis

EpiData software was used to create the database. All data were independently entered by two researchers and cross-checked to ensure accuracy. Statistical analysis was performed using SPSS 23.0 software. Normally distributed continuous data were presented as mean ± standard deviation (x̄ ± s) and were compared between groups using the independent-samples t-test. Non-normally distributed continuous data were presented as median (interquartile range) (M (IQR)) and were compared using the Mann–Whitney U test. Categorical data were presented as number (percentage) (n (%)) and were compared between groups using the χ² test. When the expected cell frequency for categorical variables was <5, Fisher’s exact test was used for intergroup comparisons. To explore associated factors for nutritional risk, variables with statistical significance in the univariate analysis were included in a binary logistic regression model for multivariate analysis. Receiver operating characteristic (ROC) curve analysis was used to evaluate the predictive efficacy of relevant indicators for nutritional risk, and the area under the curve (AUC) was calculated. Sample size was calculated using the method proposed by Obuchowski for ROC studies.²⁰ Assuming an expected AUC of 0.70, a significance level of α = 0.05 (one-sided), a power of 1 − β = 0.80, and a 1:1 ratio between groups, the calculated minimum required sample size was less than 90 cases. All statistical tests were two-sided, with p < 0.05 considered statistically significant.

Results

Comparison of general data between the two groups

Based on nutritional risk screening results, the 90 patients were divided into a no nutritional risk group (n = 66) and a nutritional risk group (n = 24). χ² test analysis showed no statistically significant differences in the distribution of sex, educational level, or marital status between the two groups (p > 0.05), although descriptive data showed a trend toward an increasing proportion of unmarried individuals and higher education levels with increasing nutritional risk scores. Furthermore, the difference in calf circumference between the two groups was statistically significant (p < 0.05) (Table 1).

Table 1.

Comparison of baseline data between the two groups of patients.

Item	Category	No nutritional risk group (MUST = 0)(n = 66)	Nutritional risk group (MUST ≥ 1)(n = 24)	Statistic	p-value
Sex	Male	25 (37.9%)	4 (16.7%)	3.626	0.057
	Female	41 (62.1%)	20 (83.3%)
Educational level	Primary or below	1 (1.5%)	0 (0.0%)	Fisher's exact test	0.438
	Secondary	35 (53%)	9 (37.5%)
	College or above	30 (45.5%)	15 (62.5%)
Marital status	Unmarried	37 (56.1%)	19 (79.2%)	Fisher's exact test	0.088
	Married	24 (36.4%)	5 (20.8%)
	Divorced/widowed	5 (7.6%)	0 (0.0%)
Age (years)		33 (23, 49)	30 (23, 34)	1.789	0.074
BMI (kg/m²)		24.03 (22.67, 26.44)	19.05 (18.59, 19.92)	–	–
Calf circumference (cm)		35.8 (34.0, 39.4)	32.0 (31.0, 34.8)	3.997	<0.001

Data are presented as n (%), mean ± standard deviation, or median (interquartile range). Percentages were calculated based on the total number of cases in each group (no nutritional risk group, n = 66; nutritional risk group, n = 24) to reflect the compositional distribution of each category within each group.

Bold text indicates statistical significance (p < 0.05).

MUST: Malnutrition Universal Screening Tool.

Comparison of PANSS scores and peripheral blood indicators between the two groups

Results from the independent-samples t-test (or Mann–Whitney U test) showed that compared with the no nutritional risk group, patients in the nutritional risk group had significantly higher PANSS total scores and general psychopathology scale scores, with statistically significant differences (p < 0.05). Among the peripheral blood indicators, hemoglobin (HGB), prealbumin (PAB), RBP, aspartate aminotransferase (AST), creatinine (Cre), BUN, sodium (Na⁺), serum ferritin (Fer), triglycerides (TG), and TT levels showed statistically significant differences between the nutritional risk group and the no nutritional risk group (p < 0.05) (Table 2).

Table 2.

Comparison of PANSS scores and peripheral blood indicators between the two groups of patients.

Category	Indicator	No nutritional risk group (MUST = 0)(n = 66)	Nutritional risk group (MUST ≥ 1)(n = 24)	Statistic	p-value
PANSS	Total score	93.83 ± 6.75	100.75 ± 4.65	2.785	0.007
	Positive subscale	24.32 ± 4.12	27.50 ± 3.63	1.907	0.06
	Negative subscale	18.83 ± 3.74	20.13 ± 2.17	0.646	0.52
	General psychopathology	50.68 ± 3.42	53.13 ± 3.64	2.257	0.026
Blood routine	WBC (×10⁹)	6.36 (5.43, 7.83)	5.93 (5.09, 6.82)	1.446	0.148
	NEUT %	57.51 ± 10.38	54.06 ± 9.33	1.432	0.156
	HGB (g/L)	133.11 ± 14.6	121.42 ± 15.37	3.313	0.001
Liver and kidney Function	TP (g/L)	66.02 ± 5.32	64.21 ± 6.11	1.369	0.175
	ALB (g/L)	45.75 ± 2.01	40.07 ± 2.79	0.660	0.511
	PAB (mg/L)	228.15 ± 44.17	195.33 ± 30.04	3.362	0.001
	RBP (mg/L)	39.20 (32.55, 45.60)	28.00 (25.60, 34.78)	4.389	<0.001
	AST (U/L)	19 (11, 32)	11.5 (7.75, 15)	3.444	0.001
	ALT (U/L)	16 (13, 25)	14 (13, 16.25)	1.718	0.086
	Cre (µmol/L)	64.42 ± 13.17	55.32 ± 8.69	3.671	0.001
	BUN (mmol/L)	3.86 (3.17, 5.23)	3.16 (2.63, 3.55)	2.603	0.009
	K⁺ (mmol/L)	3.73 ± 0.32	3.77 ± 0.33	0.452	0.652
	Na⁺ (mmol/L)	140.02 ± 1.79	138.86 ± 1.90	2.578	0.012
	Cl⁻ (mmol/L)	104.30 ± 2.17	104.44 ± 2.08	0.260	0.796
	Ca²⁺ (mmol/L)	2.25 ± 0.11	2.23 ± 0.12	0.919	0.361
	CO₂ (mmol/L)	26.04 ± 2.03	25.34 ± 2.76	1.092	0.284
Anemia-related indicators	Fer (ng/mL)	88.76 (55.73, 186.84)	35.30 (19.10, 61.41)	3.887	<0.001
	Vit B₁₂ (pmol/L)	295.5 (232.0, 352.5)	276.5 (195.25, 527.0)	0.437	0.665
	Folate (nmol/L)	12.35 (9.28, 16.43)	11.05 (7.66, 14.10)	1.172	0.241
	Hcy (µmol/L)	11.45 (10.40, 13.10)	12.40 (10.33, 14.53)	0.976	0.329
Blood lipids	TG (mmol/L)	1.08 (0.8, 1.87)	0.79 (0.57, 0.97)	2.762	0.006
	TC (mmol/L)	4.49 ± 0.10	4.07 ± 1.10	1.567	0.121
	HDL-C (mmol/L)	1.10 (0.94, 1.33)	1.21 (1.08, 1.54)	1.762	0.078
	LDL-C (mmol/L)	2.43 ± 0.72	2.17 ± 0.70	1.375	0.173
Thyroid function	TSH (mIU/L)	2.04 (1.59, 3.27)	1.91 (1.06, 2.63)	1.128	0.259
	T3 (nmol/L)	1.63 ± 0.28	1.50 ± 0.26	1.787	0.078
	T4 (nmol/L)	95.12 ± 17.53	91.08 ± 13.87	1.056	0.297
Tumor markers	AFP (ng/mL)	2.54 (1.69, 3.57)	2.11 (1.60, 2.43)	1.505	0.132
	CEA (ng/mL)	1.54 (0.95, 2.17)	1.25 (0.90, 1.67)	1.101	0.271
Coagulation function	PT (s)	11.72 ± 0.88	11.65 ± 0.87	0.331	0.742
	TT (s)	13.85 (13.08, 14.93)	14.55 (13.83, 15.38)	2.373	0.018
	APTT (s)	33.55 (28.98, 35.63)	31.45 (30.18, 35.20)	0.235	0.814
	DD (ng/mL)	73.5 (47.25, 128.75)	86.0 (53.25, 131.50)	0.589	0.556

Data are presented as mean ± standard deviation or median (interquartile range).

Bold text indicates statistical significance (p < 0.05).

AFP: alpha-fetoprotein; ALB: albumin; ALT: aminotransferase; APTT: activated partial thromboplastin time; AST: aminotransferase; BUN: blood urea nitrogen; CEA: carcinoembryonic antigen; Cre: creatinine; DD: D-dimer; Fer: ferritin; HDL-C: high-density lipoprotein cholesterol; HGB: hemoglobin; LDL-C: low-density lipoprotein cholesterol; MUST: Malnutrition Universal Screening Tool; NEUT: neutrophil fraction; PAB: prealbumin; PANSS: Positive and Negative Syndrome Scale; PT: prothrombin time; RBP: retinol-binding protein; TC: total cholesterol; TG: triglycerides; TP: total protein; TSH: thyroid-stimulating hormone; TT: thrombin time; WBC: white blood cell count.

Binary logistic regression analysis of associated factors for nutritional risk in patients with schizophrenia

To control for confounding factors, variables with statistical significance in the univariate analysis were included in a binary logistic regression model. Multivariate analysis showed that the PANSS total score, prealbumin (PAB), and serum ferritin (Fer) were independent associated factors for nutritional risk in patients with schizophrenia (p < 0.05) (Table 3).

Table 3.

Binary logistic regression analysis of nutritional risk in patients with schizophrenia.

Factor	β value	SE	Wald value	p-value	OR value	95% CI
PANSS total score	0.145	0.051	8.053	0.005	1.156	1.046–1.278
PAB	−0.103	0.008	4.636	0.031	0.902	0.866–0.958
Fer	−0.116	0.005	5.122	0.024	0.890	0.802–0.981

The β values for PAB and Fer were negative, indicating a protective effect (higher levels were correlated with lower nutritional risk), consistent with OR < 1.

Bold text indicates statistical significance (p < 0.05).

CI: confidence interval; Fer: ferritin; OR: odds ratio; PAB: prealbumin; PANSS: Positive and Negative Syndrome Scale; SE: standard error.

Diagnostic value assessment of the nutritional risk prediction model

Variables with statistical significance in the multivariate logistic regression analysis (PANSS total score, PAB, and Fer) were included to construct a combined prediction model. The predicted probability of this model was used as the test variable, and the presence or absence of concurrent nutritional risk was used as the state variable to plot the ROC curve. The results showed that the AUC for this combined model in identifying concurrent nutritional risk in patients with schizophrenia was 0.831 (95% confidence interval: 0.730–0.931). At the optimal cutoff value of 0.311, the sensitivity was 75.0% and the specificity was 80.3%, indicating good predictive performance of the model. To further evaluate the predictive value of each individual indicator, ROC curve analysis was performed using PANSS total score, PAB, and Fer as the test variables. The results showed that the combined model had better predictive performance than any single indicator (Table 4).

Table 4.

Predictive performance of individual indicators and the combined model for nutritional risk.

Indicator	AUC	95% CI	Optimal cutoff	Sensitivity	Specificity
Combined model	0.831	0.730–0.931	0.311	0.750	0.803
Fer	0.769	0.655–0.882	40.645	0.667	0.864
PAB	0.732	0.620–0.844	193.5	0.583	0.773
PANSS total score	0.669	0.547–0.791	94.5	0.792	0.515

AUC: area under the curve; CI: confidence interval; Fer: ferritin; PAB: prealbumin; PANSS: Positive and Negative Syndrome Scale; SE: standard error.

Discussion

Schizophrenia, as a highly disabling chronic disease, has become a major public health challenge requiring global attention.²¹ It not only imposes long-term suffering on individual patients but also places a heavy burden on family functioning and societal healthcare resources.²² Patient prognosis depends not only on the control of psychiatric symptoms but also on nutritional status, which is a key factor affecting overall bodily function and has become a crucial issue that cannot be ignored in the clinical management of patients with schizophrenia. Therefore, systematically screening for potential associated factors of nutritional risk and constructing an efficient and feasible predictive indicator system have important clinical and public health value for optimizing clinical nutritional management strategies and improving long-term patient outcomes.

In this study, the incidence of nutritional risk was 26.7%, further confirming that populations with mental illness are at high risk of malnutrition, consistent with several previous studies.^5,6,23 Notably, statistically significant differences in calf circumference were observed between the nutritional risk group and the no nutritional risk group, suggesting that physical examination is a crucial component of clinical nutritional management for patients with schizophrenia. Calf circumference, as an objective indicator reflecting whole-body muscle mass and fat reserves, and its reduction along with lower BMI, indicate insufficient body reserves in the nutritional risk group. This may be related to the disease itself and metabolic disturbances triggered by psychiatric symptoms, ultimately leading to deterioration of nutritional status.

The PANSS total score, as a core indicator of psychiatric symptom severity,²⁴ was confirmed in this study to be associated with increased nutritional risk. Negative symptoms in schizophrenia (e.g. avolition and social withdrawal) may lead to decreased self-feeding ability and reduced physical activity levels, thereby affecting muscle synthesis and energy metabolism balance.²⁵ Additionally, the stress state induced by psychiatric symptoms can increase basal metabolic rate and energy expenditure, significantly amplifying nutritional risk under this dual effect.²⁶ These behavioral and metabolic pathways may collectively contribute to the increased nutritional risk observed in patients with higher PANSS scores. Regarding biochemical indicators, the level of PAB, as a sensitive marker reflecting short-term nutritional status, decreases in patients with recent inadequate protein intake or impaired synthesis.²⁷ Protein deficiency can not only worsen the condition but also increase the risk of adverse complications such as infections.^28,29 Fer is the body’s main iron storage protein and is involved not only in iron metabolism but also in immune regulation. Decreased Fer levels reflect a long-term imbalance in nutrition and immunity.^30,31

Based on these findings, this study innovatively integrated the core disease severity indicator for patients with schizophrenia (PANSS total score) with biochemical indicators related to short-term (PAB) and long-term (Fer) nutritional status to construct a predictive model for nutritional risk. The AUC reached 0.831, indicating good predictive performance and providing a concise and efficient quantitative tool for the early identification of patients at high nutritional risk in clinical settings. However, this study has certain limitations. First, the sample was derived from a single-center inpatient population, which may have introduced selection bias and may not fully represent all populations with schizophrenia. Second, the study did not include potential associated factors such as medication regimens and disease duration; the mechanisms through which these variables affect nutritional risk require further exploration.

Conclusion

In summary, by integrating quantitative assessment indicators of psychiatric symptoms with easily obtainable peripheral blood biochemical markers, this study constructed a clinically practical predictive model for nutritional risk, offering a new approach for the early nutritional management of patients with schizophrenia. Clinicians can combine patients’ PANSS scores, routine biochemical test results (PAB and Fer), nutritional physical examinations, and the MUST nutritional risk screening tool for a multidimensional comprehensive assessment to promptly identify high-risk patients and initiate individualized nutritional interventions, thereby breaking the vicious cycle and improving patient prognosis. Future research could expand the sample size, conduct multicenter cohort studies, and include more clinical variables such as medication type and disease duration to further optimize the accuracy and generalizability of the predictive model. Simultaneously, targeted nutritional intervention programs based on this model could be designed, and their actual benefits in improving nutritional status, controlling psychiatric symptoms, and enhancing quality of life could be verified through randomized controlled trials, thereby providing more comprehensive evidence-based support for the integrated management of schizophrenia.

Footnotes

Acknowledgments

None.

Author contributions

Ting-Ting Jiang led the overall design of the study, defined the research protocol and technical roadmap, organized the preliminary clinical data, and developed the statistical analysis framework. Ting-Ting Jiang independently drafted the manuscript, participated in the screening of enrolled samples and discussion of the research findings, verified data authenticity by cross-checking original medical records, laboratory reports, and database entries, and reviewed and approved the final version of the manuscript for publication.

Zhu-Ma Jin was primarily responsible for research data collection, including recording demographic information and clinical symptom scores (Positive and Negative Syndrome Scale, PANSS) for enrolled patients. Zhu-Ma Jin was responsible for data entry into the EpiData database and implemented double data entry with cross-verification by two independent personnel to minimize entry errors and ensure data accuracy. She also assisted in organizing original laboratory reports and supplementing/updating research materials, reviewed the final manuscript, and approved its publication.

Hua Jin participated in nutritional risk screening using the MUST for hospitalized patients and in the collection of relevant clinical indicators. Hua Jin assisted in organizing serum biochemical data (e.g. prealbumin and ferritin) and verified the consistency between laboratory results and medical records. She also participated in the literature review and formatting revision of the initial manuscript, including adjustment of heading hierarchies, standardization of reference formatting, and unification of figure/table numbering, conducted supplementary proofreading of references; and approved the final revised version before submission.

Chong-Yang Han oversaw optimization of the research protocol and preparation of ethical review documents to ensure adherence to the principles of the Declaration of Helsinki. Chong-Yang Han independently performed statistical analyses using SPSS 23.0, including univariate analysis, binary logistic regression, and ROC curve analysis. Chong-Yang Han critically revised the key academic content of the manuscript, adjusted the logical structure of the discussion section, coordinated the overall research process, reviewed the final version intended for publication, assumed responsibility for external communication as the corresponding author, and approved the final version of the manuscript for publication.

Data availability statement

Data are available from the corresponding author upon reasonable request.

Declaration of conflicting interests

None declared.

Ethics statement

Due to the retrospective nature of the study and the use of de-identified clinical data, the requirement for informed consent was waived by the Ethics Committee of Nanjing Brain Hospital. The study was conducted in accordance with the principles of the Declaration of Helsinki (2024 version).

Funding

None.

ORCID iDs

Ting-Ting Jiang

Chong-Yang Han

References

Tandon

Keshavan

Nasrallah

HA.

Schizophrenia, ‘just the facts’ what we know in 2008. 2. Epidemiology and etiology. Schizophr Res 2008; 102: 1–18.

Van Os

Kapur

Schizophrenia. Lancet 2009; 374: 635–645.

Jahrami

Faris

MAE

Saif

, et al. Assessing dietary and lifestyle risk factors and their associations with disease comorbidities among patients with schizophrenia: a case–control study from Bahrain. Asian J Psychiatr 2017; 28: 115–123.

Bou Khalil

Hachem

Richa

Eating disorders and schizophrenia in male patients: a review. Eat Weight Disord 2011; 16: e150–e156.

Risch

Hotzy

Vetter

, et al. Assessment of nutritional status and risk of malnutrition using adapted standard tools in patients with mental illness and in need of intensive psychiatric treatment. Int J Environ Res Public Health 2022; 20: 109.

Sugai

Suzuki

Yamazaki

, et al. High prevalence of underweight and undernutrition in Japanese inpatients with schizophrenia: a nationwide survey. BMJ Open 2015; 5: e008720.

, et al. A case report of Wernicke’s encephalopathy associated with schizophrenia. Front Psychiatry 2021; 12: 657649.

Seeman

MV.

Eating disorders and psychosis: seven hypotheses. World J Psychiatry 2014; 4: 112–119.

Marx

Moseley

Berk

, et al. Nutritional psychiatry: the present state of the evidence. Proc Nutr Soc 2017; 76: 427–436.

10.

Bhat

Chiu

Jeste

DV.

Nutrition and geriatric psychiatry: a neglected field. Curr Opin Psychiatry 2005; 18: 609–614.

11.

Abayomi

Hackett

Assessment of malnutrition in mental health clients: nurses’ judgement vs. a nutrition risk tool. J Adv Nurs 2004; 45: 430–437.

12.

Mueller-Stierlin

Mötteli

Hotzy

, et al. Feasibility of a multifaceted nutrition-risk screening tool for mental health settings: the NutriMental screener. Br J Nutr 2024; 135: 559–566.

13.

Cui

McGrath

Burne

THJ

, et al. Vitamin D and schizophrenia: 20 years on. Mol Psychiatry 2021; 26: 2708–2720.

14.

Yang

Yan

, et al. Causal relationship between B vitamins and neuropsychiatric disorders: a systematic review and meta-analysis. Neurosci Biobehav Rev 2025; 170: 106068.

15.

Faugere

Maakaron

Achour

, et al. Vitamin D, B9, and B12 deficiencies as key drivers of clinical severity and metabolic comorbidities in major psychiatric disorders. Nutrients 2025; 17: 1167.

16.

Bossuyt

Reitsma

Bruns

; STARD Groupet al. STARD 2015: an updated list of essential items for reporting diagnostic accuracy studies. BMJ 2015; 351: h5527.

17.

Cortes

Yañez

Capitán-Moyano

, et al. Evaluation of different screening tools for detection of malnutrition in hospitalised patients. J Clin Nurs 2024; 33: 4759–4771.

18.

Chrástecká

Blanař

Pospíchal

Risk of malnutrition assessment in hospitalised adults: a scoping review of existing instruments. J Clin Nurs 2023; 32: 3397–3411.

19.

Kay

Fiszbein

Opler

LA.

The positive and negative syndrome scale (PANSS) for schizophrenia. Schizophr Bull 1987; 13: 261–276.

20.

Obuchowski

NA.

Computing sample size for receiver operating characteristic studies. Invest Radiol 1994; 29: 238–243.

21.

Asher

Fekadu

Hanlon

Global mental health and schizophrenia. Curr Opin Psychiatry 2018; 31: 193–199.

22.

Wang

Shi

Guan

, et al. A systematic review of methods and study quality of economic evaluations for the treatment of schizophrenia. Front Public Health 2021; 9: 689123.

23.

Kim

, et al. Factors related to malnutrition in community-dwelling patients with schizophrenia. Perspect Psychiatr Care 2019; 55: 415–423.

24.

Lim

Peh

Yang

, et al. Large-scale evaluation of the positive and negative syndrome scale (PANSS) symptom architecture in schizophrenia. Asian J Psychiatr 2021; 62: 102732.

25.

Kamoi

Mifune

Soriano

, et al. Association between dynapenia/sarcopenia, extrapyramidal symptoms, negative symptoms, body composition, and nutritional status in patients with chronic schizophrenia. Healthcare (Basel) 2024; 13: 48.

26.

Prabakaran

Swatton

Ryan

, et al. Mitochondrial dysfunction in schizophrenia: evidence for compromised brain metabolism and oxidative stress. Mol Psychiatry 2004; 9: 684–697, 643.

27.

Devoto

Gallo

Marchello

, et al. Prealbumin serum concentrations as a useful tool in the assessment of malnutrition in hospitalized patients. Clin Chem 2006; 52: 2281–2285.

28.

Robinson

Trujillo

Mogensen

, et al. Improving nutritional screening of hospitalized patients: the role of prealbumin. JPEN J Parenter Enteral Nutr 2003; 27: 389–395; quiz 439.

29.

Chertow

Goldstein-Fuchs

Lazarus

, et al. Prealbumin, mortality, and cause-specific hospitalization in hemodialysis patients. Kidney Int 2005; 68: 2794–2800.

30.

Iriarte-Gahete

Tarancon-Diez

Garrido-Rodríguez

, et al. Absolute and functional iron deficiency: biomarkers, impact on immune system, and therapy. Blood Rev 2024; 68: 101227.

31.

Kotla

Dutta

Parimi

, et al. The role of ferritin in health and disease: recent advances and understandings. Metabolites 2022; 12: 609.