Influencing factors and construction of a nomogram for post-stroke depression in patients with chronic stroke

Abstract

BACKGROUND

Post-stroke depression (PSD), a condition commonly developed in patients with chronic stroke, impairs both functional rehabilitation and daily living.

AIM

To comprehensively analyze PSD contributors in chronic phase stroke and construct a precise nomogram.

METHODS

Two hundred patients with chronic stroke admitted in over 7 years (January 2017 to January 2024), were enrolled and categorized into the PSD group (n = 96) and the non-PSD (NPSD) group (n = 104). Demographic characteristics, clinicopathological data, and biochemical indicators were collected and analyzed by univariate analysis. Significant predictors identified in the univariate analysis were subsequently incorporated into a binary logistic regression model to assess their independent effects on PSD risk. The discriminative ability/calibration of the developed PSD prediction nomogram was assessed.

RESULTS

Compared with the NPSD group, the PSD group included a higher proportion of patients aged ≥ 60 years, divorced/widowed, with an education level below senior high school, presenting with ≥ 2 comorbidities, exhibiting severe neurological impairment, and having multiple lesions. Additionally, the PSD group showed significantly higher neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR) than the NPSD group. After assigning values to significant predictors, multivariate analysis indicated that educational level (P = 0.046), NLR (P < 0.001), and PLR (P < 0.001) were independently associated with PSD in patients with chronic stroke. The developed nomogram exhibited favorable discrimination performance. The nomogram's calibration remained accurate for high-risk stratification but displayed modest inconsistencies in low- and middle-risk categories.

CONCLUSION

Education level, NLR, and PLR independently contribute to PSD in patients with chronic stroke. The constructed nomogram effectively predicts PSD risk within the range of 0.10-0.90, presenting a valuable tool for clinical monitoring and risk assessment of PSD in patients with chronic stroke.

Key Words: Stroke; Chronic phase; Post-stroke depression; Influencing factors analysis; Nomogram

Core Tip: This study centers on identifying contributors to post-stroke depression (PSD) in patients with chronic stroke and establishing a relevant nomogram. Based on the results of various analytical procedures, individuals with lower educational attainment, and high neutrophil-to-lymphocyte ratio and platelet-to-lymphocyte ratio are at greater risk of PSD. Tailored clinical interventions, along with vigilant monitoring, can effectively mitigate PSD risk in patients presenting these features. The nomogram shows strong predictive accuracy for high-risk groups; however, its reliability slightly decreases for low- and intermediate-risk groups.

INTRODUCTION

Stroke, a cerebrovascular disorder characterized by interrupted cerebral blood flow, can be broadly categorized as ischemic (80%-87% of stroke cases) or hemorrhagic (13%-20%) based on its underlying etiology[1]. The pathogenesis of ischemic stroke predominantly arises from thrombosis, cardiogenic embolism, or atherosclerosis, wherein platelet aggregation or cardiac arrest causes diminished cerebral perfusion culminating in arterial occlusion[2]. Conversely, hemorrhagic stroke is primarily associated with hypertension, cerebral amyloid angiopathy, aneurysms, or post-trauma vascular rupture[3,4]. Stroke has a constellation of clinical manifestations, ranging from motor, visual, and speech deficits to nonspecific symptoms such as headache, memory impairment, blindness, dysphagia, and dysarthria[5]. These sequelae severely impair patients’ performance of activities of daily living while imposing economic burden on their families and society[6]. Epidemiologically, stroke affects approximately 80 million individuals, resulting in 5.5 million deaths and 116 million individuals with disabilities annually, posing a global healthcare challenge[7]. Moreover, managing stroke remains a formidable challenge, as only 10% of patients fully recover, the mortality rate can reach up to 30%, and the remaining 60% experience chronic functional disabilities[8]. Among these patients, 85% experience upper limb motor dysfunction, which not only impedes daily activities but also undermines their independence, thereby predisposing them to post-stroke depression (PSD)[9,10]. PSD is the most prevalent neuropsychiatric complication following a stroke. It is characterized by persistent fatigue, depressed mood, anhedonia, social withdrawal, blunt emotions, sleep disturbances, and even suicidal ideation[11,12]. The prevalence of PSD is notably high (20%-60%)[13]. Despite its profound negative effect on functional recovery and daily living in stroke survivors, PSD often receives insufficient attention and treatment[14].

Many researchers have explored factors that affect PSD. Wang et al[15], for instance, pointed out that serum leptin and insulin-like growth factor 1 can serve as crucial indeces in recognizing PSD. Ilut et al[16] linked PSD intensity to infarct location and National Institutes of Health Stroke Scale (NIHSS) score[16]. Lee et al[17] indicated that the Hamilton depression rating scale (HAMD) score upon hospital admission can function as a prognostic factor for depression 3 months after stroke.

In light of the above, this study delves deep into PSD contributors in patients with chronic stroke and develops a visual nomogram for PSD risk quantification. The findings provide valuable insights for the early identification, prevention, and risk stratification of PSD in such a patient population. By incorporating clinical and inflammatory parameters, this study achieves superior predictive capability over single-biomarker approaches.

MATERIALS AND METHODS

General information

This retrospective study enrolled 200 patients with chronic stroke visiting Shaanxi Provincial People’s Hospital (January 2017 to January 2024). Patients were dichotomized by PSD status: PSD (n = 96) vs non-PSD (NPSD; n = 104).

Participant selection and exclusion criteria

Eligibility criteria: (1) Confirmed stroke diagnosis[18]; (2) Disease duration exceeding 1 year; (3) Age ≥ 18 years at the time of enrollment; (4) Adequate auditory and visual function for clinical evaluations; (5) Absence of significant cognitive impairment or severe aphasia; and (6) Availability of intact clinical medical records.

Exclusion criteria: (1) History of using mood stabilizers, antipsychotic medications, or antidepressants before study enrollment; (2) Pregnancy or lactation status; (3) Presence of active severe infectious diseases such as respiratory, urinary tract, or gastrointestinal infections; (4) Comorbid conditions (e.g., hematopoietic system disorders, malignant neoplasms, and connective tissue diseases); and (5) Previous radiotherapy or chemotherapy that could potentially confound blood biochemical results.

Assessment indicators

Depressive symptom assessment utilized the 24-item HAMD[19]. The severity of depression was based on HAMD scores: Suspected depression (8-20), moderate depression (20-35), and severe depression (> 35). A cutoff of 8 points distinguished the PSD group (≥ 8) from the NPSD group (< 8). Demographic characteristics, such as sex, age, marital status, and education level, were systematically collected for both groups. Clinicopathological data, such as the number of comorbid conditions, severity of neurological impairment, lesion location, and number of lesions, were also recorded. The blood biochemical parameters analyzed included the mean platelet volume (MPV), lymphocyte count, monocyte count, neutrophil count, neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), and lymphocyte-to-monocyte ratio (LMR). The reported values reflect test results from the blood work performed during the initial 24-hour hospitalization period, adhering to the protocol outlined in Chinese Guidelines for Acute Ischemic Stroke Diagnosis and Management (2023). For neurological impairment evaluation, the NIHSS[20] was utilized. Impairment levels are classified based on a 0-42 scale: Mild (1-4), moderate (5-15), and severe (> 15).

Throughout the data collection, researchers were kept uninformed about the specific HAMD scores and the purpose of the research. A blinded recheck of HAMD scores was performed by the psychological assessment team, independent neurologists did a blinded re-scoring using the NIHSS, and laboratory personnel tested blood biochemical indices in a blinded manner.

Statistical analysis

Data analyses were conducted using SPSS 19.0. Categorical data appear as frequencies (percentages) and underwent χ² testing. To identify independent PSD predictors, variables that differed significantly in the univariate analyses were incorporated into a multivariate logistic regression model. Subsequently, a nomogram was developed based on these risk factors to visualize the PSD risk. A significance criterion of P value < 0.05 was adopted.

RESULTS

Demographic characteristics of the two groups

No significant differences were observed between the PSD and NPSD groups in terms of sex distribution (P > 0.05). However, age, marital status, and educational attainment differed significantly (P < 0.05). Specifically, compared with the NPSD group, the PSD group exhibited a significantly higher proportion of patients aged ≥ 60 years, those who were divorced or widowed, and those with an educational attainment below senior high school (P < 0.05, Table 1).

Table 1 Demographic characteristics of the two patient groups, n (%).

Factors	n	PSD group (n = 96)	NPSD group (n = 104)	χ2	P value
Gender				0.329	0.566
Male	142	70 (72.92)	72 (69.23)
Female	58	26 (27.08)	32 (30.77)
Age (years)				5.592	0.018
< 60	109	44 (45.83)	65 (62.50)
≥ 60	91	52 (54.17)	39 (37.50)
Marital status				6.586	0.010
Married	159	69 (71.88)	90 (86.54)
Divorced/widowed	41	27 (28.13)	14 (13.46)
Education level				9.944	0.002
Below senior high school	106	62 (64.58)	44 (42.31)
Senior high school or above	94	34 (35.42)	60 (57.69)

PSD: Post-stroke depression; NPSD: Non-post-stroke depression.

Clinicopathological data of the two groups

Significant intergroup differences were observed regarding the number of comorbid conditions, severity of neurological impairment, and number of lesions (P < 0.05). Specifically, compared with the NPSD group, the PSD group included a significantly higher proportion of patients presenting with ≥ 2 comorbid conditions, severe neurological impairment, and multiple lesions (P < 0.05). Conversely, no marked intergroup differences were found in the lesion location (P > 0.05; Table 2).

Table 2 Clinicopathological data of the two patient groups, n (%).

Factors	n	PSD group (n = 96)	NPSD group (n = 104)	χ2	P value
Number of comorbid conditions				11.836	0.003
0	35	23 (23.96)	12 (11.54)
1	108	40 (41.67)	68 (65.38)
≥ 2	57	33 (34.38)	24 (23.08)
Severity of neurological impairment				8.144	0.017
Mild	40	12 (12.50)	28 (26.92)
Moderate	123	61 (66.67)	62 (59.62)
Severe	37	23 (23.96)	14 (13.46)
Lesion location				0.131	0.936
Frontal lobe	90	44 (45.83)	46 (44.23)
Temporal lobe	65	30 (31.25)	35 (33.65)
Basal ganglia	45	22 (22.92)	23 (22.12)
Number of lesions				5.998	0.014
Single	78	29 (30.21)	49 (47.12)
Multiple	122	67 (69.79)	55 (52.88)

PSD: Post-stroke depression; NPSD: Non-post-stroke depression.

Blood biochemical indicators of the two groups

No significant differences were detected across groups in the MPV, lymphocyte count, monocyte count, neutrophil count, or LMR (P > 0.05). However, the PSD group demonstrated significantly higher NLR and PLR than the NPSD group (P < 0.05). Detailed results are provided in Figure 1 and Table 3.

Figure 1 Blood biochemical parameters in the two patient groups. A: Mean platelet volume comparison; B: Lymphocyte levels between groups; C: Monocyte counts in both cohorts; D: Neutrophil measurements across groups; E: Neutrophil-to-lymphocyte ratio analysis; F: Platelet-to-lymphocyte ratio evaluation; G: Lymphocyte-to-monocyte ratio assessment. ^aP < 0.001 vs the non-post-stroke depression group. MPV: Mean platelet volume; NLR: Neutrophil-to-lymphocyte ratio; PLR: Platelet-to-lymphocyte ratio; LMR: Lymphocyte-to-monocyte ratio.

Table 3 Blood biochemical indicators of the two patient groups.

Factors	n	PSD group (n = 96)	NPSD group (n = 104)	t value	P value
MPV (fL)	200	9.86 ± 1.27	9.58 ± 1.49	1.424	0.156
Lymphocyte count (109/L)	200	1.58 ± 0.50	1.62 ± 0.53	0.548	0.584
Monocyte count (109/L)	200	0.40 ± 0.13	0.42 ± 0.14	1.044	0.298
Neutrophil count (109/L)	200	4.31 ± 0.54	4.06 ± 1.20	1.873	0.063
NLR	200	2.62 ± 0.83	2.03 ± 0.88	4.868	< 0.001
PLR	200	152.77 ± 31.30	130.68 ± 29.22	5.162	< 0.001
LMR	200	4.14 ± 1.08	4.16 ± 1.18	0.125	0.901

PSD: Post-stroke depression; NPSD: Non-post-stroke depression; MPV: Mean platelet volume; NLR: Neutrophil-to-lymphocyte ratio; PLR: Platelet-to-lymphocyte ratio; LMR: Lymphocyte-to-monocyte ratio.

Variable assignment

Values were assigned to variables that differed significantly in the univariate test for further analysis. Categorical variables included age, marital status, educational attainment, number of comorbid conditions, severity of neurological impairment, number of lesions (all as independent variables), and PSD (dependent variable). Continuous variables consisted of NLR and PLR, both as independent variables (Table 4).

Table 4 Assignment of variables.

Factors	Variable	Assignment
Age (years)	X1	< 60 = 0, ≥ 60 = 1
Marital status	X2	Married = 0, divorced/widowed = 1
Education level	X3	Senior high school or above = 0, below senior high school = 1
Number of comorbid conditions	X4	0 = 0, 1 = 1, ≥ 2 = 2
Severity of neurological impairment	X5	Mild = 0, moderate = 1, severe = 2
Number of lesions	X6	Single = 0, multiple = 1
NLR	X7	Continuous variable
PLR	X8	Continuous variable
Post-stroke depression	Y	No = 0, yes = 1

NLR: Neutrophil-to-lymphocyte ratio; PLR: Platelet-to-lymphocyte ratio.

Binary logistic regression analysis of risk factors for PSD

Multivariate binary logistic regression analysis identified educational level (P = 0.046), NLR (P < 0.001), and PLR (P < 0.001) as independent predictors for PSD in patients with chronic stroke. The PSD risk was 5.574 times greater in patients with educational attainment below high school relative to more educated counterparts (95%CI: 1.03-30.21). Each unit rise in NLR increases the PSD risk by 2.18-fold (95%CI: 1.45-3.27), whereas every 10-unit PLR increase correlated with a 31.7% risk elevation (adjusted OR = 1.32, 95%CI: 1.16-1.48). Detailed results are presented in Table 5.

Table 5 Binary logistic regression analysis of risk factors for post-stroke depression.

Factors	β	SE	Wald	P value	Exp (β)	95%CI
Age (years)	-1.023	0.686	2.223	0.136	0.359	0.094-1.380
Marital status	0.796	0.500	2.531	0.112	2.216	0.831-5.906
Education level	1.718	0.862	3.969	0.046	5.574	1.028-30.214
Number of comorbid conditions	0.105	0.276	0.145	0.703	1.111	0.647-1.907
Severity of neurological impairment	0.460	0.284	2.617	0.106	1.584	0.907-2.766
Number of lesions	0.169	0.694	0.059	0.808	1.184	0.304-4.615
NLR	0.780	0.207	14.175	< 0.001	2.181	1.453-3.274
PLR	0.027	0.006	19.528	< 0.001	1.028	1.015-1.040

NLR: Neutrophil-to-lymphocyte ratio; PLR: Platelet-to-lymphocyte ratio.

Construction of the nomogram

A nomogram was constructed to provide a visual representation of the developed risk prediction model. The nomogram enables PSD risk estimation by aggregating the scores assigned to each independent variable included in the model. By using 1000 bootstrap replicates for internal validation, the model achieved a concordance index of 0.806 (95%CI: 0.742-0.866), reflecting excellent predictive differentiation. The calibration plot showed strong alignment in high-risk strata but some divergence in low-medium risk segments (Figure 2).

Figure 2 Nomogram visualization of the post-stroke depression risk-prediction model and performance validation. A: Risk prediction nomogram for post-stroke depression; B: Internal validation with 1000 bootstrap replicates; C: Calibration curve evaluation. Severity means severity of neurological impairment. PSD: Post-stroke depression; NLR: Neutrophil-to-lymphocyte ratio; PLR: Platelet-to-lymphocyte ratio.

DISCUSSION

As a highly prevalent and devastating cardiovascular disorder, stroke usually triggers significant motor and neuropsychological impairments that precipitate the development of PSD[21]. Being multifactorial and complex in pathogenesis, PSD onset and progression are strongly associated with inflammatory mechanisms. These may involve microglial activation, astrocytic reactivity, nuclear factor kappa B signaling pathways, etc.[22]. This study specifically targets patients with chronic stroke to provide a comprehensive analysis of contributors to PSD occurrence in this population.

Identifying PSD symptoms and pinpointing risk factors in patients with chronic stroke not only facilitates long-term outcome prediction but also enables early and targeted therapeutic interventions[23]. The selection of patients with chronic stroke as the study cohort is justified by several key considerations. First, these patients exhibit relative clinical stability, as acute physiological stress responses and complications are typically well-managed, thereby allowing for a more precise evaluation of PSD contributors[24]. Second, the chronic phase of stroke represents a period of heightened vulnerability to PSD, during which patients often experience persistent functional disabilities, have diminished self-care capacity, and receive inadequate social support, all of which may exacerbate depressive symptoms[25]. Third, therapeutic interventions during this phase are often more effective, with tailored clinical strategies demonstrating the potential to improve recovery outcomes and the overall quality of life[26]. In this study, 96 (48.00%) of the 200 patients with chronic stroke were diagnosed with PSD, with prevalence aligning closely with the findings reported by Chau et al[27], who identified a PSD risk of 44.60%. In the analysis of demographic characteristics, compared with the NPSD group, the PSD group demonstrated a significantly higher proportion of patients aged ≥ 60 years (54.17%), divorced or widowed (28.13%), and with educational attainment below senior high school (64.58%). Regarding clinicopathological profiles, the PSD group also exhibited a greater prevalence of patients with ≥ 2 comorbidities (34.38%), moderate or severe neurological impairment (66.67% and 23.96%, respectively), and multiple lesions (69.79%) compared with the NPSD group. Furthermore, the detection of blood biochemical indicators revealed significantly high NLR and PLR in the PSD group compared with the NPSD group.

To further elucidate these associations, variables differed significantly in the univariate test—including age, marital status, educational attainment, number of comorbidities, degree of neurological impairment, lesion count, NLR, and PLR—were incorporated into a multivariate binary logistic regression model. The analysis identified educational attainment, NLR, and PLR as independent PSD predictors in patients with chronic stroke. Specifically, educational attainment below senior high school, and high NLR and PLR each increased the PSD risk in this population. Patients with educational attainment below senior high school may be more susceptible to PSD owing to limited cognitive reserve, impaired psychological stress coping mechanisms, insufficient social support, and lower rehabilitation adherence, which contribute to an increased psychological burden[28]. Conversely, individuals with higher educational attainment often prioritize the development of health literacy and building social networks. This encourages them to adopt healthy behaviors and enhances the social support they receive—both of which play a role in lowering PSD risk[29]. These findings align with the results of Ayasrah et al[30], which also identified limited education as an independent and significant risk factor for PSD in patients with stroke, corroborating our results. Previous studies have consistently demonstrated that increased NLR and PLR are strongly associated with an increased prevalence of PSD and adverse clinical outcomes[31,32]. Moreover, NLR- and PLR-mediated inflammatory processes lead to the liberation of numerous proinflammatory cytokines. This further compromises the blood-brain barrier, enabling white blood cells to penetrate more readily and thus raising the risk of stroke and PSD[33]. Supporting this, Hu et al[34] reported a significant correlation between high NLR and PLR and PSD in patients with ischemic stroke, suggesting their potential utility as biomarkers for the early prediction of PSD, which aligns closely with the findings of the present study. Another study indicated that NLR increases proportionally with depression severity, potentially serving as a biological marker for suicide susceptibility and behaviors in patients with depression. Conversely, PLR holds greater potential in predicting the prognosis of major depressive disorder[35]. However, both ratios face clinical controversies—their diagnostic accuracy for PSD may be constrained by moderate specificity/sensitivity and confounding variables such as comorbid infections or inflammatory diseases[36]. Further studies have indicated that other potential factors, such as latent infections and bacterial colonization, may increase the PSD risk. This can occur through the release of cytokines to trigger inflammatory responses and microbial metabolites crossing the blood-brain barrier[37]. In addition, numerous studies have proposed evidence-based clinical interventions to enhance the rehabilitation of patients with PSD. For instance, Hu et al[38] emphasized the importance of reducing self-defeating thoughts, improving self-efficacy, and effectively managing fatigue and pain in the design of rehabilitation programs, as these strategies significantly facilitate recovery. Similarly, Zhao et al[39] highlighted the efficacy of psychological interventions in patients with stroke and anxiety-depression comorbidity, demonstrating improvements in the activities of daily living, enhanced quality of life, alleviation of negative emotions, and amelioration of functional impairments.

Blood sampling within the first 24 hours after admission was conducted for several reasons. Primarily, it aligns with established stroke care guidelines for standardized biomarker testing windows. Furthermore, early sampling reduces potential confounding from therapeutic interventions, allowing for the optimal detection of variations in PSD-related biomarkers. A limitation is the potential oversight of time-dependent fluctuations of biomarkers, indicating the need for multi-phase sampling in future studies to better characterize biomarker trajectories.

Several limitations should be noted. The diagnostic thresholds of HAMD-24 require careful consideration when applied to patients with stroke, given its suboptimal specificity in this group. Additionally, given that the study mainly examined cases with > 1-year post-stroke, further investigation is needed to assess how inflammatory markers correlate with acute-phase PSD. Furthermore, the investigation omitted receiver operating characteristic analysis to quantify the discriminative capacity of NLR and PLR as PSD biomarkers in the chronic post-stroke period. Establishing validated cutoff values through this method would improve their clinical translational application. Moreover, the absence of social support scale integration may have led to the underestimation of the contribution of psychosocial factors. Moreover, single-timepoint biochemical data restrict longitudinal analysis, warranting multi-interval sampling in future designs to clarify temporal trends in these markers. Finally, the reliance on single-center data potentially affects broader applicability. Expanding to multicenter collaborations in future work would strengthen sample diversity and external validity. Further refinements will target these aspects for enhanced robustness.

CONCLUSION

In this study, low educational attainment, and high NLR and PLR independently increase the PSD risk in patients with chronic stroke. Clinicians should prioritize close monitoring and implement individualized, targeted interventions for patients exhibiting these risk factors to mitigate the PSD risk. Furthermore, the developed nomogram aids in stratifying PSD risk among chronic stroke survivors. Although well-calibrated for high-risk cases, its accuracy diminishes slightly in the medium-low risk demographic.