Risk factors and prediction model for depressive disorder in survivors of acute cerebral hemorrhage

Abstract

BACKGROUND

Post-intracerebral hemorrhage (ICH) depression is a prevalent and debilitating complication, adversely affecting recovery and survival. While identifying at-risk patients early is crucial, a comprehensive understanding of the specific risk factors in the acute phase remains limited. This study hypothesizes that a combination of clinical, imaging, and laboratory factors can effectively predict the onset of acute post-ICH depressive disorders. We hypothesized that advanced age, deep hematoma location, midline shift, low serum albumin, and high modified Rankin Scale (mRS) scores are independent risk factors, and a model combining these would demonstrate high predictive accuracy for acute post-ICH depression.

AIM

To investigate the risk factors and construct a prediction model for depressive disorder after acute ICH.

METHODS

This retrospective study analyzed 199 acute ICH survivors. Depression was assessed via Hamilton Depression Rating Scale 17-item version and confirmed by psychiatrists to rule out mimicking conditions/other psychiatric disorders. Univariate/multivariate logistic regression identified independent risk factors; a prediction model was built and evaluated via receiver operating characteristic (area under the curve for discriminatory power). Sample size (199) met the 180-190 minimum, estimated from 40%-60% assumed depression prevalence.

RESULTS

The depressive disorder group had older age, longer hospital stays, larger hematoma volumes, higher proportions of deep hematomas and brain midline shift, more severe brain atrophy, lower serum albumin, and worse neurological deficits (P < 0.05). Multivariate analysis identified older age, deep hematomas, brain midline shifts, low albumin, and increased mRS scores as independent risk factors (P < 0.05). The combined prediction model had an area under the curve of 0.885 (95% confidence interval: 0.832-0.926), 93.3% sensitivity, 67.0% specificity, and better predictive efficiency than single indicators.

CONCLUSION

A model incorporating age, deep hematoma, midline shift, albumin, and mRS score effectively predicts depression risk in acute ICH survivors, though external validation is required.

Key Words: Cerebral hemorrhage; Acute phase; Survivors; Depressive disorder; Influencing factors

Core Tip: This study identified five independent risk factors (advanced age, deep hematoma, midline shift, low albumin, high modified Rankin Scale score) for post-intracerebral hemorrhage depression in the acute phase. A predictive model combining these factors demonstrated high discriminatory power (area under the curve = 0.885). This model aids in the early clinical screening of high-risk individuals, facilitating timely intervention to improve long-term prognosis. Key findings emphasize the roles of structural brain injury, systemic inflammation/malnutrition (reflected by albumin), and functional dependence in depression pathogenesis after acute intracerebral hemorrhage.

INTRODUCTION

Intracerebral hemorrhage (ICH), which accounts for 28.8% of all strokes and causes about 3.41 million new cases annually with 2.89 million fatalities, is defined as bleeding brought on by the abrupt rupture of non-traumatic brain blood vessels[1]. Within 30 days of onset, the mortality rate of ICH is 40% to 50%, which is approximately twice that of ischemic stroke. At the same time, it is also the stroke subtype with the highest disability rate, with more than 70% of survivors still having functional impairment after 3 months[2]. With the widespread application of pre-hospital emergency care, intensive care unit monitoring and combined management strategies (early intensive blood pressure reduction, blood sugar control, temperature management and anticoagulation correction), more and more patients are surviving the acute phase, but survivors still face the long-term effects of multiple neurological deficits and psychological disorders. Among them, post-stroke depression (PSD) is one of the most common neuropsychiatric complications. Studies have reported that the prevalence of PSD in ICH is as high as 60%[3,4], almost twice that of the ischemic stroke group[5]. PSD not only weakens patients’ compliance with rehabilitation, prolongs hospital stay, and increases medical expenses, but is also closely related to decreased executive function, decreased ability to live daily, and increased 1-year all-cause mortality[6]. Therefore, early identification and intervention of PSD have become a key link in improving the long-term prognosis of survivors of cerebral hemorrhage[7,8]. Based on this, this study retrospectively enrolled 199 patients who received acute cerebral hemorrhage treatment in our hospital. The Hamilton Depression Rating Scale 17-item version was employed to evaluate the depressed status of the enrolled patients, and diagnoses were confirmed by a psychiatrist to enhance specificity and exclude other causes, and the demographic, imaging, laboratory, and neurological prognosis data were systematically collected. Investigating the risk factors for the development of depressive disorders in acute cerebral hemorrhage survivors and offering an evidence-based foundation for the early detection of high-risk individuals were the objectives of this study.

MATERIALS AND METHODS

General information

In this research, clinical data from 199 patients treated for acute cerebral hemorrhage in our hospital between January 2020 and January 2025 were gathered and analyzed using a retrospective research approach. According to whether the survivors of acute cerebral hemorrhage developed depressive disorders, they were split into two groups: 90 people with depressive disorder and 109 people without it. All patients and their families participating in this study had a clear and full understanding of the specific content of the research and agreed to complete the declaration of consent. In addition, this study has been rigorously reviewed by the Ethics Committee of our hospital and has been formally approved. A priori sample size calculation was performed using G*Power software. Based on an assumed medium effect size (odds ratio approximately is 2.0-3.0 for key risk factors), an alpha error of 0.05, and a power of 0.80, the estimated minimum sample size required was 180-190 participants. Our final sample of 199 meets this requirement.

Inclusion criteria: (1) Fulfilling the requirements for a cerebral hemorrhage diagnosis and being verified by imaging tests[9]; (2) Being a survivor of the acute phase (within 1 month of onset); (3) Being ≥ 18 years old; and (4) Not having previously used antidepressants or psychological interventions. Exclusion criteria: (1) Patients with severe liver, kidney, and heart problems; (2) Patients with malignant tumors; (3) Patients with systemic infectious diseases; (4) Patients with pre-existing major depressive disorder or other major psychiatric diagnoses (e.g., bipolar disorder, anxiety disorders); (5) Patients with medical conditions known to mimic depression (e.g., hypothyroidism, electrolyte imbalances) or ICH-related complications potentially causing apathy or cognitive-affective symptoms (e.g., significant post-ICH hydrocephalus); and (6) Patients with incomplete clinical data.

Data collection

The clinical data of the enrolled patients were collected, including general information, imaging indicators, laboratory indicators and related scoring scale data: (1) General information: Age, sex, length of hospital stay, history of atrial fibrillation, medication history (statins, anticoagulation); (2) Imaging indicators: Hematoma volume, hematoma location, rupture into the ventricle, cerebral edema, brain midline shift, grading of brain atrophy, subarachnoid hemorrhage; (3) Laboratory indicators: Proprotein, albumin, uric acid, random blood glucose, total cholesterol, triglycerides, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, prothrombin time, international normalized ratio, activated partial thromboplastin time, fibrinogen; and (4) Related scoring scales: Van Swieten score, Glasgow Outcome Scale (GOS) score, Mini-Mental State Examination (MMSE) score, modified Rankin scale (mRS) score.

Quality control

This study used standard spreadsheets for double-blind data collection. Within 24 hours of admission, two neurologists extracted demographic characteristics and medical history from the hospital information system and medical records, respectively, and cross-checked them. Imaging evaluation used 64-slice computed tomography (CT) scan data. The hematoma volume was first calculated semi-automatically by 3D-Slicer software, and then the imaging features such as brain atrophy and intraventricular hemorrhage were blindly evaluated by neuroimaging experts with the title of associate chief physician or above. Laboratory tests collected the first fasting venous blood after admission, and the Beckman Coulter AU5800 fully automatic biochemical analyzer was used to measure serum protein, blood sugar, blood lipids, and coagulation function. Acute phase evaluation was completed after onset. The Van Swieten score was based on magnetic resonance imaging-fluid-attenuated inversion recovery images within 7 days of onset, and the GOS, MMSE, and mRS scores were obtained 24 hours before discharge. The length of hospital stay data was directly exported from the hospital information system.

Evaluation criteria

Diagnosis of depressive disorders: The patients were routinely observed and interviewed by two qualified assessors using the 17-item Hamilton Depression Rating Scale. Additionally, all cases meeting Hamilton Depression Rating Scale 17-item version criteria (total score ≥ 7) were evaluated by a consulting psychiatrist to confirm the diagnosis of depressive disorder, rule out other psychiatric conditions (e.g., comorbid anxiety disorders, adjustment disorders), and exclude depressive symptoms attributable to general medical conditions (e.g., hypothyroidism, electrolyte imbalances) or direct physiological effects of ICH complications (e.g., hydrocephalus). After the examination, the two assessors independently scored and took the average to evaluate the severity of the disease and the treatment effect. The scale includes 17 items, each scored 0 point to 4 points or 0 point to 2 points. The higher the score, the more severe the depressive symptoms. In this study, a total score ≥ 7 points was defined as a depressive disorder[10,11].

Assessment of white matter lesions: The Van Swieten scale[12] is a standardized assessment tool based on CT imaging, which is mainly used to quantify the severity of leukoaraiosis at the level of the ventricular system. In this study, all CT scan results were independently evaluated by two professionally trained neurologists. The assessment process adopted a standardized grading method: The physician selected three consecutive CT sections in the anterior and posterior regions of the central sulcus and scored the degree of white matter lesions respectively. More severe white matter lesions were indicated by higher scores, which ranged from 0 point to 4 points and represented the sum of the scores of the two regions[13].

Prognostic assessment: The GOS is a tool employed to determine the consequences of individuals after a head injury[14]. The scale uses a 5-point scoring standard: 5 points indicate good recovery, with the patient having mild functional impairment but basically being able to take care of himself; 4 points indicate moderate disability, with the patient being able to live independently but with some activity limitations; 3 points indicate severe disability, with the patient requiring assistance from others in daily life; 2 points indicate a persistent vegetative state, with the patient only retaining sleep-wake cycles and basic eye movements; and 1 point indicates death.

Cognitive function assessment: The MMSE was employed to determine the mental abilities of the subjects[15]. This scale is a standardized cognitive function screening tool that can be completed in 5 minutes to 10 minutes. It covers five domains: Orientation (10 points), memory (6 points), attention/calculation (5 points), language (8 points), and visuospatial skills (1 point). Cognitive impairment is indicated by a total MMSE score of less than 27, and adequate cognitive function is indicated by a score of ≥ 27.

Assessment of neurological function recovery: The mRS was employed to determine the clients neurological recovery[16]. This scale is an internationally used stroke prognosis assessment tool that uses a 6-level scoring standard (0 point to 5 points), with the following specific classifications: 0 point for no symptoms; 1 point for symptoms but no significant functional impairment, and the patient can take care of himself completely; 2 points for mild disability, limited daily activities but still able to live independently; 3 points for moderate disability, requiring partial assistance but able to walk independently; 4 points for severe disability, loss of walking ability and needing care from others; 5 points for extremely severe disability, bedridden and completely dependent on others for care. The higher the score, the more severe the neurological deficit.

Statistical analysis

SPSS 21.0 was employed to examine the data. The t test was used to compare normally distributed, homoscedastic continuous variables (mean ± SD), whereas the Mann-Whitney U was used to compare the medians (interquartile range). n (%) were used to summarize categorical data, and χ² was used to compare them. To find independent predictors of post-ICH depression, variables with significant differences were incorporated into stepwise forward binary logistic regression. To evaluate model performance, MedCalc was employed to create receiver operating characteristic (ROC) curves and compute area under the curves. P < 0.05 was deemed significant.

RESULTS

Baseline information

The baseline characteristics of 199 acute ICH survivors were analyzed and it was found that the depression disorder group (n = 90) and the non-depression disorder group (n = 109) had significant differences in age, hospitalization time, hematoma volume, hematoma location, brain midline shift, brain atrophy grade, albumin and mRS score (P < 0.05), and other variables such as gender, history of atrial fibrillation, medication history, blood lipids, etc., were not distinguishing between both groups (P > 0.05; Table 1).

Table 1 Comparison of baseline data, n (%)/mean ± SD/medians (interquartile range).

Grouping1	Depression group (n = 90)	Non-depressed group (n = 109)	t/χ2	P value
Age (years)	67.18 ± 13.14	61.64 ± 14.84	2.757	0.006
Gender			0.051	0.821
Male	60 (66.67)	71 (65.14)
Female	30 (33.33)	38 (34.86)
Length of hospital stay (days)	31.98 ± 15.25	26.44 ± 14.19	2.649	0.009
History of atrial fibrillation	5 (5.56)	5 (4.59)	0.097	0.756
Medication history
Statins	23 (25.56)	24 (22.02)	0.342	0.559
Anticoagulation	37 (41.11)	46 (42.20)	0.024	0.877
Imaging indicators
Hematoma volume (mL)	88.62 (36.95, 113.80)	73.14 (49.80, 99.11)	2.001	0.045
Hematoma location			8.284	0.040
Lobe	17 (18.89)	35 (32.11)
Deep (basal ganglia/thalamus)	62 (68.89)	58 (53.21)
Infratentorial (cerebellum/brainstem)	9 (10.00)	8 (7.34)
Mixed area	2 (2.22)	8 (7.34)
Break into ventricle	25 (27.78)	35 (32.11)	0.439	0.507
Brain edema	16 (17.78)	20 (18.35)	0.011	0.917
Midline brain shift	36 (40.00)	25 (22.94)	6.753	0.009
Brain atrophy grade			7.916	0.048
0 points	37 (41.11)	65 (59.63)
1 point	29 (32.22)	27 (24.77)
2 points	23 (25.56)	17 (15.60)
3 points	1 (1.11)	0 (0.00)
Subarachnoid hemorrhage	8 (8.89)	15 (13.76)	1.145	0.285
Laboratory indicators
Preprotein (mg/L)	231.48 ± 69.02	241.79 ± 67.04	1.065	0.288
Albumin (g/L)	33.33 (30.36, 35.91)	37.94 (34.88, 41.24)	6.727	< 0.001
Uric acid (μmol/L)	284.17 ± 117.20	289.73 ± 122.50	0.325	0.745
Random blood glucose (mmol/L)	5.57 (4.88, 6.50)	5.43 (4.87, 6.37)	0.579	0.563
Total cholesterol (mmol/L)	4.26 ± 0.95	4.40 ± 0.93	0.981	0.328
Triglycerides (mmol/L)	1.59 ± 0.99	1.76 ± 0.95	1.220	0.224
High-density lipoprotein cholesterol (mmol/L)	1.01 (0.88, 1.21)	1.09 (0.96, 1.25)	2.074	0.038
Low-density lipoprotein cholesterol (mmol/L)	2.68 ± 0.74	2.71 ± 0.68	0.273	0.785
Prothrombin time (seconds)	11.78 ± 1.15	11.84 ± 1.61	0.323	0.747
International normalized ratio	1.04 ± 0.11	1.05 ± 0.15	0.518	0.605
Activated partial thromboplastin time (seconds)	30.63 ± 4.19	30.06 ± 3.55	1.046	0.297
Fibrinogen (g/L)	3.54 ± 1.00	3.47 ± 1.27	0.406	0.685
Related rating scale
Van Swieten score (points)	1.03 (0.00, 2.00)	0.86 (0.00, 1.00)	1.457	0.145
Glasgow Outcome Scale score (points)	4.51 ± 0.72	4.48 ± 0.74	0.326	0.744
Mini-Mental State Examination score (points)	26.58 ± 6.28	25.95 ± 6.44	0.692	0.490
Modified Rankin Scale score (points)	2.77 (2.00, 3.00)	1.85 (1.00, 2.00)	6.160	< 0.001

¹This table compares the baseline characteristics of acute intracerebral hemorrhage survivors in the depression group and non-depression group using appropriate statistical methods. Normally distributed continuous variables (e.g., age, expressed as mean ± SD) were compared by independent-samples t-test; non-normally distributed continuous variables [e.g., hematoma volume, expressed as median (interquartile range)] were compared by Mann-Whitney U test; categorical variables [e.g., gender, expressed as n (%)] were compared by Pearson’s χ² test. The results showed significant differences in age, hematoma volume and other indicators between the two groups (P < 0.05).

Binary logistic regression analysis

Binary logistic regression analysis was utilized on the 8 statistically significant factors in Table 1, and whether the survivors of acute cerebral hemorrhage developed depression was used as the dependent variable. After controlling for confounding factors (hospitalization time, hematoma volume, and brain atrophy grade), it was found that the independent risk factors affecting the appearance of depression in survivors of acute cerebral hemorrhage were increased age, deep hematoma, midline shift of the brain, decreased albumin, and increased mRS score (P < 0.05; Tables 2 and 3).

Table 2 Univariate logistic regression analysis.

Variable1	β	SE	Wald	OR (95%CI)	P value
Age	0.050	0.018	7.950	1.051 (1.015-1.088)	0.005
Length of hospital stay	0.035	0.018	3.780	1.036 (1.000-1.073)	0.052
Hematoma volume	0.606	0.322	3.542	1.833 (0.975-3.446)	0.059
Hematoma location
Lobe			5.854	1	0.119
Deep	1.171	0.543	4.658	3.225 (1.114-9.341)	0.031
Infratentorial	0.883	0.811	1.188	2.419 (0.494-11.847)	0.276
Mixed area	-0.345	1.227	0.079	0.708 (0.064-7.846)	0.778
Midline brain shift	1.040	0.499	4.336	2.828 (1.063-7.524)	0.037
Brain atrophy grade
0 points			3.680	1	0.298
1 point	0.909	0.498	3.328	2.482 (0.935-6.593)	0.068
2 points	0.623	0.571	1.189	1.864 (0.609-5.707)	0.275
3 points	23.400	40192.970	0.000		1.000
Albumin	-0.298	0.054	30.883	0.742 (0.668-0.824)	< 0.001
Modified Rankin Scale score	1.219	0.254	22.982	3.384 (2.056-5.571)	< 0.001

¹This table analyzes the association between each baseline indicator and depression in acute intracerebral hemorrhage survivors by univariate logistic regression. Continuous variables (e.g., albumin) and categorical variables (e.g., hematoma location) were included as independent variables, with “presence of depression” as the dependent variable. For categorical variables, “lobe” (hematoma location) and “0 points” (brain atrophy grade) served as reference groups. The results identified 8 indicators with statistical differences (P < 0.05) for subsequent multivariate analysis.

OR: Odds ratio; CI: Confidence interval.

Table 3 Multivariate logistic regression analysis.

Variable1	β	SE	Wald	OR (95%CI)	P value
Age	0.039	0.015	6.812	1.040 (1.010-1.071)	0.009
Deep hematoma	0.850	0.416	4.177	2.340 (1.035-5.288)	0.041
Midline brain shift	1.107	0.444	6.209	3.026 (1.267-7.231)	0.013
Albumin	-0.283	0.049	32.963	0.753 (0.684-0.830)	< 0.001
Modified Rankin Scale score	1.110	0.225	24.286	3.036 (1.952-4.721)	< 0.001

¹This table identifies independent risk factors for depression in acute intracerebral hemorrhage survivors by multivariate logistic regression, with confounding factors (hospitalization time, hematoma volume, brain atrophy grade) controlled. Variables with statistical differences in univariate analysis were included, and “presence of depression” was the dependent variable. The results showed that advanced age, deep hematoma, brain midline shift, low albumin level and elevated modified Rankin Scale score were independent risk factors for depression (P < 0.05).

OR: Odds ratio; CI: Confidence interval.

ROC curve analysis

The ROC results showed that serum albumin level had the strongest predictive ability, followed by mRS score, and age, deep hematoma and brain midline shift had moderate predictive efficacy. After the above five indicators were combined to construct a prediction model, the area under the ROC curve was 0.885, indicating that advanced age, deep hematoma, brain midline shift, low albumin level and increased mRS score are independent risk factors for depression in survivors of acute cerebral hemorrhage (Table 4 and Figure 1).

Figure 1 Combined receiver operating characteristic curve results.

Table 4 Receiver operating characteristic curve analysis results.

Variable1	AUC	Youden index	Optimal cutoff value	Sensitivity (%)	Specificity (%)	95%CI	P value
Age	0.608	0.178	58	75.56	42.20	0.536-0.676	0.007
Deep hematoma	0.578	0.157	1	68.89	46.79	0.507-0.648	0.022
Midline brain shift	0.585	0.171	1	40.00	77.06	0.514-0.655	0.010
Albumin	0.777	0.459	36.23	78.89	66.97	0.713-0.833	< 0.001
Modified Rankin Scale score	0.742	0.398	2	60.00	79.82	0.676-0.801	< 0.001
Combined receiver operating characteristic	0.885	0.603	0.279	93.33	66.97	0.832-0.926	< 0.001

¹This table evaluates the predictive efficacy of individual indicators and the combined model for depression in acute intracerebral hemorrhage survivors by receiver operating characteristic curve analysis. Area under the curve, sensitivity, specificity and 95% confidence interval were calculated to assess discriminatory power. The results showed that the combined model had the highest area under the curve (0.885), with sensitivity of 93.3% and specificity of 67.0%, indicating better predictive efficacy than single indicators (P < 0.001).

AUC: Area under the curve; CI: Confidence interval.

DISCUSSION

The acute phase of ICH (usually within 1 month after onset) is a critical stage for neurological recovery and psychological adaptation, and is also a high-incidence period for depressive symptoms[17]. However, there are relatively few studies on depressive disorders in the acute phase of ICH, and the influencing factors have not yet been fully clarified. The mechanism of depression in patients with ICH involves very complex pathophysiological processes, such as direct damage to the limbic system by hematoma, and secondary inflammatory response and neurotransmitter imbalance[18,19]. In addition, social and psychological factors (such as lack of social support and pre-morbid personality traits) may also play an important role in the occurrence of acute depression. Therefore, systematic analysis of the influencing factors of depressive disorders in the acute phase of ICH is of great significance for early identification of high-risk patients and implementation of targeted interventions. After controlling for potential confounding factors such as hospitalization time, hematoma volume, and degree of brain atrophy, this study found that increased age, deep hemorrhage, midline shift of the brain, decreased serum albumin, and increased mRS scores were independent risk variables for depressive disorders in survivors of acute ICH (P < 0.05).

Ageing is an independent risk variable for depressive disorder in the acute phase of ICH, and its mechanism involves multiple pathophysiological and psychosocial changes[20]. From a neurobiological perspective, the brain tissue repair capacity of elderly patients is reduced, especially the neuroplasticity of the hippocampus is reduced, which may lead to an imbalance of key neurotransmitters such as serotonin and brain-derived neurotrophic factor, directly affecting mood regulation[21,22]. While our study found no significant difference in MMSE scores between groups, the association between cognitive function assessed by MMSE and depression warrants further investigation in larger cohorts. At the same time, elderly patients often have multiple chronic diseases and are more concerned about the prognosis of the disease. In addition, their social support system may be insufficient, which makes them more likely to feel helpless and lonely, which together promotes the occurrence of depressive symptoms. Therefore, in clinical practice, older patients with ICH should undergo routine screening for depressed symptoms, and specific assessment instruments like the Geriatric Depression Scale should be taken into account.

Deep hematomas, especially hemorrhages in the thalamus and basal ganglia, significantly increase the risk of depressive disorders in the acute phase of ICH. The mechanism is mainly related to the functional impairment of specific brain regions[23]. The thalamus and basal ganglia are key nodes in the prefrontal-limbic circuit. Hemorrhages in these regions can directly damage the emotion regulation pathway. For example, abnormal connections between the anterior thalamic nucleus and the cingulate gyrus may lead to emotional processing dysfunction[24]. In addition, these deep nuclei are closely related to dopaminergic and serotonergic pathways. Neurotransmitter disorders after hemorrhage may manifest as significant anhedonia and depression. Thalamic hemorrhage may also increase the risk of hydrocephalus due to compression of the third ventricle, further leading to apathy and depressive symptoms. If deep hematomas are found during imaging examinations, especially involvement of the thalamus, one should be highly alert to the occurrence of depressive disorders and consider early intervention.

Brain midline shift is an important predictor of post-cerebral hemorrhage depression, and its influence is mainly achieved through whole-brain network damage and secondary pathological changes[25]. Midline shift indicates a significant mass effect, which may lead to abnormal connections between the bilateral cerebral hemispheres, thereby affecting the integration and regulation of emotions. Severe shift can also lead to brainstem compression, especially in the midbrain and upper pons, destroy the ascending reticular activating system, and induce core symptoms of depression such as loss of willpower[26,27]. In addition, vascular traction associated with shift may cause inadequate perfusion of distal brain tissue and aggravate psychological stress related to neurological deficits. In clinical practice, patients with midline shift exceeding 5 mm should be classified as high-risk for depression, and their psychiatric symptoms should be dynamically monitored. If necessary, neurosurgery should be consulted to evaluate the possibility of surgical intervention.

Low albumin levels are closely related to the occurrence of depressive disorders in the acute phase of cerebral hemorrhage, and its mechanism of action covers nutrition, inflammation, metabolism and other aspects[28,29]. As an important antioxidant and anti-inflammatory protein, reduced albumin levels may lead to oxidative stress and the release of proinflammatory factors. While our study did not measure specific inflammatory markers like interleukin-6 or tumor necrosis factor α, the association between hypoalbuminemia and systemic inflammation suggests that future research should incorporate these biomarkers to better elucidate the “hypoalbuminemia - inflammation - depression” pathway. In addition, low albumin levels may change drug metabolic kinetics and increase free drug concentrations. However, in the acute phase of cerebral hemorrhage, the use of antidepressants is limited, which may increase the risk of depression[30]. Beyond altered pharmacokinetics, low albumin likely reflects a state of systemic inflammation and oxidative stress, which can directly impact central neurotransmitter balance and neuroplasticity via the neuroimmune axis, providing a more direct pathophysiological link to depression in the acute ICH setting. Low albumin levels also often reflect poor overall nutritional status or swallowing disorders in patients, indirectly indicating that rehabilitation is difficult and further increasing the psychological burden. Therefore, it is important to regularly monitor depression symptoms and enhance nutritional support for individuals with low serum albumin levels.

An elevated mRS score is also a predictor of post-cerebral hemorrhage depression, and its mechanism is mainly related to functional dependence and psychological adaptation disorder. A high mRS score indicates that the patient’s daily living ability is significantly limited and needs to rely on others for care. This loss of autonomy can easily lead to a sense of uselessness and despair[31]. In addition, people with severe functional impairments tend to be pessimistic about their recovery expectations, forming a negative cognitive cycle that further aggravates depressive symptoms. For such patients, early psychological intervention and personalized rehabilitation goal setting are essential to break negative cognitive patterns and enhance confidence in treatment.

In summary, advanced age, deep hemorrhage, midline shift, low albumin level and elevated mRS score are independent risk factors for depressive disorder in the acute phase of ICH. The joint prediction model constructed based on the above variables performed well after internal validation, with an area under the ROC curve of 0.885, which provides strong support for the rapid screening of high-risk patients in the acute phase. A key limitation of this study is its single-center, retrospective design and the potential for unmeasured confounding factors, which may limit the generalizability of our findings. The homogeneity of our patient population in terms of geographic origin and treatment protocols also necessitates caution when applying the model to other settings. Future research should prioritize external validation in multicenter, prospective cohorts. Affected by the single-center, retrospective design and potential unmeasured confounding factors, and the lack of long-term follow-up data, the extrapolation and stability of the model still need to be further verified. In the future, external validation should be carried out in multicenter, prospective cohorts, systematically incorporating genetic susceptibility, inflammatory markers and social psychology indicators, and evaluating the intervention effect through randomized controlled trials; at the same time, remote digital technology can be combined to explore the “hospital-family” integrated depression prevention and treatment network, in order to improve the emotional recovery level and overall prognosis quality of ICH survivors on a larger scale.

CONCLUSION

The prediction model incorporating age, deep hematoma, midline shift, albumin, and mRS score demonstrates high discriminatory ability for identifying acute ICH survivors at risk for depressive disorder. This tool holds promise for early clinical screening. However, its generalizability is limited by the single-center retrospective design. Future multicenter prospective studies are essential to validate and refine this model, potentially incorporating additional biomarkers like inflammatory cytokines, to enhance its clinical utility and robustness.