Predictive value of preoperative shock index for postoperative outcomes in traumatic splenic rupture patients undergoing splenectomy

Abstract

BACKGROUND

Traumatic splenic rupture is the most common solid organ injury in blunt abdominal trauma, and splenectomy remains a critical intervention for patients with hemodynamic instability. The shock index (SI), a simple hemodynamic indicator, has demonstrated significant value in prognostic assessment of trauma patients. However, studies on the predictive value of SI in patients undergoing splenectomy for traumatic splenic rupture are relatively limited.

AIM

To investigate the predictive value of preoperative SI on postoperative complications, mortality, and long-term survival in patients with traumatic splenic rupture undergoing splenectomy, and to construct a prognostic prediction model based on SI.

METHODS

Clinical data of 212 patients with traumatic splenic rupture who underwent splenectomy from January 2020 to January 2025 were retrospectively analyzed. Patients were divided into low SI group (SI < 0.9, n = 78), moderate SI group (0.9 ≤ SI < 1.3, n = 89), and high SI group (SI ≥ 1.3, n = 45) based on preoperative SI (SI = heart rate/systolic blood pressure). Basic information, clinical indicators at admission, laboratory tests, imaging examinations, surgery-related indicators, and prognostic data were collected. The Clavien-Dindo grading system was used to assess postoperative complications. Univariate and multivariate logistic regression analyses were performed to identify independent risk factors for severe postoperative complications. A prognostic prediction model was constructed and its discriminative ability was evaluated using receiver operating characteristic curves. Kaplan-Meier method and Cox regression model were used for survival analysis.

RESULTS

With increasing SI, patients showed significantly decreased blood pressure, significantly increased heart rate and respiratory rate, more severe splenic injury, and significantly increased intraoperative blood loss and transfusion volume (all P < 0.001). The incidence of severe postoperative complications in the high SI group (26.7%) was significantly higher than that in the moderate SI group (12.4%) and low SI group (3.8%) (P < 0.001). Multivariate logistic regression analysis showed that SI [odds ratio (OR) = 2.34, 95% confidence interval (CI): 1.25-4.38, P = 0.008], splenic injury grade ≥ IV (OR = 2.97, 95%CI: 1.28-6.86, P = 0.011), hemoglobin ≤ 90 g/L (OR = 3.14, 95%CI: 1.40-7.07, P = 0.005), lactate level ≥ 4.0 mmol/L (OR = 3.97, 95%CI: 1.76-8.92, P = 0.001), and massive intraperitoneal hemorrhage (OR = 2.43, 95%CI: 1.02-5.78, P = 0.045) were independent risk factors for severe postoperative complications. The multivariate prediction model based on SI had an area under the curve of 0.847 (95%CI: 0.789-0.905) for predicting severe postoperative complications, which was significantly superior to SI alone (area under the curve = 0.782, P = 0.033). The optimal cutoff value of SI for predicting severe postoperative complications was 1.15, with sensitivity of 76.9% and specificity of 81.2%. Survival analysis showed that 1-year survival rates in the low, moderate, and high SI groups were 98.7%, 96.6%, and 86.7%, respectively (P = 0.015). Cox regression analysis confirmed that SI was an independent risk factor affecting patient survival (hazard ratio = 1.85, 95%CI: 1.16-2.95, P = 0.010).

CONCLUSION

Preoperative SI is an important predictor of prognosis in patients with traumatic splenic rupture undergoing splenectomy, which can effectively identify high-risk patients and guide clinical decision-making.

Key Words: Traumatic splenic rupture; Shock index; Splenectomy; Prognostic prediction; Complications

Core Tip: This study demonstrates that the preoperative shock index (SI), calculated as heart rate divided by systolic blood pressure, is a powerful predictor of postoperative complications and long-term survival in patients undergoing splenectomy for traumatic splenic rupture. By integrating SI with key clinical variables, we developed a prognostic model with strong discriminative performance. An SI threshold of 1.15 effectively identified high-risk patients. These findings support SI-based early risk stratification and provide evidence for optimizing surgical decision-making and perioperative management.

INTRODUCTION

Traumatic splenic rupture is the most common solid organ injury in blunt abdominal trauma, accounting for 40%-55% of abdominal trauma, with a mortality rate of 10%-15%[1]. The spleen has abundant blood supply, and rupture can easily lead to massive hemorrhage. Without timely treatment, it can rapidly cause hemorrhagic shock, seriously threatening patient life[2]. With the development of trauma treatment concepts, conservative splenic treatment and splenorrhaphy have gradually gained attention. However, for patients with severe splenic rupture accompanied by hemodynamic instability, splenectomy remains an important means of saving lives[3].

Accurate preoperative assessment of disease severity and prognostic risk is of great significance for formulating reasonable treatment strategies. The shock index (SI), as a simple indicator reflecting hemodynamic status, is calculated as the ratio of heart rate to systolic blood pressure and has the advantages of simple operation and easy acquisition[4]. Since first proposed by Allgöwer and Burri in 1967, SI has been widely used in the assessment of critically ill patients with trauma and hemorrhagic shock. The normal adult SI is usually 0.5-0.7; when SI ≥ 1.0, it suggests hemodynamic abnormalities, and SI ≥ 1.3 often indicates blood loss exceeding 30% of body weight[5].

Recent studies have shown that SI has important value in predicting the prognosis of trauma patients. Some scholars have found that SI is closely related to transfusion requirements, surgical complexity, length of hospital stay, and mortality in trauma patients[6]. However, research on the predictive value of SI in patients with traumatic splenic rupture, particularly its predictive effect on postoperative complications and long-term prognosis in splenectomy patients, is relatively limited. Previous studies have mostly focused on analysis of single prognostic indicators, lacking comprehensive evaluation of multi-dimensional prognostic outcomes such as postoperative complications and survival rates[7].

Based on the above background, this study retrospectively analyzed clinical data of 212 patients with traumatic splenic rupture who underwent splenectomy, aiming to explore the predictive value of preoperative SI on postoperative complications, mortality, and long-term survival, and to construct a prognostic prediction model based on SI to provide scientific evidence for clinical decision-making[8].

MATERIALS AND METHODS

Study subjects

A retrospective analysis was conducted on 212 patients with traumatic splenic rupture admitted to the Department of Emergency and Department of Surgical of our hospital from January 2020 to January 2025. This study was conducted at Fu Yang People’s Hospital (a tertiary general hospital serving as a regional trauma center, Fuyang, Anhui Province, China) and the First Affiliated Hospital of Bengbu Medical University (a provincial trauma center, Bengbu, Anhui Province, China). These centers treat approximately 2500 trauma patients annually and are equipped with 24-hour emergency surgical teams and intensive care units. This study was approved by the hospital ethics committee, and all patients or their families signed informed consent forms.

Inclusion criteria: (1) Age ≥ 18 years; (2) Traumatic splenic rupture confirmed by computed tomography (CT), ultrasound, or intraoperative exploration; (3) Complete hemodynamic data at admission, allowing accurate calculation of SI (heart rate/systolic blood pressure); (4) Underwent splenectomy treatment; and (5) Complete clinical and follow-up data with follow-up time ≥ 12 months.

Exclusion criteria: (1) Previous history of splenic diseases (such as splenomegaly, splenic cysts, hematologic diseases involving the spleen, etc.); (2) Had received spleen-related treatment at other hospitals before admission; (3) Pregnant patients; (4) Combined with severe cardiovascular and cerebrovascular diseases, malignant tumors, and other serious underlying diseases that significantly affect prognosis; (5) Missing vital sign data required for preoperative SI calculation; and (6) Patients who died within 24 hours postoperatively (unable to assess postoperative prognosis).

Grouping method: Based on the clinical significance grading standards of SI proposed by Allgöwer and Burri and combined with previous literature reports[9-12], patients were divided into: Low SI group (SI < 0.9, n = 78), moderate SI group (0.9 ≤ SI < 1.3, n = 89), and high SI group (SI ≥ 1.3, n = 45).

Data collection indicators

General information: Basic demographic characteristics and clinical background data of patients were collected: (1) Demographic characteristics: Gender, age, body mass index (BMI), grouped according to Chinese adult BMI standards; (2) Mechanism of injury: Detailed recording of trauma types, including traffic accidents, falls from height, heavy object impact, sports injuries, violent injuries, etc.; (3) Time-related factors: Time from injury to admission, time from admission to surgery; (4) Past medical history: Hypertension, diabetes, coronary heart disease, chronic kidney disease, liver disease, etc.; (5) Medication history: Anticoagulant and antiplatelet drug use; and (6) Lifestyle: Smoking history, alcohol consumption history, etc.

Clinical indicators at admission: Vital signs at admission were recorded: Systolic blood pressure, diastolic blood pressure, mean arterial pressure, heart rate, respiratory rate, body temperature. SI (SI = heart rate/systolic blood pressure) was calculated as the core predictive indicator. The Glasgow Coma Scale (GCS) was used to assess consciousness: (1) Eye opening response (E, 1-4 points); (2) Verbal response (V, 1-5 points): No verbal response 1-point, incomprehensible sounds 2 points, confused speech 3 points, inappropriate speech 4 points, good orientation 5 points; and (3) Motor response (M, 1-6 points). Total score 3-15 points, ≤ 8 points for severe consciousness disorder, 9-12 points for moderate consciousness disorder, 13-15 points for mild consciousness disorder or clear consciousness. Splenic injury severity was assessed according to the American Association for the Surgery of Trauma splenic injury grading criteria: Grade I is subcapsular hematoma < 10% surface area or capsular laceration < 1 cm depth; grade II is subcapsular hematoma 10%-50% surface area, intraparenchymal hematoma < 5 cm, or capsular laceration 1-3 cm depth not involving trabecular vessels; grade III is subcapsular hematoma > 50% surface area, intraparenchymal hematoma ≥ 5 cm, or capsular laceration > 3 cm depth involving trabecular vessels; grade IV involves segmental or hilar vascular injury producing major devascularization (> 25% of spleen); grade V is completely shattered spleen or hilar vascular avulsion with complete devascularization.

Laboratory test indicators: Laboratory test results at admission were collected: Blood routine indicators (hemoglobin, red blood cell count, hematocrit, white blood cell count, platelet count), coagulation function indicators (prothrombin time, activated partial thromboplastin time, international normalized ratio, fibrinogen), biochemical indicators (serum creatinine, blood urea nitrogen, total bilirubin, alanine aminotransferase, aspartate aminotransferase, lactate dehydrogenase), electrolyte levels (serum sodium, potassium, chloride), blood gas analysis results (pH value, arterial oxygen partial pressure, arterial carbon dioxide partial pressure, base excess, lactate level).

Imaging examination: CT examination results were independently analyzed by two radiologists with more than 5 years of abdominal imaging diagnostic experience, recording splenic rupture extent, hematoma size, intraperitoneal hemorrhage volume, and whether combined with other organ injuries. Disagreements were arbitrated by the department director. Standardized imaging assessment forms were used, with intraperitoneal hemorrhage volume classified according to CT findings as: Small amount (small amount of blood in pelvis), moderate amount (blood reaching umbilical level), large amount (blood exceeding umbilical level or whole abdominal cavity hemorrhage).

Surgery-related indicators: Surgery-related indicators were recorded: (1) Surgical timing: Emergency surgery (performed within 4 hours of admission), delayed surgery (performed more than 4 hours after admission); (2) Surgical approach: Open splenectomy or laparoscopic splenectomy; (3) Operative time (from skin incision to abdominal closure, minutes); (4) Intraoperative blood loss: Calculated using weighing method for gauze blood loss + aspirated blood volume - irrigation fluid volume (mL); (5) Intraoperative transfusion volume: Red blood cells, plasma, platelet usage; (6) Intraoperative blood pressure maintenance; and (7) Vasoactive drug use.

Main prognostic indicators: Main prognostic outcomes were assessed: (1) Postoperative complications: Standardized assessment using the Clavien-Dindo grading system, grade I for deviation from normal postoperative course without specific treatment; grade II requiring drug treatment, blood transfusion, or total parenteral nutrition; grade III requiring surgical, endoscopic, or interventional treatment (IIIA without general anesthesia, IIIB requiring general anesthesia); grade IV for life-threatening requiring intensive care unit treatment (IVA single organ dysfunction, IVB multi-organ dysfunction); grade V for death; (2) 30-day postoperative mortality; (3) Intensive care unit admission rate; (4) Mechanical ventilation time; and (5) 1-year survival rate.

Secondary prognostic indicators: Postoperative recovery was monitored: Hemoglobin level changes at 24 hours, 48 hours, 72 hours postoperatively, reoperation rate. The modified Rankin Scale was used to assess functional status at discharge: 0 points completely asymptomatic, 1 point no significant functional deficit, 2 points mild functional deficit but able to handle own affairs, 3 points moderate functional deficit requiring help but able to walk independently, 4 points moderate to severe functional deficit unable to walk independently, 5 points severe functional deficit bedridden requiring continuous care, 6 points death.

Follow-up method: Follow-up was conducted through outpatient visits combined with telephone follow-up. Telephone follow-up used standardized questionnaires implemented by trained dedicated personnel. Loss to follow-up was defined as failure to contact by phone for three consecutive times at different times and no outpatient visits.

Statistical analysis

Statistical analysis was performed using SPSS version 26.0 and R version 4.3.0 software. Normality testing was first performed (Shapiro-Wilk test). Continuous variables following normal distribution were expressed as mean ± SD, skewed variables as median (interquartile range), and categorical variables as n (%).

For comparison of continuous variables between groups: Analysis of variance (ANOVA) was used for normal distribution with equal variances, Welch test for unequal variances; Kruskal-Wallis H test for non-normal distribution. Post hoc pairwise comparisons used Bonferroni correction. Categorical variables were compared using χ² test, with Fisher’s exact test when theoretical frequency < 5.

Univariate and multivariate logistic regression analyses were used to identify independent risk factors affecting poor postoperative prognosis. Variables with P < 0.10 in univariate analysis were included in multivariate analysis, using forward method (forward likelihood ratio) for variable selection, calculating odds ratio (OR) and 95% confidence interval (CI). The liberal P < 0.10 threshold was chosen to avoid excluding potentially important predictors while using forward stepwise regression to control overfitting, with model stability validated through Bootstrap resampling and cross-validation. Hosmer-Lemeshow test was used to assess model goodness of fit.

A prognostic prediction model based on SI was constructed, with receiver operating characteristic (ROC) curves used to assess discriminative ability, calculating area under the curve (AUC) and 95%CI, and using Youden index to determine optimal cutoff value. Calibration plots were drawn to assess consistency between predicted probability and actual incidence. Survival analysis used Kaplan-Meier method to draw survival curves, with log-rank test for group comparisons. Cox proportional hazards regression model analyzed independent risk factors affecting survival, with proportional hazards assumption verified through Schoenfeld residual test. Model validation used bootstrap resampling (1000 times) for internal validation, calculating corrected C-index. Ten-fold cross-validation was used to assess model stability. Decision curve analysis was drawn to assess clinical net benefit. All statistical tests were two-sided, with P < 0.05 considered statistically significant.

RESULTS

Patient basic characteristics

This study finally included 212 patients with traumatic splenic rupture who underwent splenectomy, divided into low SI group 78 cases (36.8%), moderate SI group 89 cases (42.0%), and high SI group 45 cases (21.2%) based on preoperative SI. Comparison of basic characteristics among the three groups is shown in Table 1. The time from admission to surgery in the high SI group was significantly shorter than the other two groups (P < 0.001, ε² = 0.089), suggesting more critical condition requiring emergency surgical treatment. Dunn-Bonferroni post hoc test for pairwise comparison (corrected α = 0.0167) showed: High SI group vs low SI group (P < 0.001), high SI group vs moderate SI group (P = 0.012), moderate SI group vs low SI group (P = 0.135). There were no statistically significant differences among the three groups in baseline characteristics such as age, gender, BMI, mechanism of injury, and past medical history (all P > 0.05).

Table 1 Comparison of basic characteristics among three groups, mean ± SD/n (%)/median (interquartile range).

Indicator	Low SI group (SI < 0.9, n = 78)	Moderate SI group (0.9 ≤ SI < 1.3, n = 89)	High SI group (SI ≥ 1.3, n = 45)	Statistic	P value	Effect size
Demographic characteristics
Age (years)	42.3 ± 15.7	45.2 ± 16.9	38.6 ± 14.2	F = 2.847	0.061	η2 = 0.027
Gender (male)	56 (71.8)	67 (75.3)	38 (84.4)	χ2 = 2.831	0.243	Cramer’s V = 0.116
BMI (kg/m2)	23.4 ± 3.2	24.1 ± 3.8	22.9 ± 3.5	F = 1.842	0.161	η2 = 0.017
BMI categories				χ2 = 5.1671	0.523	Cramer’s V = 0.111
Normal (18.5-23.9)	45 (57.7)	47 (52.8)	28 (62.2)
Overweight (24.0-27.9)	28 (35.9)	35 (39.3)	14 (31.1)
Obese (≥ 28.0)	5 (6.4)	7 (7.9)	3 (6.7)
Mechanism of injury				χ2 = 12.436	0.134	Cramer’s V = 0.171
Traffic accident	42 (53.8)	53 (59.6)	31 (68.9)
Fall from height	21 (26.9)	19 (21.3)	8 (17.8)
Heavy object impact	9 (11.5)	12 (13.5)	4 (8.9)
Sports injury	4 (5.1)	3 (3.4)	1 (2.2)
Violent injury	2 (2.6)	2 (2.2)	1 (2.2)
Time-related factors
Time from injury to admission (hours)	2.1 (1.0-4.2)	2.5 (1.2-4.8)	2.8 (1.5-5.1)	H = 3.742	0.154	ε2 = 0.018
Time from admission to surgery (hours)	4.2 (2.1-7.8)	2.8 (1.5-5.2)	1.8 (1.0-3.4)	H = 18.923	< 0.001	ε2 = 0.089
Past medical history
Hypertension	18 (23.1)	24 (27.0)	8 (17.8)	χ2 = 1.542	0.463	Cramer’s V = 0.086
Diabetes	7 (9.0)	11 (12.4)	3 (6.7)	χ2 = 1.234	0.540	Cramer’s V = 0.076
Coronary heart disease	5 (6.4)	8 (9.0)	2 (4.4)	χ2 = 1.076	0.584	Cramer’s V = 0.071
Medication history
Anticoagulant use	4 (5.1)	6 (6.7)	2 (4.4)	χ2 = 0.337	0.845	Cramer’s V = 0.040
Antiplatelet drug use	8 (10.3)	12 (13.5)	3 (6.7)	χ2 = 1.543	0.462	Cramer’s V = 0.086
Lifestyle
Smoking history	32 (41.0)	38 (42.7)	21 (46.7)	χ2 = 0.432	0.806	Cramer’s V = 0.045
Alcohol consumption history	28 (35.9)	34 (38.2)	19 (42.2)	χ2 = 0.523	0.770	Cramer’s V = 0.050

¹Linear trend χ² test.

BMI: Body mass index; SI: Shock index.

Clinical indicators at admission

Comparison of clinical indicators at admission among the three groups is shown in Table 2. With increasing SI, patients showed significantly decreased blood pressure, significantly increased heart rate and respiratory rate, gradually decreased GCS scores, and more severe splenic injury (all P < 0.001). Although temperature changes were statistically significant, they had limited clinical significance (small effect size, η² = 0.038), mainly reflecting differences between low SI and high SI groups (P = 0.016). Linear trend tests showed both GCS grading and splenic injury grading presented significant dose-response relationships (all P < 0.001). After post hoc pairwise comparisons with correction (α = 0.0167), most indicators showed statistically significant differences between groups.

Table 2 Comparison of clinical indicators at admission among three groups, mean ± SD/n (%).

Indicator	Low SI group (n = 78)	Moderate SI group (n = 89)	High SI group (n = 45)	Statistic	P value	Effect size	Pairwise comparison P values3
Vital signs
Systolic BP (mmHg)	128.4 ± 15.2	106.3 ± 12.8	87.2 ± 11.6	F = 156.8	< 0.001	η2 = 0.600	P < 0.0014, P < 0.0015, P < 0.0016
Diastolic BP (mmHg)	78.6 ± 11.4	68.2 ± 10.7	56.3 ± 9.8	F = 67.4	< 0.001	η2 = 0.392	P < 0.0014, P < 0.0015, P < 0.0016
Mean arterial pressure (mmHg)	95.2 ± 11.8	80.9 ± 10.4	66.6 ± 9.2	F = 108.7	< 0.001	η2 = 0.510	P < 0.0014, P < 0.0015, P < 0.0016
Heart rate (beats/minutes)	92.3 ± 12.6	118.7 ± 14.2	142.8 ± 18.5	F = 189.3	< 0.001	η2 = 0.644	P < 0.0014, P < 0.0015, P < 0.0016
Respiratory rate (breaths/minutes)	18.2 ± 2.8	21.4 ± 3.6	26.8 ± 4.2	F = 92.6	< 0.001	η2 = 0.470	P < 0.0014, P < 0.0015, P < 0.0016
Body temperature (°C)	36.8 ± 0.6	36.6 ± 0.7	36.4 ± 0.8	F = 4.123	0.017	η2 = 0.0382	P = 0.1354, P = 0.0165, P = 0.2856
Shock index	0.72 ± 0.11	1.12 ± 0.11	1.65 ± 0.28	F = 892.4	< 0.001	η2 = 0.895	P < 0.0014, P < 0.0015, P < 0.0016
GCS score
Eye opening (E)	3.8 ± 0.5	3.6 ± 0.7	3.2 ± 0.9	F = 12.8	< 0.001	η2 = 0.109	P = 0.0244, P < 0.0015, P = 0.0186
Verbal response (V)	4.6 ± 0.8	4.2 ± 1.1	3.7 ± 1.3	F = 11.2	< 0.001	η2 = 0.097	P = 0.0154, P < 0.0015, P = 0.0456
Motor response (M)	5.8 ± 0.6	5.5 ± 0.9	5.1 ± 1.2	F = 8.9	< 0.001	η2 = 0.078	P = 0.0424, P < 0.0015, P = 0.0786
GCS total score	14.2 ± 1.4	13.3 ± 2.1	12.0 ± 2.6	F = 18.7	< 0.001	η2 = 0.152	P = 0.0064, P < 0.0015, P = 0.0036
GCS grading				χ2 = 14.281	0.001	Cramer’s V = 0.184
Mild (13-15 points)	72 (92.3)	73 (82.0)	30 (66.7)
Moderate (9-12 points)	6 (7.7)	15 (16.9)	13 (28.9)
Severe (≤ 8 points)	0 (0.0)	1 (1.1)	2 (4.4)
Splenic Injury grading				χ2 = 25.671	< 0.001	Cramer’s V = 0.246
Grade I	18 (23.1)	8 (9.0)	2 (4.4)
Grade II	32 (41.0)	25 (28.1)	6 (13.3)
Grade III	22 (28.2)	38 (42.7)	18 (40.0)
Grade IV	5 (6.4)	15 (16.9)	15 (33.3)
Grade V	1 (1.3)	3 (3.4)	4 (8.9)

¹Linear trend χ² test.

²Small effect size, limited clinical significance.

³Pairwise comparisons using Tukey honestly significant difference test, corrected α = 0.0167.

⁴Low shock index group vs moderate shock index group.

⁵Low shock index group vs high shock index group.

⁶Moderate shock index group vs high shock index group.

BP: Blood pressure; GCS: Glasgow Coma Scale; SI: Shock index.

Laboratory test indicators

Comparison of laboratory test indicators among the three groups is shown in Table 3. With increasing SI, patients showed significantly decreased hemoglobin, red blood cell count, and platelet count, more severe coagulation dysfunction, deteriorated liver and kidney function indicators, and more pronounced metabolic acidosis and elevated lactate levels (all P < 0.001). Although electrolyte changes were statistically significant, effect sizes were small (η² < 0.1) with limited clinical significance, mainly reflecting differences between low SI and high SI groups. Patients in the high SI group presented typical laboratory characteristics of hemorrhagic shock. After post hoc pairwise comparisons with correction (α = 0.0167), except for some electrolyte indicators, other indicators showed statistically significant differences between groups.

Table 3 Comparison of laboratory test indicators among three groups, mean ± SD.

Indicator	Low SI group (n = 78)	Moderate SI group (n = 89)	High SI group (n = 45)	Statistic	P value	Effect size	Pairwise comparison P values2
Blood routine indicators
Hemoglobin (g/L)	118.3 ± 18.7	96.7 ± 16.4	78.2 ± 15.8	F = 89.6	< 0.001	η2 = 0.461	P < 0.0013, P < 0.0014, P < 0.0015
Red blood cell count (× 1012/L)	3.94 ± 0.62	3.23 ± 0.55	2.61 ± 0.53	F = 87.3	< 0.001	η2 = 0.455	P < 0.0013, P < 0.0014, P < 0.0015
Hematocrit (%)	35.4 ± 5.6	28.9 ± 4.9	23.4 ± 4.7	F = 88.9	< 0.001	η2 = 0.459	P < 0.0013, P < 0.0014, P < 0.0015
White blood cell count (× 109/L)	9.8 ± 3.2	12.4 ± 4.1	15.6 ± 5.3	F = 32.4	< 0.001	η2 = 0.237	P < 0.0013, P < 0.0014, P = 0.0015
Platelet count (× 109/L)	245.6 ± 68.4	198.3 ± 58.7	128.4 ± 46.2	F = 42.8	< 0.001	η2 = 0.291	P < 0.0013, P < 0.0014, P < 0.0015
Coagulation function indicators
PT (seconds)	12.8 ± 1.6	14.2 ± 2.1	16.8 ± 2.9	F = 48.9	< 0.001	η2 = 0.319	P < 0.0013, P < 0.0014, P < 0.0015
APTT (seconds)	32.4 ± 4.8	36.7 ± 5.9	42.3 ± 7.2	F = 42.1	< 0.001	η2 = 0.287	P < 0.0013, P < 0.0014, P < 0.0015
INR	1.08 ± 0.14	1.19 ± 0.18	1.41 ± 0.24	F = 45.6	< 0.001	η2 = 0.303	P < 0.0013, P < 0.0014, P < 0.0015
Fibrinogen (g/L)	3.2 ± 0.8	2.8 ± 0.7	2.1 ± 0.6	F = 32.8	< 0.001	η2 = 0.239	P = 0.0033, P < 0.0014, P < 0.0015
Biochemical indicators
Serum creatinine (μmol/L)	78.4 ± 15.2	89.7 ± 18.6	112.3 ± 24.8	F = 48.7	< 0.001	η2 = 0.318	P < 0.0013, P < 0.0014, P < 0.0015
Blood urea nitrogen (mmol/L)	5.8 ± 1.4	7.2 ± 1.8	9.6 ± 2.4	F = 58.9	< 0.001	η2 = 0.360	P < 0.0013, P < 0.0014, P < 0.0015
Total bilirubin (μmol/L)	18.6 ± 5.2	22.4 ± 6.8	28.7 ± 8.9	F = 32.1	< 0.001	η2 = 0.235	P = 0.0023, P < 0.0014, P < 0.0015
ALT (U/L)	42.3 ± 12.8	58.7 ± 18.4	78.9 ± 24.6	F = 56.2	< 0.001	η2 = 0.349	P < 0.0013, P < 0.0014, P < 0.0015
AST (U/L)	38.9 ± 11.6	67.2 ± 20.8	95.6 ± 28.4	F = 102.8	< 0.001	η2 = 0.496	P < 0.0013, P < 0.0014, P < 0.0015
LDH (U/L)	245.8 ± 58.7	324.6 ± 72.9	428.7 ± 89.3	F = 98.4	< 0.001	η2 = 0.485	P < 0.0013, P < 0.0014, P < 0.0015
Electrolyte levels
Serum sodium (mmol/L)	139.2 ± 3.4	138.6 ± 3.8	137.1 ± 4.2	F = 4.89	0.008	η2 = 0.0451	P = 0.4383, P = 0.0064, P = 0.0635
Serum potassium (mmol/L)	4.1 ± 0.5	4.3 ± 0.6	4.6 ± 0.7	F = 10.8	< 0.001	η2 = 0.0941	P = 0.0453, P < 0.0014, P = 0.0215
Serum chloride (mmol/L)	103.8 ± 3.2	102.9 ± 3.6	101.4 ± 4.1	F = 6.47	0.002	η2= 0.0581	P = 0.1953, P = 0.0024, P = 0.0515
Blood gas analysis
pH value	7.38 ± 0.05	7.34 ± 0.06	7.28 ± 0.08	F = 39.8	< 0.001	η2 = 0.275	P < 0.0013, P < 0.0014, P < 0.0015
PaO2 (mmHg)	88.7 ± 12.4	82.3 ± 14.7	75.1 ± 16.8	F = 15.9	< 0.001	η2 = 0.132	P = 0.0063, P < 0.0014, P = 0.0215
PaCO2 (mmHg)	38.2 ± 4.6	35.8 ± 5.2	32.4 ± 6.1	F = 18.7	< 0.001	η2 = 0.152	P = 0.0093, P < 0.0014, P = 0.0035
Base excess BE (mmol/L)	-1.2 ± 2.4	-4.8 ± 3.1	-8.6 ± 3.8	F = 89.3	< 0.001	η2 = 0.460	P < 0.0013, P < 0.0014, P < 0.0015
Lactate level (mmol/L)	2.8 ± 0.9	4.6 ± 1.3	6.8 ± 1.9	F = 126.4	< 0.001	η2 = 0.547	P < 0.0013, P < 0.0014, P < 0.0015

¹Small effect size, limited clinical significance.

²Pairwise comparisons using Tukey honestly significant difference test, corrected α = 0.0167.

³Low shock index group vs moderate shock index group.

⁴Low shock index group vs high shock index group.

⁵Moderate shock index group vs high shock index group.

PT: Prothrombin time; APTT: Activated partial thromboplastin time; INR: International normalized ratio; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; LDH: Lactate dehydrogenase; PaO₂: Arterial oxygen partial pressure; PaCO₂: Arterial carbon dioxide partial pressure; BE: Base excess; SI: Shock index.

Imaging examination results

Comparison of imaging examination results among the three groups is shown in Table 4. Patients in the high SI group had more extensive splenic rupture, larger hematomas, significantly higher proportion of massive intraperitoneal hemorrhage compared to the other two groups, and higher proportion of combined other organ injuries. Linear trend tests showed splenic rupture extent, hematoma size, intraperitoneal hemorrhage volume, and combined organ injuries all presented significant dose-response relationships (all corrected P < 0.05). After Bonferroni correction (α = 0.0167), differences between groups for individual organ injuries were no longer statistically significant, but overall combined organ injury still showed significant differences.

Table 4 Comparison of imaging examination results among three groups, n (%).

Indicator	Low SI group (n = 78)	Moderate SI group (n = 89)	High SI group (n = 45)	Statistic	P value	Corrected P value	Effect size	Pairwise comparison P values2
Splenic rupture extent				χ2 = 28.341	< 0.001	< 0.001	Cramer’s V = 0.258	P = 0.0063, P < 0.0014, P = 0.0245
Localized rupture	45 (57.7)	28 (31.5)	7 (15.6)
Extensive rupture	28 (35.9)	46 (51.7)	25 (55.6)
Shattered rupture	5 (6.4)	15 (16.9)	13 (28.9)
Hematoma size				χ2 = 26.881	< 0.001	< 0.001	Cramer’s V = 0.252	P = 0.0033, P < 0.0014, P = 0.0185
Small hematoma (< 5 cm)	52 (66.7)	35 (39.3)	9 (20.0)
Medium hematoma (5-10 cm)	21 (26.9)	38 (42.7)	21 (46.7)
Large hematoma (> 10 cm)	5 (6.4)	16 (18.0)	15 (33.3)
Intraperitoneal hemorrhage volume				χ2 = 42.671	< 0.001	< 0.001	Cramer’s V = 0.317	P < 0.0013, P < 0.0014, P = 0.0095
Small amount	52 (66.7)	29 (32.6)	5 (11.1)
Moderate amount	22 (28.2)	41 (46.1)	18 (40.0)
Large amount	4 (5.1)	19 (21.3)	22 (48.9)
Combined other organ injuries	25 (32.1)	42 (47.2)	28 (62.2)	χ2 = 9.871	0.007	0.042	Cramer’s V = 0.216	P = 0.0423, P = 0.0034, P = 0.1325
Liver injury	8 (10.3)	18 (20.2)	12 (26.7)	χ2 = 5.891	0.053	0.318	Cramer’s V = 0.167	P = 0.0783, P = 0.0214, P = 0.4235
Kidney injury	6 (7.7)	12 (13.5)	8 (17.8)	χ2 = 2.851	0.240	1.000	Cramer’s V = 0.116	P = 0.2343, P = 0.0784, P = 0.5195
Pancreatic injury	3 (3.8)	7 (7.9)	5 (11.1)	χ2 = 2.131	0.345	1.000	Cramer’s V = 0.100	P = 0.3123, P = 0.1294, P = 0.5375
Intestinal injury	5 (6.4)	8 (9.0)	6 (13.3)	χ2 = 1.461	0.482	1.000	Cramer’s V = 0.083	P = 0.5673, P = 0.2134, P = 0.4685
Pelvic fracture	8 (10.3)	15 (16.9)	11 (24.4)	χ2 = 3.841	0.147	0.882	Cramer’s V = 0.135	P = 0.2343, P = 0.0424, P = 0.3245

¹Linear trend χ² test.

²Pairwise comparisons using Bonferroni-corrected paired χ² test, corrected α = 0.0167.

³Low shock index group vs moderate shock index group.

⁴Low shock index group vs high shock index group.

⁵Moderate shock index group vs high shock index group.

SI: Shock index.

Surgery-related indicators

Comparison of surgery-related indicators among the three groups is shown in Table 5. With increasing SI, more patients required emergency surgery, the proportion of open surgery increased, operative time was correspondingly prolonged, intraoperative blood loss and transfusion volume significantly increased, intraoperative blood pressure maintenance became difficult, and the proportion of vasoactive drug use significantly increased (all P < 0.01). Due to the small number of laparoscopic surgeries in the high SI group (n = 2), statistical comparison of laparoscopic to open conversion rates was not performed. Post hoc pairwise comparisons after correction (α = 0.0167) showed most indicators had statistically significant differences between groups.

Table 5 Comparison of surgery-related indicators among three groups, n (%)/mean ± SD/median (interquartile range).

Indicator	Low SI group (n = 78)	Moderate SI group (n = 89)	High SI group (n = 45)	Statistic	P value	Effect size	Pairwise comparison P values2
Surgical timing				χ2 = 25.671	< 0.001	Cramer’s V = 0.348	P = 0.0093, P < 0.0014, P = 0.0035
Emergency surgery (< 4 hours)	28 (35.9)	52 (58.4)	38 (84.4)
Delayed surgery (≥ 4 hours)	50 (64.1)	37 (41.6)	7 (15.6)
Surgical approach				χ2 = 9.847	0.007	Cramer’s V = 0.216	P = 0.0423, P = 0.0034, P = 0.1325
Open splenectomy	58 (74.4)	78 (87.6)	43 (95.6)
Laparoscopic splenectomy	20 (25.6)	11 (12.4)	2 (4.4)
Operative time (minutes)	128.4 ± 32.6	156.7 ± 41.2	167.8 ± 35.4	F = 23.1	< 0.001	η2 = 0.181	P < 0.0013, P < 0.0014, P = 0.1325
Intraoperative blood loss (mL)	485.3 ± 124.7	732.6 ± 186.4	1126.8 ± 268.9	F = 152.3	< 0.001	η2 = 0.593	P < 0.0013, P < 0.0014, P < 0.0015
Intraoperative transfusion volume
Red blood cells (U)	2.3 ± 1.8	4.6 ± 2.4	7.8 ± 3.2	F = 78.9	< 0.001	η2 = 0.430	P < 0.0013, P < 0.0014, P < 0.0015
Plasma (mL)	186.4 ± 142.3	324.7 ± 198.6	542.8 ± 256.4	F = 48.6	< 0.001	η2 = 0.317	P < 0.0013, P < 0.0014, P < 0.0015
Platelets (U)	0.2 (0-1.0)	0.5 (0-1.5)	1.0 (0-3.0)	H = 18.4	< 0.001	ε2 = 0.087	P = 0.0423, P < 0.0014, P = 0.0155
Intraoperative blood pressure maintenance				χ2 = 37.891	< 0.001	Cramer’s V = 0.423	P < 0.0013, P < 0.0014, P = 0.0065
Stable (SBP ≥ 90 mmHg)	68 (87.2)	52 (58.4)	12 (26.7)
Unstable (SBP < 90 mmHg)	10 (12.8)	37 (41.6)	33 (73.3)
Vasoactive drug use	12 (15.4)	34 (38.2)	32 (71.1)	χ2 = 29.581	< 0.001	Cramer’s V = 0.373	P = 0.0033, P < 0.0014, P < 0.0015

¹Linear trend χ² test.

²Pairwise comparisons corrected α = 0.0167, continuous variables using Türkiye honestly significant difference test, categorical variables using Bonferroni-corrected paired χ² test, non-parametric variables using Dunn-Bonferroni test.

³Low shock index group vs moderate shock index group.

⁴Low shock index group vs high shock index group.

⁵Moderate shock index group vs high shock index group.

SBP: Systolic blood pressure; SI: Shock index.

Main prognostic indicators

Comparison of main prognostic indicators among the three groups is shown in Table 6. The incidence and severity of postoperative complications in the high SI group were significantly higher than the other two groups, intensive care unit admission rate significantly increased, and mechanical ventilation time and hospital stay were prolonged (all P < 0.01). The 30-day postoperative mortality rate in the high SI group was 4.4% (2/45), which was statistically significantly different from the other two groups (P = 0.039). It should be noted that 2 additional patients in the high SI group died within 24 hours postoperatively and were excluded from this analysis per the exclusion criteria. If these patients were included, the 30-day mortality rate in the high SI group would increase from 4.4% (2/45) to 8.9% (4/45), further reinforcing the predictive value of SI. Linear trend tests showed all prognostic indicators presented significant dose-response relationships. Post hoc pairwise comparisons after correction (α = 0.0167) showed main differences were concentrated between the low SI group and high SI group.

Table 6 Comparison of main prognostic indicators among three groups, n (%)/median (interquartile range).

Indicator	Low SI group (n = 78)	Moderate SI group (n = 89)	High SI group (n = 45)	Statistic	P value	Effect size	Pairwise comparison P values2
Postoperative complications (Clavien-Dindo grading)				χ2 = 31.781	< 0.001	Cramer’s V = 0.274	P = 0.0033, P < 0.0014, P = 0.0245
No complications	48 (61.5)	32 (36.0)	8 (17.8)
Grade I	18 (23.1)	25 (28.1)	12 (26.7)
Grade II	9 (11.5)	21 (23.6)	13 (28.9)
Grade III	3 (3.8)	8 (9.0)	7 (15.6)
Grade IV	0 (0.0)	3 (3.4)	3 (6.7)
Grade V	0 (0.0)	0 (0.0)	2 (4.4)
Severe complications (≥ grade III)	3 (3.8)	11 (12.4)	12 (26.7)	χ2 = 12.581	0.002	Cramer’s V = 0.244	P = 0.0633, P < 0.0014, P = 0.0425
30-day postoperative mortality	0 (0.0)	0 (0.0)	2 (4.4)	Fisher’s exact test	0.039
ICU admission rate	15 (19.2)	42 (47.2)	34 (75.6)	χ2 = 32.171	< 0.001	Cramer’s V = 0.390	P < 0.0013, P < 0.0014, P = 0.0035
Mechanical ventilation time (hours)	8.5 (4.2-18.6)	22.4 (12.8-38.7)	48.2 (28.6-72.4)	H = 56.8	< 0.001	ε2 = 0.268	P < 0.0013, P < 0.0014, P < 0.0015
Hospital stays (days)	12.5 (8.2-18.4)	16.8 (11.2-24.6)	23.7 (15.8-35.2)	H = 32.4	< 0.001	ε2 = 0.153	P = 0.0093, P < 0.0014, P = 0.0185

¹Linear trend χ² test.

²Pairwise comparisons corrected α = 0.0167, categorical variables using Bonferroni-corrected paired χ² test, non-parametric variables using Dunn-Bonferroni test.

³Low shock index group vs moderate shock index group.

⁴Low shock index group vs high shock index group.

⁵Moderate shock index group vs high shock index group.

ICU: Intensive care unit; SI: Shock index.

Secondary prognostic indicators

Comparison of secondary prognostic indicators among the three groups is shown in Table 7. Patients in the high SI group had slower postoperative hemoglobin recovery, significantly increased reoperation rate, and significantly worse functional status at discharge (all P < 0.05). Postoperative hemorrhage was the main cause of reoperation, with the highest proportion in the high SI group. Linear trend tests showed all indicators presented significant dose-response relationships. Post hoc pairwise comparisons after correction (α = 0.0167) showed main differences were still concentrated between the low SI group and high SI group, with no statistically significant difference in reoperation rate between the low SI group and moderate SI group (P = 0.096, P =0.0167).

Table 7 Comparison of secondary prognostic indicators among three groups, mean ± SD/n (%).

Indicator	Low SI group (n = 78)	Moderate SI group (n = 89)	High SI group (n = 45)	Statistic	P value	Effect size	Pairwise comparison P values2
Postoperative hemoglobin changes (g/L)
24 hours postoperative	106.8 ± 16.4	84.2 ± 14.7	68.9 ± 13.2	F = 89.6	< 0.001	η2 = 0.461	P < 0.0013, P < 0.0014, P < 0.0015
48 hours postoperative	109.2 ± 15.8	89.6 ± 15.1	74.3 ± 14.6	F = 78.4	< 0.001	η2 = 0.428	P < 0.0013, P < 0.0014, P < 0.0015
72 hours postoperative	112.4 ± 16.2	95.7 ± 16.8	81.2 ± 15.9	F = 64.7	< 0.001	η2 = 0.382	P < 0.0013, P < 0.0014, P < 0.0015
Reoperation rate	2 (2.6)	8 (9.0)	9 (20.0)	χ2 = 8.741	0.013	Cramer’s V = 0.203	P = 0.0963, P = 0.0034, P = 0.0845
Reoperation causes				Fisher’s exact test	0.024
Postoperative hemorrhage	1 (1.3)	5 (5.6)	6 (13.3)
Abdominal infection	1 (1.3)	2 (2.2)	2 (4.4)
Pancreatic fistula management	0 (0.0)	1 (1.1)	1 (2.2)
mRS score (at discharge)				χ2 = 28.461	< 0.001	Cramer’s V = 0.259	P = 0.0063, P < 0.0014, P = 0.0215
0 points (completely asymptomatic)	52 (66.7)	38 (42.7)	12 (26.7)
1 point (no significant functional deficit)	18 (23.1)	28 (31.5)	14 (31.1)
2 points (mild functional deficit)	6 (7.7)	15 (16.9)	11 (24.4)
3 points (moderate functional deficit)	2 (2.6)	6 (6.7)	5 (11.1)
4 points (moderate to severe functional deficit)	0 (0.0)	2 (2.2)	1 (2.2)
5 points (severe functional deficit)	0 (0.0)	0 (0.0)	0 (0.0)

¹Linear trend χ² test.

²Pairwise comparisons corrected α = 0.0167 continuous variables using Türkiye honestly significant difference test, categorical variables using Bonferroni-corrected paired χ² test.

³Low shock index group vs moderate shock index group.

⁴Low shock index group vs high shock index group.

⁵Moderate shock index group vs high shock index group.

mRS: Modified Rankin Scale; SI: Shock index.

Risk factor analysis for poor prognosis

Multivariate logistic regression analysis was used to identify independent risk factors affecting severe postoperative complications (defined as Clavien-Dindo grade ≥ III complications or death within 30 days). Variables with P < 0.10 in univariate analysis were included in multivariate analysis, with results shown in Table 8. Multivariate analysis showed that SI (OR = 2.34, 95%CI: 1.25-4.38, P = 0.008), splenic injury grade ≥ IV (OR = 2.97, 95%CI: 1.28-6.86, P = 0.011), hemoglobin ≤ 90 g/L (OR = 3.14, 95%CI: 1.40-7.07, P = 0.005), lactate level ≥ 4.0 mmol/L (OR = 3.97, 95%CI: 1.76-8.92, P = 0.001), and massive intraperitoneal hemorrhage (OR = 2.43, 95%CI: 1.02-5.78, P = 0.045) were independent risk factors for severe postoperative complications.

Table 8 Multivariate logistic regression analysis of factors affecting severe postoperative complications.

Variable	Univariate analysis		Multivariate analysis
	OR (95%CI)	P value	OR (95%CI)	P value
Shock index (continuous variable)	3.68 (2.15-6.28)	< 0.001	2.34 (1.25-4.38)	0.008

OR: Odds ratio; CI: Confidence interval.

SI-based prognostic prediction model

A multivariate prognostic prediction model including SI was constructed. ROC curve analysis showed that this model had an AUC of 0.847 (95%CI: 0.789-0.905, P < 0.001) for predicting severe postoperative complications. The AUC for SI alone was 0.782 (95%CI: 0.718-0.846, P < 0.001). DeLong test showed that the multivariate model had significantly superior predictive performance compared to SI alone (Z = 2.134, P = 0.033).

According to ROC curve analysis, the optimal cutoff value of SI for predicting severe postoperative complications was 1.15 (maximum Youden index point), with sensitivity of 76.9%, specificity of 81.2%, positive predictive value of 48.8%, negative predictive value of 93.5%, and accuracy of 80.2%. Considering cost-benefit factors in actual clinical emergency department application, when the cutoff value was set at 1.0, sensitivity was 85.7% and specificity was 72.4%; when the cutoff value was set at 1.3, sensitivity was 64.3% and specificity was 89.6%.

Bootstrap resampling (1000 times) for internal validation showed a corrected C-index of 0.832 (95%CI: 0.798-0.866), indicating good discriminative ability of the model. Ten-fold cross-validation showed an average AUC of 0.826 (SE = 0.015, 95%CI: 0.796-0.856), indicating good model stability. Calibration plot showed good consistency between predicted probability and actual incidence (Hosmer-Lemeshow test, P = 0.674). Decision curve analysis showed that the SI-based prediction model had positive net benefit in the threshold probability range of 10%-60%, indicating its clinical application value.

Survival analysis

All 212 patients completed follow-up, with follow-up time of 12-60 months and median follow-up time of 28 months (interquartile range: 18-42 months). There was no statistically significant difference in follow-up time distribution among the three groups (H = 2.456, P = 0.293). During the period, 10 patients died, including 2 deaths within 30 days postoperatively (both in high SI group) and 8 deaths after 30 days (1 in low SI group, 3 in moderate SI group, 4 in high SI group).

Kaplan-Meier survival analysis showed significant differences in 1-year survival rates among the three groups (log-rank test, χ² = 8.456, P = 0.015). Pairwise comparisons using log-rank test with Bonferroni correction (α = 0.0167): 1-year survival rates were 98.7% (95%CI: 95.6%-100%) for low SI group, 96.6% (95%CI: 92.8%-99.2%) for moderate SI group, and 86.7% (95%CI: 75.8%-93.9%) for high SI group. Pairwise comparison results showed: Low SI group vs moderate SI group (P = 0.534 > 0.0167, no statistical significance), low SI group vs high SI group (P = 0.012 < 0.0167, statistically significant), moderate SI group vs high SI group (P = 0.039 > 0.0167, no statistical significance).

Cox proportional hazards regression analysis showed that SI was an independent risk factor affecting patient survival [hazard ratio (HR) = 1.85, 95%CI: 1.16-2.95, P = 0.010]. Other independent prognostic factors included age ≥ 65 years (HR = 2.46, 95%CI: 1.12-5.38, P = 0.024) and combined ≥ 2 organ injuries (HR = 3.12, 95%CI: 1.23-7.90, P = 0.016). Schoenfeld residual test verified that the proportional hazards assumption was satisfied (χ² = 4.23, P = 0.342).

DISCUSSION

This study found that with increasing preoperative SI, the incidence of postoperative complications in patients significantly increased, with the incidence of severe complications (Clavien-Dindo grade ≥ III) rising from 3.8% in the low SI group to 26.7% in the high SI group, showing a clear dose-response relationship. This result is consistent with previous studies in trauma patients[13,14]. Research has shown that SI can effectively reflect patients’ hemodynamic status and degree of blood loss. When SI ≥ 1.3, it usually indicates that blood loss has exceeded 30% of body weight, at which point the body’s compensatory mechanisms are approaching or have already decompensated[15].

Multivariate logistic regression analysis showed that SI was an independent risk factor for severe postoperative complications (OR = 2.34, 95%CI: 1.25-4.38, P = 0.008), which has important clinical significance. Compared to traditional single vital sign indicators, SI, as the ratio of heart rate to systolic blood pressure, can comprehensively reflect the compensatory status of the cardiovascular system and has higher sensitivity for hemorrhagic shock[16]. Previous studies have found that in early trauma, patients may maintain blood pressure within a relatively normal range through increased heart rate and vasoconstriction, but SI has already begun to rise, thus enabling earlier identification of hemodynamic abnormalities[17,18].

The multivariate prediction model based on SI constructed in this study showed good discriminative ability, with an AUC of 0.847 (95%CI: 0.789-0.905) for predicting severe postoperative complications, significantly superior to SI alone (AUC = 0.782). Compared with literature reports, our model’s predictive performance (AUC = 0.847) exceeds that of SI alone in trauma patients (AUC = 0.72-0.78) and outperforms modified SI (AUC = 0.79) and age SI (AUC = 0.81), confirming the added value of integrating multidimensional clinical indicators. This result indicates that combining SI with other clinical indicators can improve prediction accuracy[19]. Bootstrap resampling validation and ten-fold cross-validation both showed that the model had good stability and discriminative ability, with corrected C-index of 0.832 and average cross-validation AUC of 0.826.

According to ROC curve analysis, the optimal cutoff value of SI for predicting severe postoperative complications was 1.15, with sensitivity of 76.9% and specificity of 81.2%. Considering the actual application needs in trauma emergency departments, it is recommended to use 1.0 as the screening cutoff value (sensitivity 85.7%, specificity 72.4%) to avoid missing high-risk patients, and 1.3 as the high-risk cutoff value (sensitivity 64.3%, specificity 89.6%) to identify patients requiring more aggressive treatment[20,21]. This stratified management strategy helps optimize medical resource allocation and improve treatment efficiency.

This study found that in addition to SI, splenic injury grade ≥ IV, hemoglobin ≤ 90 g/L, lactate level ≥ 4.0 mmol/L, and massive intraperitoneal hemorrhage were also independent risk factors for severe postoperative complications. These factors together with SI constitute a prognostic assessment system for patients with traumatic splenic rupture[22]. It is worth noting that lactate level, as an important indicator of tissue hypoperfusion, has good correlation with SI, and their combined use can more accurately assess the severity of patients’ conditions[23,24].

The study results showed that patients in the high SI group had more severe splenic injury, with grade IV-V injury proportion reaching 42.2%, significantly higher than 7.7% in the low SI group. This association may be related to injury mechanism and blood loss volume, as severe splenic rupture often involves greater blood loss, leading to more unstable hemodynamics[25]. At the same time, the combined organ injury rate in the high SI group also significantly increased (62.2% vs 32.1%), suggesting that polytrauma patients have worse prognosis and require more aggressive comprehensive treatment[26].

This study found that with increasing SI, surgical complexity significantly increased. Patients in the high SI group more often required emergency surgery (84.4% vs 35.9%), had higher proportion of open surgery (95.6% vs 74.4%), and significantly increased intraoperative blood loss and transfusion requirements. These findings are consistent with international study results[27,28]. The need for emergency surgery reflects the critical nature of patients’ conditions, while open surgery, although more traumatic than laparoscopic surgery, can provide better hemostasis and surgical field of view in patients with hemodynamic instability[29].

The vasoactive drug use rate during surgery in the high SI group reached 71.1%, significantly higher than the other two groups, reflecting the complexity of hemodynamic management in these patients. Studies have shown that difficulty in maintaining intraoperative blood pressure is closely related to postoperative complications, and appropriate vasoactive drug support helps maintain perfusion of vital organs, but also suggests relatively poor patient prognosis[30].

Survival analysis showed that preoperative SI has important predictive value for patients’ long-term survival. The 1-year survival rate of patients in the high SI group was 86.7%, significantly lower than 98.7% in the low SI group. Cox regression analysis confirmed that SI was an independent risk factor affecting patient survival (HR = 1.85, 95%CI: 1.16-2.95, P = 0.010). This finding indicates that preoperative SI can not only predict early postoperative complications but is also related to patients’ long-term prognosis[31].

It is worth noting that the mortality rate in this study was relatively low, with 30-day postoperative mortality rate of only 0.9% (2/212) and 1-year total mortality rate of 4.7% (10/212), which may be related to patient selection bias and improvements in modern trauma treatment levels[32]. With the improvement of trauma center construction and promotion of multidisciplinary collaboration models, the treatment success rate for patients with traumatic splenic rupture continues to improve, but high SI patients still face higher long-term mortality risk.

The SI-based prognostic prediction model constructed in this study has important clinical application value. Decision curve analysis showed that within the threshold probability range of 10%-60%, this model has positive net benefit, indicating its utility in clinical decision-making[33]. For emergency physicians, SI, as a simple and rapidly obtainable indicator, can quickly assess prognostic risk when patients are admitted and guide the formulation of subsequent treatment strategies.

It is recommended to establish an SI-based stratified management system in clinical practice: Patients with SI < 0.9 belong to the low-risk group and may consider conservative treatment or delayed surgery; patients with SI 0.9-1.3 belong to the medium-risk group and require close monitoring and timely surgery; patients with SI ≥ 1.3 belong to the high-risk group and require immediate surgical intervention and preparation for postoperative intensive care[34]. This stratified management strategy helps optimize medical resource allocation and improve treatment efficiency and patient prognosis.

This study has some limitations that need to be acknowledged. First, as a single-center retrospective study, there may be selection bias and information bias. Notably, the exclusion of patients who died within 24 hours postoperatively (n = 2, both in the high SI group) may introduce survival bias and potentially underestimate the mortality risk associated with extremely high SI values. This exclusion criterion, while necessary to assess postoperative complications adequately, limits our ability to fully evaluate early mortality. Second, although the sample size of this study is relatively large, the sample size of the high SI group is still limited, which may affect the statistical power. Third, as a single-center study conducted in two tertiary hospitals in one province of China, the findings may have limited generalizability to other clinical settings, particularly trauma centers with different patient populations or resource availability.

Future research should consider the following directions: First, conduct multicenter prospective studies to validate the effectiveness and generalizability of the SI prediction model; second, explore combining SI with other emerging biomarkers (such as soluble suppression of tumorigenicity 2, cardiac troponin, etc.) to construct more precise prognostic prediction models; third, study the impact of different treatment strategies on prognosis of patients with different SI stratifications to provide evidence for individualized treatment; finally, utilize artificial intelligence and machine learning technologies to develop intelligent prognostic prediction systems based on multi-dimensional data.

CONCLUSION

In conclusion, preoperative SI is an important predictor of prognosis in patients with traumatic splenic rupture undergoing splenectomy, with advantages of simple operation, easy acquisition, and accurate prediction. The SI-based prognostic prediction model can effectively identify high-risk patients and guide clinical decision-making, having important clinical application value. It is recommended to widely apply SI for risk assessment and stratified management in trauma treatment to improve patient prognosis.