Predictive Value of D-Dimer Combined with the Injury Severity Score for In-Hospital Mortality in Trauma Patients

By Jiajian Peng^1,2, Minhua Kuang^1,2, Qixin Yu^1,2, Zhangrong Liang^1,2

Affiliations

1. The Eighth Clinical Medical College of Guangzhou University of Chinese Medicine, Foshan, China
2. Department of Emergency, Foshan Hospital of Traditional Chinese Medicine, Foshan, China

doi: 10.29271/jcpsp.2026.08.1076

ABSTRACT
Objective: To evaluate the predictive performance of combining serum D-Dimer levels with Injury Severity Score (ISS) for forecasting in-hospital mortality and unfavourable outcomes among trauma patients.
Study Design: An analytical study.
Place and Duration of the Study: Department of Emergency, Foshan Hospital of Traditional Chinese Medicine, Foshan, China, from January 2021 to April 2025.
Methodology: Trauma patients were divided into a survival group (n = 250) and a death group (n = 35) based on in-hospital survival status. Whole-body CT examinations were performed upon admission, and general clinical data, serum D-Dimer levels, and ISS scores were compared between the two groups. ROC curves were generated to evaluate predictive performance. Logistic regression identified independent prognostic factors, which were then used to build a nomogram model using R software.
Results: A total of 285 trauma patients were included, with a mortality rate of 12.3%. Logistic regression analysis revealed that D-Dimer (p <0.001), ISS score (p = 0.02), GCS (p = 0.06), and days of hospitalisation (p <0.001) were independent risk factors for in-hospital mortality. The AUC was 0.766 for D-dimer, 0.740 for the ISS score, and 0.811 for the combined model. The combined model demonstrated significantly better predictive performance than either individual predictor (p <0.05). The nomogram was constructed based on the results of the multivariate regression analysis and demonstrated good predictive performance.
Conclusion: D-Dimer levels and ISS scores were higher in trauma patients who died during hospitalisation.

Key Words: Trauma, D-Dimer, Injury severity score.

INTRODUCTION

Trauma refers to the structural damage of human tissues or organs caused by external forces. Its clinical symptoms vary, ranging from pain and swelling in mild cases to shock and haemorrhage in severe cases, posing a serious threat to life.¹ In 2017, trauma-related medical cases in China reached 77.1 million, with over 700,000 deaths.² Trauma is a leading cause of death and disability across all ages, especially among individuals under forty.³ During emergency care, accurately predicting in-hospital mortality risk is crucial for individualised treatment, informed decision-making, and efficient resource allocation. Given the heterogeneity in trauma outcomes, early risk stratification helps identify high-risk patients, improve interventions, and optimise prognosis.⁴

The Injury Severity Score (ISS), based on the Abbreviated Injury Scale (AIS), is widely used to assess trauma severity due to its objectivity and standardisation.^5,6 However, ISS has limitations: it requires complex scoring, accurate identification of multiple injuries, and fails to reflect real-time condition changes.⁷ These drawbacks reduce its effectiveness in rapid assessment and continuous monitoring in emergency settings.⁸ Therefore, ISS alone is insufficient for prognosis prediction, and there is a need to combine it with simpler, more sensitive indicators. 

Recent studies have explored biomarkers such as copeptin and CD5L for trauma prognosis and risk prediction.⁹ However, single biomarkers often fall short in predictive performance,¹⁰ and their high cost and sampling difficulty hinder clinical use. Combining ISS with serum markers may enhance prognostic accuracy. Among these, D-Dimer has gained attention due to its association with trauma severity and poor outcomes.¹¹ Most trauma patients arrive with elevated D-Dimer, indicating excessive fibrinolysis and increased bleeding or death risk.¹² As a common marker of hypercoagulability and hyperfibrinolysis, D-Dimer is widely used in the prognosis of critically ill patients with sepsis and thrombotic diseases, and is often elevated in trauma cases.¹³

Since coagulation abnormalities are common in trauma, D-Dimer may be valuable for prognostic assessment. Thus, combining D-Dimer with ISS may leverage the objectivity of ISS and the simplicity of D-Dimer to create a more compre-hensive and practical model for early prognosis prediction. This approach may help identify high-risk patients sooner and improve survival outcomes.

This study aimed to retrospectively evaluate the predictive value of D-dimer levels and the ISS for in-hospital mortality among trauma patients and to determine whether combining these two measures improves the accuracy of mortality prediction.

METHODOLOGY

A total of 285 trauma patients were treated at the Department of Emergency, Foshan Hospital of Traditional Chinese Medicine, Foshan, China, from January 2021 to April 2025. Retrospective data were collected by reviewing patients' electronic medical records. All medical, demographic, and laboratory data were retrieved from the hospital database. The study was approved by the Ethics Committee of the hos-pital (Approval No. KY [2023]270). This retrospective study was conducted in compliance with the Declaration of Helsinki. All patient data were anonymised and kept confidential, used solely for research and statistical analysis (Figure 1).

Inclusion criteria were trauma patients who underwent both D-Dimer testing and ISS scoring in the emergency depart-ment (ED). All types of trauma were included, encompassing both blunt and penetrating injuries, regardless of the number of injured sites or injury mechanisms. The exclusion criteria were age <14 years, injury time >24 hours, dead on arrival, automatic discharge, missing clinical or laboratory data, previous coagulation abnormality, and anticoagulants (used within 6 months before the injury).

Blood specimens were collected within 2 hours of presen-tation. Fresh venous blood was drawn in the ED into sodium citrate (9:1) vacuum tubes (Guangzhou Improve Medical Instruments Co., Ltd.). Samples were registered and prom-ptly transported to the laboratory for analysis. D-dimer levels were measured using the Werfen Hemagglutination Analyser, and all procedures were performed in accordance with the manufacturer’s instructions. The ISS score was used to assess the severity of the patient's trauma (AIS-2015 version), taking the sum of the squares of the AIS scores for the 3 highest injury sites, with ISS <9 as mild, 9-15 as moderate, 16-24 as serious injuries, and ≥25 as severe injuries, with a maximum score of 75. Whole-body CT (WBCT) examination was performed using a 64-slice X-ray (CT) scanner (Philips Medical Systems), 140kV, 500 mA. WBCT is defined as imaging of the head, cervical spine, thorax, abdomen and pelvis. The decision to perform WBCT was made by the receiving physician based on a compre-hensive assessment of the mechanism of trauma, vital signs and physical examination findings suggestive of injury.

Statistical analyses were performed using IBM SPSS^® version 26.0. Continuous variables with normal distribution were presented as mean ± standard deviation, while those not conforming to normality were expressed as median and interquartile range (P₂₅, P₇₅). Categorical variables were summarised as frequencies and percentages. Group comparisons for normally distributed data employed Student’s t-test, while non-parametric data were analysed using the Mann-Whitney U test. The chi-square test or Fisher's exact test was utilised to evaluate differences in categorical data between groups. Receiver Operating Characteristic (ROC) curves were constructed to assess the prognostic value of D-Dimer and ISS scores individually and in combination. The Youden index was used to identify optimal cut-off points by maximising the sum of sensitivity and specificity. Univariate logistic regression was first conducted to screen variables potentially associated with in-hospital mortality (p <0.1), followed by multivariate logistic regression to identify independent predictors. A nomogram model for mortality prediction was subsequently developed using the RMS package in R software (version 4.5, httpss:// www.r-project.org/). A two-tailed p-value of <0.05 was considered statistically significant.

RESULTS

A total of 338 emergency trauma patients were included from January 2021 to April 2025. After excluding 9 patients who died on admission, 3 with incomplete records, and 41 unable to complete relevant tests, 285 patients remained for analysis, including 250 survivors and 35 non-survivors, yielding a mortality rate of 12.3%. There were 203 males (71.2%) and 82 females (28.8%; Table I).

To evaluate the predictive ability of D-Dimer, ISS, and their combination for in-hospital mortality, ROC analysis was performed. The AUC was 0.766 (95% CI: 0.679-0.854)for D-Dimer, 0.740 (95% CI: 0.652-0.828) for ISS, and 0.811 (95% CI: 0.741-0.881) for the combined model, indicating improved discriminative performance with the joint model. At the optimal cut-off, D-Dimer demonstrated a sensitivity of 0.657 and specificity of 0.780, while ISS showed a sensitivity of 0.600 and specificity of 0.872 (Figure 1). The combined model achieved the highest balance between sensitivity (0.710) and specificity (0.803). DeLong’s test revealed no significant AUC difference between D-dimer and ISS (Z = 0.418; p = 0.676) or between D-dimer and the combined model (Z = –1.117; p = 0.264). However, the combined model showed a significantly higher AUC than ISS alone (Z = –2.401; p = 0.016), suggesting a superior predictive value.

Table I: Comparison of preoperative baseline characteristics between survivors and non-survivors (n = 285).

Variables	Overall (n = 285)	Survivors (n = 250)	Non-survivors (n = 35)	t/χ2/Z/F	p-values
General information	-	-	-	-	-
Age (years)	50 (36.5, 58)	49.5 (36, 57)	53 (38, 63)	-1.000	0.317
Gender (male)	203 (71.2%)	174 (69.6%)	29 (82.9%)	2.633	0.105
GCS	15 (8, 15)	15 (15, 15)	15 (8, 15)	-4.589	<0.001*
SP	121 (102, 133)	122 (102.75, 133)	111 (92, 126)	-2.177	0.029*
DP	74 (65, 83)	74.5 (64, 83.25)	70 (62, 79)	-1.926	0.597
Heart rate	87 (78, 98)	89 (78, 108)	94 (76, 120)	-0.958	0.338
AIS 0-6	-	-	-	13.668	<0.001*
Lightly hurt	72 (25.3%)	71 (28.4%)	1 (2.9%)	-	-
More seriously hurt	118 (41.4%)	95 (38.0%)	23 (65.7%)	-	-
Most seriously hurt	95 (33.3%)	84 (33.6%)	11 (31.4%)	-	-
Laboratory tests	-	-	-	-	-
D-Dimer	8.98 (3.86, 21.49)	7.42 (3.44, 18.72)	18.26 (10.02, 53.81)	-5.106	<0.001*
PLT	217.00 (171.00, 263.00)	217 (172.75, 264.75)	197 (150, 263)	-0.851	0.395
Hb	123.00 (102.00, 139.00)	123 (102, 140)	116 (101, 127)	-1.004	0.315
FIB	2.43 ± 0.79	2.48 ± 0.79	2.04 ± 0.67	9.758	0.002*
PT	11.80 (11.00, 12.50)	11.7 (10.9, 12.5)	12 (11.4, 12.3)	-1.918	0.055
APTT	28.00 (25.50, 29.90)	28.05 (25.5, 29.83)	28.2 (25.4, 30.5)	-0.486	0.627
TT	16.00 (14.10, 17.60)	16 (14.1, 17.6)	15.9 (14.2, 18.7)	-0.267	0.789
AT-III	94.35 ± 16.13	94.37 ± 16.16	94.15 ± 16.38	0.006	0.940
Ca2+	2.16 (2.06, 2.24)	2.16 (2.08, 2.24)	2.11 (1.99, 2.22)	-1.833	0.067
ISS	13 (11, 16)	12 (11, 15)	17 (13, 20)	-5.45	<0.001*
Outcomes	-	-	-	-	-
MODS	11 (3.9%)	7 (2.8%)	4 (11.4%)	6.160	0.013*
ARDS	4 (1.4%)	3 (1.2%)	1 (2.9%)	0.609	0.435
DVT	28 (9.8%)	25 (10.0%)	3 (8.6%)	0.658	0.417
Infections	7 (2.5%)	4 (1.6%)	2 (5.7%)	6.228	0.013*
Days of hospitalisation	21 (13, 32)	21.5 (14, 33)	16 (1, 31)	-2.263	0.024*
Days of ICU stay	0 (0, 3)	0 (0, 3)	0 (0, 5)	-1.307	0.191
Notes: AIS: Abbreviated injury scale; PLT: Blood platelet count; Hb: Haemoglobin; FIB: Fibrinogen; PT: Prothrombin time; APTT: Activated partial thromboplastin time; TT: Thrombin time; AT-III: Antithrombin-III; ISS: Injury severity score; GCS: Glasgow Coma Scale; SP: Systolic pressure; DP: Diastolic pressure; MODS: Multiple organ dysfunction syndrome; ARDS: Acute respiratory distress syndrome; DVT: Deep vein thrombosis; ICU: Intensive care unit. *Significant difference at p-value <0.05. Data are presented as mean (SD) or median (IQR). p-values were calculated using Student’s t-test for normally distributed continuous variables, the Mann-Whitney U test for non-normally distributed continuous variables, and the chi-square or Fisher’s exact test for categorical variables, as appropriate.

Table II: Determine the risk factors of in-hospital mortality by logistic regression analysis (n= 285).

	Β	S. E	OR	95% CI	p-values
D-Dimer	0.046	0.013	1.048	1.022, 1.072	<0.001
ISS	0.103	0.033	1.108	1.022, 1.072	0.002
GCS	-0.237	0.086	0.789	0.667, 0.935	0.006
Days of hospitalisation	-0.103	0.024	0.902	0.861, 0.946	<0.001
ISS: Injury Severity Score; GCS: Glasgow Coma Scale; OR: odds ratio; CI: 95% confidence interval. p-values were obtained from multivariate logistic regression analysis.

Figure 1: Cut-off of D-Dimer, ISS, and combined model for all-cause mortality.

Multivariate logistic regression identified D-Dimer (OR = 1.048, 95% CI: 1.022 - 1.072; p <0.001), ISS (OR = 1.108, 95% CI: 1.022 - 1.072; p = 0.002), GCS (OR = 0.789, 95% CI: 0.667 - 0.935; p = 0.006) and days of hospitalisation (OR = 0.902, 95% CI: 0.861-0.946; p <0.001) as independent predictors of in-hospital mortality (Table II).

Figure 2: Nomogram for predicting in-hospital death of a trauma patient.

A nomogram was constructed based on these predictors, allowing estimation of mortality risk by scoring each variable and projecting the total score to a risk-probability axis (Figure 2).

DISCUSSION

The condition of trauma patients is often complex and rapidly evolving, making timely and effective medical treatment crucial for preserving life and improving prognosis.¹ Early classification and severity assessment during the emergency phase are essential for guiding individualised treatment strategies, improving efficiency, and reducing mortality and disability.¹⁴ Identifying high-risk factors for in-hospital mortality further aids in optimising emergency care, minimising compli-cations, and improving outcomes.

Multivariate regression analysis in this study revealed that elevated D-Dimer levels, higher ISS scores, lower GCS, and shorter hospital stays were independent predictors of in-hospital mortality. The combined model of D-Dimer and ISS recorded at admission demonstrated superior predictive performance, with significantly higher mortality risk among patients with elevated levels of both markers. These asso-ciations remained significant after adjusting for confounders such as age and gender.

GCS was also found to be an independent predictor of mor-tality, consistent with previous studies showing its strong association with trauma outcomes.¹⁵ As a widely used triage tool, GCS effectively assesses consciousness and injury seve-rity, with meta-analyses supporting its predictive sensitivity and specificity.¹⁶ However, factors such as alcohol use or sedatives may interfere with GCS assessment. Additionally, shorter hospital stays were associated with higher mortality, which aligns with literature suggesting early deaths or compli-cations occur in more severely injured patients.¹⁷

ISS, derived from AIS, quantifies trauma severity based on anato-mical injury scores and is widely used in trauma care.⁶ Nume-rous studies have demonstrated that ISS is an independent predictor of mortality risk in trauma patients, with significantly higher morbidity and mortality rates for ISS >20 and very low survival rates for ISS >50.¹⁸ In this study, the death group had significantly higher ISS scores than the survival group, supporting ISS as a valid prognostic tool. However, ISS does not account for age, physiology, or systemic inflammation, limiting its individualisation.¹⁹ It also only considers the most severe injury in each region, potentially underestimating cumulative trauma burden.

D-Dimer, a product of fibrin cross-linking degradation, is a com-monly used laboratory indicator to evaluate the coagu-lation status of the body. D-Dimer levels are often significantly elevated in a variety of diseases such as severe trauma, infec-tion, and pulmonary embolism.²⁰ Due to its easy detection, low cost, and rapid results, D-Dimer has been widely used in clinical coagulation evaluation. Severe trauma can trigger endothelial injury and coagulation activation, leading to sustained D-Dimer elevation. This study confirmed significantly higher D-Dimer levels in non-survivors, supporting its prognostic value.

Both D-Dimer and ISS demonstrated individual predictive value for in-hospital mortality, with D-Dimer showing slightly better discrimination. The combined model achieved higher AUC values and statistically outperformed ISS alone, indicating improved predictive accuracy. While ISS remains a cornerstone in trauma assessment, its complexity and lack of dynamic responsiveness limit real-time use.⁷ D-Dimer, by contrast, offers rapid insights into systemic dysfunction and inflammation.¹³ Their combination integrates anatomical and physiological markers, enhancing risk stratification and aiding clinical decision-making.

This study has several limitations. First, it was a single-centre, retrospective analysis, which may introduce selection bias and limit the generalisability of the findings. Second, although statistical adjustments were made, residual confounding cannot be entirely excluded because of the observational nature of the study. Third, some variables had missing values or small subgroup sample sizes, which may have affected the power of statistical tests, particularly for categorical comparisons. Finally, the study focused on baseline characteristics and did not include follow-up outcomes or time-to-event data, which could provide more comprehensive insights into patient prognosis. Moreover, although this study demonstrates the prognostic significance of combining D-dimer and ISS in predicting in-hospital mortality, D-dimer itself lacks sufficient specificity and sensitivity, as its levels may be influenced by various non-specific conditions, limiting its standalone clinical utility. Future multicentre, prospective studies comparing this combined model with other established prognostic indicators—such as lactate, SOFA, or TRISS scores—are warranted to validate and expand upon the present findings.

CONCLUSION

This study demonstrates that both D-Dimer levels and ISS scores independently predict in-hospital mortality in trauma patients. Their combined use significantly improves predic-tive accuracy, yielding a higher AUC than either marker alone, indicating potential value in early risk stratification, and providing a more reliable auxiliary decision-making tool for emergency trauma management.

ETHICAL APPROVAL:
The study was approved by the Ethics Committee of Foshan Hospital of Traditional Chinese Medicine, Guangdong, China (Approval No. KY[2023]270). This retrospective study was conducted in accordance with the principles of the Declaration of Helsinki.

PATIENTS’ CONSENT:
Informed consent was not obtained due to the retrospective nature of the study design.

COMPETING INTEREST:
The authors declared no conflict of interest.

FUNDING:
This research is supported by the 2023 Self-Funded Science and Technology Innovation Project of Foshan (No. 23200010 06428).

AUTHORS’ CONTRIBUTIONS:
JP: Data collection, formal analysis, and manuscript drafting.
MK: Study conception, supervision, and critical revision.
QY: Data curation and statistical validation.
ZL: Visualisation, data checking, and manuscript editing.
All authors approved the final version of the manuscript to be published.

REFERENCES

1. Yolcu S, Sener K, Tapsiz H, Ozer AI, Avci A. Revised trauma score and CRAMS better predicted mortality in high-energy-trauma patients than early-warning score. Ir J Med Sci 2023; 192(4):1855-60. doi: 10.1007/s11845-022-03208-2.
2. Leilei D, Pengpeng Y, Haagsma JA, Ye J, Yuan W, Yuliang E, et al. The burden of injury in China, 1990–2017: Findings from the Global Burden of Disease study 2017. Lancet Public Health 2019; 4(9):e449-61. doi: 10.1016/S2468-2667(19)30 125-2.
3. Herbert HK, Hyder AA, Butchart A, Norton R. Global health: Injuries and violence. Infect Dis Clin North Am 2011; 25(3): 653-68, x. doi: 10.1016/j.idc.2011.06.004.
4. Liu XY, Qin YM, Tian SF, Zhou JH, Wu Q, Gao W, et al. Performance of trauma scoring systeISS, TRISS, and GTOS based on a systematic review and meta-analysis. Eur J Trauma Emerg Surg 2024; 50(4):1453-65. doi: 10.1007/ s00068-024-02467-1.
5. Rating the severity of tissue damage: I. The abbreviated scale. JAMA 1971; 215(2):277-80. doi: 10.1001/jama.1971. 03180150059012.
6. Baker SP, Oʼ Neill B, Haddon W, Long WB. The injury severity score: A method for describing patients with multiple injuries and evaluating emergency care. J Trauma 1974; 14(3):187-96.
7. Li H, Ma YF. New injury severity score (NISS) outperforms injury severity score (ISS) in the evaluation of severe blunt trauma patients. Chin J Traumatol 2021; 24(5):261-5. doi: 10.1016/j.cjtee.2021.01.006.
8. Kilgo PD, Meredith JW, Hensberry R, Osler TM. A note on the disjointed nature of the injury severity score. J Trauma 2004; 57(3):479-85; discussion 486-7. doi: 10.1097/01.ta.0000 141024.96440.7c.
9. Hsein YC, Wu IJ, Tan J, Huang SS, Lu KT, Su CH, et al. Serum levels of copeptin predict adverse outcomes and improve risk prediction of TRISS and MGAP scores in patients with multiple trauma: A single-centre prospective cohort study. J Trauma Acute Care Surg 2023; 94(2):336-43. doi: 10.1097/ TA.0000000000003793.
10. Liu H, Li H, Han L, Zhang Y, Wu Y, Hong L, et al. Inflamma-tory risk stratification individualises anti-inflammatory pharma-cotherapy for acute type A aortic dissection. Innovation (Camb) 2023; 4(4):100448. doi: 10.1016/j.xinn.2023.100 448.
11. Zhang Y, Luo S, Xie Y, Wang Y, Fang Y, Wang S, et al. Postoperative intra-abdominal hypertension predicts worse hospital outcomes in children after cardiac surgery: A pilot study. Interdiscip Cardiovasc Thorac Surg 2024; 38(2): ivae019. doi: 10.1093/icvts/ivae019.
12. Sartini S, Spadaro M, Cutuli O, Castellani L, Sartini M, Cristina ML, et al. Does antithrombotic therapy affect outcomes in major trauma patients? A retrospective cohort study from a tertiary trauma centre. J Clin Med 2022; 11(19):5764. doi: 10.3390/jcm11195764.
13. Nakama R, Yamamoto R, Izawa Y, Tanimura K, Mato T. Serum D-dimer level as a biomarker for identifying patients with isolated injury to prevent unnecessary whole-body computed tomography in blunt trauma care. Scand J Trauma Resusc Emerg Med 2021; 29(1):12. doi: 10.1186/s13049-020-00 815-9.
14. Schipper IB, Simmen HP, Pfeifer R. Trauma systems in Europe / hospital categories. Eur J Trauma Emerg Surg 2025; 51(1):143. doi: 10.1007/s00068-025-02828-4.
15. Chien YC, Chiang WC, Chen CH, Sun JT, Jamaluddin SF, Tanaka H, et al. Comparison of on-scene Glasgow Coma Scale with GCS-motor for prediction of 30-day mortality and functional outcomes of patients with trauma in Asia. Eur J Emerg Med 2024; 31(3):181-7. doi: 10.1097/MEJ.0000000 000001110.
16. Ahmadi S, Sarveazad A, Babahajian A, Ahmadzadeh K, Yousefifard M. Comparison of glasgow coma scale and full outline of unresponsiveness score for prediction of in-hospital mortality in traumatic brain injury patients: A systematic review and meta-analysis. Eur J Trauma Emerg Surg 2023; 49(4):1693-706. doi: 10.1007/s000 68-022-02111-w.
17. Spencer AL, Hosseinpour H, Nelson A, Hejazi O, Anand T, Khurshid MH, et al. Predicting the time of mortality among older adult trauma patients: Is frailty the answer? Am J Surg 2024; 237:115768. doi: 10.1016/j.amjsurg.2024.05.009.
18. Wollner G, Weihs V, Frenzel S, Aldrian S, Negrin LL. Musculus psoas major morphology - A novel predictor of mortality in elderly polytraumatized patients. BMC Emerg Med 2023; 23(1):13. doi: 10.1186/s12873-023-00783-0.
19. Jin H, Zhang Y, Zhang Q, Ouyang L, Li X, Zhang Y, et al. Comparison of injury severity score (ISS) and new injury severity score (NISS) in the evaluation of thoracic trauma patients: A retrospective cohort study. Emerg Med Int 2024; 2024:4861308. doi: 10.1155/2024/4861308.
20. Sun C, Xi F, Li J, Yu W, Wang X. Longitudinal D-dimer trajectories and the risk of mortality in abdominal trauma patients: A group-based trajectory modelling analysis. J Clin Med 2023; 12(3):1091. doi: 10.3390/jcm12031091.