Admission FDP and D-Dimer Predict Functional Outcomes in Closed TBI: A Single-Center Retrospective Cohort Study

1. Introduction

Traumatic brain injury (TBI) is a global leading cause of death and disability. The 2021 Global Burden of Disease study recorded 20.84 million new and 37.93 million prevalent TBI cases worldwide, leading to 5.48 million years lived with disability [1]. Maas et al. reported TBI carries the highest incidence among neurological illnesses and brings both acute damage and long-term chronic sequelae [2]. Closed TBI, mostly induced by traffic crashes and falls, causes diverse clinical outcomes; many patients recover rapidly, yet a large proportion sustain persistent functional deficits that damage life quality and social productivity [3]. Thus early screening for high-risk TBI patients is vital for personalized clinical management.

Coagulopathy is a unique acute TBI complication. Damaged brain tissue secretes tissue factor-rich microvesicles into blood, activating extrinsic coagulation and secondary hyperfibrinolysis to break hemostatic balance [4]. Zhang et al. clarified this pathological cascade: local brain injury triggers systemic coagulation disorders, mixing hypercoagulability and hyperfibrinolysis to worsen intracranial and systemic hemorrhage [5]. Clinical data indicate coagulopathy occurs in 10–97% of TBI patients, raising mortality risk 9-fold and adverse outcome odds 30-fold [6].

D-dimer (DD) and fibrin degradation products (FDP) are sensitive fibrinolytic markers. DD peaks within 3 hours post-TBI then declines gradually [7]. Hayakawa et al. verified hyperfibrinolysis after TBI correlates with hematoma expansion, with DD and FDP reflecting this dynamic process [8]. Multiple cohorts have validated DD’s prognostic value for death: a 1,338-patient multicenter trial found elevated admission DD independently predicted in-hospital complications and long-term mortality [9]; a Japanese national trauma database also linked high DD to chronic neurological dysfunction [10]. In contrast, clinical evidence supporting FDP as a prognostic marker remains limited. Kaufman et al. first identified an inverse correlation between admission FDP and GOS in severe closed TBI, with predictive performance comparable to GCS [11], while a 204-patient cohort further confirmed FDP independently predicts poor prognosis among mild TBI individuals [12].

Four critical research gaps remain unaddressed in existing literature. First, most prior studies adopted mortality or GOS as primary endpoints rather than discharge mRS, a scale enabling finer grading of post-discharge functional impairment. Second, relevant evidence predominantly focuses on severe mixed TBI populations, with few data specific to mild-to-moderate isolated closed TBI. Third, head-to-head comparisons of the independent predictive capacity of DD and FDP within identical cohorts are scarce. Fourth, the association between the two fibrinolytic markers and hospital length of stay (LOS) has rarely been explored, restricting understanding of their outcome-specific predictive roles.

Against this background, the present single-center retrospective cohort study analyzed five routine admission coagulation indicators (PT, TT, FBG, FDP, DD) and their independent associations with discharge mRS among closed TBI patients. We adopted three sequentially adjusted regression models, restricted cubic spline (RCS) and subgroup analyses to systematically compare the prognostic performance of DD and FDP, and further explored their correlation with LOS to distinguish their predictive specificity for different clinical outcomes. Briefly, our work intends to supply clinical evidence supporting early risk stratification using routine fibrinolytic markers. Notably, our analysis reveals that admission FDP acts as a more robust independent predictor of post-discharge functional disability than D-dimer, with a detectable nonlinear threshold effect, while neither marker correlates with hospitalization duration—findings that fill the abovementioned research gaps through direct comparative assessment of the two biomarkers in mild-to-moderate closed TBI.

2. Materials and Methods

2.1. Study Design and Participants

This single-center retrospective cohort enrolled consecutive closed TBI inpatients admitted to the 941st Hospital between September 2023 and December 2025. Inclusion criteria: 1) clinical and imaging-confirmed mild/moderate/severe closed TBI or closed epidural hematoma; 2) complete admission coagulation tests; 3) pre-injury and discharge mRS records. Exclusion criteria: 1) unidentifiable patient ID; 2) over 30 missing core variables; 3) uncorrectable admission data errors.

2.2. Data Collection

All data were extracted from electronic medical records, including:

Demographics: age, sex, BMI, blood type, smoking and drinking duration/volume.

Injury data: injury mechanisms, admission time, primary diagnosis, binary comorbidities (scalp injury, fracture, intracranial hemorrhage, concussion, etc.).

Admission vital signs: temperature, pulse, respiratory rate, SBP, DBP.

Laboratory tests: blood routine, five coagulation indicators, liver/kidney function, serum electrolytes.

Medical history: hypertension, coronary heart disease, diabetes and disease course; pre-injury and discharge mRS.

2.3. Outcome Definitions

Primary outcome: discharge mRS, which was categorized into investigator-defined ordinal three tiers (0 asymptomatic, 1 mild dysfunction, ≥2 obvious disability) as well as standard binary grouping (mRS≥2 = poor outcome, mRS<2 = good outcome). Secondary outcome: continuous LOS value.

2.4. Exposure Variables

PT, TT, FBG, FDP, DD were core exposures. Log1p transformation was applied to FDP and DD to correct right-skewed distribution before regression.

2.5. Statistical Analysis

Continuous data: median (IQR), Kruskal–Wallis test for intergroup comparison. Categorical data: count (percentage), Fisher’s exact test. All comparisons stratified by discharge mRS tiers.

Spearman correlation was used to assess associations between lab indices, LOS and mRS, visualized via heatmaps (*P<0.05, **P<0.01, ***P<0.001).

Ordinal proportional odds regression matched 3-tier mRS. The proportional odds assumption was tested using the Brant test, and no significant violation was detected (all P > 0.05), supporting the use of the proportional odds model. Binary logistic regression matched dichotomized mRS. Three adjustment schemes: Model 1 crude; Model 2 adjusted for age + sex; Model 3 further adjusted for primary diagnosis and baseline mRS. Results shown as OR with 95% CI.

RCS with four knots assessed nonlinear dose-response of DD/FDP, output overall and nonlinear P values via Wald test.

Subgroup stratification: age (<65/≥65), sex, hypertension history, injury mechanisms, TBI severity. Firth penalized logistic regression solved small-sample separation bias; interaction P calculated via likelihood ratio test.

Negative binomial regression analyzed DD/FDP and LOS (adjusted for age, sex, injury mechanisms, hemorrhage, fracture), reported as incidence rate ratios (IRRs) with 95%CI.

All analyses used R 4.4.2, two-sided α=0.05.

2.6. Ethical Approval

This study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of The 941st Hospital of the Joint Logistics Support Force of the Chinese People's Liberation Army (approval number: 941-IRB-2026005). Informed consent was waived because of the retrospective study design and use of anonymized clinical data.

No generative artificial intelligence tools were used in the design, execution, data analysis, figure generation, or writing of this manuscript.

Restrictions apply to the availability of these clinical patient-level data. Individual electronic medical record data cannot be publicly shared owing to hospital data-protection regulations; aggregated analytical datasets may be available upon reasonable request to the corresponding author with formal ethical approval.

3. Results

3.1. Baseline Characteristics

A total of 140 patients with closed TBI were stratified by discharge mRS into three groups (Table 1).

Patients with poorer functional outcomes were significantly older (P = 0.018) and more likely to be male (P = 0.016). Pulse rate (P = 0.021) and respiratory rate (P = 0.001) increased progressively with worsening outcomes, while blood pressure showed no significant difference (P > 0.05).

Among coagulation markers, D-dimer and FDP showed the most significant differences across groups and increased progressively with worsening outcomes (both P < 0.001); prothrombin time and fibrinogen also showed significant differences (P = 0.026 and P = 0.029, respectively). In the mRS ≥2 group, white blood cell count (P = 0.011) and neutrophil count (P = 0.003) were elevated, while lymphocyte count (P = 0.024) was decreased. Blood urea nitrogen levels increased with rising mRS scores (P = 0.037), whereas other biochemical indicators were comparable across groups (all P > 0.05).

Regarding clinical characteristics, there were significant differences between groups in the distribution of initial diagnoses (P < 0.001) and smoking history (P = 0.045), while there were no significant differences in injury mechanisms, length of hospital stay, comorbidities, or other clinical variables (all P > 0.05).

3.2. Correlation Analysis of Laboratory Parameters and Functional Outcome

Spearman rank correlation analysis (Figure 1A) revealed that D-dimer (rₛ = 0.40, P < 0.001) and FDP (rs = 0.38, P < 0.001) showed the stronger correlations with discharge mRS score among all laboratory parameters, with a moderate positive correlation between the two markers themselves (rs = 0.57, P < 0.001).

Neutrophil count (rs = 0.29, P < 0.01), white blood cell count (rs = 0.25, P < 0.01), blood urea nitrogen (rs = 0.23, P < 0.05), fibrinogen (rs = 0.20, P < 0.05), monocyte count (rs = 0.20, P < 0.05), and prothrombin time (rs = 0.14, P < 0.05) were also significantly and positively correlated with discharge mRS score, whereas lymphocyte count was inversely correlated (rs = -0.24, P < 0.05). Other biochemical indices and LOS showed no significant correlation with discharge mRS. Within coagulation parameters, FDP was positively correlated with white blood cell count (rs = 0.19, P < 0.05) and neutrophil count (rs = 0.27, P < 0.001), and inversely correlated with lymphocyte count (rs = -0.32, P < 0.001), suggesting a potential link between coagulation activation and systemic inflammation.

These patterns were further corroborated by violin plots of log1p-transformed coagulation parameters (Figure 1B), in which D-dimer and FDP displayed a clear stepwise elevation across mRS groups, reaching substantially higher levels in the mRS ≥ 2 group compared with both the mRS = 0 and mRS = 1 groups (both P < 0.001). PT and FBG followed a similar directional trend, though with comparatively smaller effect sizes.

3.3. Ordinal Logistic Regression: Coagulation Parameters and Functional Outcome

To evaluate the independent predictive value of each coagulation parameter for discharge functional outcome, we fitted three progressively adjusted proportional odds models with discharge mRS (0 / 1 / ≥2) as the ordinal outcome (Table 2).

In the unadjusted model (Model 1), D-dimer (OR = 5.611, 95% CI: 2.690–11.706, P < 0.001), FDP (OR = 2.538, 95% CI: 1.700–3.789, P < 0.001), and fibrinogen (OR = 6.390, 95% CI: 1.209–33.764, P = 0.029) were significantly associated with higher discharge mRS. PT and TT were not significant in Model 1 (both P > 0.05). After adjustment for age and sex (Model 2), D-dimer (OR = 5.067, 95% CI: 2.383–10.774, P < 0.001) and FDP (OR = 2.370, 95% CI: 1.569–3.579, P < 0.001) retained strong significance, while the association for fibrinogen was attenuated to non-significance (OR = 4.655, 95% CI: 0.839–25.828, P = 0.078). In the fully adjusted model (Model 3, additionally adjusted for primary diagnosis, systolic blood pressure, and pre-injury mRS), FDP remained the only coagulation parameter with a statistically significant independent association with functional outcome (OR = 1.749, 95% CI: 1.019–3.002, P = 0.042). The effect of D-dimer was markedly attenuated and lost significance (OR = 2.373, 95% CI: 0.916–6.145, P = 0.075), indicating that its prognostic contribution was substantially explained by injury severity and baseline functional status. PT, TT, and FBG were not significant in any model. These results identify FDP as the most robust independent predictor of functional outcome at discharge among all coagulation parameters examined.

3.4. Dose–Response Relationships: Restricted Cubic Spline Analysis

To explore potential non-linear dose–response relationships, RCS analyses with four knots were conducted for D-dimer and FDP (Figure 2).

For D-dimer (Figure 2A), both the unadjusted and age- and sex-adjusted models revealed significant positive monotonic associations with discharge mRS (P overall < 0.001 for both), with no evidence of non-linearity (P non-linear = 0.552 and 0.632, respectively), suggesting an approximately log-linear relationship. In the fully adjusted model, however, the association was substantially attenuated and no longer statistically significant (P overall = 0.236), accompanied by a flattened OR curve and markedly widened confidence intervals.

For FDP (Figure 2B), significant positive associations were similarly observed in the unadjusted and minimally adjusted models (P overall < 0.001 for both). Of note, Model 2 revealed significant non-linearity (P non-linear = 0.029), with the OR rising gradually at lower FDP values before accelerating at moderate-to-high concentrations — a pattern consistent with a threshold effect. As with D-dimer, the association was attenuated in the fully adjusted model (P overall = 0.197).

Taken together, both markers demonstrated clear positive dose–response relationships with functional outcome in the unadjusted and minimally adjusted models. The non-linear pattern observed for FDP raises the possibility of a threshold beyond which its association with poor outcome becomes more pronounced.

3.5. Binary Logistic Regression and Subgroup Analyses

Binary logistic regression analyses of the association between coagulation parameters and unfavorable functional outcome (mRS ≥2) are presented in Table 3.

Consistent with ordinal regression findings, D-dimer and FDP were significantly associated with poor outcome in Models 1 and 2 (D-dimer: Model 1 OR = 4.280, 95% CI: 1.783–10.271, P = 0.001; Model 2 OR = 4.510, 95% CI: 1.670–12.182, P = 0.002; FDP: Model 1 OR = 2.334, 95% CI: 1.488–3.661, P < 0.001; Model 2 OR = 2.246, 95% CI: 1.357–3.716, P = 0.001). In the fully adjusted model, both associations were attenuated to non-significance (D-dimer: OR = 1.550, 95% CI: 0.387–6.210, P = 0.536; FDP: OR = 1.483, 95% CI: 0.804–2.738, P = 0.207). FBG, TT, and PT were non-significant in all models. Binary and ordinal regression results were concordant in Models 1 and 2 but diverged in Model 3, where FDP retained significance in the ordinal model but not in the binary model. This divergence likely reflects the reduced statistical power of binary logistic regression when the unfavorable-outcome group is small, as binary categorization discards within-group ordering information that the proportional odds model exploits. The ordinal model is therefore designated as our primary analysis. Subgroup analyses are presented in Figure 3.

For D-dimer (Figure 3A), the overall OR was 4.10 (95% CI: 1.79–10.06, P < 0.001), with significant associations observed in male patients, the <65-year age group, patients without hypertension, and those with falls as the injury mechanisms (all P < 0.05). For FDP (Figure 3B), the overall OR was 2.29 (95% CI: 1.49–3.62, P < 0.001), with significant associations across both age groups, in males, patients without hypertension, and those with falls, road traffic accidents, and mild TBI (all P < 0.05). Interaction tests for both markers were non-significant across all subgroups (all P for interaction > 0.05), indicating consistent predictive effects irrespective of patient characteristics.

3.6. Coagulation Parameters and Length of Stay

Results of negative binomial regression analyses for LOS are presented in Table 4.

After adjustment for age, sex, injury mechanisms, primary diagnosis, intracranial hemorrhage, and fracture, neither D-dimer (IRR = 1.08, 95% CI: 0.904–1.300, P = 0.385) nor FDP (IRR = 1.04, 95% CI: 0.941–1.150, P = 0.452) was significantly associated with LOS. These findings stand in contrast to the primary outcome analyses and suggest that the predictive utility of D-dimer and FDP is outcome-specific, reflecting functional recovery status rather than overall disease severity as expressed through hospitalization duration.

Covariate effects were highly consistent across both models. Injury mechanisms were the predominant determinant of LOS: compared with road traffic accident, falls on the same level (DD model IRR = 0.756, 95% CI: 0.611–0.935, P = 0.010; FDP model IRR = 0.754, 95% CI: 0.608–0.935, P = 0.010) and assault injuries (DD model IRR = 0.628, 95% CI: 0.481–0.819, P < 0.001; FDP model IRR = 0.624, 95% CI: 0.477–0.816, P < 0.001) were associated with significantly shorter hospital stays, likely reflecting the greater injury complexity and medico-legal considerations associated with road traffic accidents. Age, sex, injury severity, intracranial hemorrhage, and fracture did not reach statistical significance in either model (all P > 0.05).

4. Discussion

This cohort of 140 closed TBI patients confirmed admission FDP as the only independent coagulation marker for discharge mRS after full confounder adjustment. DD’s predictive value vanished after controlling baseline function and diagnosis. FDP showed a nonlinear threshold effect in minimally adjusted models, while neither marker correlated with hospital stay, indicating outcome-specific predictive capacity.

Intergroup comparison demonstrated that DD/FDP levels rose alongside functional disability, consistent with fibrinolysis activation aggravating neurological damage [5,13]. Ordinal regression confirmed FDP’s stable independent predictive ability across three adjustment schemes, while DD’s effect was explained by underlying injury severity. RCS first identified FDP’s threshold risk growth in mild-to-moderate TBI.

Existing literature mainly focused on DD’s link to hemorrhage and death [14,15,16,17,18,19,20], while FDP’s prognostic evidence was limited [17,21,22,23,24,25,26]. Few head-to-head cohort comparisons of DD and FDP used mRS as primary endpoint. Nakae et al. also found fibrinolytic markers differentiated prognosis better than routine coagulation tests [7]; our study further confirmed FDP’s superior robustness after comprehensive correction.

Although both are fibrinolytic products, FDP and DD differ in pathophysiological significance: FDP reflects degradation of both fibrinogen and fibrin (primary and secondary fibrinolysis) [27], whereas DD specifically marks cross-linked fibrin breakdown (secondary fibrinolysis) [28]. Thus FDP provides a more comprehensive assessment of fibrinolytic activation and stronger predictive power. In acute TBI, damaged brain releases TF, activating extrinsic coagulation and generating fibrin [29]; TF may also enter circulation via disrupted blood-brain barrier, causing consumptive coagulopathy [29]. Concurrently, hypoperfusion and endothelial injury release t-PA, inducing hyperfibrinolysis [17,25,26,30,31]. Both markers rise, but FDP captures the entire cascade from fibrinogen consumption to cross-linked fibrin lysis, while DD reflects downstream events after thrombin generation and correlates more with “injury severity” [32]. After adjustment for diagnosis and baseline mRS, DD’s effect attenuates, suggesting DD mediates severity; FDP retains independent predictive power due to comprehensive fibrinolytic assessment.

Injury mechanisms were the predominant determinants of LOS, rather than fibrinolytic markers. LOS is disturbed by non-clinical administrative factors, while DD/FDP specifically reflect brain tissue damage. Subgroup tests validated stable predictive value across demographic and injury subgroups, and FDP effectively stratifies risk even for mild TBI patients. The nonlinear threshold of FDP requires large multicenter verification before clinical application.

FDP testing is cheap and routine in all hospitals. Clinicians should monitor admission FDP for closed TBI patients; high FDP values warrant frequent neuroimaging and serial coagulation tests, even with normal GCS and vital signs. FDP cannot replace GCS and imaging for standalone judgment.

Strengths: strict closed TBI inclusion reduced extracranial injury confounding; dual ordinal/binary mRS endpoints; progressive adjustment + RCS nonlinear modeling; direct comparison of five coagulation indices.

Limitations: Several limitations should be acknowledged. First, the single-center retrospective design may introduce selection bias and limit generalizability. Second, the relatively small sample size—particularly in the mRS ≥2 group—reduced statistical power for subgroup analyses and precluded propensity score matching. Third, only a single admission measurement of coagulation parameters was available; serial monitoring over time would have better captured the dynamic evolution of fibrinolytic dysfunction and its relationship to neurological recovery. Fourth, follow-up was restricted to discharge mRS, without 3- or 6-month functional assessments, preventing evaluation of the markers' longer-term prognostic utility. Future multicenter prospective studies with longitudinal sampling and extended follow-up are warranted to validate our findings.

5. Conclusions

This study confirms that, in patients with closed TBI, admission FDP levels independently predict poorer discharge mRS, outperforming D-dimer and other coagulation markers after comprehensive adjustment for age, sex, diagnostic classification, systolic blood pressure, and baseline functional status. A significant nonlinear dose–response relationship was observed in the age- and sex-adjusted model, suggesting a potential predictive threshold, though external validation in a larger cohort is required. Neither FDP nor D-dimer was significantly associated with length of stay, indicating that their predictive role is outcome-specific and points to neurological impairment rather than general disease complexity.

From a clinical translational perspective, FDP is a routine, cost-effective, and rapidly measurable laboratory parameter widely available in most emergency settings. Its integration into the initial emergency workup of closed TBI patients could serve as an adjunctive screening tool to identify individuals at heightened risk of poor functional recovery—even when GCS appears preserved—thereby facilitating earlier neurological surveillance and more individualized treatment planning. However, FDP should be interpreted in conjunction with GCS, neuroimaging, and other clinical findings rather than as a standalone decision-making criterion. These findings support the use of admission FDP as an adjunctive biomarker for early functional prognostication in closed TBI; multicenter prospective studies with longitudinal sampling and extended follow-up are warranted to further validate its clinical utility.