Immediate Postoperative Coagulopathy Predicts the Long-term Survival of Traumatic Brain Injury Patients: a Retrospective Cohort Study


 Objective: To assess the association between immediate postoperative coagulopathy and the long-term survival of traumatic brain injury (TBI) patients undergoing surgery, as well as to explore predisposing risk factors of immediate postoperative coagulopathy.Methods: This retrospective study included 352 TBI patients from January 1, 2015, to April 25, 2019. The log-rank test and a Cox proportional hazard model were conducted to assess the relationship between immediate postoperative coagulopathy and the long-term survival of TBI patients. Furthermore, a multivariate logistic regression model was performed to identify the underlying risk factors for postoperative coagulopathy.Results: Of the 352 patients analyzed, the median age was 50 (41,60) years, and 82 (23%) patients were female. By May 26, 2019, 117 (33.24%) patients had died, 195 (55.40%) had survived, and 40 (11.36%) had been lost to follow-up. The median follow-up time was 773 days. In the log-rank test, immediate postoperative coagulopathy was significantly associated with the survival of TBI patients (P = 0.002). A Cox proportional hazard model identified immediate postoperative coagulopathy (HR, 1.471; 95% CI, 1.011-2.141; P = 0.044) as an independent risk factor for survival following TBI. According to multivariate logistic regression analysis, abnormal ALT and RBC at admission, intraoperative infusion of crystalloid solution > 2900 mL, infusion of colloidal solution > 1100 mL and intraoperative bleeding > 950 mL were identified as independent risk factors for immediate postoperative coagulopathy.Conclusions: Those who suffered from immediate postoperative coagulopathy due to TBI were at higher risk of poor prognosis than those who did not.


Introduction
Traumatic brain injury (TBI) is one of the leading causes of death and disability worldwide, and it represents a global health concern and nancial burden [1,2]. The main causes of early death in trauma victims are acidosis, hypothermia and coagulopathy, which are related to and in uence each other. Thus, this vicious circle is often referred to as the " trauma triangle of death " [3,4]. Trauma-induced coagulopathy manifests as a state of hypercoagulopathy trending towards thrombosis [5] and a state of hypocoagulopathy with progressive intracranial hemorrhage and increased systemic bleeding [6,7].
Studies have shown that trauma-induced coagulopathy is common in TBI patients [8][9][10] and the incidence of coagulation disorders has substantial heterogeneity, ranging from 7-54% [11,12]. Reasons for this variation include different techniques and de nitions used, the heterogeneity of the patients and the various testing times [13]. Secondary coagulopathy after TBI represent an important factor for unfavorable prognosis [14,15], resulting in a nine-fold higher risk of death and a 30-fold higher risk of poor prognosis than that in TBI patients without secondary coagulation disorders [7,9,16]. Mortality in TBI patients with coagulopathy is also highly heterogeneous, ranging from 22-66% [17,18]. TBI patients with coagulopathy tend to suffer from delayed or progressive intracranial hemorrhage, as well as from microvascular thrombosis [19,20].
Retrospective and observational studies have focused on coagulation upon admission or the presence of any coagulation disorders during the period of hospitalization [21,22]. A multicenter study described the course of coagulopathy in patients with isolated TBI, and associated it with computed tomography (CT) characteristics and outcomes [15]. Previous studies mainly focused on the coagulopathy on admission, however, the association between coagulopathy in perioperative period, especially immediate postoperative coagulopathy, and long-term survival of TBI patients has not been well explored. It is of great signi cance to explore this relationship in view of the fact that many TBI patients require surgical interventions, and it has been well established that perioperative management impacts coagulation functions. We therefore investigated for whether immediate postoperative coagulopathy was related to the long-term survival of these patients. Furthermore, we investigated the predisposing risk factors that may cause immediate postoperative coagulopathy, so that these risk factors could be controlled and managed to avoid coagulopathy.

Study protocol
This study was approved by the institutional Eethics Bboard of Tangdu Hospital (No. 201909-13) and was registered with ClinicalTrials.gov (NCT04322721).

Cohort
The clinical data of 447 consecutive TBI patients were retrospectively collected from January 1, 2015, to April 25, 2019, in Tangdu Hospital of the Fourth Military Medical University. We included patients who recorded an intracranial injury as the main diagnosis or the coexisting diagnoses upon admission.
Intracranial injuries were identi ed by WHO ICD-11 (NA07.0 to NA07.9). The inclusion criteria were any patients with mild, moderate, or severe TBI who entered the operating room for surgery, including for decompressive craniectomy and the evacuation of an intracranial hematoma. Other inclusion criteria were that the interval from injury to admission was less than 24 hours, an absence of other severe extracranial injuries, and an extracranial Abbreviated Injury Scale (AIS) score < 3. TBI patients who did not undergo surgery or who did not survive before undergoing any surgery were excluded from the study. TBI patients who had hemorrhagic or ischemic cerebrovascular disease within half a year or who had other systemic diseases, such as uremia, cirrhosis, or malignant tumors, were also excluded.

Indicators
We recorded the coagulation function at admission and on the rst day after operation. Coagulopathy was de ned as an aPTT > 40 sec and/or an INR > 1.2 and/or a platelet count < 100 x 10 9 per liter [23,24]. Patient demographic characteristics and CT ndings were also retrospectively adjusted to evaluate the relationship between perioperative coagulation and long-term outcome. The demographic data collected included age; gender; admission AIS (head) ; admission Glasgow Coma Scale (GCS), which was categorized as moderate (GCS 9-13) or severe (GCS ≤8); pupil reaction at admission, and CT ndings of midline shift, subarachnoid hemorrhage and effaced basal cisterns. Biochemical tests upon admission included serum glucose (GLU), hemoglobin (GB), aspartate aminotransferase (AST), alanine aminotransferase (ALT), and red blood cells (RBCs). GLU > 151 mg/dL, GB < 120 g/L, AST ≥ 40 U/L, ALT ≥ 40 U/L, and RBCs (men > 5.5×10 12 /L or < 4.0×10 12 /L, and women> 5.0 ×10 12 /L or < 3.5×10 12 /L) were regarded as abnormal values. Furthermore, we reviewed the intraoperative factors, including the preoperative shock index, infusion of crystalloid solution and colloidal solution, ratio of infusion plasma to RBCs, perioperative-bleeding, net uid input, total surgery time and surgical approach. The net uid input was calculated as the sum of the crystalloid solution and colloid solution and infusion of blood products, and was then subtracted from the amount of intraoperative blood loss and urine. The shock index was categorized as normal (< 1.0), mild (1.0-1.5), moderate or severe shock (> 1.5). The crystalloid solution, colloidal solution, intraoperative-bleeding, net uid input, and total surgery time were analyzed as categorical variables, which were classi ed using ROC curve analysis based on postoperative coagulation.

Statistics
Follow-up was conducted by telephone or inpatient or outpatient review, and the nal follow-up date was May 26, 2019. Patients were followed up every six months. We used the Kaplan-Meier estimator to estimate the survival rates and performed log-rank tests to compare survival rates between the survival group and death group. A Cox proportional hazard model was used to estimate the hazard ratio and 95% con dence interval (CI). Furthermore, univariate and multivariate logistic regression analysis were performed to explore the underlying factors that could lead to perioperative coagulation dysfunction. A P value < 0.05 was regarded as signi cant. Statistical analysis was carried out using SPSS 20.0 (IBM, New York, NY).

Patients Demographics and Clinical Data
Of the 352 patients analyzed, the median age was 50 (41,60) years, and 82 (23%) patients were female. Overall, 175 (49.72%) patients suffered from immediate postoperative coagulopathy. Other variables are shown in Tables 1 and 2. Next, we compared the differences in variables at admission and the perioperative period between patients with immediate postoperative coagulopathy and those without. The differences in variables at admission between patients with and without immediate postoperative coagulopathy are listed in Table  1 Immediate postoperative coagulopathy and survival By May 26, 2019, 117 (33.24%) patients had died, 195 (55.40%) had survived, and 40 (11.36%) had been lost to follow-up. In the log-rank test, immediate postoperative coagulopathy was signi cantly associated with the survival of TBI patients (P = 0.002). The survival analysis showed that the 1-year survival rate of total TBI patients was 0.676, and the 3-year survival rate was 0.628 ( Figure 1). The 1-year survival rate of TBI patients suffering from postoperative coagulopathy was 0.595, and the 3-year survival rate was 0.551. Accordingly, the 1-year survival rate of TBI patients without postoperative coagulopathy was 0.759 and the 3-year survival rate was 0.708 ( Figure 1). Multivariate Cox regression analysis identi ed immediate postoperative coagulopathy (adjusted HR, 1.471; 95% CI, 1.007-2.133; P = 0.046) as an independent risk factor for survival following TBI (Table 3), which indicated that immediate postoperative coagulopathy is of great clinical importance and worthy of further attention.

Discussion
This study analyses of the relationship between outcomes of TBI patients requiring surgery and immediate postoperative coagulopathy using a survival analysis, and it analyzed the potential risk factors causing immediate postoperative coagulopathy. We found that immediate postoperative coagulopathy (adjusted HR, 1.466; 95% CI, 1.007-2.133; P = 0.046) was identi ed as an independent risk factor for survival following TBI. The 1-year survival rate of TBI patients suffering from immediate postoperative coagulopathy was 0.595, whereas that of patients without immediate postoperative coagulopathy was 0.759. The risk factors that contributed to immediate postoperative coagulopathy included abnormal ALT and RBC at admission, infusion of crystalloid solution > 2900 mL, infusion of colloidal solution > 1100 mL, and intraoperative bleeding > 950 mL.
In the log-rank test, we found that immediate postoperative coagulopathy was signi cantly associated with the survival of TBI patients (P = 0.002). In the Cox regression model analysis, immediate postoperative coagulopathy was identi ed as an independent risk factor. Age, GCS, AIS (head) , pupil reaction, effaced basal cisterns, and coagulopathy at admission were adjusted. The prognostic model, IMPACT (International Mission for Prognosis and Clinical Trial), has shown that three most prognostic predictors: age, GCS, and pupillary reactivity [25], which all were adjusted in our prognostic model. Older age was analyzed as a signi cant predictor of prognosis in TBI [26,27], which was also showed in our study. Undoubtedly, the GCS score and AIS (head) , which described the severity of TBI patients, are crucial factors affecting long-term survival, as previous shown [27]. In the log-rank test, coagulopathy at admission was identi ed as a risk factor in our study. However, it showed no signi cance in the Cox regression model analysis. Meanwhile, postoperative coagulopathy was an independent risk factor for long-term survival, which indicated that postoperative coagulopathy is of great clinical importance and worthy of further attention.
A previous study showed that delayed or sustained trauma-induced coagulopathy was more frequently associated with unfavorable outcome than early, short-lasting coagulopathy [15]. Postoperative coagulopathy, as a delayed or more speci c form, was also related to prognosis in our study. We also found that the incidence of postoperative coagulopathy was much higher than that of coagulopathy at admission (50.64% vs 18.59%), which further indicated that postoperative coagulopathy might have profound clinical signi cance. We did not include preoperative coagulation dysfunction because of the particularity of patients with traumatic brain trauma undergoing surgery. Most of these patients requiring surgery were operated on within a few hours of admission. Preoperative coagulation function will overlap to some extent. In light of this, it is reasonable that we did not include preoperative coagulation function.
Overall, 312 TBI patients underwent surgery and completed follow-up, with a 1-year survival rate of 0.676 and a 3-year survival rate of 0.628. Interestingly, we found that the 1-year survival rate decreased rapidly to 0.676, while the 3-year survival rate decreased by only 0.048 (Fig. 1). This suggests that TBI patient deaths in those who underwent surgery were concentrated in the rst year. In other words, if the patient made it through the rst year, his or her chances of survival were high. The 1-year and 3-year survival rates of TBI patients suffering from postoperative coagulopathy were 0.595 and 0.551, respectively, which were accordingly below the overall survival rate. In contrast, the 1-year and 3-year survival rates of TBI patients with normal coagulation were 0.759 and 0.708, respectively, which were accordingly above the overall survival rate (Fig. 1).
Since immediate postoperative coagulopathy was identi ed as an independent risk factor for long-term survival, we next discussed the underlying risk factors for coagulopathy. Abnormal ALT and RBC at admission were identi ed as independent risk factors for immediate postoperative coagulopathy, which suggested that abnormal liver function and hypovolemia might lead to immediate postoperative coagulopathy. Interestingly, some perioperative factors, including the infusion of crystalloid solution > 2900 mL, colloidal solution > 1100 mL or their sum > 3450 mL, and the net-uid-input > 2425 mL, were risk factors for postoperative coagulation disorders. On the one hand, the infusion of large amounts of uid results in the dilution of coagulation factors in the plasma. On the other hand, whether the input of a large amount of uid during surgery will cause the further release of brain-derived molecules to the circulatory system of the whole body and further aggravate coagulation disorders remains a question that deserves additional attention and research, because of the basic experimental evidence that pro-coagulant molecules (such as tissue factors, phosphatidylserine, cardiolipin, and vWF) are released from damaged brain tissue [28][29][30]. In this study, there were no de nite ratios between crystal infusion and colloid infusion, and there was no signi cant difference between the two groups (data not shown).
Interestingly, intraoperative infusion of colloidal solution > 1100 mL or intraoperative infusion of crystalloid solution > 2900 mL were identi ed as independent risk factors in the multivariate logistic regression analysis. Previous studies have shown that large amounts of colloid and can affect body coagulation functions [31][32][33], which was also con rmed in this study. Furthermore, we found that when the intraoperative input of colloidal uid was more than 1100 mL, the impact on postoperative coagulation was signi cant. This threshold was derived from the ROC curve. However, this conclusion was drawn from our retrospective study. Therefore, it is necessary to carry out prospective clinical trials to study the infusion of colloids in patients with craniocerebral trauma.
There are currently no clear guidelines for uid management during craniocerebral trauma surgery, although there are some other guidelines for surgical uid management [31,34,35]. How to maintain the balance between wet and dry conditions is a question we all need to consider, especially in TBI, because we do not know whether the high perfusion of uid during surgery will cause secondary damage to the brain. Just as our research provides evidence for the management of uids during surgery, more evidence-based prospective experiments are needed to explore this issue. We need a model or algorithm to calculate the most appropriate uid intake for patients with craniocerebral trauma that can maintain cerebral and systemic perfusion as well as avoid secondary injury to the brain caused by elevated cranial pressure owing to high perfusion. This appropriate uid intake will maximize the bene t of patients.
There were several limitations in our study. First, this was a retrospective cohort study. Retrospective articles are not a substitute for prospective randomized controlled trials. Second, other data that we did not recorded may have affected our analysis. For example, information about prehospital treatment could not be recorded, as this was a retrospective article. Third, only APTT, INR, and platelet counts were recorded and regarded as crucial factors to de ne coagulopathy, and other coagulation-related variables were not enrolled. However, most of the current research on TBI-induced coagulopathy includes these three factors; thus, it is reasonable to some extent. This study focused on the relationship between immediate postoperative coagulopathy and the long-term survival of TBI patients, and the role of late postoperative coagulopathy is worth of further exploration.

Conclusion
Those who suffered from immediate postoperative coagulopathy due to TBI had a higher risk for poor prognosis than those who did not. Critical care should be given to immediate postoperative coagulopathy, and more emphasis should be placed on managing the underlying risk factors.

Declarations
The datasets used and analyzed in the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate
This study was approved by the institutional ethics board of Tangdu Hospital (No. 201909-13) and was registered with ClinicalTrials.gov (NCT04322721), and all procedures followed were in accordance with the ethical standards. The informed consent was obtained from all participants.

Consent for publication
Not applicable.

Declaration of con icting interests
All authors declare that they have no con ict of interest.