Age-related differences in the impact of coagulopathy in patients with isolated traumatic brain injury: An observational cohort study

BACKGROUND Although age and coagulopathy are well-known predictors of poor outcome after traumatic brain injury (TBI), the interaction effect of these two predictors remains unclear. OBJECTIVES We assessed age-related differences in the impact of coagulopathy on the outcome following isolated TBI. METHODS We conducted a retrospective observational study in two tertiary emergency critical care medical centers in Japan from 2013 to 2018. A total of 1036 patients with isolated TBI (head Abbreviated Injury Scale ≥ 3 and other Abbreviated Injury Scale < 3) were selected and divided into the nonelderly (n = 501, 16–64 years) and elderly group (n = 535, age ≥65 years). We further evaluated the impact of coagulopathy (international normalized ratio, >1.2) on the outcomes (Glasgow Outcome Scale-Extended [GOS-E] scores, in-hospital mortality, and ventilation-free days) in both groups using univariate and multivariate models. Further, we conducted an age-based assessment of the impact of TBI-associated coagulopathy on GOS-E using a generalized additive model. RESULTS The multivariate model showed a significant association of age and TBI-associated coagulopathy with lower GOS-E scores, in-hospital mortality, and shorter ventilation-free days in the nonelderly group; however, significant impact of coagulopathy was not observed for all the outcomes in the elderly group. There was a decrease in the correlation degree between coagulopathy and GOS-E scores decreased with those older than 65 years. CONCLUSION There was a low impact of coagulopathy on functional and survival outcomes in geriatric patients with isolated TBI. LEVEL OF EVIDENCE Therapeutic study, Level IV.

T raumatic brain injury (TBI) is a major health complication responsible for 282,000 hospitalizations and 56,000 deaths in 2013 in the United States alone. 1 A large Japanese trauma registry study found that 40.2% of severe TBI patients were 60 years or older. 2 Although coagulopathy is caused by the traumatic injury itself and the subsequent blood dilution related to fluid resuscitation, 3 Traumatic brain injury is an independent risk factor for developing coagulopathy characterized by excessive fibrinolysis due to extensive tissue factor (TF) release from the injured brain. It has been previously reported that two-thirds of patients with severe TBI present with coagulation system abnormalities upon arrival at the emergency department (ED). Moreover, they often significantly affect the clinical course of the patients through the progression in intracranial hemorrhage, extracranial hemorrhage, or both. [4][5][6] Although TBI has been reported to be more common among younger people, 7 an increasing proportion of elderly patients with TBI are being admitted to trauma centers in the majority of developed countries. 8 Given the differences in physiologic reserves and functional capacities between elderly and younger patients, age is considered a significant risk factor for trauma death. 9 Therefore, it is important to understand and compare the characteristics of elderly and nonelderly patients with TBI to provide adequate trauma care practice.
Given the age-based differences in the outcomes of trauma victims, we hypothesized that TBI-associated coagulopathy could have different effects on patient outcomes between elderly and nonelderly patients. We aimed to determine age-related differences in the impact of TBI-associated coagulopathy on patient outcomes following isolated TBI.

Study Design and Setting
This observational study aimed to evaluate differences in the impacts of TBI-associated coagulopathy on the patient outcomes between elderly and nonelderly patients with isolated TBI. We retrospectively analyzed patients with isolated TBI admitted between April 1, 2013, and March 31, 2018, to two tertiary emergency care hospitals in Japan (Tokyo Medical and Dental University Hospital of Medicine or Matsudo City General Hospital).
This study complied with the principles of the 1964 Declaration of Helsinki and its later amendments. This study was approved by the institutional review board of Tokyo Medical and Dental University and Matsudo City Hospital (2019-217). The requirement for informed consent from each patient was waived given the retrospective design of the study and its use of anonymized patient and hospital data.

Study Population
We consecutively enrolled patients with isolated blunt severe TBI. The exclusion criteria were as follows: (1) patients younger than 15 years; (2) patients with cardiac arrest upon arrival at the ED; (3) patients who suffered unsurvivable injury (i.e., head Abbreviated Injury Scale [AIS] = 6); (4) patients who were transferred from other hospitals; (5) patients who were transferred to other hospitals within 24 hours after admission; and (6) patients with a history of taking anticoagulants or antiplatelet agents given their significant effect on clotting function independent from that of TBI. Further, we excluded patients with missing or insufficient clinical data for analysis.

Data Collection
We retrospectively collected the following information from patients'  10 ; AIS of the head and neck, face, chest, abdomen, pelvis and extremities, and surface 11 ; Injury Severity Score (ISS); Revised Trauma Score (RTS) calculated using the GCS, SBP, and RR upon arrival at the ED 12 ; functional status at hospital discharge evaluated by GCS-Extended (GOS-E) scores; whether the patient underwent emergency neurosurgical intervention and/or received transfusion within 24 hours of admission. Head injury types were categorized into four classifications based on radiologic findings: acute subdural hematoma, acute epidural hematoma, traumatic subarachnoid hemorrhage, and intracerebral hematoma/contusion. Further, we collected laboratory results including the international normalized ratio. Blood samples were obtained from the patients immediately upon arrival in the ED.

Outcome Measures and Definitions
We defined the primary endpoint as the functional outcome at hospital discharge assessed using the GOS-E scores. The GOS-E is a scoring system that classifies functional outcome into eight stages: (1) dead, (2) vegetative state, (3) low severe disability, (4) upper severe disability, (5) low moderate disability, (6) upper moderate disability, (7) low good recovery, and (8) upper good recovery. 13 We defined the secondary endpoints as the in-hospital mortality and ventilator-free days (VFD). 14 We defined isolated severe TBI as an injury with an AIS head score of 3 or greater and an AIS score less than 3 for other body parts. Based on a previous report that an age older than 65 years is an independent predictor of death in trauma patients, 15 we defined elderly as 65 years or older. We defined coagulopathy as an INR greater than 1.2 based on previous reports. [16][17][18] We defined VFD as previously defined. 14

Statistical Analysis
We divided the enrolled patients into two age categories: specifically, the nonelderly (age = 16-64 years) and elderly group (age, ≥65 years). In univariate analysis, we used Student's t test or Mann-Whitney U test to compare continuous variables and the χ 2 test or Fisher's exact to compare categorical variables as appropriate. Categorical variables were reported as number (percentage), while continuous variables were reported as a median (interquartile range) as appropriate. The GOS-E score was classified as the ordinal variable.
First, we used a multivariable ordered logistic regression model to evaluate the interaction between TBI-associated coagulopathy and age on the GOS-E score with age as a continuous or categorical variable to determine whether age had an effect on the impact of coagulopathy on the GOS-E score. We incorporated age, sex, GCS score, SBP, RR, AIS of the head, ISS, and CCI as variables in the multivariate model, which were a priori selected, and are clinically plausible and well-known confounders in epidemiologic studies on trauma, and based on the number of outcomes (the 10 events per variable rule). Second, we used the univariate and multivariate ordered logistic regression models to evaluate the impact of TBI-associated coagulopathy on the outcomes in the nonelderly and elderly group. Finally, we used a generalized additive model to visualize the impact of coagulopathy on GOS-E scores according to age, on the assumption of the nonlinear change according to aging. We incorporated the aforementioned variables into the multivariate model.
We performed all statistical analyses using the R software (version 3.5.1; R Foundation for Statistical Computing, Vienna, Austria). Moreover, we used a command to add statistical functions frequently used in biostatistics. All reported p values were two-sided, and p values less than 0.05 were considered to be statistically significant. Figure 1 shows the patient selection diagram. Among 1,370 potentially eligible patients, we analyzed 1036 patients; among them, 501 (48.4%) and 535 (51.6%) patients were in the nonelderly and the elderly group, respectively. Table 1 presents the baseline characteristics. We observed coagulopathy in 506 (48.8%) patients (192 [38.3%], nonelderly group; 314 [58.7%], elderly group). The overall mortality rate was 19.4% (201/1036), while the mortality rate in patients with coagulopathy upon arrival was 28.9% (146/506). Although the elderly group had higher CCI, there were no significant differences between the two groups in the GCS, RR, SBP, RTS, ISS, or AIS scores of each body region, including the head AIS. The Elderly group was more likely to suffer a ground level fall, whereas the nonelderly group was more likely to suffer an assault, traffic accident, or fall from a relatively higher altitude than that of the elderly group. There were no significant differences between the two groups in the type of head injury. Table 2 provides the univariate analysis results for the outcomes between the elderly and nonelderly groups. Compared with the nonelderly group, the elderly group had significantly lower GOS-E scores (median, 8; interquartile range [IQR], 7-8 vs. median, 7; IQR, 5-8; p < 0.001) and shorter VFD (median, 28 days; IQR, 25-28 days vs. median, 28 days; IQR, 20-28 days; p = 0.008). Compared with the nonelderly group, there was a numerical, but not significant, increase in the in-hospital mortality rate in the elderly group (88 [17.6%] in the nonelderly group vs. 113 [21.1%] in the elderly group, p = 0.163).   In the entire study cohort, the p values for the interaction term of age and TBI-associated coagulopathy on the GOS-E score, in-hospital mortality, and VFD were less than 0.001, less than 0.001, and 0.002, respectively, which indicated a significant effect of age on the impact of TBI-associated coagulopathy on the outcomes. All of the p values for interaction term of age category (elderly or nonelderly group) and coagulopathy were less than 0.001, which indicated a significant between-group difference in the impact of TBI-associated coagulopathy on the outcomes. Table 3 presents the multivariate analysis results regarding the impact of coagulopathy on the outcomes in the elderly and nonelderly groups. After adjusting for age, sex, GCS score, SBP, RR, AIS of the head, ISS, and CCI, we found a significant association of TBI-associated coagulopathy with lower GOS-E scores and in-hospital mortality and shorter VFD in the nonelderly group. However, we did not observe a significant impact of TBI-associated coagulopathy on all the outcomes in the elderly group. Figure 2 shows the generalized additive model plots demonstrating the adjusted difference in the impact of coagulopathy on the GOS-E score. The impact of coagulopathy on the GOS-E score was at an approximately constant level until about 65 years old where it continued to decrease with age.

DISCUSSION
In this retrospective observational study, we evaluated age-based differences in the impact of coagulopathy on the functional and survival outcomes in 1,036 patients with isolated TBI. After adjusting for potential confounders, we found a lower impact of TBI-associated coagulopathy on GOS-E scores and in-hospital mortality in the elderly group compared with the nonelderly group. Notably, the impact of TBI-associated coagulopathy on GOS-E scores decreased steadily with increasing age after the age of 65 years.
The presence of coagulopathy upon ED admission has been reported to be a strong predictor of patient outcomes and overall TBI prognosis. 5,19 Specifically, compared with patients without coagulopathy, patients with coagulopathy have a 9 and 30 times higher risk of mortality and unfavorable functional outcome, respectively. 3, 18 Brohi et al. 20 postulated that tissue injury alone did not result in trauma-induced coagulopathy unless coupled with tissue hypoperfusion. They suggested that coagulopathy was led by the hypoperfusion-induced increase of thrombomodulin levels, thus converting thrombin from fibrin generation to the activation of protein C. However, the pathogenesis of the coagulopathy post-TBI has been a subject of debate; it has been suggested that TBI was likely to induce coagulopathy through independent mechanisms in the absence of tissue hypoperfusion, shock, or hemodilution. 21,22 Since the underlying mechanism of coagulopathy causation could not elucidated our clinical data, further research is warranted to reveal the differences in the mechanism between TBI-associated coagulopathy and trauma-induced coagulopathy.
The course of TBI-associated coagulopathy and the subsequent increase in intracranial hemorrhage has been reported to reflect the rapid progression from hypercoagulable and hyperfibrinolysis states to a hypocoagulable state within a few hours after head injury until the clotting factors are consumed. 23,24 The mechanism underlying the initial post-TBI hypercoagulable and hyperfibrinolysis state has been reported to  involve extensive TF release by the damaged brain into circulation, which results in coagulation pathway activation and the subsequent consumption of coagulation factors. 3,25 Further, alternative proposed mechanisms include the release of endogenous tissue-type plasminogen activator and urokinase-type plasminogen activator from contusional brain tissue 26 or the depletion of alpha-2-plasmin inhibitor. 27 Plasmin, which is the major effector for fibrin clot lysis, is converted from plasminogen by both tissue-type plasminogen activator and urokinase-type plasminogen activator. Although the aforementioned mechanisms could render patients with TBI prone to have bleeding tendencies, they are probably an oversimplification of a much more complex series of events that occur either simultaneously or sequentially after TBI.
Although there was no between-group difference in the ISS and RTS scores, there was a higher proportion of patients with coagulopathy upon arrival and lower impact of coagulopathy on outcomes in the elderly group compared with the nonelderly group. We excluded patients with a history of taking anti-coagulants or anti-platelet agents; however, the various physiologic derangements and coexisting diseases in elderly patients might affect their coagulation function. Moreover, they are various modifications of the blood coagulation system that accompany the aging process. The plasma levels of some factors (such as fibrinogen, factor VII, and factor VIII) have been reported to increase with aging. 28,29 Further, increased TF expression and availability, which could augment factor VII, have been reported in elderly patients. 30 These age-related changes could induce the hypercoagulable state and secondary hyperfibrinolysis. 28,29 Therefore, the risk for coagulopathy, which was determined using the aforementioned markers, could be lower in elderly patients. However, since these procoagulant proteins are normally present in plasma in large excessive amount, the pathophysiologic significance of the relatively small aging-related changes remains unclear. Although quantitative evaluation was not possible, age-related differences associated with the ability to participate in physical therapy or the rehabilitation might have also affected the functional outcome at hospital discharge.
Our findings could potentially influence the treatment strategy in elderly patients with TBI presenting with coagulopathy. Early intervention for coagulopathy in patients with TBI has been reported to be independently associated with favorable survival and neurological outcomes. 31 Although the administration of adequate blood product amounts is the most common intervention for TIC, there have been reports warning against the excessive use of blood products given the potential risk of adverse events caused by blood transfusion. 32 Larger transfusion amounts have been reported to be associated with mortality in trauma patients even after adjusting for injury severity 33 ; further, this effect might be even more pronounced in geriatric patients. Our findings indicate a potential risk of employing a blood transfusion strategy based on coagulation function in geriatric TBI patients. Specifically, the disadvantage of blood transfusion might outweigh its benefit and the limited effect of coagulopathy on outcomes among elderly patients might mislead clinicians.
Several limitations should be considered in the interpretations of our findings. First, since this was a two-center retrospective observational cohort study, there are limited generalizability and inevitable limitations regarding residual confounding including patients' detailed time course before the ED arrival. Second, there is a considerable risk of type II error given the limited sample size. Third, more than half of our patients suffered multifocal intracranial hemorrhage, which impeded us from performing sub-group analysis based on the brain injury pattern. Finally, specific devices for monitoring blood clotting function, such as thromboelastography (functional fibrinogen by TEG; Haemonetics Corporation, Braintree, MA) or thromboelastometry (ROTEM, FIBTEM, TEM International GmbH, Munich, Germany) [34][35][36] were not routinely used in the study-participating hospitals. Since we defined the coagulopathy as INR greater than 1.2 alone, any derangement in coagulopathic parameters other than INR may affect outcomes post-TBI, there was a potential concern in underestimating the risk of morbidity and mortality. However, since it has been shown in existing literature that coagulation profile-evaluated coagulopathy worsens the TBI outcome by increasing intracranial hematomas and cerebral contusions that exacerbate secondary brain injuries, [37][38][39] TBI-associated coagulopathy measured by INR may possibly be a major predictor in patients with TBI. Nonetheless, to the best of our knowledge, this is the first study to evaluate age-related differences in the impact of coagulopathy on the outcomes. Our findings indicate the need for a different age-based interpretation regarding the presence of TBI-associated coagulopathy in patients with TBI. There is a need for further research to confirm our results and develop an age-related therapeutic strategy.

CONCLUSION
Among the patients with isolated TBI, there was a lower impact of coagulopathy on the functional and survival outcomes at discharge in geriatric patients than that in younger patients. There is a need for further research to investigate the mechanisms underlying this difference and to develop an optimal therapeutic strategy for geriatric patients with TBI. AUTHORSHIP W.T. participated in the study design, data collection, drafting of the article, and the statistical analysis. A.E. and H.K. participated in the statistical analysis and helped draft the article. K.M. and Y.O. participated in the study conception and design, data collection, and drafting of the article. All the authors have read the article, discussed the results, and approved this submission.