Between January 2018 and September 2021, 254 patients with symptomatic CSDH were treated surgically at a single tertiary institute, in whom CSDH was diagnosed using initial brain computed tomography (CT) or magnetic resonance imaging scans. Patients whose postoperative hematoma changes were not traced because of the follow-up loss or emergent surgical or systemic complications were excluded. The data extracted from the medical records comprised demographic features, clinical presentation, preoperative Markwalder’s grading score (MGS) and Glasgow coma scale (GCS), history of the cardio-cerebrovascular disease (CCVD) with antithrombotics (AT) medications (e.g., anti-platelets and anti-coagulants), surgical and systemic complications, postoperative administration of TXA and/or statins, and adverse events during adjunctive medical treatment.

In all patients, the surgery was a single burr-hole trephination with the placement of a subdural catheter for hematoma drainage. No craniotomy or twist drill surgery was performed. Although detailed surgical techniques varied among neurosurgeons, five neurosurgeons performed surgery following our institutional guideline : to place a subdural catheter immediately after the puncture of the pseudo-membrane of the CSDH through a single burr-hole site without routine saline irrigation. Typical CSDH patients underwent three postoperative CT scans : immediately after surgery, 24 to 48 hours after surgery before removing the subdural drainage catheter, and 1 week after surgery to confirm hospital discharge. None of the patients was prescribed steroids during the treatment period. Follow-up CT scans were obtained 2–3 weeks after the discharge and 1-month interval thereafter.

No additional adjunctive medical therapy was performed during the early study period. However, during the late study period, administration of oral TXA was considered only in the patients without a history of CCVD to avoid unexpected thrombotic adverse events. Since using oral TXA adjuvant for the first time in March 2020, 88 patients were enrolled, of which 33 patients were not subjected to the oral TXA due to a history of CCVD. Of the remaining patients, 75% (41 of 55 patients) were treated with oral TXA as an adjunctive medication. Patients who refused adjunctive medications were not treated with TXA. Therefore, among patients with no history of CCVD during the late period of the study, majority of patients received oral TXA 750 mg daily until the primary or secondary endpoint (TXA group), and the remaining patients were observed without TXA medication (non-TXA group). Whereas patients with a CCVD history and AT medication were grouped into the AT group and analyzed for comparison with the other two groups. All AT medications were discontinued after surgery and were not resumed until the hematoma was resolved. Patient selection and grouping are summarized in Fig. 1.

The primary endpoint of this study was the recurrence of CSDH, defined as the re-appearance or re-accumulation of CSDH requiring neurosurgical interventions, including repeat surgery, middle meningeal artery embolization, and intensive medical treatments. The decision regarding the clinical situation was made by experienced neurosurgeons.

The secondary endpoint was the resolution of CSDH. It was defined as complete or near-complete resorption of the CSDH with the recovery of neurological deficits. Radiologically, complete resorption was considered when the thickness of the CSDH in the follow-up CT scan was less than 5 mm. The near-complete resorption of CSDH was determined when CSDH was stationary in subsequent examinations, even if the hematoma thickness was close to 5 mm or more. In addition, the time to resolution, defined as the period between surgery and resolution of CSDH, was used in the Kaplan-Meier method. The serial thickness of the CSDH was also investigated to demonstrate changes in the CSDH over time.

The data from initial and follow-up images were extracted including hematoma location (unilateral or bilateral), classification of CSDH (homogeneous, laminated, separated and trabecular), midline shift, and maximal thickness of CSDH [16]. To minimize measurement error, two neurosurgeons examined the thickness of CSDH according to the following agreement : 1) the maximal thickness was measured on a perpendicular line to the brain cortex (frontal or parietal lobe) from the inner table of the skull in axial, coronal and sagittal images; 2) sequential measurements of thickness were performed on the same axis of initial measurement as possible; and 3) the site of pneumocephalus was avoided to be measured. The paired values of the CSDH thickness, assessed independently by two examiners, were statistically evaluated.

Statistical analyses were performed using R software (version 4.1.2; R Project for Statistical Computing, Vienna, Austria) and SPSS software (version 18.0.0; SPSS Inc., Armonk, NY, USA) with a 2-sided p-value of 0.05 as the significance threshold in all tests. Categorical variables were presented as the number of patients with percentages. Normally distributed continuous variables were reported as mean with standard deviations, and non-parametric variables as median with interquartile ranges (IQRs). Three-group comparisons were conducted using Fisher’s exact test, one-way ANOVA, and the Kruskal-Wallis test. Tukey’s test was adopted as a post-hoc test comparing all possible pairs of mean values of groups to determine which specific pairs of groups were significantly different. The logistic regression analysis was used for determining risk factors for the primary outcome. The Kaplan-Meier method was used for survival analyses, and the log-rank test to compare treatment was conducted. For the primary and secondary endpoints, data censoring occurred when patients died of unrelated problems of study, and patients remained free of the entity being studied at the end of the follow-up period. The Cox proportional hazards model was used to perform the univariate and multivariate analyses. The measurement error of CSDH thickness was evaluated using the intraclass correlation coefficient (ICC). The CSDH thickness values were presented with line graphs as median values with IQR at each time points series (initial, 1 week, and 1, 2, 3, 4, and 6 months).

CSDH recurred in 16 patients (6.7%). Among these, 12 patients underwent repeat surgery, three underwent middle meningeal artery embolization, and one was managed conservatively. Multiple interventions were required in three patients. The recurrence rates were 2.4% in the TXA group (n=1), 7.0% in the non-TXA group (n=8), and 8.2% in the AT group (n=7). The recurrence rate was the lowest in the TXA group, but the difference was not statistically significant (Fig. 2A). Recurrence-free survival was similar among the three groups (Fig. 2B). The univariate and multivariate analyses revealed that patient demographics including age, sex, medical history, preoperative MGS, bilateral hematoma and surgery, preoperative hematoma thickness, and the use of TXA and/or statins were not related to recurrence (Table 2). However, the homogeneous type of CSDH was a negative risk factor for recurrence in the multivariate analysis (p=0.015; hazard ratio [HR], 0.259; 95% confidence interval [CI], 0.087–0.766). In patients with recurrence, the trabecular type (n=7) was the most prevalent hematoma type, followed by the homogeneous (n=5), laminar (n=2), and separated (n=2) types.

Resolution of CSDH was achieved in 206 patients (85.8%), and the crude rates were not statistically different among TXA (39 of 41 patients, 95.1%), non-TXA (97 of 114 patients, 85.1%), and AT (70 of 85 patients, 82.4%) groups (Fig. 2A). However, the median estimated time to resolution was significantly faster in the TXA group (51 days; 95% CI, 45.0–57.0) than the non-TXA (109 days; 95% CI, 102.3–115.7) and AT (88 days; 95% CI, 66.7–109.4) groups (log-rank test, p<0.001; Fig. 3A). In the multivariate analysis, TXA administration was a significant positive factor for achieving CSDH resolution (p<0.001; HR, 3.323; 95% CI, 2.231–4.948; Table 3). In addition, less residual hematoma at discharge (i.e., maximal thickness of CSDH at one week after surgery) was also a significant prognostic factor for CSDH resolution (p=0.004; HR, 0.943; 95% CI, 0.906–0.981).

The maximal thickness values of CSDH over the follow-up period were measured in 1004 CT scans, and the ICC was 0.843 (p=0.03; 95% CI, 0.452–0.933). CT scans at initial presentation and 1 week (a total of 480 CT scans) after surgery were available for all patients, and the number of reviewed CT scans decreased over time because patients reached primary or secondary endpoints or were lost to follow-up. In the TXA group, only two CT scans were available at 4 months and none at 6 months after surgery. The thickness was comparable among the three groups at the initial time and 1 week after surgery. However, 1 month after surgery, CSDH thickness in the TXA group decreased abruptly and showed a significant difference from that in the other groups (p<0.001, Fig. 3B). The difference in the thickness between the TXA group and other groups was significant until the 4 months when almost all patients in the TXA group achieved CSDH resolution. Meanwhile, the non-TXA and AT groups showed almost same patterns of CSDH resolution.

Our study was retrospective and limited by a small sample size and biases of a single-institution study. There was bias in patient selection, resulting in an imbalance in potential confounding variables including age, previous medication, and bilateral surgery, although these factors were not related to the primary and secondary outcomes in the multivariate analyses. In addition, possible confounding factors such as surgical methods, irrigation during surgery, open or close drainage, drug interaction and other demographic and radiological features that may affect outcomes should be considered. To validate and generalize the study results, further studies should be performed including multicenter studies and randomized trials as well as physiological experiments. Furthermore, the TXA group comprised patients without CCVD history. These patients were selected because of concerns about the possibility of TXA-associated thromboembolism, although several studies have reported the safety of TXA [3,17,18]. A subsequent trial is planned at our institution to validate the effectiveness and safety of adjunctive TXA regardless of CCVD history. Another weak point of this study was that the TXA administration was decided by the neurosurgeon’s preference rather than by definite selection criteria. In addition, statins, another potentially beneficial drug in reducing CSDH recurrence, was administered more frequently to the patients in the TXA group [4,7,21,23,24,28]. TXA administration was conducted relatively recently in this cohort, and a more detailed investigation of drug history for the TXA group may be the reason for the imbalance. Although multivariate analysis revealed no relationship between statin administration and study outcomes, combined effects of TXA and statins need to be investigated. Therefore, caution should be exercised when interpreting the results of this study.