Introduction

Patients with acute subdural hematoma (ASDH) are recommended to undergo surgical evacuation in case of a thickness of hematoma greater than 10 mm or a midline shift greater than 5 mm on a computed tomography (CT) scan, regardless of their Glasgow Coma Scale (GCS) score [2]. However, the initial surgical indication may be actually based on the patients’ GCS score and pupillary examination, as well as CT findings [7]. Therefore, patients with a large hematoma or focal neurological deficit can be treated conservatively on admission. Some patients with ASDH who are initially treated conservatively require subacute surgery [1, 4,5,6, 8, 9, 12, 16, 18]. However, little is known about the characteristics or factors related to the requirement for surgery due to deterioration of patients’ clinical and/or CT findings in the subacute phase after conservative treatment. Identifying the factors related to subacute worsening of patients who are treated conservatively could be helpful for managing these patients more appropriately.

In this study, we evaluated the clinical results and determined the predictive factors for the need for subacute surgery in patients with ASDH treated conservatively on admission.

Methods and materials

Patients and data collection

This study was approved by the Institutional Review Board of Saitama Medical University International Medical Center (IRB No. 18-304). From May 2007 to June 2018, 568 patients diagnosed with ASDH were admitted to our hospital. Patients with disturbance of consciousness and anisocoria at admission had acute surgery performed, while the other patients were initially treated conservatively. All patients were treated with maximum effort and no treatment was withheld during their hospitalization. Among these patients, we excluded 146 patients who underwent acute surgery, 23 patients under 16 years of age, 93 patients who had a hematoma only around the falx cerebri or tentorium cerebelli, 36 patients with insufficient data, and 70 patients who died just after admission. After exclusion, we retrospectively reviewed the medical records of the remaining 200 patients initially treated conservatively. We defined the periods from onset as follows: acute phase, ≤ 3 days; subacute phase, 4–20 days; and chronic phase, ≥ 21 days, based on McKissock’s classification [4,5,6, 9, 18]. Next, we divided patients into two groups according to their clinical course. A total of 17 (8.5%) patients underwent surgery because of deterioration of their clinical and/or CT findings in the subacute phase (subacute surgery group) and 183 (91.5%) did not undergo surgery (nonsubacute surgery group). Among the patients in the nonsubacute group, 178 (89%) patients did not undergo surgery during their hospitalization, while five (2.5%) underwent burr hole surgery or minimal craniotomy to evacuate a chronic hematoma.

Using patients’ medical records, we evaluated the following: patients’ age; sex; mechanism of trauma; medical history, including antithrombotic therapy; GCS score on admission; platelet count; prothrombin time–international normalized ratio; activated partial thromboplastin time; mean blood pressure; presence of headache or focal neurological deficit; and modified Rankin Scale (mRS) score on discharge. We divided patients’ initial level of consciousness into three groups: GCS 14–15, 9–13, and 3–8, and patients’ history of antithrombotic therapy into five types: single antiplatelet therapy, dual antiplatelet therapy, oral anticoagulation therapy, single antiplatelet therapy with oral anticoagulation therapy, and no antithrombotic therapy.

Radiological evaluation

From initial brain CT, we collected the following radiological data: degree of midline shift, hematoma thickness, hematoma volume, hematoma density, brain atrophy, and the presence of combined lesions. Hematoma volume was calculated as length (cm) × width (cm) × depth (cm)/2. We defined homogenous density as uniformly high density compared with cerebral gray matter and mixed density as a mixture of high and low density compared with cerebral gray matter. To assess brain atrophy, we measured frontal horn index (FHI), cella media index (CMI), and Sylvian fissure ratio (SFR). FHI is the ratio of the maximum distance between the frontal horns and the inner table diameter on the same line. CMI is the ratio of the biparietal diameter of the skull to the maximum external diameter of the central part of the lateral ventricles. SFR is the average of the maximum width of the two Sylvian fissures or of the Sylvian fissure on the unaffected side, divided by the transpineal inner table diameter.

Statistical analysis

We used Pearson’s chi-square test, the Mann–Whitney U test, the unpaired Student’s t test, Wilcoxon’s signed-rank test, and the bivariate correlation test for statistical analysis. A p value < 0.05 was considered statistically significant. All statistical analyses were performed using SPSS (version 24.0; IBM Corp., Armonk, NY, USA).

Results

In the subacute surgery group, the mean (± standard deviation) duration from onset to surgery was 12.7 ± 3.5 days. Among the 17 patients in the subacute surgery group, eight patients underwent craniotomy, and nine patients underwent burr hole surgery. In this group, 12 patients had a GCS score of 14–15 at admission and their level of consciousness and/or focal neurological deficit deteriorated subacutely. Furthermore, disturbance of consciousness persisted in three patients after admission because of seizures, alcohol abuse, or high fever. One patient with disturbance of consciousness at admission recovered temporarily after treatment for seizures, but his level of consciousness deteriorated again in the subacute phase. The other patient was admitted at 9 days after onset with deterioration of consciousness. CT revealed that hematoma density decreased in 13 patients and was unchanged in four patients before surgery. No patients developed new bleeding after admission, but all patients had significantly larger midline shift (p = 0.002), hematoma thickness (p = 0.001), and hematoma volume (p = 0.015) on CT immediately before surgery. The ratio of midline shift to hematoma volume also increased significantly compared with the initial CT findings (p = 0.007). Four patients recovered well after surgery and were discharged, while 13 patients were transferred to other hospitals for intensive rehabilitation care.

Table 1 shows a comparison of patients’ characteristics and clinical findings between patients undergoing and not undergoing subacute surgery. No significant difference was found for age (p = 0.718), sex (p = 0.354), mechanism of trauma (p = 0.302), history of antithrombotic therapy (p = 0.130), history of hypertension (p = 0.254), diabetes mellitus (p = 0.336), ischemic heart disease (p = 0.527), renal dysfunction (p = 0.233), stroke (p = 0.346), liver dysfunction (p = 0.141), alcohol abuse (p = 0.141), platelet count (p = 0.123), prothrombin time–international normalized ratio (p = 0.214), activated partial thromboplastin time (p = 0.107), mean blood pressure (p = 0.909), and the presence of headache (p = 0.432). We also found no significant difference in the level of consciousness on admission between the two surgical groups (p = 0.344). However, the subacute surgery group had a significantly higher ratio of patients with focal neurological deficits at admission (p = 0.004) and unfavorable mRS scores at discharge (p = 0.001) than did the nonsubacute surgery group.

Table 1 Patients’ characteristics and clinical findings
Full size table

Table 2 shows patients’ initial CT findings on admission. Patients in the subacute surgery group had significantly greater midline shift (p < 0.001), thicker hematomas (p < 0.001), and larger hematoma volume (p < 0.001) compared with those in the nonsubacute surgery group. Regarding brain atrophy, no significant difference was found for FHI (p = 0.235), while CMI was significantly smaller (p < 0.001), and SFR was significantly larger (p = 0.019) in the subacute surgery group compared with the nonsubacute surgery group. Hematomas with mixed density occurred significantly more frequently in patients in the subacute surgery group (p < 0.001). Conversely, no significant difference was found for combined lesions, brain contusion (p = 0.544), subarachnoid hemorrhage (p = 0.133), fracture (p = 0.547), pneumocephalus (p = 0.532), intraventricular hemorrhage (p = 0.558), and epidural hematoma (p = 0.485).

Table 2 Patients’ initial computed tomographic findings on admission
Full size table

Illustrative case

A 70-year-old man was transferred to our hospital with headache, slight consciousness disturbance, and elevated blood pressure. Initial brain CT showed ASDH with mixed density and slight midline shift (Fig. 1a). He was treated conservatively because his GCS score was 15, and symptoms had improved. However, 9 days after admission, his consciousness had gradually deteriorated to a GCS score of 11 over 2 days. Ten days after admission, brain CT showed brain swelling and worsening of the midline shift, although the hematoma thickness was not remarkably unchanged; hematoma density had decreased, but the hematoma retained a mixed feature (Fig. 1b). He underwent immediate craniotomy to remove the hematoma, and his consciousness improved soon after surgery. His postoperative course was uneventful and he recovered well. Fourteen days after surgery, he was transferred to another hospital to continue rehabilitation.

Fig. 1
figure 1

Computed tomography (CT) on admission showing a mixed-density acute subdural hematoma and slight midline shift (a). Ten days after admission, CT showed brain swelling and worsening of the midline shift; the density of the hematoma had decreased, but the hematoma retained the mixed-density feature (b)

Full size image

Discussion

It is well known that some patients with ASDH treated conservatively require surgery because of subacute clinical deterioration [1, 4,5,6, 8, 9, 12, 16, 18]. These patients show temporary acute resolution of the hematoma followed by hematoma enlargement and/or brain swelling associated with decreased hematoma density. Brain swelling tends to be severe regardless of the degree of hematoma enlargement, but no quantitative examinations were described in previous reports. Our data showed that patients in the subacute surgery group had a significantly increased ratio of midline shift to hematoma volume before surgery compared with initial CT measurements.

The present study showed that large hematoma, brain atrophy, and hematoma density were useful predictors of subacute surgery in patients with ASDH treated conservatively. Patients in our subacute surgery group had significantly greater midline shift, thicker hematomas, and larger hematoma volume than those in the nonsubacute surgery group, and these findings are consistent with the results from previous reports [1, 12, 16, 18]. Akamatsu et al. reported that patients requiring subacute surgery had significantly greater midline shift, thicker hematomas, and larger hematoma volume than those receiving conservative treatment [1]. Mathew et al. reported that patients requiring surgery for enlarging hematomas had significantly thicker and larger-volume hematomas than those without enlarging hematomas receiving conservative treatment [12]; similar results were found in other studies [16, 18]. Additionally, focal neurological deficit, which reflects the mass effect of subdural hematoma, has become a predictor of subacute surgery.

Patients who can be treated conservatively on admission despite having a large-volume hematoma appear more likely to have brain atrophy [12]. We measured FHI, CMI, and SFR as indicators of brain atrophy, in this study. CMI was significantly smaller in our subacute surgery group and had a negative correlation with hematoma thickness or volume (correlation coefficient − 0.35 and − 0.38, respectively). These observations might reflect strong compression of the lateral ventricle by the subdural hematoma. In comparison, SFR was significantly larger in the subacute surgery group and had a positive correlation with age (correlation coefficient 0.42), consistent with previous reports showing that patients requiring subacute surgery are likely to have brain atrophy [4, 5, 12]. Son et al. also reported that SFR was larger in patients with enlarging subdural hematomas, subacutely [16]. Patients with brain atrophy rarely show neurological symptoms because they have enough space for the hematoma to enlarge [16]; however, in atrophic brains, bridging veins are easily injured, even with mild trauma [12]. Son et al. reported that a hematoma in the parietal region, where many cortical vessels or dural sinuses are contained in the wide subdural space, was a predictor of poor clinical course in patients with ASDH [16]. These findings suggest that brain atrophy affects subdural hematoma enlargement; therefore, brain atrophy is considered a meaningful predictor of subacute subdural hematoma enlargement.

Mixed-density hematomas were observed more frequently in patients in our subacute surgery group. Hematoma density may be closely related to the mechanisms of subacute subdural hematoma enlargement. Various mechanisms have been proposed to explain subacute hematoma enlargement, namely, transudation or exudation of blood components from the outer membrane to the subdural space [6], rebleeding from injured cortical arteries [6, 9], subacute cortical hyperperfusion secondary to acute hypoperfusion [3], concurrent ASDH and chronic subdural hematoma (“acute-on-chronic subdural hematoma”) [16], and delayed resolution of thick ASDH [16]. Increased cerebrospinal fluid (CSF) in the subdural space secondary to injury of the arachnoid membrane by trauma is also a proposed mechanism [5, 6, 9, 15,16,17,18]. Yamada and Natori stated that the arachnoid membrane was especially difficult to repair in patients with brain atrophy [18]. Izumihara et al. reported that CSF inflow into the subdural space was the main cause of hematoma enlargement [5]; the authors intraoperatively observed a liquid hematoma close to the brain and the absence of an inner membrane [5]. CSF inflow into the subdural space is a possible mechanism producing mixed-density hematomas seen on CT [10, 16]. Proposed mechanisms for mixed-density hematomas include hyperacute hematomas secondary to continuous active bleeding or coagulopathy and acute-on-chronic subdural hematomas [16]. The latter have been found frequently in older patients with brain atrophy [11, 16, 19]. In our study, patients with mixed-density hematomas tended to be older and had brain atrophy with larger SFR (data not shown), suggesting that the presence of a mixed-density hematoma on initial CT is a risk factor. Considering these mechanisms, inflow of CSF or acute-on-chronic subdural hematomas are the probable mechanisms of subacute hematoma enlargement.

In the present study, we found no significant difference for a history of antithrombotic therapy between our two groups. In patients with ASDH, antithrombotic therapy is closely related to hematoma enlargement [1, 4, 5, 13, 16, 18]. Izumihara et al. reported that more than half of patients with ASDH requiring subacute or chronic surgery had a history of antiplatelet therapy [4, 5]. Son et al. also reported that the use of anticoagulants was a significant prognostic factor for hematoma development, although the type of drugs used was not mentioned [16]. Conversely, Yamada and Natori observed no significant difference for history of antithrombotic therapy between patients with and without subacute hematoma enlargement [18]. Antithrombotic therapy may lead to hematoma development with mild symptoms at onset and with microscopic bleeding caused by trauma leading to enlargement of the subdural hematoma [1, 13]. However, little information is available regarding the type of antithrombotic drugs. Regarding anticoagulants, warfarin-related ASDH has been reported [13, 14], but the effect of direct oral anticoagulants remains to be elucidated. In our study, anticoagulant drugs were used in 14 patients (7.7%; warfarin for 11 patients, dabigatran for one patient, and rivaroxaban for two patients) in the nonsubacute surgery group, whereas these drugs were used in two patients (11.8%; both receiving warfarin) in the subacute surgery group. Further studies are needed to reveal how and what type of direct oral anticoagulants affect the clinical course of ASDH.

In the present study, patients with disturbance of consciousness and anisocoria at admission underwent acute surgery to save their lives. Even if patients have a large hematoma or focal neurological deficit at admission, surgery can be withheld or postponed until the chronic phase if possible in some of them without disturbance in consciousness, which applies to the present study. This is the case especially in those who have a risk of general anesthesia, such as old age or systemic complications. In the present study, adopting a more aggressive attitude for acute surgery might have improved the outcome of patients with ASDH because the postoperative course of most patients in the subacute surgery group was unfavorable. However, performing acute surgery for all patients who are compatible with the indication for surgery according to international guidelines would be difficult. Therefore, careful follow-up of clinical findings or CT images is important, especially in patients with adverse prognostic factors.

There are several limitations in our study. First, potential selection bias may have resulted from the single-center retrospective study design. Second, evaluation of patients’ neurological deficits or activities of daily living was not performed by the same observer; therefore, subjective variations between observers were included to some extent in the evaluation. However, the risk may be limited because we used a prospectively collected database in this study. Third, although mRS score is usually used to assess the patients’ outcome at 6–9 months after onset, we adopted it as an outcome scale at discharge because following the clinical courses of all patients after discharge was difficult.

Conclusion

Large hematoma, brain atrophy, and hematoma density are possible useful predictors of the need for subacute surgery in patients with ASDH initially treated conservatively. We recommend careful follow-up of clinical findings or CT images in patients with adverse prognostic factors, even if their initial symptoms are mild.