Indications for Surgery

• An acute subdural hematoma (SDH) with a thickness greater than 10 mm or a midline shift greater than 5 mm on computed tomographic (CT) scan should be surgically evacuated, regardless of the patient's Glasgow Coma Scale (GCS) score.
• All patients with acute SDH in coma (GCS score less than 9) should undergo intracranial pressure (ICP) monitoring.
• A comatose patient (GCS score less than 9) with an SDH less than 10-mm thick and a midline shift less than 5mmshould undergo surgical evacuation of the lesion if the GCS score decreased between the time of injury and hospital admission by 2 or more points on the GCS and/or the patient presents with asymmetric or fixed and dilated pupils and/or the ICP exceeds 20 mm Hg.

Timing

• In patients with acute SDH and indications for surgery, surgical evacuation should be performed as soon as possible.

Methods

• If surgical evacuation of an acute SDH in a comatose patient (GCS < 9) is indicated, it should be performed using a craniotomy with or without bone flap removal and duraplasty.

SDH represents one type of intracranial mass lesion, and surgical management attempts to define the subset of patients who would benefit from surgical evacuation of an acute SDH. SDH are diagnosed on a CT scan as extracranial, hyperdense, crescentic collections between the dura and the brain parenchyma. They can be divided into acute and chronic lesions. Herein, "acute SDH" is defined as an SDH diagnosed within 14 days of traumatic brain injury (TBI).

Incidence

Studies conducted after the introduction of CT scanning report an incidence of acute SDH between 12 and 29% in patients admitted with severe TBI. Combining several publications, acute SDH was diagnosed in 21% of 2870 patients8, 26, 27, 36. When including mild, moderate, and severe head injuries, 11% (360 of 3397 patients) present with SDH21, 28. The mean age for this combined group is between 31 and 47 years, with the vast majority of patients being men6, 11, 17, 21.

Mechanism

The mechanism of injury responsible for the development of an SDH differs between age groups. Most SDH are caused by motor vehicle-related accidents (MVA), falls, and assaults. In one study, 56% of SDH in the younger group (18-40 yr) were caused by MVA and only 12% were caused by falls, whereas, in the older groups (>65 yr), these mechanisms were responsible for 22% and 56% of SDH, respectively15. Falls have been identified as the main cause of traumatic SDH in two studies looking specifically at patients older than 75 and 80 years3, 16. Studies with comatose patients describe MVA as the mechanism of injury in 53 to 75% of SDH. This indicates that MVA causes more severe injury, possibly because of high-velocity accidents and diffuse axonal injury18, 26, 36.

Clinical Presentation

Between 37 and 80% of patients with acute SDH present with initial GCS scores of 8 or less4, 6, 21, 28, 33. A lucid interval has been described in 12 to 38% of patients before admission but there is no conclusive evidence that this correlates with outcome1, 9, 31, 34, 36. The definition of lucid interval is vague. Authors interpret the lucid interval differently and analysis of its frequency requires documentation during the prehospital phase. Pupillary abnormalities are observed in 30 to 50% of patients on admission or before surgery6, 10, 28, 33.

Mortality

Studies looking at patients from all age groups with GCS scores between 3 and 15 with SDH requiring surgery quote mortality rates between 40 and 60%10, 11, 13, 14, 17, 27, 39. Mortality among patients presenting to the hospital in coma with subsequent surgical evacuation is between 57 and 68%7, 12, 18, 20, 26, 36.

Associated Injures

Only 30 to 40% of SDH requiring surgery are isolated lesions21, 28. In the majority of cases, the SDH is associated with other intracranial and extracranial injuries. Contusions and intracerebral hematomas are the most frequently associated intracranial abnormalities. Associated intracranial and extracranial lesions have been reported in larger series to occur in 47 to 57% of patients presenting with GCS scores between 3 and 154, 11, 15, 28 and in 65 to 82% of patients with GCS scores less than 1018, 26. In patients with SDH, contusions and fractures are frequent associated injuries16, 19, 25, 31, 36, 37. An associated subarachnoid hemorrhage has been observed in 14 to 25% of patients with SDH4, 28 and epidural hematomas are observed in 6 to 14% of patients4, 28. Significant extracranial injuries are observed in 18 to 51% of patients and the majority of these cases include facial fractures, extremity fractures, and thoracic and abdominal trauma5, 6, 11, 28, 31. Because of the frequent association of SDH with parenchymal injury, surgical management decisions should take into consideration the recommendations for both lesion types.

A MEDLINE computer search using the following keywords for the years 1975 to 2001 was performed: "traumatic brain injury" or "head injury" and "subdural" or "intradural" and "hematoma" or "hemorrhage." The search was narrowed by including the keywords "surgical treatment" or "surgery" or "operation" or "craniotomy" or "craniectomy" or "craniostomy" or "burr holes" and excluding "chronic" and "spinal." These searches combined yielded 161 articles. The reference lists of these publications were reviewed and an additional 18 articles were selected for analysis. Case reports, publications in books, and publications regarding penetrating brain injuries, or spinal or chronic SDH were not included. Chronic SDH was defined as an SDH occurring or diagnosed more than 14 days after trauma. Articles were excluded if the diagnosis of SDH was not based on CT scanning, or if subgroups of patients who did not undergo CT scanning were not clearly identified. Publications with fewer than 10 patients or publications that did not include information on outcome were excluded. Of these 179 articles, 21 were selected for analysis.

Indication for Surgery

The decision to operate on an SDH is based on the patient's GCS score, pupillary exam, comorbidities, CT findings, age, and, in delayed decisions, ICP. Neurological deterioration over time is also an important factor influencing the decision to operate. Trauma patients presenting to the emergency room with altered mental status, pupillary asymmetry, and abnormal flexion or extension are at high risk for either an SDH and/or an epidural hematoma compressing the brain and brainstem.

CT Parameters

Many investigators have tried to define a relationship between CT parameters, such as hematoma volume, clot thickness, midline shift (MLS), and patency of the basal cisterns, and outcome. Two studies using multivariate analysis to identify factors affecting outcome from SDH found contradictory results. Howard et al.15 reported on 67 patients, with GCS scores between 3 and 15, who were undergoing surgery, and found a significant correlation between poor outcome and the volume of the SDH and the MLS. The volume of the SDH, the MLS, and mortality were significantly greater in older patients. van den Brink et al.33 found no difference in hematoma volumes, MLS or status of the basal cisterns when comparing surgical patients, who had a GCS of 3 to 15, and favorable versus unfavorable outcome. Zumkeller et al.39 investigated CT scan parameters in 174 patients with SDH and a GCS between 3 and 15 undergoing surgery. The findings revealed a 10% mortality rate in patients with a clot thickness of less than 10 mm, and a 90% mortality for patients with clots thicker than 30 mm. For an MLS greater than 20 mm, there was a steep increase in mortality. Both parameters correlated well with the Glasgow outcome score (GOS). In a mixed group of patients treated with or without surgery, Servadei et al.28 also found a correlation between outcome and clot thickness, MLS, and status of the basilar cisterns. Kotwica and Brzezinski18 found a significant relationship between MLS and outcome in 200 patients with GCS scores lower than 10, who were undergoing surgery for SDH. In summary, there seems to be a relationship between CT parameters and outcome, but it is difficult to determine specific threshold values.

Surgical Versus Nonoperative Treatment of SDH

The decision for nonoperative versus surgical management of SDH is influenced by the GCS score; CT parameters, such as MLS, SDH clot thickness and volume, and patency of the basal cisterns; and the salvageability of the patient (i.e., whether the primary injury is so extensive that evacuation of the SDH will not make a difference in outcome). On the basis of the reviewed literature, a clot thickness greater than 10 mm or a MLS greater than 5 mm are suggested as critical parameters for surgical evacuation of an acute SDH, regardless of the GCS.

Wong37 tried to identify parameters that would predict the failure of initial nonoperative management. No treatment protocol was defined. Six of 31 patients with GCS scores between 6 and 15 who were initially treated without surgery required a later craniotomy because of neurological deterioration (performed within 3 d). The authors found that an MLS greater than 5 mm in patients with a GCS score of lower than 15 on the initial CT scan was significantly related to the failure of nonoperative treatment. Hematoma volume and thickness of the hematoma were not predictive. Good outcome was achieved in all patients.

Matthew et al.22 reviewed the data on 23 patients with GCS scores between 13 and 15 who were initially treated nonoperatively. No criteria were defined for nonoperative management. All patients had an isolated SDH and all were observed in the neurosurgical intensive care unit. Six patients required delayed (mean, 14 d) evacuation of their SDH. Significant differences in clot thickness and hematoma volume were found between the operative and the nonoperative groups. In addition, all patients with an initial hematoma thickness greater than 10 mm required surgery. Finally, Servadei et al.27 developed a protocol to select comatose patients with SDH for nonoperative management. The criteria used to select comatose patients for nonoperative treatment were clinical stability or improvement during the time from injury to evaluation at the hospital, hematoma thickness less than 10 mm and MLS less than 5 mm on the initial CT scan, and ICP monitoring in the neurosurgical intensive care unit. Surgery was performed if the ICP exceeded 20 mm Hg. Fifteen of 65 comatose patients with SDH were treated nonoperatively. Of these, two patients were identified that required delayed surgery based on increasing ICP and the development of intracerebral hematomas. Good outcome was achieved in 23% of the patients in the surgery group and 67% of the patients in the nonoperative group. The authors concluded that nonoperative treatment can be safely used for a defined group of comatose patients with SDH.

Age and Salvageability

Increasing age is a strong independent factor in prognosis from severe TBI, with a significant increase in poor outcome in patients older than 60 years of age2. Among patients with acute SDH, there is also a tendency for older patients to have a poorer outcome, especially those patients presenting with low GCS scores3, 16, 18, 19, 36. In comatose patients with GCS scores less than 9 who underwent craniotomy for SDH, Wilberger et al.36 found that age older than 65 years was statistically correlated with poorer outcome. In patients with GCS scores less than 10 undergoing surgery for SDH, Kotwica and Brzezinski18 found that there was a statistically significant difference in 3-months outcome between younger patients (18-30 yr of age, 25% mortality) and older patients (>50 yr, 75% mortality). Three smaller studies looked specifically at patients between 70 and 100 years of age with an admission GCS (one study) or preoperative GCS (two studies) equal to or less than 9. The 49 patients from these three studies all underwent surgery. Forty-eight patients died and one had a poor outcome (severely disabled or vegetative)3, 16, 19. No patient older than the age of 75 years who preoperatively was extensor posturing, flaccid to pain, or had unilateral or bilateral fixed and dilated pupils made a good recovery (GOS, 3-5)16. In 23 comatose patients aged 66 years and older who presented with an acute SDH, Howard et al.15 found that 17 died and the others survived in a vegetative state or with severe disabilities.

Functional outcomes in older patients with low GCS scores have also been reported. However, these articles did not document whether patients showed signs of cerebral herniation. Hatashita et al.14 reported 9 deaths in 12 patients older than 65 years who presented with GCS scores between 4 and 6 and underwent surgery for SDH, as compared with 34% for those aged 19 to 40 years. Two older patients survived with a GOS of 4 or 5. In another publication, 1 of 28 comatose patients older than 65 years made a functional recovery after craniotomy for SDH36. Although some studies that included patients with all GCS scores undergoing surgery for SDH found a relationship between age and outcome15, 21, 28, other authors failed to describe such a relationship13, 17, 26, 33, 39. Three studies using multivariate analysis in patients operated on for SDH did not identify age as an independent predictor of outcome15, 26, 33. In summary, there is a relationship between poor outcome and age, low GCS, and signs of herniation, but it is not possible to predict death on the basis of old age and poor GCS with certainty.

Timing of Surgery

The time from injury to entering the operating room is one of the few factors that can be affected by intervention. Unfortunately, the relationship between time from injury to operation and outcome is difficult to study because patients who are operated on soon after TBI tend to have more severe injuries than those who undergo delayed surgery. Therefore, outcome in patients operated on a short time after injury is frequently worse when compared with patients undergoing delayed surgery. Furthermore, time from TBI to surgery may not be as important as time from clinical deterioration or onset of cerebral herniation to surgery. The literature supports the statement that the length of time from clinical deterioration to operative treatment of an SDH is significantly related to outcome. Haselsberger et al.13 studied the time interval from onset of coma to surgery in 111 patients with SDH. Thirty-four patients were operated on within 2 hours after onset of coma. Of those patients, 47% died and 32% recovered with good outcome or moderate disability. However, 54 patients who underwent surgery longer than 2 hours after the onset of coma had a mortality of 80% and only 4% had a favorable outcome. These differences were statistically significant.

Seelig et al.26 studied the delay to surgery in 82 patients with SDH who were all comatose on admission. They found a 30% mortality rate in patients operated on within 4 hours after injury and a 90% mortality in patients who had surgery more than 4 hours after injury. The mean time for evacuation was 390 ± 39 minutes in patients who died and 170 ± 18 minutes in patients who made a functional recovery. Multivariate analysis identified time to surgery as one of the factors determining outcome from SDH. The weaknesses of this study are that a proportion of patients did not undergo CT scanning and that SDH was diagnosed using air ventriculography. In comatose patients undergoing surgery for SDH, Wilberger et al.36 found that the time interval from TBI to surgery was 374 ± 31 minutes for patients who died and 280 ± 26 minutes for patients who made a functional recovery. Mortality in patients undergoing surgery within 4 hours of injury was 59% versus 69% in patients operated on after 4 hours. A statistically significant difference could only be found in patients who underwent surgery after 12 hours, in which case, mortality rose to 82%.

Sakas et al.24 looked at outcome from surgery for intracerebral hematoma, epidural hematoma, and SDH in 40 patients who developed bilateral pupillary abnormalities during their hospital course. The authors found a significant relationship between the time from onset of bilateral pupillary abnormalities and 6-months outcome. Patients who had surgery more than 3 hours after herniation had a higher morbidity and mortality than those undergoing surgery earlier (mortality, 63% versus 30%.

Most studies focusing on the time between injury and surgery did not find a correlation with outcome15, 17, 18, 21, 28, 32. Some investigators even reported that early surgery was associated with worse results than delayed surgery6, 14, 29. As mentioned, this may be related to the fact that most investigators do not control for other variables affecting outcome, such as prehospital hypotension, hypoxia, GCS score, and associated intracranial lesions. In 82 patients undergoing surgery for SDH, Dent et al.6 found that time to surgery of less than 4 hours was associated with a significantly lower rate of functional outcome when compared with surgery delayed for longer than 4 hours (24% versus 51%). Mortality was approximately 30% in both groups. The authors also found that patients who underwent surgery within 4 hours were more likely to have obliterated basal cisterns and showed a tendency for lower GCS scores and more associated intracranial injuries, suggesting a more severe TBI.

The only large study with patients with low GCS scores (GCS < 10) that did not find a relationship between early surgery and better outcome was the study by Kotwica and Brzezinski18. In that study, mortality was approximately 60% in all patients, regardless of whether they had surgery within 4 hours or between 4 and 16 hours after TBI. A detailed analysis reveals that although GCS scores were the same, almost 90% of patients undergoing early surgery had associated intracranial lesions. Associated lesions were found in 78% and 64% of patients surviving the first 5 and 12 hours, respectively. This indicates that patients with early surgery had more severe injuries. In summary, there is evidence that patients who undergo surgery within 2 to 4 hours after clinical deterioration have a better outcome than those who undergo delayed surgery.

Surgical Technique

Different surgical techniques have been advocated for the evacuation of an SDH. The most commonly used techniques are:

• Twist drill trephination/craniostomy procedures.
• Burr hole trephination.
• Craniotomy with or without dural grafting.
• Subtemporal decompressive craniectomy.
• Large decompressive hemicraniectomy, with or without dural grafting.

Most investigators do not specify the type of surgical treatment used for evacuation of the SDH and, if they do, they usually do not address the effectiveness of the procedure. Except for two studies14, 30 no papers were found looking at the impact of procedure type on outcome. The choice of operative technique is influenced by the surgeon's expertise, training, and evaluation of the particular situation. Some centers treat all SDH with decompressive craniectomies18, 23, whereas other centers used solely osteoplastic craniotomies36. Most studies report a mixture of procedures depending on the clinical and radiographic evaluation13-15, 17, 38, or combined approaches in the same patient, i.e., subtemporal decompression plus subsequent craniotomy26 or craniotomies with contralateral decompressive craniectomies in some children31. One study evaluated decompressive hemicraniectomies for the treatment of selected patients with SDH30.

Only two investigators addressed the effect of the operative technique on outcome from SDH. Hatashita et al.14 looked at 3-months GOS in 60 patients with GCS scores between 3 and 15 admitted for SDH evacuation. All patients underwent surgery. The authors performed 24 burr holes, 25 craniotomies, 8 craniotomies with dural grafting, and 3 decompressive craniectomies. In patients with GCS scores between 4 and 6, the authors found a statistically significant increased mortality and reduced functional recovery rate in patients undergoing burr hole trephination versus craniotomy. Koc et al.17 compared craniotomy, craniotomy with dural grafting, and decompressive craniectomy in 113 patients with GCS scores between 3 and 15 undergoing SDH evacuation. Seventeen patients underwent decompressive craniectomy and all died. No other significant differences were found between treatment groups. The results of all of these studies have to be viewed with caution because groups undergoing different types of surgical treatment were not comparable.

In patients with an acute SDH, clot thickness or volume and the MLS on the preoperative CT correlate with outcome. In studies analyzing CT parameters that may be predictive for delayed surgery in patients undergoing initial nonoperative management, an MLS greater than 5 mm or a clot thickness greater than 10 mm on the initial CT scan emerged as significant prognostic factors (see Appendices for measurement techniques). Therefore, patients with SDH presenting with a clot thickness greater than 10 mm or an MLS greater than 5 mm should undergo surgical evacuation, regardless of their GCS. Patients who present in a coma (GCS < 9) but with an SDH with a thickness less than 10 mm and an MLS less than 5 mm can be treated nonoperatively, providing that they undergo ICP monitoring, they are neurologically stable since the injury, they have no pupillary abnormalities, and they have no intracranial hypertension (ICP > 20 mm Hg). Because of the frequent association of SDH with parenchymal injury, surgical management decisions should take into consideration the recommendations for both lesion types.

• Craniotomy versus decompressive craniectomy and dural grafting for the initial evacuation of SDH. Effect of different prehospital ambulance systems on timing of surgery and outcome from SDH.
• Incidence and impact of prehospital hypotension and hypoxia on outcome from SDH.
• Identification of subgroups that do not benefit from surgery: older patients with low GCS scores, pupillary abnormalities, and associated intracerebral lesions.
• Prospective evaluation of the treatment option for comatose patients (GCS < 9) presented above: does operating on all comatose patients, regardless of their hematoma thickness and MLS lead to a better outcome than following the treatment option presented above.