Division of Neurosurgery, St. Michael s Hospital, University of Toronto, Ontario, Canada

Transcription

1 J Neurosurg 112: , 2010 The natural history of brain contusion: an analysis of radiological and clinical progression Clinical article Hu s s e i n Al a h m a d i, M.D., Sh o b h a n Vac h h r a j a n i, M.D., a n d Mi c h a e l D. Cu s i m a n o, M.D., Ph.D., F.R.C.S.C., Division of Neurosurgery, St. Michael s Hospital, University of Toronto, Ontario, Canada Object. Although brain contusions are a common neurosurgical condition, surprisingly little has been written about their natural history. The purpose of this study was to identify factors that predict radiological and clinically significant progression of this pattern of traumatic brain injury in patients who did t initially require surgery. On the basis of their results and the available literature, the authors suggest a management algorithm. Methods. The authors performed a retrospective review of clinical and radiological records of consecutive patients with brain contusions who initially underwent conservative treatment. Significant radiological progression was defined as a 30% increase in contusion size on CT scans. Statistical analysis was performed to identify clinical and radiological predictors of CT contusion progression, the significance of progression, and predictors of clinical outcome. Results. Of 98 patients identified with brain contusions who initially received conservative treatment, 44 (45%) had significant progression on CT, and 19 (19%) required surgical intervention. The initial size of the contusion and the presence of subdural hematoma were the only statistically significant predictors of CT progression in the multivariate analysis (p = and 0.05, respectively). Four patients required delayed contusion evacuation (3 had radiological progression on follow-up scans). Good Glasgow Coma Scale (GCS) scores on presentation and younger age were predictors of eventual discharge from the hospital (OR 1.471, CI , p < and OR 0.949, CI , p = 0.011, respectively). No patients with an initial GCS score of 15 or an initial contusion size < 14 ml required delayed evacuation. Conclusions. Contusion progression is a common phemen that is seen more commonly in larger contusions. Patients with large contusions and low initial GCS scores are at risk for delayed deterioration. A proposed management algorithm for patients with contusions initially treated conservatively may help practitioners identify the best course of treatment. (DOI: / JNS081369) Ke y Wo r d s traumatic brain injury brain contusion contusion progression natural history Tr a u m at i c brain injury is the number one cause of death in trauma victims who reach the hospital alive. 1 Among this population, intracerebral hemorrhage, or brain contusion, is a common finding. Contusions can range in appearance from the more solid hematoma to a classic salt-and-pepper appearance (Fig. 1). They tend to show a progressive increase in mass effect on repeated imaging. 6 8,15,17 19,21,23 This change can be a result of expansion of the hematoma, appearance of perihematomal edema, or even the development of new lesions that were t present on initial scans (Fig. 2). 20,21 Several factors, including the severity of the injury, coagulopathy, the need for cardiopulmonary resuscitation in the field, old age, short duration between injury and the first CT scan, multiple hematomas, midline shift, Abbreviations used in this paper: GCS = Glasgow Coma Scale; ICP = intracranial pressure; SDH = subdural hematoma; TBI = traumatic brain injury. and surgical decompressive procedures, have been associated with radiological progression. 11,18,23,26 Although some authors have recently studied the progression of brain contusions and its predictors, few have examined the clinical relevance of progression, and very few management recommendations have been published based on available natural history data. As a result, there is wide variability internationally in the treatment of these patients, with some patients receiving several CT scans per day to monitor progression and others virtually ne. In a systematic review of the literature on the value of routine CT scanning, Wang et al. 25 showed that indications for routine CT scanning remain unclear, and called for future studies that would stratify cases by severity of TBI, clearly define inclusion and exclusion criteria, quantify progression of injury on CT, and determine factors predictive of injury progression. The aim of the present study was to evaluate which factors best predicted radiological and clinical progression and the ultimate need for 1139

2 H. Alahmadi et al. Fig. 1. Computed tomography images showing different patterns of contusions. surgical intervention and outcome in a group of patients with brain contusions who underwent conservative treatment initially. Methods Inclusion and Exclusion Criteria We reviewed the records of all patients with history of trauma and CT diagsis of cerebral contusion/traumatic intracerebral hemorrhage treated at St. Michael s Hospital (downtown Toronto adult trauma center) between May 2004 and June Patients were included if they had an initial CT scan at our institution that confirmed the diagsis of contusion, received conservative treatment initially, and had at least 1 other CT scan during the acute hospitalization. Patients who underwent craniotomy after the first scan to address their contusions were excluded, as were those who were admitted but never underwent scanning or surgery during the hospital stay. Patients who were discharged home directly from the emergency department were also excluded. Patients who underwent external ventricular drain placement initially or in whom surgery was performed only for an extraaxial hematoma and t the contusion were included. Data Collection We collected relevant clinical data including results from presenting and follow-up neurological examinations, history of antiplatelet therapy, any intensive care unit or surgical intervention, and discharge disposition (death, home, rehabilitation, or long-term care facility). Investigation data included international rmalized ratio, prothrombin time, activated partial thromboplastin time, and the characteristics of the initial and follow-up CT scans with timing of follow-up scans. We reviewed all CT scans and calculated the size of the contusion hematoma volume using the formula ABC/2 where A, B, and C represent the largest dimensions of the hematoma measured at perpendicular angles to one ather. This formula is easy to use and its accuracy has been validated. 14 For heterogeneous contusions with a salt-and-pepper appearance, the entire heterogeneous Fig. 2. Computed tomography images showing contusion progression between the initial (left) and the follow-up scan (right). mass was measured. Contusions were also classified as lobar or deep. Frontal, temporal, parietal, or occipital contusions were considered lobar, whereas contusions in the diencephalon, brainstem, or cerebellum were classified as deep. Coexisting extraaxial hematomas were recorded. We also reviewed the radiologist s report in all cases to confirm the changes in contusion size. Outcome Criteria Three outcome measures were assessed, 1 radiological and 2 clinical: Increase in Hematoma Size. A significant increase in contusion size was defined as an enlargement of 30% of the original hematoma size on CT studies. This criterion was chosen because a similar cut-off value was used previously in the literature. 18 Neurosurgical Intervention. Neurosurgical intervention was defined as the requirement for any one of the fol lowing: ICP monitoring, craniotomy and evacuation of hematoma, or decompressive craniectomy. We also re corded whether patients required a tracheosotomy, because this may reflect a poor short-term neurological recovery. Discharge Disposition. Final neurological status (measured with the GCS) when leaving our institution was ted. The patient s discharge status was classified as to whether the patient was discharged home, or to ather institution, or whether the patient died. Statistical Analysis Statistical analysis was performed using commercially available software (SAS for Windows, SAS Institute). Descriptive statistics were generated for demographic data. The Pearson chi-square test was performed to examine the univariate association of radiographic contusion progression with any neurosurgical intervention, patient disposition, and the need for tracheostomy. Too few patients with contusions required evacuation to test significant statistical association between contusion progression and contusion evacuation. Student t-test was used to compare the mean GCS score at admission and 1140

3 The natural history of brain contusion discharge, and the mean GCS score change during hospitalization in patients who progressed and those who did t. Finally, multivariable logistic regression models were built using the Harrell method to determine independent predictors at admission of radiographic progression and patient disposition. The small number of patients undergoing surgical evacuation of their contusions prevented construction of a multivariate regression model to predict evacuation. Statistical significance was set at a probability value of Results Patient Characteristics Our initial search of patients with brain injuries admitted to our unit yielded 1154 patients, 98 of whom met our inclusion criteria. Demographic and baseline characteristics are shown in Table 1. Seventy-nine patients were male (81%), and patient ages ranged from 17 to 86 years. The mean ± SD admission GCS score was 10.8 ± 3.8 with an average improvement to 13.8 ± 2.7. The average initial contusion volume was 14.4 ± 18.3 ml, and the median initial volume was 8 ml (1 92 ml). Patient Outcomes With respect to the outcomes of interest, 44 patients (45%) displayed radiographic progression on CT scans during their initial hospital course, and 19 (19%) required subsequent surgical intervention. There were slightly more patients discharged to ather institution or who died than were discharged home (57 vs 43%). Only 10% of patients required a tracheostomy. Clinical and radiological outcomes are summarized in Table 2. The timing of follow-up scans ranged from 5 hours to 4 days postinjury (Table 3). Radiographic progression of contusions was significantly associated with the need for neurosurgical interventions, excluding tracheostomy (p = ), but t with patient disposition (p = ) or the need for tracheostomy (there were too few patients to perform statistical analysis). There was significant difference in the mean outcome GCS score in patients who had progression on CT compared with scores in patients who did t. The mean GCS score and mean change in GCS score over the hospital course was t statistically different between the patients with CT progression and those without. Multivariable logistic regression models were developed to search for independent predictors of radiographic progression and patient disposition. The volume of the contusion at admission and the presence of SDH were the only independent predictors of radiographic progression (p = and 0.05, respectively; Table 4). Extradural hematoma and subarachid hemorrhage were t predictors of radiological progression. Finally, presentation GCS score independently predicted the probability of being discharged home in mul tivariate analysis (OR 1.471, 95% CI ; p < 0.001). Patient age similarly predicted disposition although less so than presentation GCS score (OR 0.949, 95% CI , p = 0.011; Table 5). TABLE 1: Summary of patient characteristics Variable No. of Patients (%) sex male female coagulation rmal abrmal antiplatelet therapy initial GCS score deep contusions present contusion size <10 ml ml ml >50 ml 79 (81) 19 (19) 90 (92) 8 (8) 6 (6) 92 (94) 45 (46) 23 (23) 29 (30) 9 (9) 89 (91) 58 (59) 22 (22) 11 (11) 7 (7) Delayed Surgery Group Initial conservative treatment failed in 19 patients, and delayed surgery was required (Table 6). Only 2 patients with initial GCS scores > 13 required delayed surgery (both had GCS scores of 15). In these 2 patients, the contusions were < 10 ml, and repeated scans showed change in contusion size. Both patients underwent craniotomies for enlarging epidural hematomas during the first 48 hours and were eventually discharged home from the hospital. No patient who required delayed contusion evacuation had an initial GCS score > 13. In the other 17 patients who required delayed surgery, 7 had only ICP monitor insertion and 10 patients had craniotomies (in 2 of these the bone flap was t replaced). Four of the craniotomies were done for evacuation of ei- TABLE 2: Summary of outcome characteristics Variable No. of Patients (%) Intervention outcome GCS score disposition home to ather institution or died tracheostomy 19 (19) 79 (81) 87 (89) 6 (6) 5 (5) 42 (43) 56 (57) 10 (10) 88 (90) 1141

4 H. Alahmadi et al. TABLE 3: Timing of follow-up scans Timing of Follow-Up Scan ther the brain contusion alone or the brain contusion and extraaxial hematoma. The remaining craniotomies were done for extraaxial hematomas in 5 patients, and elevation of depressed skull fracture in 1. Four patients had contusions that required evacuation. The contusions ranged in size from 14 to 92 ml, and initial GCS scores in these patients were All 4 patients underwent surgery within the first 48 hours of hospital presentation. Three had significant contusion progression and 1 had progression. However, in this patient the coexisting SDH had increased in size and the treating neurosurgeon decided to evacuate the SDH and underlying contusion. All the contusions that required delayed surgery had coexisting SDHs. Discussion No. of Patients w/in 1st 24 hrs 37 2nd day 27 3rd day 13 4th day 21 Radiological Progression The results of our study demonstrated that about half of contusions managed conservatively would progress radiologically over time in hospital. This is similar to the rates of progression reported by other authors: 51% by Oertel et al., 18 44% by Stein et al., 23 and 51% by Narayan et al. 17 However, our results were higher than the 16.4% reported by Yadav and colleagues 26 and 38% reported by Chang et al. 7 On average, the criterion for radiological progression used by Yadav and colleagues demanded approximately a doubling in contusion size. This strict criterion would explain the lower rate of progression in that study. In contrast, Chang et al. reported mostly small contusions 1 ml, which may explain the apparently more benign behavior of contusions in their series. We used a TABLE 4: Results of the regression analysis of CT progression* Variable p Value contusion size presence of EDH presence of SAH presence of SDH 0.05 sex age anticoagulation antiplatelet therapy initial GCS score deep contusion * EDH = epidural hematoma; SAH = subarachid hemorrhage. TABLE 5: Regression analysis of patient disposition* Variable p Value OR (95% CI) initial GCS score < ( ) contusion size ( ) sex ( ) anticoagulation ( ) antiplatelet therapy ( ) deep contusion ( ) progression on CT ( ) age ( ) * There were too few patients with tracheostomies to include this variable in the model (the validity of the model would be compromised). cut-off value for significant clinical progression of 30%. Oertel and associates used a criterion of at least 25% increase in hematoma dimensions. Using strict criteria such as a 30% enlargement to define radiological progression minimizes potential errors in identifying contusion progression, given the potential difficulties that can arise with repeated imaging because of different angulation of CT scan cuts across the contusion and across different CT scanners within a single institution or at several institutions. We suggest that future studies should use a similar percentage criterion. We found that initial contusion size and the presence of a coexisting SDH were the only statistically significant predictors of CT progression (p = and 0.05, respectively). The direct correlation between the initial size of contusion and progression has been shown by other authors. 7,8 The association between SDH and contusion progression has also been reported before. 7 One potential explanation is that some of these SDHs might be secondary to a burst lobe from an underlying large contusion that is more likely to enlarge. Our sample size, the strict criteria we used for a relevant enlargement of the contusion, or the inclusion and exclusion criteria may have limited our ability to identify other factors like antiplatelet therapy and coagulopathy as significant predictors of progression. Understanding the clinical significance of this radiological phemen and its impact on treatment in these patients is important because t all patients with radiological progression show clinical deterioration or require surgical intervention. Patients with radiological progression in our series were more likely to require delayed surgical intervention consisting of intracerebral pressure monitor insertion or craniotomy (OR 4.254, 95% CI , p = ). In reviewing the charts of the 4 patients who underwent delayed contusion evacuation, we ted that 3 showed radiological contusion progression. The association between radiological progression and a worse outcome has been previously reported. In the early 1990s, Stein et al. 23 demonstrated that radiological progression of traumatic lesions is associated with a worse outcome, a correlation also found in later series. 8,19 The insensitivity of CT progression in predicting discharge outcome could have been the result of our small sample 1142

5 The natural history of brain contusion TABLE 6: Characteristics of patients who required delayed surgery Characteristic No. of Patients (%) timing of surgery w/in 1st 24 hrs w/in 2nd 24 hrs after 2 days initial GCS score intervention craniotomy ICP monitor placed coexisting SDH size, confounding factors, or an inadequate cut-off point for important progression. In the present study, a poor GCS score and older age on presentation were inversely related to the chance of going home from the hospital (OR 1.471, CI , p = and OR 0.949, 95% CI , p = , respectively), in keeping with the literature as kwn progsticators of a poor outcome. 2,3 From our analysis of the charts of patients who underwent delayed surgery, at least 3 important conclusions can be drawn about the need to perform surgery in patients with contusions who initially undergo conservative treatment. First, our data indicate that small contusions are less likely to enlarge and subsequently require surgical intervention. No patient with a contusion < 14 ml needed delayed evacuation. This finding indicates that routine extensive resource utilization such as repeat CT scanning and prolonged intensive care unit stay for patients with contusions < 10 ml is unlikely to change treatment with respect to surgery, suggesting that this practice needs critical reevaluation. Our findings still suggest that patients who present in good clinical condition with brain contusions initially must be assessed and reassessed clinically. Dunn and colleagues 10 showed that 7.8% of patients who scored 3 or higher on the verbal component of the GCS at admission died. Those who were verbally responsive at admission but later died were older, had more severe extracranial injuries as reflected by the injury severity scale score, a worse abbreviated injury score for the head region, a more altered level of consciousness, and reached the operating theater more quickly than those who survived. Extraaxial hematomas and multiple intracranial pathological conditions were also more frequent in those who died than in those who lived. Dunn et al. described a subset of patients who, despite speaking at one point, deteriorated rapidly and irreversibly despite rapid operative treatment. It was thought that advances in head injury care and trauma care were unlikely to be of benefit in those patients with malignant intracranial hematoma. Narayan et al. 17 reviewed their experience with 56 patients with contusions. Five patients needed evacuation 7 (37) 4 (21) 8 (42) 2 (11) 6 (32) 11 (58) 12 (63) 7 (37) 12 (63) 7 (37) after initial conservative management, and all of them had initial contusion size > 5 ml. Our findings and those of Narayan et al. are at odds with those of Chang et al., 7 who found that 6 of the 10 patients in whom initial conservative treatment failed had initial contusion sizes of 1 ml. The series by Chang et al. included 1 patient who underwent evacuation for a cerebellar hemorrhage of 1 ml and ather patient who had surgery for a frontal contusion with an initial size of 1 ml that progressed to a total size of 4 ml. Because these contusions seem smaller than what would be generally considered to require surgery, 5 a detailed individual review of these cases is needed to exclude idiosyncratic factors unrelated to contusion size that led to surgery in these patients. The second conclusion from our study is that contusion progression is more likely to happen in the first 2 days after the injury. No patient required delayed contusion evacuation after the first 48 hours. Narayan et al. 17 also found that deterioration most often occurs early after injury. The association between radiological progression and shorter time between injury and initial scan 18 also supports this conclusion. Thus, intensive monitoring for most patients beyond that point is unlikely to be very fruitful if the patient has been radiologically and clinically stable to that point. The last conclusion with respect to surgery for this population is that all patients with contusion who eventually required evacuation of the contusion had abrmal GCS scores on presentation. Although 2 patients with rmal GCS scores required evacuation of epidural hematomas, patient with a rmal GCS score and a brain contusion alone required evacuation. In addition, every patient who required contusion evacuation also had an associated SDH. This suggests that patients with low GCS scores and those with extraaxial hematomas warrant more intensive vigilance, whereas those with rmal GCS scores without extraaxial hematomas are less likely to benefit from resource-intensive vigilance. Management Algorithm One aim of our study was to use our findings and prior reports in the literature to develop a management algorithm to assist in the management of brain contusions because unguided routine serial CT scanning has been shown to have effect on patients treatment. 4,9,13,24 Although such algorithms would ideally arise from highlevel clinical evidence such as randomized controlled trials, it is unlikely that such data will arise in the near future. One purpose of suggesting an algorithm at this stage is to stimulate discussion among neurosurgeons internationally as to the optimal treatment for these patients and to stimulate well-designed prospective studies. Given the resources involved in caring for patients with brain contusions, such an algorithm has the potential to guide limited resources to areas where they are most likely to have a significant impact. Although such an algorithm may have more relevance in lower-income countries, it would also apply to any middle- or high-income nation with a limited resource base. No management algorithm will apply in every situation: every set of data, including ours, has limitations. 1143

6 H. Alahmadi et al. Algorithms do t replace good clinical judgment in individual cases, and surgeons must act when action is required. However, there appears to be a convergence of results from a number of studies at the present time indicating that small contusions in patients with a good initial GCS score and the absence of clinical deterioration in the first 48 hours are very unlikely to progress to require surgery. Stein et al. 22 studied the cost-effectiveness of a model comparing the strategy of routine serial scanning with that of awaiting clinical deterioration before repeating the CT scan in patients with mild TBI and an intracranial abrmality. They demonstrated that the routine serial scanning model had only slightly more effectiveness at a higher cost only in 20-year-old patients. The cost-effectiveness of the strategy declined exponentially as the patient age increased. The results of our study give support to using a less resource-intensive approach in the younger patient with an isolated small brain contusion and stable clinical condition. Our data suggest that older patients with worse GCS scores, and those with large contusions and SDH are most likely to progress, and may warrant a more resource-intensive monitoring strategy that includes repeated CT scanning and intensive nursing supervision. Figure 3 shows a suggested management algorithm for the care of adult patients with brain contusions. Supplementing this algorithm with ICP monitoring and measuring the trends in serum levels of biomarkers of brain injury 12 may add to the sensitivity of this algorithm in identifying patients at risk for further deterioration and in predicting outcome. The use of serum biomarkers in this context remains, however, largely experimental, and further validity studies will be necessary before their incorporation into any management algorithms, particularly those in which serial CT scanning is avoided. The authors of future studies should also assess the validity and cost-effectiveness of such an algorithm. The limitations of our study include its retrospective design, and therefore its lack of a consistent protocol for the timing of CT scans after admission. This may have influenced our ability to detect radiological progression. However, Narayan et al., 17 working with a very strict prospective protocol in a well-funded multicenter study, could only recruit 63 patients and, therefore, had limited external validity. Despite this, their conclusions were similar to ours. Thus, we see a value in the retrospective results involving a larger number of patients presented here. Ather limitation of our study is that it represents a subset of all patients presenting with brain contusion. We excluded patients whose clinical condition was such that the patient obviously required immediate surgery for significant mass effect or were moribund and expected to die. We also excluded patients with contusions so small that the involved clinicians judged it unnecessary to repeat their CT scans. We only included patients about whom clinicians had some clinical or radiological concern and so decided to repeat brain imaging. This makes it quite distinct from other studies in the literature that were frequently only radiological in nature. Using these criteria may have limited the statistical power of some of our statistical analyses, but ensured that we had repeated brain imaging in patients with moderate-sized contusions that had a potential for various outcomes. Ather limitation of our study relates to our outcome variables and their short-term nature. Finally, our data are limited to adults and so cant be applied in children with brain contusions. Future studies should address present and past limitations, and also include international data on cost-effectiveness. Conclusions Progression of brain contusion on CT scans is a com- Fig. 3. Flow chart showing the algorithm for the treatment of patients with brain contusions and a rmal level of consciousness. 1144

7 The natural history of brain contusion mon finding, and larger contusions at admission are more likely to progress. Patients with a poor initial GCS score and large contusions are at an increased risk of requiring delayed surgical intervention. Routine serial imaging for patients with small contusions and rmal GCS score is less likely to alter their management, especially beyond 48 hours. Larger prospective studies that include analysis of the cost and benefit of intensive monitoring and imaging paradigms are required. Disclaimer The authors report conflict of interest concerning the materials or methods used in this study or the findings specified in this paper. References 1. Acosta JA: Lethal injuries and time to death in a level I trauma center. J Am Coll Surg 186: , The Brain Trauma Foundation, The American Association of Neurological Surgeons, The Joint Section on Neurotrauma and Critical Care: Age. J Neurotrauma 17: , The Brain Trauma Foundation, The American Association of Neurological Surgeons, The Joint Section on Neurotrauma and Critical Care: Glasgow coma scale score. J Neurotrauma 17: , Brown CV, Weng J, Oh D, Salim A, Kasotakis G, Demetriades D, et al: Does routine serial computed tomography of the head influence management of traumatic brain injury? A prospective evaluation. J Trauma 57: , Bullock MR, Chesnut R, Ghajar J, Gordon D, Hartl R, Newell DW, et al: Surgical management of traumatic parenchymal lesions. Neurosurgery 58 (3 Suppl):S25 S46, Bullock R, Golek J, Blake G: Traumatic intracerebral hematoma which patients should undergo surgical evacuation? CT scan features and ICP monitoring as a basis for decision making. Surg Neurol 32: , Chang EF, Meeker M, Holland MC: Acute traumatic intraparenchymal hemorrhage: risk factors for progression in the early post-injury period. Neurosurgery 58: , Chieregato A, Fainardi E, Morselli-Labate AM, Antonelli V, Compagne C, Targa L, et al: Factors associated with neurological outcome and lesion progression in traumatic subarachid hemorrhage. Neurosurgery 56: , Dharap SB, Khandkar AA, Pandey A, Sharma AK: Repeat CT scan in closed head injury. Injury 36: , Dunn LT, Fitzpatrick MO, Beard D, Henry JM: Patients with a head injury who talk and die in the 1990s. J Trauma 54: , Flint AC, Manley GT, Gean AD, Hemphill JC III, Rosenthal G: Post-operative expansion of hemorrhagic contusions after unilateral decompressive hemicraniectomy in severe traumatic brain injury. J Neurotrauma 25: , Guzel A, Er U, Tatli M, Aluclu U, Ozkan U, Duzenli Y, et al: Serum neuron-specific elase as a predictor of short-term outcome and its correlation with Glasgow Coma Scale in traumatic brain injury. Neurosurg Rev 31: , Kaups KL, Davis JW, Parks SN: Routinely repeated computed tomography after blunt head trauma: does it benefit patients? J Trauma 56: , Kothari RU, Brott T, Broderick JP, Barsan WG, Sauerbeck LR, Zuccarello M, et al: The ABCs of measuring intracerebral hemorrhage volumes. Stroke 27: , Lobato RD, Cordobes F, Rivas JJ, de la Fuente M, Montero A, Barcena A, et al: Outcome from severe head injury related to the type of intracranial lesion. A computerized tomography study. J Neurosurg 59: , Lobato RD, Gomez PA, Alday R, Rivas JJ, Dominguez J, Cabrera A, et al: Sequential computerized tomography changes and related final outcome in severe head injury patients. Acta Neurochir (Wien) 139: , Narayan RK, Maas IR, Servadei F, Skolnick BE, Tillinger MN, Marshall LF: Progression of traumatic intracerebral hemorrhage: a prospective observational study. J Neurotrauma 25: , Oertel M, Kelly DF, McArthur D, Boscardin WJ, Glenn TC, Lee JH, et al: Progressive hemorrhage after head trauma: predictors and consequences of the evolving injury. J Neurosurg 96: , Servadei F, Murray GD, Penny K, Teasdale GM, Dearden M, Iantti F, et al: The value of the worst computed tomographic scan in clinical studies of moderate and severe head injury. Neurosurgery 46:70 77, Servadei F, Nanni A, Nasi MT, Zappi D, Vergoni G, Giuliani G, et al: Evolving brain lesions in the first 12 hours after head injury: analysis of 37 comatose patients. Neurosurgery 37: , Soloniuk D, Pitts L, Lovely M, Bartkowski H: Traumatic intracerebral hematoma: timing of appearance and indications for operative removal. J Trauma 26: , Stein SC, Fabbri A, Servadei F: Routine serial computed tomographic scans in mild traumatic brain injury: when are they cost effective? J Trauma 65:66 72, Stein SC, Spettell C, Young G, Ross SE: Delayed and progressive brain injury in closed-head trauma: radiological demonstration. Neurosurgery 32:25 31, Velmahos GC, Gervasini A, Petrovick L, Dorer DJ, Doran ME, Spaniolas K, et al: Routine repeat head CT for minimal head injury is unnecessary. J Trauma 60: , Wang MC, Linnau KF, Tirschwell DL, Hollingworth W: Utility of repeat head computed tomography after blunt head trauma: a systematic review. J Trauma 61: , Yadav YR, Basoor A, Jain G, Nelson A: Expanding traumatic intracerebral contusion/hematoma. Neurol India 54: , 2006 Manuscript submitted October 26, Accepted May 11, Please include this information when citing this paper: published online July 3, 2009; DOI: / JNS Address correspondence to: Michael D. Cusima, M.D., Ph.D., Division of Neurosurgery, St. Michael s Hospital, 30 Bond Street, Toronto, Ontario, Canada M5W 1B6. mountain@smh. toronto.on.ca. 1145