THOMAS G. SAUL, M.D., THOMAS B. DUCKER, M.D., MICHAEL SALCMAN, M.D., AND ERIC CARRO, M.D.

J Neurosurg 54:596-600, 1981 Steroids in severe head injury A prospective randomized clinical trial THOMAS G. SAUL, M.D., THOMAS B. DUCKER, M.D., MICHAEL SALCMAN, M.D., AND ERIC CARRO, M.D. Department of Neurosurgery, University of Maryland, School of Medicine, Baltimore, Maryland This is a prospective randomized study of the efficacy of steroid therapy in patients with severe head injury. One hundred patients were randomized into two equal groups: the steroid group received 5 mg/kg/day of methylprednisolone, and the nonsteroid group received no drug. The groups were similar in their clinical features. All patients received a standardized therapeutic regimen. The patients were also classified as early responders or nonresponders to the overall treatment protocol without regard to steroid administration, on the basis of change in Glasgow Coma Scale score during the first 3 days of admission. There was no statistically significant difference in the outcome of the steroid and nonsteroid group at 6 months. Of the responders who were on steroids, 74% had good outcomes or were disabled, compared with 56% of the responders who did not receive steroids. In the nonresponder group, the patients on steroids were actually associated with a worse outcome than those who did not receive steroids: 75% of the nonresponders who received steroids were dead or vegetative, compared to 56% of those who were not receiving steroids. The data suggest that: 1) the effect of steroids may be different for different patient groups; 2) in order to identify these patients, a sensitive coma scale is needed; and 3) a rational approach to steroid therapy in head-injured patients may be to start all patients on steroids, but to discontinue their use in patients identified as not benefiting from steroid therapy. KEY WORDS 9 head injury 9 intracranial pressure 9 Maryland Coma Scale 9 prognosis 9 steroid therapy I N 1961, Galicich and French 8 documented the beneficial effect of steroids in the management of cerebral edema due to primary and metastatic brain tumors. Subsequently, this effect has been confirmed by other investigators, and steroids are now a universal part of the treatment of intracranial neoplasms? 4,15 However, the efficacy of steroid therapy in the management of patients with severe head injury remains controversial. The results of both experimental and clinical investigations have been contradictory. Studies can be found that support the beneficial effect, no effect, and, in some instances, the adverse effect of steroid therapy. 1,3-Sa-l~ In spite of these apparent contradictions, most neurosurgeons elect to include steroids in their protocol for management of severely head-injured patients. Becker has stated: "The patient would receive high dose steroid therapy (despite the lack of clear evidence of its beneficial effect)... -2 Tindall maintains: "Realizing that there is no definite proof that dexametha- sone, even in megadoses, benefits patients with head injury... I would, nevertheless, institute therapy with dexamethasone. ''2~ Such statements and the continued widespread use of steroids in this setting implies that there is a persisting clinical impression that steroids do have some beneficial effect in certain patients. This prospective randomized clinical trial of steroids was performed and analyzed in an attempt to resolve some of these contradictions associated with steroid therapy. Clinical Material and Methods The patients entered in this study included those admitted to the University of Maryland Shock-Trauma Unit during an 18-month period, who had sustained craniocerebral trauma and were admitted within 6 hours of injury. These patients had no other body systems injured and had a Glasgow Coma Scale (GCS) score of seven or less. 19 596 J. Neurosurg. / Volume 54 / May, 1981

Steroids in severe head injury In addition to the GCS, the patients were assigned a score based on the Maryland Coma Scale. 17 This is an index used for patients with central nervous system trauma. It has two purposes: 1) on admission, it is used to predict outcome; and 2) combined with subsequent scores in a simple formula, it is used to indicate clinical course. The scoring is based on an index that describes simple levels of intracranial dysfunction, with attention to the cranial nerves in the fine details of movement, and that categorizes levels of consciousness. The index contains the three variables from the GCS, automatically calculated so that the data can be compared from center to center. A patient who has been assessed receives both an absolute score and a percentage of total function (47 points). If it is not possible to test all parameters, then the "neuro index" is recorded for as many as are tested and the percentage is calculated on the basis of the total possible score. At regular intervals, subsequent scores are compared to the initial one, the difference in scores representing the patients' clinical course; such data are also easily converted into a recovery index. By being more comprehensive, this scale enables us to measure a patient's daily changes and to more fully evaluate treatment protocols. All of these patients received the same standardized treatment protocol, which included immediate neurological assessment, followed by endotracheal intubation and hyperventilation. Computerized tomography or carotid angiography were used to diagnose intracranial mass lesions. Patients with mass lesions underwent immediate operation and removal of the lesions. Intracranial pressure monitoring was performed in all patients via an intraventricular catheter or a subarachnoid bolt. One hundred patients were randomized into two groups of 50. The steroid group received 250 mg of methylprednisolone intravenously as a loading dose, followed by 125 mg every 6 hours. On occasion, dexamethasone, in equivalent doses, was used according to the same scheme. The nonsteroid group received no drug. All patients had their outcomes assessed at 6 months, based on the Glasgow Outcome Scale devised by Jennett and Bond? x good recovery (resumption of normal life, possibly with minor neurological and psychological deficits); disabled (includes severe disability and moderate disability); or vegetative (awake, but without cerebral cortical function); and dead. In addition to the above, the patients were also classified as either early responders or nonresponders to the overall treatment protocol. This classification was applied to the entire group of 100 patients without regard to whether they were receiving steroids r not. A protocol responder was defined as a patient whose GCS score was 8 or higher within the first 3 days: that is, he had improved from an admission GCS score of 7 or less, up to at least 8. A protocol nonresponder was defined as a patient whose GCS score remained at 7 or TABLE l Clinical features of steroid and nonsteroid groups Steroid Nonsteroid Features Group Group no. of cases 50 50 average age (yrs) 32 30 craniotomies no. 7 5 percent 14 10 uncontrolled intracranial pressure no. 5 6 percent 10 12 TABLE 2 Incidence of protocol responders and nonresponders* Protocol Steroid Nonsteroid Response Group Group responders 38 34 nonresponders 12 16 *X 2 =0.446; p 20.25. less during the first 3 days: that is, his GCS score had not changed or it had worsened since admission. Therefore, we have classified our patients in two ways. First, by randomization into steroid and nonsteroid groups and then independently, as early responders or nonresponders to the total treatment protocol based on the changes of their GCS scores within the first 3 days of admission. We then looked at the 6-month outcome of our patients with each of these classifications in mind. Comparison of Groups Results Both groups were similar in their clinical features (Table 1). The average age for the entire group was 31 years, with a range from 16 to 72 years. The age of patients in the steroid group averaged 32 years, while the average age in the nonsteroid group was 30 years. Seven (14%) of the steroid patients underwent emergency craniotomy, as did five (10%) of the nonsteroid patients. Removal of an intracranial hematoma was the most common procedure, but two of the 12 operated on had surgery for depressed skull fractures. Uncontrollable intracranial hypertension was encountered in five (10%) of the steroid group and in six (12%) of the nonsteroid group. In addition, the incidence of protocol responders and nonresponders was comparable in both groups (Table 2), there being 38 and 12, respectively, for the steroid group, and 34 and 16 for the nonsteroid group. This is another indica- J. Neurosurg. / Volume 54 / May, 1981 597

T. G. Saul, T. B. Ducker, M. Saleman and E. Carro Outcome TABLE 3 Outcomes in the nonsteroid group at 6 months Protocol Protocol Total Responders Nonresponders Cases good l0 3! 3 disabled 9 4 13 vegetative 12 3 15 dead 3 6 9 total cases 34 16 50 Outcome TABLE 4 Outcomes in the steroid group at 6 months Protocol Protocol Total Responders Nonresponders Cases good 22 1 23 disabled 6 2 8 vegetative 9 2 11 dead 1 7 8 total cases 38 12 50 tion that the steroid and nonsteroid groups were similar as to the type of patient and the severity of injuries. Outcomes Related to Steroid Administration Table 3 shows the 6-month outcomes of the nonsteroid group as a whole and also broken down into protocol responders and nonresponders. The total group had 13 good results, 13 disabled, 15 vegetative, and nine deaths. Table 4 shows the same information for the steroid group, which had 23 good outcomes, eight disabled, 11 vegetative, and eight deaths. For statistical purposes, the outcomes were combined into a good or disabled group and a vegetative or dead group (Table 5). The steroid group had 31 good or disabled and 19 vegetative or dead, compared to 26 and 24, respectively, for the nonsteroid grot]p. There were no significant statistical differences in the outcomes of the two groups (p = 0.22). Outcomes of Responders and Nonresponders Related to Steroid Administration When we analyze the outcomes of the protocol responders and nonresponders and examine how the steroid and nonsteroid patients did within these groups, some interesting statistical trends appear. Table 6 shows the outcomes of the protocol responders. Of the 38 patients in this group who received steroids, 28 (74%) were disabled or had good outcomes, compared to only 19 out of 34 (56%) of the patients in this group who did not receive steroids. Examining the protocol nonresponders (Table 7), the patients who received steroids were associated with a worse outcome than the nonsteroid patients. In this nonresponder group, 12 patients received steroids, and only three (25%) were disabled or had good outcomes, compared to seven out of 16 (44%) of the patients who did not receive steroids. Furthermore, nine out of 12 (75%) of the steroid nonresponders were vegetative or dead, compared to nine out of 16 (56%) of the nonsteroid nonresponders (p = 0.19, according to the Fisher exact probability test). Discussion Several comments about this series of patients are warranted. First, there is a much lower incidence of intracranial surgical lesions in this series than in others reported in recent literature. In addition, the mortality rate (17%) is quite low. This may be due to the exclusion of patients with multiple injuries and the low incidence of surgical lesions. Furthermore, in order to be included in the study, the patients had to live 3 days, excluding all patients with injuries that brought about death in the first 72 hours. In other words, this is a very select patient population, which will affect mortality rates. These data are not offered as a definitive statement on steroid efficacy in severe head injuries. The patient numbers and the lack of statistical significance prevents this. However, we believe the data may support several important facts. The question of steroid efficacy has never been clear, and perhaps the answer is not "yes" or "no," but rather "yes" in certain patients. In other words, the effect of steroids may be different for different patient groups, that is, those TABLE 5 Outcomes in the steroid versus nonsteroid groups* Group Good or Vegetative Disabled or Dead steroid group 31 19 nonsteroid group 26 24 *X 2 ---- 0.653; p ---- 0.22. 598 TABLE 6 Outcomes of protocol responders in relation to steroids* Steroid Nonsteroid Outcome Group Group No. Percent No. Percent disabled or good 28 74 19 56 vegetative or dead 10 26 15 44 *X 2 = 1.785; p = 0.10. J. Neurosurg. / Volume 54 / May, 1981

Steroids in severe head injury TABLE 7 Outcomes of protocol nonresponders in relation to steroids* Steroid Nonsteroid Outcome Group Group No. Percent No. Percent disabled or good 3 25 7 44 vegetative or dead 9 75 9 56 *p = 0.19, Fisher exact probability test. who are protocol responders as opposed to protocol nonresponders. The difference in patient response to steroids may be a reflection of many factors, including host factors of each patient, the extent of the initial brain injury, and the responsiveness of that injury to ongoing therapeutic regimens. In order to identify these different patient groups, a sensitive "neuro index" may be necessary. 13a~ We have not found the initial assessment of the patient to be useful in predicting the steroid effect. Our experience seems to indicate that while the GCS is good for identifying acute severe outcomes, it is too limited to measure daily variations on which to base changes in a given therapeutic protocol. The Maryland Coma Scale appears to be more helpful in this respect. For example, there were at least two patients who were nonresponders according to the GCS, but who were improving as measured on the Maryland Coma Scale, and these patients actually had good outcomes at 6 months after receiving steroid therapy? 7 Increased intracranial pressure (ICP), when it occurred, was treated the same in all patients. The occurrence of elevated ICP was no different in either the steroid- or the nonsteroid-treated patients. Hyperventilation, mannitol, and cerebrospinal fluid drainage were all used to control elevated ICP when pressure was over 25 torr. We were not using barbiturate therapy at the time of this study. We considered ICP uncontrollable if it was above 25 mm/hg and remained so in spite of therapy. We believe that whatever beneficial effects steroids produce, they are unable to alter very high levels of ICP. This supports previous reports by Gudeman, et al., 8 and Pitts, et al. ~8 The variable response to steroids observed in different patient groups may partially account for the differences in steroid efficacy reported previously. In order to establish the efficacy of steroids in head injuries, one must conduct studies in which the patient's ongoing response to the entire regimen can be precisely assessed. Based on the data, our regimen for steroid therapy in patients with severe head injuries is as follows: we begin steroid treatment in all such patients (either dexamethasone 1 mg/kg/day, or methylprednisolone 5 mg/kg/day); on the 3rd day, we evaluate the patient's overall response as determined by our "neuro index;" if the patient has improved, we continue steroids for 7 to 10 days; if there has been no improvement, we discontinue the steroids abruptly. Further studies from multiple centers and with larger numbers of patients are warranted to confirm or disprove these concepts. References 1. Alexander E Jr: Medical management of closed head injuries. Clin Neurosurg 19:240-250, 1972 2. Becker DP: Viewpoint: Severe closed head injury. Neurosurgery 4:277-278, 1979 3. Benson VM, McLaurin RL, Foulkes EC: Traumatic cerebral edema: an experimental model with evaluation of dexamethasone. Arch Neurol 23:179-186, 1970 4. Cooper PR, Moody S, Clark WK, et al: Dexamethasone and severe head injury. A prospective double-blind study. J Neurosurg 51:307-316, 1979 5. Faupel (3, Reulen H J, Mfiller D, et al: Double-blind study on the effects of severe closed head injury, in Papplus HM, Feindel W (eds): Dynamics of Brain Edema. Berlin/Heidelberg/New York: Springer-Verlag, 1976, pp 337-343 6. (3alicich JH, French LA: The use of dexamethasone in the treatment of cerebral edema resulting from brain tumors and brain surgery. Am Pract 12:169-174, 1961 7. (3obiet W, Bock W J, Liesegang J, et al: Treatment of acute cerebral edema with high dose of dexamethasone, in Beks JW, Bosch DA, Brock M (eds): Intracranial Pressure III. Berlin/Heidelberg/New York: Springer- Verlag, 1976, pp 231-235 8. (3udeman SK, Miller JD, Becker DP: Failure of highdose steroid therapy to influence intracranial pressure in patients with severe head injury. J Neurosurg 51: 301-306, 1979 9. Gutterman P, Shenkin HA: Prognostic features in recovery from traumatic decerebration. J Neurosurg 32: 330-335, 1970 10. Itani AL, Ducker TB: Effects of high and low doses of steroids on head injuries. Surg Forum 27:478-480, 1976 l 1. Jennett B, Bond M: Assessment of outcome after severe brain damage. A practical scale. Lancet 1:480-484, 1975 12. Kobrine AI, Kempe L(3: Studies in head injury. Part II. Effect of dexamethasone traumatic brain swelling. Surg Neurol 1:38-42, 1973 13. Langfitt TW: Comment on prognosis of patients with severe head injury. Neurosurgery 4:283-289, 1979 14. Maxwell RE, Long DM, French LA: The effects ofglucosteroids on experimental cold-induced brain edema. Gross morphological alteration and vascular permeability changes. J Neurosurg 34:477-487, 1971 15. Meinig (3, Aulich A, Wende S, et al: The effect of dexamethasone and diuretics on peritumor brain edema: comparative study of tissue water content and CT, in Pappius HM, Feindel W (eds): Dynamics of Brain Edema. Berlin/Heidelberg/New York: Springer-Verlag, 1976, pp 301-305 16. Pitts LH, Kaktis JV: Effect of megadose steroids on ICP in traumatic coma, in Shulman K, Marmarou A, Miller JD, et al (eds): Intracranial Pressure IV. Berlin/Heidelberg/New York: Springer-Verlag, 1980, pp 638-642 J. Neurosurg. / Volume 54/ May, 1981 599

T. G. Saul, T. B. Dueker, M. Saleman and E. Carro 17. Salcman M, Schepp RS, Ducker TB: Calculated recovery rates in severe head trauma. (In preparation) 18. Sparacio RR, Lin T-H, Cook AW: Methylprednisolone sodium succinate in acute cranio-cerebral trauma. Surg Gynecol Obstet 121:513-516, 1965 19. Teasdale G, Jennett B: Assessment of coma and impaired consciousness. A practical scale. Lancet 2:81-84, 1974 20. Tindall GT: Viewpoint: Severe closed head injury. Neurosurgery 4:277-278, 1979 21. Tornheim PA, McLaurin RL: Effect of dexamethasone on cerebral edema from cranial impact in the cat. J Neurosurg 48:220-227, 1978 22. Yen JK, Bourke RS, Nelson LR, et al: Numerical grading of clinical neurological status after serious head injury. J Neurol Neurosurg Psychiatry 41:1125-1130, 1978 Address reprint requests to." Thomas G. Saul, M.D., Division of Neurological Surgery, University of Maryland School of Medicine, 22 South Greene Street, Baltimore, Maryland 21201. 600 J. Neurosurg. / Volume 54/ May, 1981