Indian J Pediatr 1988; 55:739-748
Head trauma Joseph A. Weinberg
Depap~ment of Emetge/zey, Universityof Tennessee, LeBonheur, Children's Medical Cento; Memphis, Tennessee, USA Epidemiology Head injuries represent a common and serious pediatric problem. Trauma is the leading cause of death in the United States of America in children over one year of age; and cerebral injury is the major cause of morbidity and mortality. Approximately eight hundred thousand children sustain head injuries annually and one hundred thousand require hospitalization. 1 There are four to six thousand deaths and fifteen to eighteen thousand children who require prolonged hospitalization.
Children are different Children respond differently to head trauma than adults. The child's calvarium is thinner, softer, more flexible, and absorbs forces better without breakage. Often, significant intracranial pathology is not associated with overlying skull fractures. The fontanelle and
expanding sutures may protect the brain from some of the effects of swelling and limit epidural hematomas to the dural attachments at the suture line. The child is top heavy, lacks stability and judgement, and therefore is accident prone and inclined to strike his head when involved in trauma. Children often have intense and rapid reactions to trauma and are difficult to evaluate due to limited verbal abilities. As such, careful observation is necessary in evaluating the injured child. Regardless of age or size of the patient, the goal in any head injury is to prevent ~econdary damage to the brain. Primary brain injury incurred at moment of impact cannot be altered. However, it is imperative that further secondary injuries due to hypoxia, hypotensiou, ischemia, or increased intracranial pressure should be prevented with aggressive management.
Epidural hematomas If the patient is awake and alert in the emergency department with either no history of loss of consciousness or a brief period of unconsciousness then the primary injury caused little neuronal
Reprint requests: Dr. Joseph A. Weinberg, Assistant Professor of Pediatrics, Medical Director, Department of Emergency, LeBonheur Children's Medical Center, Memphis, Tennessee 38103, USA. 739
740
THE INDIANJOURNALOF PEDIATRICS
damage. One must then assess the risk of the child developing a lesion, (an epidural or subdural hematoma), which can cause increased intracranial pressure through a mass effect. The major danger would be herniation of the brain with ischemia of the brain stem and ultimately death. Classically epidural hematomas are associated with fractures of the skull across the course of the middle meningeal artery and often follow a relatively trivial blow. This results in the clinical syndrome of uncal herniation. There is an initial loss of consciousness followed by a lucid interval when the patient appears normal. This is followed by increasing headache due to displacement of the dura and large vessels. As the expanding clot displaces brain substance, there is compression of the midbrain and its vascular supply leading to progressive loss of consciousness, unilateral pupillary dilatation (ipsilateral to the hematoma in 80% of cases), and contralateral hemiparesis. At this point, time is critical as there may be only one to two hours before the patient becomes decerebrate and then, as the medulla is progressively compromised, the patient develops bradycardia, apnea, and dies. Unfortunately, children often do not follow this classical pattern. Fifty percent of epidural hematomas in children are not associated with fractures. Bradycardia may be the only sign without any history of loss of consciousness and is always worrisome. If the epidural bleeding is venous in origin, vomiting, papilledema, and headache may not appear for several
Vol. 55, No. 5
days after the trauma. The hematoma may be missed on initial studies and require repeat CT scans in patients with persistent symptoms or secondary clinical deterioration? Subdural hematomas
Subdural hematomas (SDH) in children under one year of age may be chronic or acute and are often associated with child abuse. The infant may be comatose or apneic or may be irritable and vomiting. Retinal hemorrhages are an important clue to the underlying SDH. The pupils may be fixed, dilated, and the infant often has seizures. There may be cutaneous bruises, but often there are no signs of obvious trauma. Acute interhemispheric subdural hematomas are common in the "shaken" baby syndrome. This lesion has been thought to be caused by violent shaking of an infant causing linear and rotational shearing injuries to the brain. However, this mechanism has recently been questioned) In the older child, subdural hematomas are usually associated with acceleration/ deceleration injuries, as in motor vehicle accidents. These are usually associated with loss of consciousness. Rarely there may be a lucid interval but usually these patients have severe injuries with significant cerebral edema. This is followed by progressive deterioration with signs of herniation as described above for epidural hematomas. However, because there is often underlying cerebral laceration or contusion the prognosis is
741
WEINBERG : HEADTRAUMA
less favorable for subdural hematomas than for epidural hematomas even if recognized and operated promptly. Following minor head injury many children will have, in sequence; a period of post traumatic stupor, a period of normal behavior, and a period of lethargy, all known as the pediatric concussion syndrome. 4 The pupils fluctuate in size but remain equal. There may be focal seizures or cortical blindness, a positive Babinski, hypertonia, and vomiting. Usually, the patient recovers with no further progression of symptoms. However, rarely herniation and death has been reported to occur. With more severe trauma, the child remains comatose or has persistent alteration in the level of consciousness. This implies widespread dysfunction of both hemispheres and/or the brain stem and includes: contusions and lacerations, intracranial hematomas, subarachnoid hemorrhage, cerebral edema, and increased intracranial pressure. The CT (computerized tomography) scan has been of great value in evaluating these lesions. Most children with mild symptoms can be observed. However, if there is progression of symptoms, focal neurological deficits or a depressed skull fracture, then a CT scan should be obtained. All comatose patients should be thoroughly studied. Any infant with prolonged loss of consciousness, full fontanelle, bradycardia, or splitting of the sutures should also have a CT scan: If the CT scan shows a significant lesion with a mass effect causing a midline shift,
then surgery is indicated. Rarely a patient may deteriorate so rapidly, that emergency burr holes (or a subdural tap in infants) may be indicated before a CT is obtained. 6 However, it is preferable to obtain a CT first (if possible). Also, one should obtain a CT scan after such an emergency procedure for definitive diagnosis.
Skull X-rays The role of skull X-rays in evaluating head trauma patients is controversial and a number of authors have developed "high yield" criteria to define their usage: The presence of a fracture rarely effects the treatment and course of the pediatric patient, s,9 However, in adults, skull fracture may be a significant marker for subsequent neurologic deterioration, t~ In pediatrics, the major value of these films is detecting an unsuspected depressed skull fracture, documenting fracture in suspected abuse, 13 and following "growing" fractures in infants) a However, primary evaluation, resuscitation, and/or the CT scan should never be delayed to obtain plain radiographs of the skull. Leonidas 9 has set forth criteria as guidelines for obtaining skull X-rays in children. The NIH recognized the special risk in children under the age of two in setting these criteria) 4 Children with fractures that cross the groove of the middle meningeal artery, occipital fractures 15, and basilar skull fractures are at a higher risk for complications.
742
T I l E INDIAN J O U R N A L OF PEDIATRICS
Parental expectations and legal considerations ol~ten lead to the ordering of unnecessary skull films. The parental concerns can be assuaged with a careful explanation of the importance of neurologic evaluation versus skull films. The reason for not obtaining an X-ray can be documented on the chart. CT scan anti MRI
Magnetic resonance imaging (MRI) is currently not practical in the acute setting. Moreover, CT reliably demonstrated hemorrhage in the first three days after injury) 6 MRI may be useful in demonstrating lesions in patients who have persistent neurologic abnormalities, including coma. Emergency department management
If the patient has minimal symptoms or no progression of symptoms, observation is all thzl_ is needed and this can often be done at home. Generally, the child should be admitted lor observation if there was loss of consciousness for greater than fifteen minutes, focal neurologic signs or progressive loss of neurologic function, post traumatic seizures, repeated vomiting, skull fractures, diastasis of the sutures, and inadequate outpatient supervision. The situation is quite different in patients with serious head injuries or who develop mass lesions. The care required to prevent secondary injury to these patients is more extensive. Our goal should be first to prevent and then treat the
Vol. 55, No. 5
ongoing processes in the traumatized brain that will lead to ischemia thus further damaging neurones. Neurones will be injured either by anoxia or ischemia resulting from inadequate delivery of oxygen to meet metabolic demands. Oxygen delivery is determined by the oxygen carried in the blood (a function of hemoglobin concentration and saturation) and cerebral blood flow. One cannot routinely directly measure the cerebral blood flow. Cerebral perfusion pressure (mean arterial pressure - intracranial pressure) can be measured and provides a means of approximating the impact of an injury or a treatment modality on blood flow. Thus control of intracranial pressure (ICP) and support of arterial pressure are key factors in these patients. Bruce has shown that the most common damage in the seriously injured, comatose child is diffuse brain swelling) 8 The predominance of this diffuse injury as opposed to mass lesions in children may be due to the mechanism of injury; motor vehicle accidents causing a much lower incidence of mass lesions. 19 Motor vehicle accidents are by far the most common cause of serious head injury in children, causing approximately two-thirds cases in our own series? ~ Indeed, Gennarelli, et al have stated that outcomes vary in patients with the same GCS score based on the underlying CNS lesion .21 ABCs The first priority is the basic ABCs of life support: airway, breathing, and
743
WEINBERG : IIEAD TRAUMA
circulation. Hypoxia and particularly hypcrcardia cause marked rises in ICP. Brain perfusion, is significantly compromised by hypotension. 6 Only aftcr the airway and circulation are stabilizcd should particular attention be paid to thc CNSproblems. It is worth rcmembering that shock is ahnost never due to head injury alone. The only exception to this rulc is an infant with subgaleal or cpidural blccding and extensive blecding from scalp lacerations. Otherwise, the patient must be evaluated for internal or spinal cord injury to account for the onsct of shock. The possibility of an associated spinal injury should always bc kept in mind while transporting or examining the patient, especially with motor vehicle and swimming accidents. The neck should bc protected by a neck immobilizcr or traction. Signs of spinal injurics include: hypotcnsion; local tenderness in the ncck or spine; flushed, warm, dry skin; flaccid paraplegia cspccially with flaccid rectal sphincter; urinary rctcntion; priapism; diaphragmatic breathing; flexion of forearms on thc chest; and a sensory deficit to pain. The paticnt should bc thoroughly examined for signs of basilar skull fractures, that is, blood or CSF from the nose or ear. Raccoon's sign (periorbital ccchymoses), BaUlc's sign (ccchymoscs over the mastoid bone), or facial nerve palsy. A :nini neurologic exam can bc quickly performed. This consists of CXalnining tile level of consciousness, lhc fimdi, pupils and ~.xtraoccuk~r move--
ments, the corneal reflex and facial grimace, respiratory pattern, motor responses and tone, deep tendon reflexes and plantar responses. By doing this, the clinician can quickly decide if there arc signs of a focal lesion, need for a CT scan, and assign a Glasgow Coma Scale Score (GCS). If the patient is to be transported, the neck and airway must be protected. 22 If possible, the patient should be intubated and the cuff inflated (in children 8 years of age and older) to prcvcnt aspiration. Airway should bc rechecked in the emergency departmcnt, lntubation is pcrformcd if not accomplished in the prchospital sctting as this is critical in paticnts with a GCS < 8, inadequatc ventilation, or difficult to maintain airway. A large borc intravenous linc is placed and blood drawn for complete blood count, coagulation profile, electrolytes, glucose, blood gases, osmolality, and type/cross match blood. Increased intracranial pressure (ICP) Trcatmemt for increased intracranial pressure should be started immediately. The initial important step is to hypervcntilate the patient to keep the pCO z in the 25-30 Tort range, An experienced clinician can perform a rapid sequence intubatkm to prevent a risc in ICP during procedure. Atropine 0.02 mg/kg, pentothal 3-5 mg/kg and succinylcholinc ~ mg//kg or pancuronium 0.t mg/kg arc given to relax the patient. Lidocz~inc, 1.5 mg/kg, given inlravcnously
744
THE INDIAN JOURNAL OF PEDIATRICS
can blunt the rise in ICP. Cricoid pressure is applied to occlude the esophagous and an sndotracheal tube is placed followed bya gastric tube. The use of muscle relaxants provides less traumatic intubation. However, this procedure should be attempted by skilled personnel. Mannltoi 0.5 g/kg and Lasix 1 mg/kg are given if the patient is deteriorating provided this will not compromise the patient's blood pressure. The role of steroids is controversial but they are probably of no benefitS and may be harmful.~ Once the patient is stabilized, a CT scan is obtained to rule out a mass lesion and assess brain swelling. Surgical intervention is performed as indicated. However a surgical lesion is much less common in children than adults. 19 After the scan or operation, the patient should be observed in an Intensive Care Unit. If the circulation is adequate, then euvolemic dehydration is maintained with D 5 1/2 NS at 1/2 to 2/3 maintenance. The patient is positioned with his head up 30 degrees and kept in midline. General supportive care includes prophylaxis for stress ulcers. Dilantin is used for seizure prophylaxis without affecting the neurologic examination. Volatile anesthetics such as halothane should be avoided as they cause cerebral vasodilatation. Nutritional care is vital using either tube feedings or total parenteral nutrition after initial stabilization is complete, as children are subject to the same metabolic stress as
Vol. 55, No. 5
adults. ~ Later in the course, the medical team must watch for common complications such as disseminated intravascular coagulationZT,occult abdominal trauma2s, nonketotic hyperglycemia, diabetes insipidus, the syndrome of inappropriate antidiuretie hormone secretion, a n d seizures. Hypertension can be controlled using Lasix and vasodilators. Blood pressure may be rising in response to an increase in ICP. Therefore, treating hypertension should be done with great caution and only after assuring that ICP is normal. A central venous or pulmonary artery line may be necessary to monitor fdling pressures and cardiac output. ICP treatment
Initial treatment of ICP consists of hyperventilation, paralysis with pancuronium 0.1-0;2 mg/kg/hr, and sedation with morphine, valium or pentothal. Osmolar therapy with mannltoi (0.5-1.0 g/kg) is used initially if the serum osmolality is less than 320. An intracranial pressure monitor is inserted in patients with a GCS of 3 - 5 or in patients with GCS < 7 and shock. A ventricular catheter, subaraclmoid bolt, or epidural monitor is utilized depending on local preference. It is impossible to reliably detect rises in ICP by clinical examination. Continual monitoring allows rational treatment of the patient, detects occult pressure rises, prevents over treatment, allows venting of CSF to treat pressure rises (if a ventricular catheter is used) and allows one to measure the effect
745
WEINBERG : HEADTRAUMA
of other treatment modalities (especially PEEP). ~ The patient is monitored until at least 72 hours post-injury or there has been no rise in ICP above 20 for twentyfour hours. Early in the patient's course, any rise above 20 mm Hg should be treated unless the rise is transient with procedures, for example, suctioning. All painful procedures should be performed with the patient maximally sedated. Acute rises in pressure are treated with manual hyperventilation. Mannitol 0.25 g/kg, pentothal 2-5 mg/kg IV, valium 0.2 mg/kg, or Lasix 1 mg/kg can also be utilized3~ If conventional therapy fails, then high dose barbiturate therapy, hypothernfia, or radical surgical decompressions can be considered. 31 However, the efficacy of these extreme measures in the head trauma patient is questionable. 32z3 The outcome Several studies in adults 3z-u have shown that the use of early intubation, CT diagnosis, intracranial pressure monitoring, and aggressive control of increased ICP can effect a significant improvement in morbidity and mortality in severe head trauma cases. The establishment of a similar aggressive comprehensive approach to pediatric trauma has also been reported to improve outcomeY ~6 However, these studies have been criticized as being uncontrolled and unblinded. However, a prospective study at the Medical College of Virginia has
demonstrated improved outcomes in patients with treatable increased ICP. This study also demonstrated siotmificantly better outcomes in children than adults with severe head'injuries, especially with diffuse brain injuries. 19 Outcome can be correlated 37with almost all deaths occurring in the 5% of children with GCS greater than or equal to 833 The fatality rate in these children with severe injuries has varied from 9% ~ to 59%. 39 While children have less mortality and disability than adults with severe head injury 19, the injuries still represent major public health problems all over the globe)9 Severe head injuries cause lasting cognitive defects and emotional disturbances in children. 4~ There has been concern about the sequelae of minor head injury since the report of persistent defects in adults by Rimel. 41 These impairments have been reported to resolve by three month# 2 and to correlate w:,L lesions found on MRI scans?3 Similar studies are needed in children as minor head injuries represent 90 to 95% of pediatric head trauma. Summary Society continues to pay a high price for the toll taken by head trauma. Attention must be paid to preventing these injuries, especially involving motor vehicle accidents and bicycle accidents. Current treatment of serious head injuries demands skilled airway management in the field; rapid transfer to Pediatric Trauma Centers; meticulous attention to
746
TI IE INDIAN JOURNAL OF PEDIATRICS
oxygenation, ventilation, and systemic perfusion during the resuscitation phase; rapid identification and early surgical m a n a g e m e n t of m a s s lesions; and aggressive control of intracranial pressure. Our hopes lie in better programs for prevention, improved understanding of the dylmmics of the ongoing neurologic injury both at the organ system and cellular levels, and improved programs in neurorehabilitation.
References 1. Fife D. Head injury with and without hospital admission: Comparisons of incidence and short-term disability. AJPI1 1987; 77:810-812 2. Milo R, Razon N, Schiffer J. Delayed epidural hematoma, a review. Acta Neurochh" 1987; 84:13-23 3. Duhaime AC, Gennarclli TA, Thibault LE, Bruce DA, Margulics SS, Wiser R. The shaken baby syndrome: a clinical pathological, and biochemical study. J geurosttzg 1987; 66:409-415 4. Snoek ,IW, Minderhoud SNI, Wilmink JT. Delayed deterioration following mild head injury in children. Brahl 1984; 107:15-36 5. Bruce DA, Schut L. The value of CAT scanning following pediatric head injury. Clin Pediatr 1980; 19:719-725 6. Andrews BT, Levy ML, Pitts LH. Implications of systcmic hypotension for the neurological examination in patients with severe head injury. Sulg Nettrol 1987; 28: 419422 7. McClcan PM, Joseph LP, Rudolph H. Skull film radiography in the management of head
Vol. 55, No. 5
trauma.Ann Emerg Med 1984; 13:607-611 8. Harwood-Nash DC, Hendrick EB, Hudson AR. The significance of skull fractures in children. Radiology 1971; 101:151-155 9. Leonidas JC, Ting W, Binkiewicz A, Vaz R, Scott RM, Pauker SG. Mild head trauma in children: when is a roentgenogram necessary. Pediaoics 1982; 69:139-143 10. Jones ,iJ, Jeffreys RV. Relative risk of alternative admission policies for patients with head injuries. Lancet 1981; ii: 850-853 11. Fischer RP, Carlson ,i, Perry .IF. Postconcussive hospital observation of alert patients in a primary trauma center. J Trauma 1981; 21:920-924 12. Dacey RG, Alves WM, Rimel RW, Winn HR, Jane ,IA. Neurosurgical complications after apparently minor head injury. J Nettrosttlg, 1986; 65:203-210 13. Meservy C,I, Towbin R, McLaurin RL, Myers PA, Ball W. Radiographic characteristics of skull fractures resulting from child abuse. Am J Roent 1987; 149: 173-175 14. Masters SJ, McClean PM, Arcarese JS et al. Skull X-ray examinations after head trauma. Recommendations by a multidisciplinary pancl and validation study. New Engl J Med 1987; 316:84-91 15. Young HA, Schmidck HH. Complications accompanying occipital skull fractures. J Trauma 1982; 22:914-920 16. Consensus Conference. Magnetic resonance imaging. J Am Med Assoc 1988; 259: 21322138 17. Wilbergcr ,IE, Decb Z, Rothfus W. Magnetic resonance imaging in cases of scvcre hcad injury. Nettrosulgety 1987; 20: 571-576
WEINBERG: HF_.ADTRAUMA
18. Bruce DA, Alavi A, Bilaniuk L c t al. Diffuse cerebral swelling following head injuries in children: the syndromes of "malignant brain edema". ]Neut~sung 1981; 54:170-178 19. Alberico AM, Ward JD, Choi SC, Marmarou A, Young HF. Outcome after severe head injury: rclationship to mass lesions, diffuse injury, and ICP course in pediatric and adult patients. J NeUl'OSttng 1987; 67:648-656 20. Bysani GK, Stidham GL, Wcinbcrg JA. Spectrum of severe pediatric multiple trauma. PediatrRes 1988; 23: 229A 21. Gennarelli TA, Spiclman GM, Lagfitt TW et al. Influence of the type of intracranial lesion on outcome from scvcre head injury. A multicentcr study using a new classification. J Netuvsttng 1982; 56:26-32 22. Gentleman D, Jcnnctt B. Hazards of interhospital transfer of comatose head-injured patients. Lancet 1981; 11:852-855 23. Braakman R, Schoutcn H J, Blaauw-van Ellishoeck M et al. Mcgadose steroids in severe he~d injury. J Nettrosttng 1983; 58: 326-330 24. DeMaria EJ, Reichman W, Kenncy PR, Armitage JM, Gann DS. Scptic complications of corticosteroid administration after central nervous system trauma. Am~ Sung 1985; 202:248-252 25. Ford EG, Jennings LM, Andrassy RJ Steroid treatment of head injuries in children: the nutritional consequences. Cunent Sungety 1987; July-August: 311-313 26. Phillips R, Ott L, Young B, Walsh J. Nutritional support and measured energy expenditure of the child and adolescent with head injury, l Neurosung 1987; 67:846-851 27. Miner ME, Kaufman HH, Graham SH, Haar FH, Gildenberg PL. Disseminated intravascular coagulation fibrinolytic syndrome following head injury in children: frequency and prognostic implications. J Pediatr 1982; 100:687-691 28. Butterworth JF, Miller JD, Maull KI, Becker DP. Detection of occult abdonfinal
747
trauma in patients with severe head injuries. Lancet 1980; ii: 759-762 29. Shapiro HM, Marshall LF. Intracranial pressure responses to PEEP in head-injured patients. J Ttruuna 1978; 18:254-256 30. Marshall LF, Smith RW, Rauscher LA, Shapiro HM. Mannitol dose requirement in brain-injured patients. J Neulosung 1978; 48: 169-172 31. Marshall LF, Smith RW, Shapiro HM. The outcome with aggressive treatment in severe head injuries. J Neutosung 1979; 50:20-25 32. Miller JD, Butterworth JF, Gudeman SK et al. Furthcr experience in the management of severe head irjury. J Neungsung 1981; 54: 289-299 33. Ward JD, Baker DP, Miller D et al. Failure of prophylactic barbiturate coma in the treatment of severe head injury. J Neurosung 1985; 62:383-388 34. Bowers SA, Marshall LF. Outcome in 200 consecutive cases of severe injury treated in San Diego County: A prospective analysis. Neusostu~,ety 1980; 6:237-242 35. Colombani PM, Buck JR, Dudgeon DL, Miller D, Hailer JA. One year experience in a regional pediatric trauma center. J Pediatr Stug 1985; 20:8-13 36. Kaiser G, Pfenninger J. Effect of neurointensive care upon outcome following severe head injuries in childhooda preliminary report. Neutopediaoics 198~; 15:68-75 37. Wigstyl J, Sutcliffe A J, Alpar EK. Early prediction of outcome following head injury in children. J Pediatr Sung 1987; 22:127-129 38. Bruce DA, Raphaely RC, Goldberg AI et al. Pathophysiology, treatment, and otttcome following severe head injury in .children. Childs Btr~bz 1979; 5:174-191 39. Kraus JF, Fife D, Conroy C. Pediatric brain injuries: The nature, clinical course, and early outcomes in a defined United States population. Pediatrics 1987; 79:501-507 40. Filley CM, Cranberg LD, Alexander MP, Hart EJ. Neurobehavioural outcome after
748
THE INDIAN JOURNAL OF PEDIATRICS
dosed head injury in childhood and adolescence.Arch Neurol 1987; 44:194-198 41. Rimel RW, Giordan B, Barth JT et al. Disability caused by minor head injury. Neurosurgay 1981; 9:221-228 42. Levin HS, Mattis S, Ruff RM et al. Neurobehavioral outcome following minor head injury: A three-center study. J
Vol. 55, No. 5
Neurosurg 1987; 66:234-243 43. Levin HS; Amparo E, Eisenberg HM et al. Magn e.t,i'r resonance imaging and computerhzed tomograph in relation to the neurobehavioral sequelae of mild and moderate head injuries. J Neurosurg 1987; 66:706-713
STRATEGIES FOR CONTROL OF MALARIA WITHINA PRIMARY ttE A L T H CARE PROGRAMME Two drug strategies for the control of malaria in children aged 3-59 months have been compared in a rural area of the Gambia treatment of presumptive episodes of clinical malaria with chloroquine by village health workers, and treatment combined with fortnightly chemoprophylaxis with 'Maloprim' (pyrimethamine/dapsone) which was also given by village health workers. Treatment alone did not have any significant effect on mortality or morbidity from malaria. In contrast, treatment and chemoprophylaxis reduced overall mortality in children aged 1-4 years, mortality from probable malaria, and episodes of fever associated with malaria parasitaemia. A high level of compliance with chemoprophylaxis was obtained and no harmful consequences of chemoprophylaxis were observed. Abstracted from : Greenwood BM et al, Lancet 1988; 1 : 1121-1127