J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.35.5.698 on 1 October 1972. Downloaded from Journal ofNeurology, , and Psychiatry, 1972, 35, 698-706

Delayed traumatic intracerebral haemorrhage

GOPAL BARATHAM AND WILLIAM G. DENNYSON From the Department of Surgical , University of Edinburgh

SUMMARY Twenty-one out of 7,866 head were complicated by the development of delayed intracerebral haematomata. The age distribution of patients with this condition closely resembled that of patients with subdural haematomata and differed sharply from patients with extradural haemorrhage. This finding, combined with the fact that the two conditions often coexisted, suggests the possibility of similar aetiological factors operating in their production. The producing the was often minor and the larger haematomata appeared to be associated with longer 'asympto- matic' intervals. The neurological deterioration was in most instances clearly the result of an increase in . When possible, angiography followed by definitive was the most satisfactory method of management and multiple burr holes even when combined with needling of the hemisphere yielded unsatisfactory results. The distribution of tended to confirm their traumatic origin. On no occasion was there a vascular abnormality to account for the haemorrhage and, despite the fact that the ages of most patients were in the seventh and eighth decades, the

incidence of degenerative vascular disease was small. Contusional injury causes a local failure of theProtected by copyright. mechanisms that regulate cerebral blood flow. , , and venous congestion pro- duce cerebral hyperaemia which encourages gradual haematoma formation particularly at the sites of injury. This explains not only the situation of the lesions but also the latency between the trauma and their development.

The in severe head injuries with pact. The trend in these patients is one of extensive cerebral contusions, cerebral lacera- recovery, complete or partial, from the con- tions, generalized neuronal disruption, or pri- cussional injury, followed by a deterioration in mary lesions ofthe -stem is notoriously bad. the level of consciousness or the development of Such cases are recognizable by the unrelenting neurological deficits which were initially absent; which exists from the moment of injury. such patients are clinically indistinguishable Therapeutic attention has therefore been directed from those with extradural and subdural towards those patients who develop intracranial haematomata. Though the incidence of such

;space-occupying lesions some time after the cases is small, our interest in them is occasioned http://jnnp.bmj.com/ initial . These tend to be recovering by the fact that they are potentially treatable by from the coma due to the concussional injury removal of the compressing haematoma though, (even if the classical is not always because of destruction of cerebral tissue, treat- present) and to deteriorate secondarily because ment is less likely to be as effective as in those of a rise in intracranial pressure. The lesion situations where the compressive lesions are which most commonly produces this picture is entirely extracerebral.

haem,orrhage into the extradural or subdural Bollinger (1891) originally described the con- on September 30, 2021 by guest. space. The majority of patients with a traumatic dition of delayed intracerebral haemorrhage intracerebral haemorrhage, usually resulting associated with trauma, and proposed the term from severe cerebral lacerations, are profoundly 'traumatische Spat-Apoplexie'. The original unconscious from the time of injury. In a small description and that of subsequent workers minority, however, the intracerebral haemor- (Michel, 1896; Bailey, 1904; Symonds, 1940; rhage appears to occur some time after the im- Grant and Asutin, 1950; McLaurin and McBride, 698 Delayed traumatic intracerebral haemorrhage 699 J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.35.5.698 on 1 October 1972. Downloaded from

1956; Morin and Pitts, 1970) remain unsatisfac- tory for various reasons which will be elaborated in the discussion. A major weakness has been that the term 'apoplexy' does not constitute a definite pathological entity. This series is, there- in which haematomata Total head fore, limited to those cases )5 [ injuries occurred primarily into the substance of the In [ 1960-69 in brain. un 0 E60 0 METHODS E I40 Subdural Patients were selected from the total admissions to z haematoma the traumatic wards of the Department of Surgical II20.° Neurology between the years 1960-69. All persons O Extradural who, associated with trauma to the head, suffered a haemorrhage loss of consciousness, a period of or com- 6 Delayed plained of symptoms referable to the central nervous 3 * intracerebral 0. . 0 0) ) haematoma system were classified as head injuries and were a> ) 0) as 0J) + admitted for observation. In all cases it was possible Cao _-)N4 0D u0D00D to obtain a reasonably accurate idea of the severity _ (s4 Cr) < LA ID r> ofthe trauma and the initial duration ofunconscious- Age in decades ness. Consciousness was considered as being regained FIG. 1. Age distribution of head injuries and of the when it was possible to obtain rational answers to various complicating haematomata. simple questions and all patients who did not achieve this level of lucidity were excluded. Protected by copyright. During the 10-year period 7,866 patients with head fairly evident that the age incidence of extradural injuries were admitted to this department. Intra- with the age incidence cerebral haematomata were diagnosed in 96 either at haemorrhage corresponds operation or necropsy, and ofthese, 21 had, following of all head injuries, the highest frequency being definite trauma to the head, a transient period of in the second and third decades. The distribution recovery ('the asymptomatic interval') followed by a of delayed traumatic intracerebral haemorrhage, secondary deterioration. A detailed study of the case however, more closely resembles that of sub- records and necropsy reports of these 21 patients dural haematoma, the largest incidence being in was made. Initially an attempt was made to divide the seventh and eighth decades. The striking our patients into two groups, one with and the similarity between the age distribution of sub- other without evidence of cardiovascular disease. In dural and delayed intracerebral haematomata, six instances there was no clinical or necropsy evi- and the sharp distinction between the age dence of cardiovascular disease and these would incidence of these two conditions and that of conform exactly to Bollinger's criteria for the diagnosis of 'Spat-Apoplexie'. As these six, how- extradural haemorrhage suggests the possibility ever, differed in no other way from the remaining that similar causal mechanisms may operate in http://jnnp.bmj.com/ 15 it was felt that any distinction between the the production of subdural and delayed intra- groups was arbitrary, if not artificial, and the entire cerebral haematoma. 21 patients were analysed as a single group. NATURE OF HEAD INJURY The severity of the initial head injury was assessed both in terms of RESULTS the duration of coma as well as from the point of

AGE The oldest patient in this series was 83 view of the trauma producing it. Those patients on September 30, 2021 by guest. years old, and though the youngest was 5 years with an initial period of of old, the majority of patients were in the seventh under five minutes were classified as minor; decade. The age distribution of all head injuries those unconscious for up to an hour were was compared with that of extradural haemor- classified as moderate; those unconscious for rhage, subdural haematoma, and delayed trau- over an hour were considered major. Twelve matic intracerebral haematoma (Fig. 1). It is cases were considered to have had minor head 700 Gopal Baratham and William G. Dennyson J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.35.5.698 on 1 October 1972. Downloaded from injuries, and nine moderate ones. In no instance TABLE 2 did the initial coma last for over an hour. Though ASYMPTOMATIC' INTERVAL AND HAEMATOMA SIZE it is impossible to assess the severity of impact scientific accuracy, it was felt that some Duration of Number ofpatients with asymptomatic' interval rough clinical impression of this was possible. Haematoma Haematoma Haematoma The impact was classified as being mild, moder- under 3 cin 4-6 cin over 7 cm ate, or severe, depending on the circumstances 1-5 hr I 0 0 6-11 hr 0 0 1 of the accident. For example, falls from the same 12-24 hr 2 2 1 level were classed as mild, collisions between 1-2 days 0 1 3 3-7 days 0 0 l pedestrians and motor vehicles were deemed 8-14 days 1 1 0 severe, and those between pedestrians and pedal Over 14 days 0 1 0 cycles were considered moderate. In only three instances were the impacts considered severe; eight were moderate, and 10 mild. These findings substantiate the view that intracerebral haemato- slowly developing haematomata would allow mata can occur after a minor head injury caused compensatory mechanisms to occur within the by a trivial accident. intracranial cavity and would eventually reach a greater size than those in which the was ' ASYMPTOMATIC ' INTERVAL This was the interval more rapid. This group would therefore be between the initial recovery of consciousness and associated with a longer 'asymptomatic' interval. the secondary deterioration. The shortest was The wide dispersion of the figures is probably a one and a half hours and the longest 19 days reflection of the multiplicity of factors which (Table 1). When this period was under 24 hours operate in determining the final size of theProtected by copyright. most patients were not strictly asymptomatic and haematoma cavity. The haematomata were divided into two groups according to their size, the smaller having diameters of 4 cm or less and TABLE 1 the larger with diameters greater than this. It appeared (Table 3) that the larger lesions were DURATION OF 'ASYMPTOMATIC' INTERVAL associated with an 'asymptomatic' interval of 'Asynmptomatic' interval Patients 24 hours or more. (no.) 1-5 hr 2 6-11 hr 3 12-24 hr 5 TABLE 3 1-2 days 4 3-7 days 4 'ASYMPTOMATIC' INTERVAL AND HAEMATOMA SIZE 8-14 days 2 Over 14 days I 'Asymptomatic' interval Number ofpatients Haematomna Haematoma http://jnnp.bmj.com/ under 4 cm over 4 cnm Under 24 hr 3 1 had minor as a sequel to Over 24 hr 1 10 for which they were still under observation. They were all, however, well enough to answer simple questions, had no neurological deficits, and did headache of any severity. An attempt not have on September 30, 2021 by guest. was made to correlate the duration of the NEUROLOGICAL DETERIORATION Depression in 'asymptomatic' interval with the size of the the level of consciousness was the commonest haematoma as measured at necropsy. Necropsy manifestation of neurological deterioration and material was used, as estimates of haematoma occurred in 13 patients. On six occasions this size at operation are notoriously inaccurate was sufficiently rapid to produce deep coma (Table 2). It seemed possible that the more before operative intervention. Headache of in- Delayed traumatic intracerebral haemorrhage 701 J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.35.5.698 on 1 October 1972. Downloaded from creasing severity occurred in four patients but that obstruction of the sinus predisposed to- never as a solitary symptom; in three it was wards haematoma formation by causing intra- associated with a reduction in the level of con- cranial venous congestion. In the second patient sciousness and in the fourth with an oculomotor (described in detail later), evacuation of a tense palsy. Only two patients developed a subdural haematoma may have provoked en- and in both it was of fairly sudden onset. In no largement of an intracerebral one. In 15 of the 21 instance did occur. With respect to the patients, however, it was not possible to suggest rarity of pareses and the absence of seizures, the a cause for delayed haematoma formation. This, findings of the present series differ from that of however, does not exclude the possibility of McLaurin and McBride (1956). transient and undetected episodes of hypoxia Deepening coma, increasing headache, and acting as a precipitating factor. oculomotor paresis are indicative of a rising intracranial pressure. In all but one instance-a RADIOLOGICAL STUDIES Ten patients had patient with a cerebellar haematoma-the rise in fractures and six of these subsequently developed pressure was supratentorial. The natural se- haematomata in relation to the fracture sites. In quence of events with an enlarging supratentorial five of these there was evidence of cerebral con- mass is uncal herniation and, if this is not tusion or laceration underlying the fracture site expeditiously and adequately relieved, . implicating the direct or 'coup' mechanism of Eight patients were found at necropsy to have injury. Two patients with posteriorly placed asymmetrical uncal and transfalcine herniation. fractures developed haematomata and con- In five of these an antemortem diagnosis of tusions of the frontal lobes. These were clearly intracerebral haematoma was made and the examples of the 'contre-coup' type of injury.

persistence of herniation at necropsy implies Eight patients were investigated by carotid Protected by copyright. that treatment was either ineffective or tardy. An angiography. In four the haematomata were attempt was made to discover possible causes for satisfactorily located. In three instances angio- the development of haematomata or the expan- graphy revealed subdural collections which were sion of existing lesions. In particular, attention assumed to be the cause of the vascular dis- was directed towards the possibility of hypoxic placements. There is no doubt that the compo- episodes producing cerebral hyperaemia and site nature of the lesion interfered with the enhancing haematoma formation. In four accuracy of angiographic diagnosis. In a single instances with or without an case the intracranial pressure was high enough element of venous engorgement may have to prevent filling of the internal carotid artery occurred. In two this was the result of myo- beyond the level of the siphon. Ventriculography cardial insufficiency, one patient having an was undertaken only once, in a patient with an anginal attack immediately preceding the onset unsuspected cerebellar haematoma. of coma and the other providing at necropsy Though angiography provided the diagnosis evidence of recent myocardial . On in only half the patients thus investigated, one occasion necropsy revealed chronic bron- because of the anterior location of most of the http://jnnp.bmj.com/ chitis, emphysema, and right ventricular hyper- lesions (Fig. 2), it would appear to be the most trophy and, though there was no clinical descrip- suitable diagnostic procedure. It has been tion of a hypoxic episode, both hypoxia and suggested (McLaurin and McBride, 1956) that venous engorgement were both chronically ventriculography is the investigation of choice as present. A fourth patient had at necropsy diffuse it can be combined with exploratory burr holes. anoxic changes of the , kidney, It is not possible to say on the experience of the

and myocardium though no cause for this could present series whether or not this is so. It is on September 30, 2021 by guest. be found. In a further two patients it was likely, however, that in the presence of cerebral possible to suggest a cause of haematoma oedema ventricular cannulation may be difficult formation. The first had a of the and ventricular filling insufficient to be of left lateral sinus underlying a fracture of the diagnostic value. temporal bone. Deterioration occurred about 36 Echoencephalography is attractive because it hours after the head injury, and it is probable is swift, innocuous, and can be easily repeated. A J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.35.5.698 on 1 October 1972. Downloaded from 702 Gopal Baratham and William G. Dennyson

shift of the midline was demonstrated in 10 hemisphere, may provide inadequate exposure patients with supratentorial lesions, while in the for an operative diagnosis. Subdural collections remaining 10 the echo was midline or the of one kind or another were found on 15 examination inconclusive. A midline echo does occasions and possibly confused the picture. On not, therefore, exclude the presence of a haema- eight occasions, in patients with acute subdural toma and must be interpreted in conjunction collections, the coexisting intracerebral lesion with other available data. was overlooked, the neurological deterioration being attributed entirely to the presence of the surface haematoma. Eleven patients were found to have a thin film of subdural blood, three had 2 massive solid haematomas, and in one the collec- Frontal 1 1 tion was of the chronic fluid variety. This patient illustrates several points and will be described in detail.

0 CASE REPORT 00 0. F.B., a 79 year old man, was admitted to our wards on 17 July 1969. Twelve days previously he had fallen while drunk and was admitted to a general Temporal surgical unit. He rapidly regained consciousness and after being observed for three days was discharged home free of any neurological deficits. He remained well for 11 and then was days noticed to be slightlyProtected by copyright. confused. On the following day he became pro- gressively more drowsy and was admitted to the FIG. 2. Distribution of haematomata. Five occurred neurosurgical unit. At the time of admission he at multiple sites. would not obey simple commands but responded to painful stimuli by withdrawal movements which were less vigorous on the left side. Both plantar responses were extensor. Echoencephalography It is not possible to suggest an ideal diagnostic showed a right-to-left shift of 1 cm. Biparietal and procedure for the location of traumatic intra- bifrontal burr holes were immediately made. On the cerebral haematomas for this will depend on right side there was a large fluid subdural collection several factors and must eventually be deter- of dark altered blood which was under considerable mined by the urgency of the clinical situation. It tension. No abnormality was noted on the left side. does appear, however, that carotid angiography The subdural haematoma was evacuated and the is most likely to provide information about the irrigated with saline. Though the gyri were shrunken, indicating an absence of oedema site of the lesion a though the presence of surface and some degree of atrophy, the brain expanded haematoma may confuse the diagnosis. rapidly. A silver clip was attached to the cortex so as http://jnnp.bmj.com/ to provide a radiological guide to the size of the OPERATIVE MANAGEMENT All 21 patients had subdural space in the event of reaccumulation. some form of operative treatment. In seven this After surgery his level of consciousness improved was limited to burr holes and in only two of and he was obeying simple commands. Four hours these was an operative diagnosis made. Fourteen later, however, he rapidly lapsed into coma, respond- underwent either a craniotomy or craniectomy ing to painful stimuli only by decerebrate posturing. Radiographs of the skull showed the silver clip and the diagnosis was made or confirmed at on September 30, 2021 by guest. on the operation on nine occasions. In four of these a abutting inner table of the skull, indicating preoperative had established that the subdural haematoma had not reaccumulated angiogram the and this was confirmed by reopening the burr holes. presence of an intracerebral mass and the Needling the produced darkish blood, craniotomy was entirely elective. The figures, followed by fresh blood mixed with CSF. It was felt however, imply that burr holes, even when that he had suffered a massive periventricular multiple and combined with needling of the haemorrhage and no further treatment was attemp- J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.35.5.698 on 1 October 1972. Downloaded from Delayed traumatic intracerebral haemorrhage 703 ted. He died shortly after surgery and necropsy moment of impact and similar pathological revealed a massive haemorrhage in the right parietal processes presumably operate at all of these lobe with rupture into the lateral ventricle. sites, even if Bollinger's original supposition of The subdural fluid evacuated initially was un- trauma producing a diffuse is doubtedly the result of the head injury 12 days pre- viously. It is possible that sudden removal of the not acceptable to-day. surface collection provoked expansion of the deep haematoma, particularly as the cavity was found at operation to contain both altered and fresh blood. TABLE 4 The finding of a substantial surface haematoma SITUATION OF MULTIPLE LESIONS and the absence of brain swelling undoubtedly led to the deeper lesion being overlooked initially. Whether Sites Patients or not an episode of hypoxia could have occurred (no.) during the period of coma and whether the hyper- R. frontal + R. cerebellar I R. frontal + R. temporal 2 aemia associated with this precipitated further R. temporal + L. temporal 1 haemorrhage must remain a matter for speculation. R. frontal + basal ganglia I Total 5 PATHOLOGICAL FINDINGS Fifteen of the 21 patients died and were subjected to necropsy examination. In every case an assessment was made of the severity and distribution of athero- Two patients had haematomata in the basal matous disease, particular attention being paid ganglia and this finding is generally accepted as to its occurrence in the coronary and cerebral proof of their hypertensive aetiology (Johansson, vessels. A special search was made for 1961). Courville (1962) has, however, described Protected by copyright. and arteriovenous malformations which could symmetrical basal gangliar haemorrhages and account for the haemorrhage. In only one suggested a traumatic aetiology because of the instance was an unruptured of the bilaterality of the lesions. Although both patients anterior communicating artery discovered and with basal gangliar haemorrhage were elderly, in this was not thought to be relevant to the neither was there any clinical or necropsy evi- haematoma located in the parietal lobe. All dence of either atheromatous or hypertensive haematomata were substantial- that is, none disease. In both, the history of trauma was were 'slit' haemorrhages and their size was sufficiently definite to preclude the possibility of expressed in respect of their greatest diameter. a spontaneous hypertensive haemorrhage result- This ranged from 2-5 cm to 15 cm with a mean of ing in incidental trauma, one patient being 6 cm and generally, therefore, formed significant struck by a girder while at work and the other space-occupying lesions. An obvious feature of run down by a car while crossing a street. the distribution of lesions (Fig. 2) is the fre- Further, in neither did a period of , quency with which the frontal and temporal lobes as could possibly result from raised intracranial were involved. This finding is in keeping with pressure, occur before the neurological deteriora- http://jnnp.bmj.com/ that of earlier workers (Courville and Blomquist, tion. A feature of interest is that one of the 1940; Browder and Turney, 1942; Courville, gangliar lesions was associated with an ipsi- 1962). A noticeable difference between the lateral frontal haematoma and that both con- present and previous reports is the absence of ditions appeared to be of the same age at haematomata in the occipital region and the necropsy. The pathological mechanism pro- frequency of multiple lesions which were found ducing these lesions, however, remains obscure. on five occasions (Table 4). Though multiple The two cerebellar haematomata are also of on September 30, 2021 by guest. lesions have been reported (De Jong, 1942) they interest. The first resulted from an occipital blow do not appear to be a common finding (Bollinger, which produced a fracture of the occiput and an 1891; McLaurin and McBride, 1956; Morin and underlying laceration of the . The Pitts, 1970). The infrequency of multiple haema- second was found in association with a frontal tomata in earlier series is surprising, as more haematoma caused by a blow high on the fore- than one area of the brain is damaged at the head. J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.35.5.698 on 1 October 1972. Downloaded from 704 Gopal Baratham anid Willianm G. Dennyson

The association between the haematomata and haemorrhage was not well elucidated at the end or lacerations was studied. On of the last century and it is, therefore, injudicious 10 occasions haematomata were found in to attempt to adhere rigidly to the criteria pro- association with contusions or lacerations. In posed by Bollinger as Morin and Pitts (1970) only two of these was there a possible cause for have done. the neurological deterioration. Of five patients A further problem arises when the interval without obvious superficial focal brain injury it between the trauma and the 'apoplectiform was possible to suggest a cause for haematoma episode' is prolonged. Unless a definite patho- formation on three occasions. This finding logical sequence of events can be established, the suggests that under certain conditions small relevance of the trauma must remain uncertain. traumatic haemorrhages in the can In fact the trauma, particularly when minor, may increase sufficiently to become of clinical have been purely fortuitous and remembered significance. The suggestion that haematomata only because of the development of symptoms not associated with contusional injury have a referable to the head. Further, the relatively high better prognosis than those with such injury incidence of 'Spat-Apoplexie' in the young may (McLaurin and McBride, 1956) could not be merely reflect the fact that this group is more substantiated as only necropsy material was prone to injury than the more senior members of used in separating patients into two groups. the community. Although the tendency to contusional injury Two forms of delayed intracerebral haemor- increases with ageing and degenerative vascular rhage have been described after trauma. The disease, in only eight patients was there evidence first occurs within the substance ofthe brain, and of degenerative changes affecting both the the second primarily within the ventricular coronary and cerebral vasculature. Minckler system. It is well established that blood stainingProtected by copyright. (1971) has suggested the possibility of sudden of the occurs after head increase of size of contusional haemorrhages in injuries (Russell, 1932) and that this can result alcoholics. On two occasions this may have been even after minor trauma (Grahmann and the case. In the first, a known alcoholic, precipi- Schmidt, 1964), though it is difficult to relate tous deterioration occurred 24 hours after traumatic subarachnoid haemorrhage to the injury. In the second, the deterioration, though delayed intraventricular haemorrhage which equally sudden, occurred 12 hours after a fall occurs in association with trauma. Further, it is and was caused by a haematoma 9 cm in diam- not impossible that different pathological mech- eter. This woman, though not an overt alcoholic, anisms may operate in producing ventricular, as was found at necropsy to have evidence of opposed to intracerebral, haemorrhage. This is Wernike's encephalopathy. particularly so as trauma is known to precipitate haemorrhage from arteriovenous malformations (Ameli, 1968). Such lesions produce haemor- DISCUSSION rhage within the CSF pathways and may be Apoplexy is a term of dubious validity as it does difficult to demonstrate at necropsy, especially http://jnnp.bmj.com/ not denote a specific pathological entity. By when they are small. Because of these con- common consent it has come to include all con- siderations this study has been limited to patients ditions caused by '. . . acute vascular lesions of with haemorrhage into the substance of the the brain' (Dorland, 1965). Unfortunately, brain of indubitable traumatic origin. several different diseases fall into this category The mechanism producing cerebral contusions and the tendency to group them all as a single after blunt head injuries has been well studied entity has served only to confuse accurate (Gurdjian, Lissner, Hodgson, and Patrick, on September 30, 2021 by guest. diagnosis. The first of Bollinger's four cases 1966), and Courville (1962) has divided the appeared to have had a subdural haematoma traumatic haematomata so produced into those and the third secondary brain-stem haemorrhages associated with superficial cerebral contusions resulting from raised intracranial pressure due to and those which occurred independently in the an internal . Undoubtedly, the white matter. Though the work of Courville and pathology of the various forms of intracranial Gurdjian explains the occurrence and situation J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.35.5.698 on 1 October 1972. Downloaded from Delayed traumatic intracerebral haemorrhage 705 of the haematoma, it does not explain the ruption of bridging corticodural veins, is an latency between the injury and the development added factor in subdural haemorrhage (Krau- of symptoms. Bollinger's original submission land, Mallach, Mussoni, and Spitz (1962), and that trauma produced progressive cerebral this probably explains the similarity of age softening and that haemorrhage occurred when distribution and the coexistence of the two con- this involved the wall of a is difficult ditions. Injury, whether a contusion or a to prove. The vasoparalytic theory of Evans and laceration, produces the prerequisites for haemor- Scheinker (1946) suffers from the same weakness rhage within the brain substance. The likelihood though both would satisfactorily explain the of this is increased by conditions of hypoxia, period of latency. hypercapnia, or venous congestion. The bleeding It is well recognized that both hypoxia and is rarely brisk, differing in this respect from that hypercapnia can increase cerebral blood flow. of middle meningeal haemorrhage, and probably Cairns (1939) showed that the brain swelling consists of oozing from venules. The larger which occurred with carbon dioxide retention haematoma cavities appear to take a greater and the trauma of retraction was caused by time to develop and are associated with a longer cerebral vasodilatation. More recently, animal 'asymptomatic interval'. Eventually the haema- experiments have demonstrated that trauma toma produces a rise in intracranial pressure produces an increase in regional cerebral blood and may rupture into the subdural space or into flow, not only in the traumatized area but also in the . It is possible that de- the brain surrounding it, and that the hyperaemia compression of a surface collection may provoke in the surrounding brain persists for an in- further haemorrhage into an existing deeper definite period after the trauma (Brock, 1968). It haematoma, though the view of Lafforgue

has further been suggested (Langfitt, Kumar, (1904) that trauma produced bleeding into an Protected by copyright. and Millar, 1971) that the increased arterial already existing spontaneous haematoma cavity pressure consequent upon hypoxia can, because is difficult to prove. Though the operative results of a local loss of autoregulation, be transmitted in this series are not encouraging, the slow to the veins, thus facilitating diapedesis. The development and the absence of neurological demonstration (Langfitt, Weinstein, and Kassell, deficits before those from raised intracranial 1966) that the vasodilatation occasioned by pressure supervene, imply that the condition is carbon dioxide retention may persist even after potentially treatable. Failure to recognize the resumption ofnormal respiration is also relevant. syndrome undoubtedly accounts for the high A combination of factors, therefore, operates in mortality in this series as an antemortem the production of the delayed variety of trau- diagnosis was made on only nine occasions. matic intracerebral haemorrhage. Focal trauma Clinicians have for some time recognized that produces increased flow within and around traumatic haemorrhage into the brain substance traumatized brain. Hypoxia, carbon dioxide can occur some time after a head injury and Lin, retention, and any cause for elevation of venous Cook, and Browder (1958) have stressed the pressure will increase this tendency, thus pro- difficulties involved in distinguishing clinically http://jnnp.bmj.com/ ducing ideal conditions for the gradual develop- between the deep haematoma and the more ment of an intracerebral haematoma. It is now commonly occurring superficial variety. Neither possible to formulate a sequence of events the incidence of this syndrome nor the relation- suggesting aetiological mechanisms and offering ship between its pathogenesis and its clinical a correlation between pathological findings and presentation appears to have been adequately clinical manifestations. explored.

Trauma of a trivial nature can produce on September 30, 2021 by guest. in It is a pleasure to acknowledge our gratitude to marked displacements the atrophic of Professor F. John Gillingham and Mr. John Shaw elderly persons. These result in contusions or for their valuable guidance, to Dr. K. Hermann for lacerations of usually the frontal and temporal her assistance and advice with the German literature, regions but whose exact site will depend on the to Mr. Phillip Harris and Mr. Edward Hitchcock for site of the impact. A similar mechanism can permission to use their cases, and to Miss V. Picton- produce subdural bleeding and, through dis- Phillipps who assisted with processing the data. 706 Gopal Baratham and William G. Dennyson J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.35.5.698 on 1 October 1972. Downloaded from

REFERENCES Johansson, S. H. (1961). Hypertensive and normotensive intracerebral haemorrhage, pp. 90-94. Acta Psychiatrica Ameli, N. 0. (1968). Brain angiomata and intracerebral et Neurologica Scandinavica, 36, Suppl. 150. haematoma. Progress in Brain Research, 30, 427-431. Krauland, W., Mallach, H. J., Missoni, L., and Spitz, W. U. Bailey, P. (1904). Traumatic apoplexy. Medical Record, 66, (1962). Subdurale Blutungen aus isolierten Verletzungen 528-529. von Schlagadern on der Hirnoberflache dur Stumpfe Bollinger, 0. (1891). Ueber traumatische Spat-Apoplexie. Gewalt. Virchows Archiv fur pathologische Anatomie und Ein Beitrag zur Lehre von der Hirnerschutterung. Inter- Physiologie, 336, 87-98. nationale Beitrdge zur Wissenschaftlichen Medicin. Fest- Lafforgue, E. (1904). Hemorragies intra-craniennes trauma- schrift, Rudolf Virchow. Berlin A. Hirschwald. 2, 457-470. tiques evoluant en deux temps. Bulletin Medical (Paris), 18, Brock, M. (1968). Experimental 'luxury perfusion' in the 875-878. cerebral cortex of the cat. Progress in Brain Research, 30, Langfitt, T. W., Kumar, V. S., and Miller, J. D. (1971). 125. Cerebral oedema caused by anoxia and changes in intra- Browder, J., and Turney, M. F. (1942). Intracerebral cranial vascular pressures. In Brain and Blood Flow. Pro- hemorrhage of traumatic origin: its surgical treatment. ceedings of the 4th International Symposium on the Regula- New York State Journal of Medicine, 42, 2230-2235. tion of Cerebral Blood Flow, pp. 386-392. Edited by R. W. Cairns, H. (1939). Raised intracranial pressure: hydro- R. Russell. Pitman Medical: London. cephalic and vascular factors. British Journal of Surgery, Langfitt, T. W., Weinstein, J. D., and Kassell, N. F. (1966). 27, 275-294. Vascular factors in head injury. Contribution to brain- Courville, C. B., and Blomquist, 0. A. (1940). Traumatic swelling and intracranial hypertension, pp. 172-194. , with particular reference to its Edited by W. F. Caveness, and E. A. Walker. Head Injury. pathogenesis and its relation to 'delayed traumatic Conference Proceedings. Lippincott: Philadelphia. apoplexy'. Archives of Surgery, 41, 1-28. Lin, T. H., Cook, A. W., and Browder, E. J. (1958). Intra- Courville, C. B. (1962). Traumatic intracerebral hemorrhages, cranial hemorrhage of traumatic origin. Medical Clinics of with special reference to the mechanics of their production. North America, 42, 603-610. Bulletin of the Los Angeles Neurological Society, 27, 22-38. McLaurin, R. L., and McBride, B. H. (1956). Traumatic Dorland, W. A. N. (1965). Dorland's Illustrated Medical intracerebral . Review of 16 surgically treated Dictionary. 24th edn., p. 120. Saunders: Philadelphia. cases. Annals of Surgery, 143, 294-305. Evans, J. P., and Scheinker, I. M. (1946). Histologic studies Michel, E. (1896). Ein Beitrag zur Frage von der sogenannten of the brain following head trauma. II. Post-traumatic traumatischen Spatapoplexie. Wiener klinische Wochen- petechial and massive intracerebral hemorrhage. Journal of scrift, 9, 789-793.

Neurosurgery, 3, 101-113. Minckler, J. (Editor) (1971). Pathology ofthe Nervous System,Protected by copyright. Grahmann, H., and Schmidt, H. (1964). Liquoruntersuchun- p. 1729. Vol. 2. McGraw-Hill: New York. gen bei frischen gedeckten traumatischen Hirnschaden. Morin, M. A., and Pitts, F. W. (1970). Delayed apoplexy Zentrablatt fur Neurochirurgie, 24, 282-289. following head injury ('traumatische Spat-Apoplexie'). Grant, F. C., and Austin, G. M. (1950). The surgical treat- Journal of Neurosurgery, 33, 542-547. ment of spontaneous and traumatic intracerebral hemor- Russell, W. R. (1932). Discussion on the diagnosis and rhage. American Journal of Medical Science, 219, 237-241. treatment of acute head injuries. Proceedings of the Royal Gurdjian, E. S., Lissner, H. R., Hodgson, V. R., and Patrick, Society of Medicine, 25, 751-757. L. M. (1966). Mechanism of head injury. Clinical Neuro- Symonds, C. P. (1940). Delayed traumatic intracerebral surgery, 12, 112-128. haemorrhage. British Medical Journal, 1, 1048-1051. http://jnnp.bmj.com/ on September 30, 2021 by guest.