Child's Nerv Syst (1992) 8:411-416 m NS �9 Springer-Verlag 1992

Management of brain abscess in children: review of 130 cases over a period of 21 years

]smail H. Tekk6k and Aykut Erbengi

Department of Neurosurgery, Hacettepe University School of Medicine, Sihhiye, Ankara 06100, Turkey

Received October 25, 1991/Revised April 26, 1992

Abstract. The data on 130 children with brain abscesses treated over 21 years (1970-1990) were analyzed retro- spectively. The whole group included four infants. Chronic ear infection and cyanotic congenital heart dis- ease were the most common predisposing factors. In in- fants, meningitis and/or ventriculitis were dominant in the etiopathogenesis. Cases were evaluated according to the treatment received and also according to time peri- ods. More than half of the patients (n = 74) in this series were treated by primary or secondary excision. Comput- ed tomography (CT) facilitated the diagnosis and helped the planning of treatment. Aspiration gained increasing credit after the advent of CT. Microorganisms could be identified in 54% of the cultured specimens. Staphylococ- ci, streptococci and Proteus were the dominating mi- croorganisms. Penicillin and chloramphenicol have long been the mainstay of antimicrobial therapy but have re- cently been replaced by third-generation cephalosporins and sulbactam-ampicillin combinations. Overall mortali- ty was 15.5% but showed a decline from 30% in the pre-CT era to 6% in the last 5 years and to zero in the last three. Neither the location nor associated heart disease contributed to the mortality, but mortality among infants was as high as 50%.

Key words: Brain abscess - Children - Congenital heart disease - Infant - Infection - Mortality

The introduction of computed tomography (CT) and de- velopment of more effective antibiotic regimens have cer- tainly contributed to the reduced mortality from brain abscess [2, 3, 5, 8, 10, 11, 14, 17, 22]. Mortali ty from brain abscess was up to 61% in the preantibiotic era [2], which was reduced to 3 0 - 4 0 % after antibiotics came into use [4, 13] and to about 10% after the advent of CT [5]. A mortality as low as 5% has been reported only recently [3, 101.

Correspondence to: I.H. Tekk6k

Since the first operation for a brain abscess by Morand in 1768, methods of surgical intervention have also evolved [19]. Tube drainage, marsupialization, and the migration method of Kahn are now of historical in- terest. Excision, aspiration, and the open evacuation technique of Maurice-Williams [11] are the currently used techniques, and only recently stereotactic methods have been used for aspiration purposes [10].

Children with brain abscess appear to be an exception- al group, since the prevalence of the predisposing factors differs from that among adults, and the regenerating ca- pacity of the developing brain may contribute positively to the neurological outcome of the disease. The purpose of this paper is to analyze the epidemiology, symptom- atology, bacteriology, surgical features, and the long- term outcome of patients under the age of 18 treated at the department of Neurosurgery, Hacettepe University Medical School, over a 21-year period, which encompass- es periods before and after the availability of CT scan- ning.

Materials and methods

One hundred and thirty cases of brain abscess in children (aged under 18 years) treated at our institution between January I, 1970, and December 31, 1990, have been retrospectively studied. During this time 80 adult patients with brain abscess appear in our records, making a total of 210, out of which childhood abscesses constituted 62%. During the same period a total of 3278 patients of all ages were treated for space-occupying lesions, the abscess cases making up 6.4% of this number.

The files, X-rays or X-ray reports, treatment charts, operating notes and laboratory reports were reviewed. CT became available at our institution at the end of 1975, so the study period was divided into time periods accordingly.

Inclusion/exclusion criteria were as follows:

1. All patients were admitted to or transferred to and treated in the neurosurgery department.

2. Patients with subdural empyema with infection extending intra- cerebrally were excluded.

3. In all patients the diagnosis was reached before death.

412

4. The presenting neurological state was not taken into account and all patients with "brain abscess" on their operating notes were included.

5. "Operative mortality" was defined as death while the patient was hospitalized following surgery.

6. For the nonsurgically treated patient(s) the criterion was the CT s c a n .

Results

Epidemiology

Table 1. Main routes of infection (n= 130)

Route Specific origin n %

Direct spread Mastoiditis-otitis 28 21.5 (n = 50) Sinusitis-cellulitis 3 2.3

Sinus thrombosis 2 1.5 Meningitis 6 4.6 Penetrating head injuries 11 8.5

Septic embol i Congenital heart disease 26 20.0 (n = 28) Lung (empyema) 1 0.7

Immune deficiency (IgA) 1 0.7

Undetermined 52 40.0

Table 2. Congenital heart diseases in 26 cases of brain abscess

Cardiac pathology n

Tetralogy of Fallot 16

Ventricular septal defect Alone 2 With pulmonary stenosis 1 With atrial septal defect + transposition of great 1

arteries + tricuspid atresia With atrial septal defect + pseudotruncus 1

Double outlet right ventricle 1

Transposition of great arteries Alone 1 With pulmonary stenosis 1

Atrial septal defect With single ventricle 1 With tricuspid insufficiency + pulmonary stenosis 1

The number of pediatric abscesses seen per year over the entire period ranged between 2 and 14 (mean 5 per year), with 20 cases prior to the advent of CT and 110 after.

The youngest patient was 2 months old. There were four infants (less than 12 months of age) in the whole series (3%). Of 130 cases, 70 were male and 60 female. Forty-five percent overall were aged 3 - 8 years and 30% were in the 6- to 8-year age group. The incidence of brain abscess decreased between the ages of 8 and 18 years.

Predisposing factors

The most common predisposing factor was adjacent cra- nial infection (mastoiditis, otitis, sinusitis, cellulitis, and sinus thrombosis, 33 patients), followed closely by cyan- otic congenital heart disease (CCHD; 26 patients) (Table 1). Among the 26 patients with CCHD, tetralogy of Fallot was the most common pathology (Table 2). Other predisposing factors were penetrating head injury (11 pa- tients), meningitis (6 patients), pulmonary infection (empyema; 1 patient). Only one patient in this series had a documented immunodeficiency. No predisposing fac- tor could be identified in 52 patients.

Total 26

Table 3. Presenting symptoms (n = 130)

Symptoms n %

Presenting symptoms and signs

The most common presenting symptom was headache, occurring in 61 patients. Hyperpyrexia occurred in 39 patients and seizures in 27 patients (Table 3).

Neurological examination was completely normal in 10, while 9 children were comatose on presentation. Pa- pilledema was a finding in 75 children (61%). It occurred in 55% of patients with a supratentorial and in 76% of patients with an infratentorial abscess.

Location of the abscess

The abscess was supratentorial in 113 cases out of 130. The frontal and temporal lobes were involved in 55%. Of 28 orogenic abscesses, 13 were located in the supra- and 1 t in the infratentorial compartment, while 4 appeared in both. In children with CCHD the parietal lobes were most commonly involved. There was a striking left-sided predilection (77%) in this subgroup. Seventeen children harbored multiple abscesses (13%).

Headache 61 46.9 Nausea-vomitting 47 36.1 Fever 39 30.0 Seizures 27 20.7 Weakness on one side 25 19.2 Mental status change 25 19,2 Speech disturbance 7 5.3 Squint 5 3.8 Facial asymmetry 3 2.3 Unsteadiness 3 2.3

Table 4. Surgical treatment (n = 129)

1st Operation 2nd Operation 3rd Operation Total as as as cured by

Aspiration 2 Aspiration 22 "~

Aspiration 71 /~ "~ Excision 3 "~ Excision 13

Excision Aspiration 1 -

58 /~ N Excision 1

Aspiration 55

Excision 74

413

Table 5. Cultured microorganisms. Oto., orogenic; CCHD, cyanotic congenital heart disease; Men., meningitis; Head, head injury; Sin., sinusitis

Bacteria Route of infection

Oto. CCHD Men. Head Sin. (n = 28) (n = 26) (n = 6) (n = 11) (n = 3)

Staphylococcus aureus 19 6 Proteus mirabilis 5 5 Anaerobic Streptococcus 5 1 B-hemolytic Streptococcus 4 A-hemolytic Streptococcus 1 Streptococcus pneumonia 1 1 Klebsiella 1 Pseudomonas aeruginosa 1 1 Enterobaeter 1 1 Salmonella 1 Various coliform bacteria 1 Bacteroides fragilis 2

4 3

Total 41 17 5 2 4 3

Cultivation rate (%) 60.7 19.2 33.3 36.3 100.0

100

, ~ �9 Aspiration --o-- Excision /

- - / / 80 - " ~ - . ~ .

4o \\ 0

0 i 1970-1975 1976-1980 1981-1985 1986-1990

Periods Fig . 1. E v o l u t i o n o f t h e s u r g i c a l t e c h n i q u e

Surgery

All patients but one (n= 129) underwent one or more surgical intervention(s). The initial surgical procedure was aspiration in 71 and excision in 58 patients (Table 4). The size of abscess cavity was monitored by repeated X-rays taken after injection of a contrast material suspen- sion (Pantopaque) during initial aspiration of the pus. CT scanning and successive radionuclide studies were mainly used for this purpose after 1976.

Nearly half (n = 35) of the patients initially treated by aspiration needed a second operation, whereas only two patients (3.4%) initially treated by excision needed fur- ther surgery. Overall, 74 patients underwent primary (n = 58) or secondary (n = 16) excision, while in 55, aspi- ration alone (once or more) was the mode of treatment.

An analysis of surgical preference according to time periods shows a trend towards aspiration from 10% dur- ing the pre-CT era to 90% during the last 5-year period (Fig. 1).

Two patients with supra- (2/113) and 2 patients with infratentorial (2/17) abscesses needed CSF diversion, while mastoidectomy by the ear, nose, and throat sur-

T a b l e 6. M e d i c a l t r e a t m e n t (n = 130)

Medication Time periods and % of cases

1970- 1976- 1981- 1986- 1975 1980 1985 1990 (n=20) (n=50) (n=28) (n=32)

C o r t i c o s t e r o i d s 30 4 0 68 72 Pen i c i l l i n 75 78 82 62

C h l o r a m p h e n i c o l 70 74 85 53 G e n t a m y c i n 4 0 22 10 9 A m i k a c i n - 19

M e t h i c i l l i n 10 18 7 3 M e t r o n i d a z o l e - - 7 4 0

S u l b a c t a m a n d - - - 31 a m p i c i l l i n

C e p h a l o s p o r i n - 6 11 25 O r n i d a z o l e - - - 31

geons was performed after abscess surgery in 16 out of 28 patients with mastoiditis and/or otitis.

Bacteriology

Microorganisms were identified in 41 of 76 cultured spec- imens (54%). In 9 patients the culture revealed growth of more than one organism. Pus smear was the only identi- fication method in 12. The records revealed no data in 42 patients. The most common organisms were Staphylococ- cus spp. (19 patients), aerobic and anaerobic Streptococ- cus spp. (i1 patients), and Proteus mirabilis (5 patients; Table 5). The culture rate of otogenic pus was markedly higher than that of CCHD (17/28 vs. 5/26). All Proteus growing cultures were of those from the otogenic abscess- es. Anaerobic streptococci, coliform bacteria, and Bac- teroides grew from cardiogenic abscesses.

414

Table 7. Surgical mortality related to treatment (n= 129). Op., Operated; *, not significant (P>0.05) (Fisher's e x a c t )~2 test)

Surgical technique Before CT After CT Overall

Op. Dead % Op. Dead % Op. Dead %

Aspiration 2 0 0 53 8 15.0 55 8 14.5 *

Excision 18 6 33.3 56 6 10.7 74 12 16.2"

Total 20 6 33.3* 109 14 12.8" 129 20 15.5

Table 8. Surgical mortality in relation to abscess location, systemic disease, and number of abscesses. Z z test, P > 0.05, not significant

n Mortality %

Location Supratentorial 112 17 15 Infratentorial 17 3 18

Systemic pathology Cong. heart disease 26 4 15.3 Rest 103 16 15.5

Multiple abscess 16 2 12.5 Single abscess 113 18 15.9

Antibiotic therapy

Penicillin (97 patients, 75%) and chloramphenicol (92 patients, 71%) were the mainstay of antimicrobial thera- py over the 21-year period. Methicillin gained some pop- ularity between the years 1976 and 1980, but lost it again soon after cephalosporins came into use. Metronidazole and ornidazole were actively included in the regimens after 1985. A cefotaxim and sulbactam-ampicillin combi- nation was widely used in recent years (Table 6). Before 1980, penicillin and streptomycin were often instilled into the abscess cavity.

Use of steroids

Corticosteroids were given to 68 patients (52%), mainly for the control of edema, which was reduced to nil over 5 -8 days. The use of corticosteroids did vary significant- ly in the different time periods: while 37% received dex- amethasone in first half of our study period (1970-1980), the number increased to 70% during the second half (1981-1990; Table 6). Steroids did not contribute nega- tively to mortality figures; 62 of the 110 survivors (56%) were given steroids, whereas only 6 of the 20 dead (33%) received such treatment.

Surgical mortality

Eight of 55 patients who underwent aspiration(s) as the sole treatment died (Table 7). Two of these presented with a Glasgow Coma Scale score (GCSS) of 3 and another two with GCSS of 6-7. Of the remaining, three had CCHD while another was an infant aged 2 months.

Overall mortality was 15.5% (20 patients). When the mortality rates were compared by time period, the rate had fallen from 30% in the pre-CT era to 6% for the last 5-year period (Fig. 2). The difference between the mortal- ity rates of the pre- and post-CT era was not statistically significant (Fisher's exact Z 2 test), but the numbers for the pre-CT era are too small for conclusions to be drawn. There have been no deaths (0%) among 21 patients treat- ed over the last 3 years.

Two of four infants died (50%). Table 8 outlines the statistical analysis of surgical mortality according to lo- cation, presence of systemic disease, and multiplicity of cavities (all with Z 2 test). The location of the abscess did not make a statistically significant contribution (P > 0.05) to the mortality rate (Table 8). Similarly, surgi- cal mortality was recalculated for the congenital heart disease group and such heart disease was not found to be a significant contributory factor (P > 0.05). The mortality rate for patients with multiple abscess was again com- parable with that of the whole group, without a statisti- cally significant difference (P>0.05).

Length of hospital stay

In the aspiration group the mean hospital stay was 54 days, markedly longer than in the excision group (23 days).

Follow-up

Of the 110 patients discharged, 69 patients (63%) were available for follow-up during a period ranging between

35

3O

o~" 25

>" 20

10

I I 1970-1975 1976-1980 1981 - 1985 1986-1990

Periods

Fig. 2. Surgical mortality according to 5-year time periods

415

6 months and 16 years. Twenty-one were followed for 3 years or more.

Long-term follow-up was evaluated according to the sequelae and the surgical treatment received. Patients available for long-term follow-up and treated by aspira- tion(s) only (n = 5) were all neurologically normal with- out clinical epilepsy at a mean follow-up time of 4 years, while only 50% of the children treated by primary or secondary excision (n = 16) appeared neurologically nor- mal, with a 31% rate of epilepsy.

Discussion

Brain abscess, in spite of all the advances in diagnostic procedures, surgical techniques, and antimicrobial thera- peutics, is still a life-threatening condition. Considering their susceptibility to infections and trauma as well as the natural prevalence of CCHD in children, this age group deserves a separate evaluation to uncover the specific conditions of the pathogenesis and outcome.

There was a striking peak of occurrence of a brain abscess between the ages of 4 and 8 years, and a steeper peak between 6 and 8 years, consistent with that in the literature [7, 12]. The mean age for different etiologies, however, varied in our series, it was 8 for the CCHD group and l l for children with otogenic abscess.

We, like others [2, 3, 4, 14, 22], found chronic otitis media leading to mastoiditis to be the most common predisposing factor (21%). The occurrence of otogenic abscesses showed a slow but gradual decrease over a 21- year period [13, 14].

CCHD as a predisposing factor appeared to occur only in 4 - 5 % of other series [2, 3, 6, 8, 13], except for two with 17% [9] and 26% [12], respectively, which accords with our figure of 20%. However, CCHD seemed to show an increase as a predisposing factor in the present series (10% during the first half versus 34.4% during the second half). Of 26 children with CCHD, 6 had under- gone previous cardiovascular surgery but an abscess de- veloped between 2 months and 9 years after the opera- tion. A palliative shunt procedure like the Blalock-Taus- sig, which increases the life expectancy of patients with CCHD, does not seem to affect the possibility of develop- ment of a brain abscess [7]. The fact that the left common carotid artery arises directly from the aortic arch may be held responsible for the left-sided predilection of abscess- es in our CCHD group, since the organism from the cardiopulmonary circulation will enter the cerebral circu- lation in a more direct way, avoiding a subclavian-com- mon carotid bifurcation as in the right.

Meningitis-ventriculitis appeared as a cause unique for the pediatric age group and this especially holds true for three out of four infants in our series. The proportion of children with no predisposing factor - 40% - may reflect the child's vulnerability to subclinical infections, leading to hematogenous dissemination without the fac- tor being realized initially.

Headache was the major presenting symptom in ap- proximately half of the patients, consistent with literature data for all ages. What is striking was the fact that hyper-

pyrexia was the presenting symptom in only 30% of the children. This may suggest a rather slow or inactive active host reaction of the children to the infection. The occur- rence of seizures in our series (21%) was in accordance with that of the literature [14].

Relatively few children were admitted and treated in the pre-CT era, but these were included for comparison. There is no doubt that CT has revolutionized neurosurgi- cal practice for brain abscesses. CT allows early diagnosis (even the stage of the infective process), more accurate localization than angiography or isotope scans, and it can show multiple lesions. The detection of more cases with multiple abscesses was only made possible by CT [9]. CT also offers us immediate follow-up.

CT not only emerged as a diagnostic tool but also served us as a guide for treatment. With the availability of CT scans the role of aspiration has increased, as Ka- gawa et al. [7] once predicted. Certainly there are risks involved with aspiration techniques. Rupture of the ab- scess into the ventricle or leakage into the subarachnoid space should be considered [11]. A less likely risk is an intracerebral hematoma. The advantages of excision over aspiration are the immediate eradication of the intracra- nial sepsis and shortened postoperative medical treat- ment and length of hospital stay [11].

Though there was no statistical difference in the over- all mortality associated with the two methods in our se- ries or in others [10, 13, 22], it is necessary to mention that the zero mortality over the last 3 years (n = 2~) is related only to treatment by aspiration, and that long-term se- quelae are significantly lower in the aspiration group.

The condition of the patient on admission was one major factor determining the outcome: a view shared by many [13, 14]. Five of nine children who presented in a comatose state died, while only one of ten alert children could not be saved. Factors like use of steroids, location, or multiplicity of the abscess or associated CCHD that may have contributed to mortality were investigated, but no such contribution could be deduced. Mampalam and Rosenblum [10] attributed four of their eight deaths to CCHD, stating that these patients have marked and chronic hypoxemia and are vulnerable to further alter- ations in oxygenation or in fluid balance. Raimondi et al. [16] shared the same view with a mortality rate of 47.4%. This fact partially holds true for our three patients with CCHD, who were treated by aspiration to minimize the general risk. This means that it is not that the aspiration caused their death, but that their poor state allowed only treatment by aspiration [17]. We still believe that, even in patients with CCHD, a surgical specimen should be sought. Another point, as suggested by Beller et al. [2] and later by Mampalam and Rosenbhim [10], is the theo- retical contribution of steroids to a higher mortality; in actual fact its effect appeared to be the reverse of this, and dexamethasone seemed to save more lives in our series [15].

The only patient in this series who was treated nonsur- gically was a female who had thalamic and periventricu- lar multiple abscesses and had a favorable outcome after 12 weeks of medical treatment. Nonsurgical therapy may be initiated in patients with a stable neurological state

416

who harbor deeply seated, multiple abscesses, especially if the clinician has easy access to a CT scanner [9]. The medical approach should not subrogate the role of neuro- surgical management [18].

Microorganisms were identified in just over half of the cultured specimens, which is a quite low rate for a refer- ence centre like ours, but considering the fact that 20 out of 35 patients with no growth received preoperative an- tibiotics during their initial stay in pediatric wards, a sterile culture rate of 46% seems justifiable. With newer microbiological techniques like selective cultivation and prolonged anaerobic incubation, a 90% positive culture rate can be achieved [3, 6]. Szuwart and Bennefeld con- cluded that the occurrence of anaerobic bacteria is matched by a decreased occurrence of sterile suppuration [20].

The bacteriological spectrum of this series was not different from many of the reviews in the literature [2, 4, 6, 10, 14, 19]. Staphylococci (46%), streptococci (15%), and Proteus (12%) were the predominant aerobic bacte- ria. Anaerobic organisms were usually anaerobic strepto- cocci and rarely Bacteroides. Investigating by time peri- od, Staphylococcus abscesses appeared to decrease, while streptococci, especially the anaerobic subtype, have emerged as agents over the last 5 years, a view previously expressed by Nielsen et al. [14]. Anaerobic streptococci were the most common agents in cardiogenic abscess in the present series, as in the experience of Raimondi et al. [16]. It is interesting to note that there was only one patient identified to have a Bacteroides abscess in this series, in spite of an increasing incidence of this species as reported in recent reviews [3, 22]. Proteus grew solely in specimens from otogenic abscesses, as reported earlier [22].

The selection of antibiotics before 1980 largely de- pended on what was available and so penicillin + chloramphenicol was employed in most of the cases. Metronidazole received popularity after 1980 and be- came a standard with its excellent bactericidal effect to virtually all anaerobes [6, 8]. In recent years, new and promising protocols are being created here at Hacettepe and used efficiently. Cefotaxim combined with sulbac- tam-ampicillin (SAM) and metronidazole or ornidazole as the standard is one such regimen that proved highly successful in the last eight patients treated by aspiration alone with no mortality (unpublished data). An open prospective study at our institution was also recently car- ried out to evaluate the efficacy of SAM in the treatment of brain abscess in patients of all ages (n = 19), and SAM was found to be highly effective as a single agent for treating brain abscesses with (n = 15) or without (n = 4) intervention [1].

References

1. Akova M, Ozcebe O, Gullu I, Unal S, Ozgen T, Akalin E (1991) Sulbactam/ampicillin for the treatment of brain abscesses. Pro- ceedings of 17th International Congress of Chemotherapy, June 23-28, 1991, Berlin, Germany

2. Belier AJ, Sahar A, Praiss I (1973) Brain abscess. Review of 89 cases over a period of 30 years. J Neurol Neurosurg Psychiatry 36:757-768

3. Donald FE, Firth JL, Holland IM, Hope DT, Ispahani P, Punt JAG (1990) Brain abscess in the 1980's. Br J Neurosurg 4:265- 272

4. Garfield J (1969) Management of supratentorial intracranial abscess. A review of 200 cases. Br Med J 2: 7-11

5. Gupta SK, Mohanty S, Tandon SC, Asthana S (1990) Brain abscess: with special reference to infection by Pseudomonas. Br J Neurosurg 4:279-285

6. Ibrahim AWM, A1-Rajeh SM, Chowdhary UM, Ammar A (1990) Brain abscess in Saudi Arabia. Neurosurg Rev 13:103- 107

7. Kagawa M, Takeshita M, Yato S, Kitamura K (1983) Brain abscess in congenital cyanotic heart disease. J Neurosurg 58:913-917

8. Keet PC (1990) Cranial intradural abscess management of 641 patients during the 35 years from 1952 to 1986. Br J Neurosurg 4:273-278

9. Kratimenos G, Crockard HA (1991) Multiple brain abscess: a review of fourteen cases. Br J Neurosurg 5:153-161

10. Mampalam TJ, Rosenblum ML (1988) Trends in the manage- ment of bacterial brain abscesses. A review of 102 cases over 17 years. Neurosurgery 23:451-458

11. Maurice-Williams RC (1983) Open evacuation of pus. A satis- factory surgical approach to the problem of brain abscess. J Neurol Neurosurg Psychiatry 46:697-703

12. McGreal DA (1962) Brain abscesses in children. Can Med As- soc J 86:261-268

13. Morgan H, Wood MW, Murphey F (1973) Experience with 88 consecutive cases of brain abscess. J Neurosurg 38:698-704

14. Nielsen H, Gyldensted C, Harmsen A (1982) Cerebral abscess. Aetiology and pathogenesis, symptoms, diagnosis and treat- ment. A review of 200 cases from 1935-1976. Acta Neurol Scand 65:609-622

15. Quartey GRC, Johnston JA, Rozdilsky B (1976) Decadron in the treatment of cerebral abscess. J Neurosurg 45:301-310

16. Raimondi AJ, Matsum0to S, Miller RA (1965) Brain abscess in children with congenital heart disease. J Neurosurg 23: 588- 595

17. Rosenblum ML, Hoff JT, Norman D, Weinstein PR, Pitts L (1978) Decreased mortality from brain abscess since the advent of computerized tomography. J Neurosurg 49:658=668

18. Rosenblum ML, Hoff JT, Norman D, Edwards MS, Berg BO (1980) Nonoperative treatment of brain abscess in selected high-risk patients. J Neurosurg 52:217-225

19. Stephanov S (1988) Surgical treatment of brain abscess. Neuro- surgery 22:724-730

20. Szuwart U, Bennefeld H (1990) Bacteriological analysis ofpyo- genic infections of the brain. Neurosurg Rev 13: I 13-118

21. Williams B (1990) Cerebral abscess (Editorial). Br J Neurosurg 4:248-251

22. Yang S-Y (1981) Brain abscess. A review of 400 cases. J Neuro- surg 55:794-799