Brain abscess in the computed tomography era: A 10-year experience from Auckland, New Zealand

Abstract

Notes were reviewed for 68 patients with brain abscess diagnosed at Auckland Hospital, Auckland, New Zealand between 1978 and 1988. Mean age was 30 years (range one week to 74 years). There were 48 men and 40% were Maori or Pacific Island Polynesians. Seventy-two per cent of patients had headache, 54% had fever and 72% had lateralizing neurological signs. Thirty-one per cent of abscesses were associated with contiguous infection (otic, sinus, dental). Forty-four per cent were in the frontal lobe. Two abscesses were sterile; 197 bacterial isolates were cultured from the remainder. Fifty-four per cent contained obligate anaerobes, which were the only isolates in 22%. Streptococcus anginosus was the single most common isolate present in 22% of the abscesses. Amoxycillin plus metronidazole provided cover for approximately 95% of the total isolates on the basis of sensitivity testing. Treatment was with surgery and antibiotics in all but three patients, who were cured with antibiotics alone. Sixty per cent had a definitive regimen of penicillin (or ampicillin/amoxycillin) and/or metronidazole, always intravenous initially but subsequently often orally. Median duration of antibiotic treatment was 57 days (range 28 to 206). Seventy-five per cent had initial aspiration, 9% open drainage and 7% were excised initially. Seventy-one per cent had a good functional outcome. Mortality was 8.8%. Factors associated with a poor outcome were trauma as a cause, and delays after admission of more than seven days to diagnosis and/or operation.

Brain abscess is a serious infection of the central nervous system which in the past has been associated with significant morbidity and mortality. Even after the introduction of antibiotics, mortality remained high at around 30 to 50% (1–6).

In general, mortality through the last two decades has fallen to around 10% (7–9), although some recent studies show surprisingly high mortalities of 40% (10) and 31% (11). The advent of computed tomography scanning and the better selection and use of antibiotics, with increasing recognition of the prevalence of anaerobic organisms in brain abscesses, are usually credited with the good trends in mortality (7,9,12,13).

We review our experience with brain abscess since computed tomography scanning began in this hospital, to evaluate our management, morbidity and mortality data, and to identify both favourable and unfavourable prognostic factors.

PATIENTS AND METHODS

The neurosurgical unit at Auckland Hospital is a referral centre for a population of about 1.5 million people living in the northern part of New Zealand and Pacific island countries (ie, Rarotonga, Samoa, Fiji and Niue). The study period was from October 1978 to December 1988, the beginning being chosen to coincide with the introduction of head computed tomography scanning at the hospital.

The diagnosis of brain abscess was confirmed by computed tomography scan and operation or autopsy microbiology with or without histology in all but three patients, in whom it was made on the basis of clinical presentation, computed tomography scan appearance and response to antibiotics. Clinical files for demographic data and information on clinical presentation, investigations, management and outcome were examined. Different patient parameters were correlated with outcomes to identify prognostic factors.

Microbiological techniques:

All brain abscess material was Gram stained and cultured for aerobic and anaerobic bacteria. Identification and sensitivity testing were carried out by the routine methods used over this 10-year period in the Clinical Microbiology Laboratory.

Statistics:

Statistical tests performed as appropriate were: Students t test, χ² test, exact probability test of Fisher and the Kolmogorov-Smirnov two-tailed test.

RESULTS

Sixty-eight patients with brain abscess were identified during the study period. The ages ranged from one week to 74 years with a mean age of 30 years. There were 48 male and 20 female patients. Maoris and Polynesian Pacific Islanders accounted for 28% and 12% of cases, respectively, compared with 11% and 8% in the total Auckland population. Incidence and mortality are shown in Figure 1; there was no change in either over the 10 years.

Figure 1.

Incidence and mortality of brain abscess in Auckland Hospital from 1978 to 1988: *1978 numbers for three months only

Abscess site and source:

Abscess sites are tabulated in Table 1. Multilocular abscesses were classified as one abscess. The frontal lobe was the most common site. Ten per cent of patients had two or more abscesses.

TABLE 1.

Site of abscess (n=71)^*

Site	Number	Percentage
Frontal	32	45
Temporal	17	24
Parietal	13	18
Occipital	5	7
Basal ganglia	2	3
Pituitary	1	1.5
Cerebellum	1	1.5
Total	71	100

^*Of 68 patients, 61 had solitary abscesses, five had two abscesses, for a total of 71 abscesses (two patients excluded with two abscesses)

The most probable source for the abscesses is shown in Table 2 along with their relative frequencies. Contiguously spreading infection from the ears, teeth or sinuses was the principal source, accounting for 31% of the abscesses. In 22%, no obvious source could be identified.

TABLE 2.

Source or predisposing cause (n=68)

Causes	Number	Percentage
Contiguous infection	2	31
Otic	(11)	(17)
Dental	(5)	(7)
Sinus	(5)	(7)
Trauma	11	1
Neurosurgery	2	3
Metastatic	19	28
Cyanotic congenital heart disease	(5)	(8)
Lung infection	(7)	(10)
Other sites	(7)	(10)
Unknown	1	2
Total	68	100

Clinical presentation:

The most common symptoms were: headache, 72%; nausea or vomiting, 48%; seizure, 33%; fever, 33%; and behavioural change, 21%. Others such as lateralizing neurological symptoms, visual or speech disturbance, confusion or ataxia were each seen in fewer than 20%. On admission, however, one or more localizing signs were present in 72% (44 of 61) of the patients. We excluded seven patients with abscess secondary to trauma because of prior confounding neurological signs.

Fever was present in 54%, meningism in 43%, hemi motor or sensory signs in 39%, cranial nerve paresis in 31%, visual field defect in 30% and papilledema in 26%. Mental status on admission was classified retrospectively using the classification described by Garfield in 1969 (2) for all except two patients (Table 6).

TABLE 6.

Determinants of outcome

	Outcome		Significance
	Good n=48	Poor n=20
Age (years)	29.8±17.4	30.1±20.0	None
Sex
Male	35	13	None
Female	13	7
Trauma
Associated	4	7	P=0.02
Not associated	44	13
Admission data
Mental status (n=66)
Fully alert and oriented (n=35)	26	9	None
Drowsy and disoriented but reponding to commands (n=23)	16	7
Responding only to painful stimuli (n=7)	4	3
No response (n=1)	1	0
Temperature
≥38.0°C	33	11	None
<38.0°C	13	4
Time interval between admission and:
Diagnosis
≥7 days	4	7	P=0.003
<7 days	43	8
Intravenous antibiotic
≥7 days	4	3	None
<7 days	43	11
Operation
≥7 days	9	7	P=0.05
<7 days	35	6

Outcome: Good indicates no or minor sequelae only, eg, minor motor, sensory, visual defects not interfering with daily activity, convulsions totally controlled by anticonvulsant medicines; Poor indicates all other outcomes including death

Laboratory tests:

Erythrocyte sedimentation rate was greater than 30 mm/h in 71% (32 of 45) and white blood cell count greater than 11.000×10⁹ cells/L in 66% (38 of 58). Blood cultures were positive in 14% (six of 42).

Forty-three per cent of patients had a lumbar puncture before surgery. Cerebrospinal fluid showed a very wide range of abnormality; four specimens were entirely normal. Median and ranges were: white blood cells, 508 (zero to 33,600) ×10⁶/L (n=22); glucose, 3.4 (zero to 4.7) mmol/L (n=24); protein, 1.3 (0.2 to 5.2) g/L (n=24). Four of 28 grew organisms that were present in the brain abscess.

Computed tomography scanning:

All 68 patients had head computed tomography scans, but the present study excluded from the following time-related analysis five patients in whom brain abscess developed as a complication of other illness while in hospital. Fifty-seven per cent of patients had a computed tomography scan within 24 h of hospital admission and 84% (53 of 63) within seven days. Of the 10 patients with scans after seven days, nine were from regional areas.

Contrast enhancement was used routinely during the study period. In 84% (53 of 63) of patients, the first computed tomography scan correctly diagnosed brain abscess or cerebritis. Of the 10 not diagnosed, six scans were interpreted as cerebral tumour, one as hematoma and one as infarct. The remaining two were initially considered normal but repeat scans 22 and 24 days later revealed brain abscess.

Abscess microbiology:

Brain abscess tissue was cultured from 65 of the 68 patients. Two tissue specimens were sterile and had no organisms on Gram stain, while 197 bacterial isolates were identified from the remaining 63 specimens (Table 3).

TABLE 3.

Organisms isolated and their relevant sensitivities (n=197)

Aerobic	Number (%)	Anaerobic	Number (%)
Streptococci (eg, viridans streptococci, Streptococcus anginosus) (100% sensitive to benzylpenicillin)	31 (16)	Gram-negative bacilli (eg, Bacteroides species, Fusobacterium species) (87% (60 of 69) sensitive to benzylpenicillin, 99% (68 of 69) sensitive to metronidazole)	69 (35)
Staphylococci (38% sensitive to benzylpenicillin, 100% sensitive to flucloxacillin)	13 (6.5)	Streptococci (eg, Peptostreptococcus species) (100% sensitive to benzylpenicillin, 87% (13 of 15) sensitive to metronidazole)	24 (12)
Enterobacteriaceae (eg, Escherichia coli, Proteus mirabilis) (64% (7 of 11) sensitive to amoxycillin)	11 (6)	Gram-positive nonspore forming bacilli (eg, Lactobacillus species, Propionibacterium species) (100% sensitive to benzylpenicillin, 73% (8 of 11) sensitive to metronidazole)	24 (12)
Fastidious Gram-negative bacilli (eg, Haemophilus species, Eikenella corrodens) (50% (4 of 8) sensitive to benzylpenicillin, 92% (12 of 13) sensitive to amoxycillin)	13 (6.5)	Clostridia (100% sensitive to benzylpenicillin, 89% (8 of 9) sensitive to metronidazole)	9 (4.5)
		Others (2 of 3 sensitive to benzylpenicillin; 2 of 3 sensitive to metronidazole)	3 (1.5)
Total	68 (35)		129 (65)

Fifty-one per cent (32 of 63) of the culture-positive abscesses had single pathogens. Single pathogens were associated with traumatic (73%) or metastatic (64%) sources. Fourteen per cent had either two or three isolates and 35% had four or more. In 22% (14 of 63) of the abscesses only obligate anaerobes grew, in 46% (29 of 63) only aerobes.

Two-thirds of the total isolates (129 of 197) were obligate anaerobes with anaerobic Gram-negative bacilli being the most common. The most common single isolate was Streptococcus anginosus, comprising 7% of all the isolates and present in 22% of the brain abscesses.

Anaerobic organisms were found in brain abscesses from all sources, predominately otic, dental, sinus and metastatic sources. Aerobic streptococci were the most common aerobes isolated, especially important in brain abscesses of unknown, sinus, trauma and dental sources. Staphylococci were rare other than in abscesses associated with trauma, neurosurgery or bacteremia. Aerobic Gram-negative bacilli were seen with all sources. Clostridia were present only in abscesses consequent on trauma.

Antibiotic sensitivities:

Sixty-two per cent (38 of 64) of the aerobes tested were sensitive to benzylpenicillin and 81% (59 of 68) sensitive to amoxycillin.

Ninety-three per cent (115 of 124) of the anaerobic organisms tested were sensitive to benzylpenicillin and amoxycillin: all 14 of the Bacteroides fragilis group and two (of 12) other unspeciated anaerobic Gram-negative bacilli were resistant to benzylpenicillin and amoxycillin. Ninety-three per cent (89 of 96) of the anaerobes tested were sensitive to metronidazole.

MANAGEMENT

Two misdiagnosed patients had no appropriate treatment. In all but three of the remaining 66 patients, treatment comprised antibiotic administration and surgery.

Antibiotic management:

Overall, 60% of patients (39) received benzylpenicillin or amoxycillin (or ampicillin) and/or metronidazole as their definitive regimens. Eleven per cent received chloramphenicol alone or with metronidazole and 19% received a variety of other antibacterial regimens depending on the isolated organisms.

The median duration of antibiotic treatment was 57 days, ranging from 28 to 206 days, but 95% received antibiotics for between 28 and 96 days. Intravenous antibiotics were given for a median 24 days (range five to 65). Twelve patients (four with Staphylococcus aureus alone) were treated with intravenous antibiotics alone for a median 42 days (range 28 to 83). Three patients were treated with antibiotics without surgery and all were cured without sequelae.

Twenty-three patients (34%) received fewer than 21 days intravenous treatment and of these, eight (12%) (detailed in Table 4) received only five to 10 days intravenous treatment before continuing with oral treatment. This group of 23 patients was similar in characteristics to the remainder of the patients; 18 had a good outcome and five a poor outcome (including one death). For all of these patients, intravenous antibiotic treatment was followed with a minimum 28-day oral course (median 44 days, range 28 to 84). S aureus brain abscess was not treated with oral antibiotics.

TABLE 4.

Details of patients receiving intravenous treatment for five to 10 days only

Age (years)	Sex	Symptom duration (days)	Abscess site	Source/predisposing factors	Operation	Bacterial isolates	Antibiotic data
							Antibiotic	Duration (days)
								Intravenous	Oral	Outcome
18	M	21	Occipital	Lung	Aspiration (×2)	Streptococcus anginosus, Bacteroides urealyticus, Fusobacterium nucleatum. Fusobacterium necrophorum, other anaerobic Gram-negative bacteria	Benzylpenicillin and chloramphenicol	5		Good
52	M	14	Temporal	Mastoiditis	Drainage (×1) Aspiration (×1)	Proteus mirabilis, Klebsiella species, Bacteroides thetaiotamicron, Peptostreptococcus magnus, Peptostreptococcus asaccarolyticus, other anaerobic Gram-negative bacteria	Phenoxymethyl penicillin		42
1	M	1	Frontal	Periorbital	Aspiration (×1)	Haemophilus influenzae	Ampicillin and chloramphenicol and metronidazole	10		Good
51	M	7	Temporal	Odontic	Aspiration (×3)	Streptococcus anginosus, Haemophilus paraphrophilus	Cotrimoxazole and metronidazole		49
6	M	7	Temporal	Cyanotic congenital heart disease	Drainage (×1)	Streptococcus anginosus	Ampicillin and chloramphenicol	9		Good
56	M	6	Temporal	Odontic	Excision (×1)	Haemophilus paraphrophilus	Amoxycillin		56
48	F	22	Parietal	Otic	Aspiration (×3)	Fusobacterium nucleatum	Chloramphenicol	10
35	M	10	Frontal	Ondontic/sinus	Aspiration (×2)	Actinobacillus actinomycetemcomitans, Haemophilus aphrophilus, Fusobacterium nucleatum	Amoxycillin and metronidazole		49	Good
							Chloramphenicol	10		Good
							Phenoxymethol penicillin		84
							Benzylpenicillin and chloramphenicol	9		Good
							Chloramphenicol		28
							Amoxycillin and metronidazole	5		Good
							Metronidazole		56
							Chloramphenicol and metronidazole	10	28	Good

^*Outcome: Good indicates no or minor sequelae only, eg, minor motor, sensory, visual defects not interfering with daily activity, convulsions totally controlled by anticonvulsant medicines

Surgical management:

Surgery comprised aspiration, open drainage or excision (Table 5). Abscess excision was chosen as initial management in five patients; one required later aspiration. Overall, patients had a mean of 2.5±1.3 neurosurgical procedures (usually repeat aspirations).

TABLE 5.

Surgical management and outcome

		Outcome
		Poor (death)	Good
	Number of patients
No operation
Undiagnosed	2	2 (2)	0
Untreated	1	1 (1)	0
Antibiotic only	3	0 (0)	3
Operation and antibiotic
Aspiration	51 (later excision 8)	14 (2)	37
Open drainage	6 (later excision 0)	2 (1)	4
Excision	5 (later aspiration 1)	1 (0)	4
Total	68	20 (6)	48

Outcome: Good indicates no or minor sequelae only, eg, minor motor, sensory, visual defects not interfering with daily activity, convulsions totally controlled by anticonvulsant medicines; Poor indicates all other outcomes including death

MORTALITY

The total mortality from brain abscess over this 10-year period was 8.8% (six of 68). Two were misdiagnosed as tumours and the patients died with no definitive treatment. One child died after treatment was withheld because of abscess-induced gross frontal lobe destruction, one adult with a cerebellar abscess died of respiratory arrest seven days after admission despite appropriate treatment, one died from chloramphenicol marrow aplasia, and another from unexplained recurrent abscesses with multiple changing organisms over an 18-month period. In summary, only three patients diagnosed and appropriately treated died directly of brain abscess or its treatment, a mortality of 4.3% (three of 65).

MORBIDITY

Convulsions were diagnosed in 23% of the 62 surviving patients, and 50% (31 patients) took anticonvulsants. Twenty-nine per cent had visual impairment, 18% motor or sensory deficits and 45% had varying degrees of either memory loss, cognitive impairment or headache.

Seventy-one per cent (48 patients) were considered to have a good functional outcome, ie, no sequelae or minor sequelae; convulsions controlled by anticonvulsants were included in this category. Twenty-nine per cent (20 patients) had a poor outcome: six died, one remained in coma and 13 had major morbidity (ie, difficulty in control of convulsions, blindness and hemiplegia). One patient with S aureus brain abscess relapsed. One child had two anatomically distinct brain abscesses several weeks apart, secondary to lung suppuration.

PREDICTORS OF OUTCOME

Functional outcomes as described were related to a number of patient variables in order to identify potential prognostic factors. Some of them are tabulated in Table 6. There was no correlation between the presence of focal signs, mental status, erythrocyte sedimentation rate or white blood cell count on admission, and eventual outcome. Patients in whom the diagnosis (P=0.0025) or operation (P=0.05) occurred more than seven days after admission had a worse prognosis than those diagnosed or operated on earlier. Those with brain abscess secondary to trauma also had a significantly worse prognosis (P=0.023). Starting antibiotic before or later than seven days after admission did not alter outcome (P=0.9).

DISCUSSION

Demographic features of the patients in our series were similar to those reported elsewhere. Average age, preponderance of men, range of predisposing factors and inability to determine an underlying source in about 20%, were similar to the findings of others (3–6,10,14).

Only 10% of our patients had multiple abscesses, which is at the lower end of most reports which range from 2 to 30% (3–6,15). The higher figures have usually been reported in more recent series in which localization has been by computed tomography scan (10,15, 16–18) (although that cannot explain our low percentage). The frontal lobe as the predominant site is a common finding (5,14,15). On average, patients had symptoms for 14 days before admission, two-thirds had headache and two-thirds had focal neurological signs. Seizure was a presenting symptom in a third and only half were febrile on admission. The symptoms and signs of brain abscess are dominated by those of a space-occupying lesion, and fever is surprisingly often absent for a bacterial disease (4,6,10,11,19).

The first computed tomography scan gave the correct diagnosis in 84% and seven of the nine false negative scans were in the first five years of this review. This overall sensitivity is similar to the overall classification accuracy of 86% reported by Coulam et al in 1980 (20). White blood cell count, erythrocyte sedimentation rate and cerebrospinal fluid findings are insensitive and nonspecific (4,5,6,10,19). One of our patients deteriorated soon after lumbar puncture which suggested a causal relationship. As has been stated by others (2–6,10,19), lumbar puncture is contraindicated where this disease is suspected.

Only 3% of the abscesses cultured in this study were sterile, at the lower end of values reported in the 1980s, which ranged from 0 to 19% (10,11,21). The predominant role of anaerobic organisms, first described in 1963 (12), was reconfirmed in this series, with 65% of all isolates being anaerobic and 54% of brain abscesses containing anaerobes.

Parenteral antibiotic and surgery are widely regarded as the principal components of optimal management. The antibiotic debate is about which drug or drugs to use, when to start, what the duration should be and the need for parenteral as opposed to oral drugs. The surgical debate is about aspiration, open drainage or excision, and whether surgery is invariably needed. No controlled data exist for any of these controversies (22–26).

Aspiration was the most common surgical technique used in this study and the low mortality certainly supports this as an appropriate strategy. We treated only three patients with antibiotics alone; surgical inaccessibility, small size and multiplicity were the reasons for these decisions and are those commonly stated by others (15,26,27). All three survived without sequelae.

Choice of definitive regimens over the 10-year period reflected the bacteria isolated and their sensitivities, and knowledge about penetration of antibiotics into the brain (28). As with all retrospective series, it is not possible to evaluate the comparative efficacies of these definitive regimens. We used penicillins and/or metronidazole in 70% of patients and could have used them in more. A combination of benzylpenicillin plus metronidazole would provide antibiotic cover for about 85% of all isolates in this study, and amoxycillin plus metronidazole about 95% of all isolates. In a more clinically useful way, if abscesses related to trauma and neurosurgery are excluded, amoxycillin and metronidazole is an appropriate combination regimen in 92% of cases. Our use of chloramphenicol over this decade has been phased out, principally because of its toxicity.

As a result of these findings, therefore, we now recommend that patients with brain abscess, other than those secondary to trauma or after neurosurgery, begin intravenous amoxycillin 12 g/day and intravenous metronidazole 0.5 g 12-hourly until surgery and culture results allow targetted selection. Those with abscesses secondary to trauma or neurosurgery should also receive intravenous flucloxacillin 12 g/day for presumptive S aureus treatment.

Recently it has been suggested that patients who have excision should have three to four weeks of intravenous treatment, those aspirated four to six weeks intravenously and those treated with antibiotics alone a minimum of four weeks (10). In this series, 23 patients were treated with less than 21 days of intravenous antibiotic treatment which was always followed by oral treatment. Only one later death occurred and 18 patients made a good recovery. Thus cure with shorter intravenous courses than those recommended is possible. We cannot critically define those who might receive shorter intravenous treatment and subsequent oral treatment because of the retrospective nature of this evaluation. Nevertheless, stable clinical status, aspirated abscesses with organism identity and sensitivity established, and appropriate oral antimicrobial agents which enter brain tissue, appear to be important requirements.

We used oral metronidazole 400 mg tid, because oral doses have been shown to reach effective levels in brain abscess pus (29). Most recently we have used amoxycillin in large doses orally (median of 3.0 g/day) for both penicillin and amoxycillin sensitive bacteria, because it is the best absorbed oral penicillin. We reiterate however that optimal regimens – drugs, dose, route and duration – remain unestablished.

We found 23% of the survivors to be significantly impaired, in the sense that they needed major family or institutional support. Using the two outcome groups we defined, we found few parameters predicted particular outcomes. Trauma as a cause, and diagnosis or operation more than seven days after admission, were associated with a poorer outcome. The latter two variables are not independent.

The most important finding of this study is the low overall mortality of 8.8% for patients with brain abscess in this city in the past 10 years following the use of computed tomography scanning. That this can fall to 4.3% where diagnosed brain abscess can be actively treated is noteworthy. Because of the low mortality seen at this hospital, it was not possible to show significant correlations between our collected data and mortality. Nevertheless, patients who were undiagnosed or untreated died: untreated brain abscess is a lethal disease.