OUP user menu

Bacterial brain abscess: microbiological features, epidemiological trends and therapeutic outcomes

C.‐H. LU, W.‐N. CHANG, Y.‐C. LIN, N.‐W. TSAI, P.‐C. LILIANG, T.‐M. SU, C.‐S. RAU, Y.‐D. TSAI, C.‐L. LIANG, C.‐J. CHANG, P.‐Y. LEE, H.‐W. CHANG, J.‐J. WU
DOI: http://dx.doi.org/10.1093/qjmed/95.8.501 501-509 First published online: 1 August 2002


Background: Despite the advent of modern neurosurgical techniques, new antibiotics, and powerful imaging technologies, brain abscess remains a potentially fatal central nervous system infection.

Aim: To determine the epidemiological trends, prognostic factors, and outcomes of bacterial brain abscess, to improve the therapeutic strategy for this disease.

Design: Retrospective hospital‐based epidemiology study.

Methods: Over a period of 15 years (1986–2000), 123 patients were retrospectively identified as having brain abscesses at Kaohsiung Chang Gung Memorial Hospital. To compare changes over time, the appearance of disease among our patients was divided into two time periods: 1986–1993 and 1994–2000.

Results: The prevalence rate of brain abscesses caused by Gram‐negative organisms significantly increased in the second study period. Viridans streptococci and Klebsiella pneumoniae were the two prevalent pathogens associated with haematogenous spread. Metastatic septic abscess, a devastating complication of K. pneumoniae septicaemia, frequently occurs in diabetic patients, with a high mortality rate. Viridans streptococci were the most prevalent pathogens from infection in paranasal sinusitis, but no fatality occurred. In recent years, head trauma and/or post‐neurosurgical states have become important predisposing factors, and nosocomial infections also play an important role.

Discussion: Despite the availability of new antibiotics and the development of better neurosurgical techniques, therapeutic outcomes of brain abscess showed no significant change when comparing the two study periods, and only the presence of septic shock influenced outcome.


Despite the advent of modern neurosurgical techniques, new antibiotics, and new powerful imaging technologies, brain abscess remains a potentially fatal central nervous system (CNS) infection.1–,4 In 1973, Kao et al.5 reported the clinical features of 26 cases of brain abscess with surgical intervention; however, a large series study of brain abscess in Taiwan is still lacking. A hospital‐based study provides accurate information about localization of brain abscess, predisposing factors, clinical features, the prevalence rate of implicated bacterial pathogens, and causes of fatality. In this study, we investigated differences between nosocomial and community‐acquired infections, predisposing factors, clinical and neuroimaging findings, complications, and therapeutic outcomes, in order to improve the therapeutic strategies for this potentially fatal disease.


Over a period of 15 years (January 1986 to December 2000), 123 patients were retrospectively identified as having brain abscesses at Kaohsiung Chang Gung Memorial Hospital. The largest medical centre in southern Taiwan, this facility is a 2482‐bed acute‐care teaching hospital, which provides both primary and tertiary referral care services. Southern Taiwan consists of two cities and three counties (Kaohsiung Municipality and Hsien, Tainan City and Hsien, and Pingtung Hsien) with a population of approximately 5 441 000 as of October 1998.6 The annual population increase for Taiwan has been reported to be 7.52%.6 To compare changes over time, the appearance of disease among our patients was divided into two time periods: the first time period was 8 years (January 1986–December 1993) and the was 7 years (January 1994–December 2000).

Criteria of bacterial brain abscess for inclusion were: (i) characteristic computerized tomography (CT) and/or magnetic resonance imaging (MRI) findings; (ii) evidence of brain abscess seen during surgery or histopathological examination; and (iii) classical clinical manifestations including headache, fever, localized neurological signs and/or consciousness disturbance.7–,9 All materials from bacterial brain abscesses were cultured for aerobic and anaerobic bacteria, mycobacteria, and fungi. Aerobic culture media included blood, chocolate, EMB, CNA media, and thioglycolate broths. Anaerobic cultures were processed in Brucellae agar plate, BBE/PEA media, and thioglycolate broths. BHI agar, mycobiotic media, SDA media, and SCG media were used for fungal isolation. Lowenstein‐Jensen media and middle 7H11 media were used for mycobacterial isolation. Organisms were identified by API (Analytab Products), RapID‐ANA II identification system (Innovative Diagnostic Systems), and ID 32 C (VITEK, bioMérieux Vitex). Patients were considered to have mixed infections if at least two bacterial organisms were isolated from the initial cultures. Patients who were initially treated at other hospitals but subsequently transferred to our hospital for further therapy were also included in this study, with initial clinical data collected at those hospitals used for analysis.

Brain abscesses were defined as nosocomial according to the 1988 guidelines of the Centers for Disease Control.7 Brain abscesses related to head trauma with skull fracture or neurosurgical procedures, were classified as post‐neurosurgical. Other patients who presented with no distinctive characteristics, and/or who had not undergone invasive procedures, were classified as having the spontaneous form. Patients with evidence of brain abscesses not due to bacterial pathogens were excluded from this study.

The mental status for patients was classified into two groups: (i) clear consciousness; and (ii) inattention, confusion, clouded consciousness, stupor, or coma. Surgical intervention and/or antibiotic therapy were the mainstays of our treatment. Antibiotic susceptibility was determined using the Kirby‐Bauer disc diffusion method (Mueller‐Hinton II agars; Becton Dickinson). Appropriate antimicrobial therapy was defined as the administration of one or more antimicrobial agents, which demonstrated effectiveness against bacterial pathogens, as determined from susceptibility tests, and which were capable of passing through the blood‐brain barrier in adequate amounts. Surgical treatment consisted of either aspiration or excision of the abscess. Aspiration was used to aspirate the content of the abscess with a ventricular catheter via a burr hole or small craniotomy, which left the capsule alone. Craniotomy and resection of the abscess was defined as evacuation. In our institution, those patients presenting in good neurological condition with a well‐formed abscess in the right hemisphere underwent evacuation, whereas those presenting significant surgical risks or those with an abscess that was deep‐seated or in the left hemisphere underwent aspiration. Recovery was evaluated 3 months after patient discharge.

Data, including initial clinical manifestations, acquisition of infection, type of infection, numbers of abscesses, various treatment regimens, and fatalities in the two study periods, were analysed using the χ2 test or Fisher's exact test. Data for age, and interval between onset and diagnosis between fatal and non‐fatal groups were compared using Student's t‐test. Stepwise logistic regression was used to evaluate the relationship between clinical factors and the mortality rate adjusting for other potential confounding factors. All analyses were conducted using SAS (1990).10


The 123 patients included 92 males (mean age 43 years; range 1 month–80 years) and 31 females (mean age 41 years; range 2 months–73 years) (Figure 1). Of these 123 patients, 103 had community‐acquired infections, while the other 20 were diagnosed with nosocomial infections. Sixty‐nine cases appeared in the first time period (January 1986–December 1993), with the other 54 in the second time period (January 1994–December 2000) (Table 1). The ratios of community‐acquired to nosocomial meningitis for these two time periods were 58:11 and 44:10, respectively.

Associations for portal of entry and causative pathogens are listed in Table 2. Portal of entry for infection in 94 culture‐positive brain abscesses included haematogenous spread (n=32), postneurosurgical states (n=17), contiguous infection from parameningeal foci such as an otogenic origin (n=13), paranasal sinusitis (n=9), and unknown (n=24). In this study, Klebsiella pneumoniae and viridans streptococci were the two most prevalent pathogens associated with haematogenous spread, accounting for 52% (15/29) of the episodes. Staphylococcus aureus, K. pneumoniae, and viridans streptococci were the three common pathogens in postneurosurgical states, accounting for 47% (7/15) of the episodes. Viridans streptococci were the most prevalent pathogens from paranasal sinusitis infections, while Proteus species were the common pathogens of otogenic origin, in brain abscesses with polymicrobial infection. The mean interval between onset of symptoms to detection of brain abscess was 21 days (range 1–700 days).

In the first study period, 58% (40/69) of cases involved a single pathogen, with viridans streptococci being the most prevalent (23%), followed by Bacteroides species (18%), S. aureus (10%), and Corynebacterium species (10%). Fourteen percent (10/69) of patients had mixed infections, and 26% (18/69) of patients had a negative culture. In the second study interval, 70% (38/54) of cases were infected by a single pathogen, with viridans streptococci again the most prevalent (29%), followed by K. pneumoniae (26%). Eleven percent (6/54) of patients had mixed infections, and 20% (11/54) of patients had a negative culture.

Clinical manifestations of these 123 patients are listed in Table 3. Fever was found in 71 patients and headache in 68. Forty‐nine patients were admitted with disturbed consciousness. Twenty‐one patients suffered septic shock, 13 of whom subsequently died. Seizures and/or status epilepticus occurred in 23 patients, of whom five died. Concomitant bacterial meningitis was found in 29 patients. Of these 29, 24 had community‐acquired infections and five had acquired the infections nosocomially. Four of the five nosocomially‐infected patients had undergone neurosurgical procedures as the underlying condition; the other had had a stroke. Of these 29 cases with concomitant bacterial meningitis, five were identified as K. pneumoniae infection and revealed multiple septic abscesses on further examination, including liver abscess, renal abscess, lung abscess, paraspinal abscess, endophthalmitis, pyomyositis, purulent pericarditis, and septic arthritis. All five of these patients were diabetic. Other clinical manifestations in these 123 cases included hemiparesis, stiff neck, facial palsy, hemiparaesthesia, and disturbance of speech and vision.

The locations of brain abscesses of these 123 cases were supratentorial in 118 cases and infratentorial in five (Table 4). In total, 99 (80.5%) had a single brain abscess and 24 (19.5%) had multiple brain abscesses. The most common sites for brain abscess were the frontal lobe (33%), followed by the temporal lobe (20%) and temporo‐parietal lobe (10%). The left hemisphere was involved more than the right with 63 on the left, 47 on the right, two on the midline, six bilaterally on both hemispheres, and five on the infratentorial area.

Antimicrobial therapy, with or without surgical intervention (aspiration or total excision), was the cornerstone of treatment in these 123 patients. Seventeen patients received antimicrobial therapy alone, four of whom had multiple pyogenic brain abscesses; the other 13 were associated with poor systemic conditions. Penicillin G with chloramphenicol was the mainstay of initial empiric antimicrobial treatment of bacterial brain abscess unless Staphylococcus or Gram‐negative bacilli were more likely to be the responsible pathogens. The choice of final antibiotics was guided by the final culture results. With this regimen of therapy, 12 patients experienced recurrence of brain abscesses, and they received repeated aspiration and/or total excision of brain abscesses. In total, 102/123 patients survived. Mortality rates for the first and second study periods were 19% and 15%, respectively (Figure 2). Nine of 67 patients who received both antimicrobial therapy and total excision died. Six of 33 patients who received both antimicrobial therapy and aspiration died. All six patients who received antimicrobial therapy and aspiration with subsequent total excision survived, while there were six fatalities among the 17 patients who received antimicrobial therapy alone. Mortality rates of patients by group of causative organism were as follows: Gram‐negative bacilli, 27% (7/26); Streptococcus spp., 0% (0/21); Staphylococcus spp., 33% (3/9); Corynebacterium spp., 25% (1/4); anaerobic pathogens, 17% (3/18); mixed bacterial species, 13% (2/16); and negative cultures, 14% (4/29) (Table 1). Of the 21 patients who died, 13 had fulminant clinical courses and deterioration of systemic conditions, three had intracerebral haemorrhage after surgical interventions, and four had brain swelling with subsequent herniation. Among the 102 survivors, 64 resumed normal lives or had mild mental abnormalities, 27 had focal neurological deficits with moderate disability, and 11 had severe disability or were in a persistent vegetative state.

The potential prognostic factors of these 123 patients are listed in Table 5. According to the statistical analysis, only the presence of septic shock (p<0.001) had a significant influence on therapeutic outcomes. Variables, including age at infection, sex, acquisition of infection, type of infection, type of treatment, headache, fever, septic shock, seizure, concomitant meningitis, mental status, and the interval between onset and diagnosis, were used in the stepwise logistic regression, and the presence of septic shock (p=0.0001) was the only factor predictive of mortality.

Figure 1.

Age distribution of patients with brain abscess.

View this table:
Table 1

Causative organisms, January 1986 through December 2000

View this table:
Table 2

Relation between predisposing factors and causative pathogens

View this table:
Table 3

Clinical manifestations

View this table:
Table 4

Localization of brain abscesses

Figure 2.

Numbers of fatal and non‐fatal cases of bacterial brain abscess, distributed over 15 years.

View this table:
Table 5

Prognostic factors


The causative pathogens of bacterial brain abscess vary with time period, geographic distribution, age, underlying medical and/or surgical conditions, and mode of infection.1–,5 The relative frequency of mixed bacterial infections and limitations of anaerobic culture techniques also influence the pathogen prevalence in different studies.1–,14 A study from China reported that Gram‐negative bacilli, with Proteus, Enterobacter and Alcaligenes species being common, accounted for 21% (82/400) of the implicated pathogens of brain abscess from 1952 to 1972.14 In one early study of brain abscess from Taiwan, Gram‐negative bacilli, Staphylococcus species, and Streptococcus species accounted for 40% (10/25) of the episodes,5 and these three groups of pathogens also accounted for 46% (56/123) of the pathogens implicated in this study. Of the Streptococcus species, viridans streptococci were the most frequently found, at 13% (9/69) and 22% (12/54) of the implicated pathogens in the first and second study time periods, respectively. Although there was a change in the relative frequency of pathogens, viridans streptococci remained the most prevalent pathogens in both time periods. Among the Gram‐negative bacilli, K. pneumoniae, E. coli, P. aeruginosa, and other rare Gram‐negative organisms accounted for more than 21% of implicated pathogens. There was a significantly increased prevalence of Gram‐negative organisms of from 13% (9/69) in the first to 31% (17/54) in the second time period.

This study shows that clinical presentations and underlying disorders vary greatly in brain abscesses due to different species. K. pneumoniae was the most prevalent pathogen associated with haematogenous spread or postneurosurgical states. Multiple metastatic septic abscesses were common in patients with brain abscess due to K. pneumoniae. Although the reason is unknown, this association of Klebsiella infection with metastatic abscesses has been noted in other reports from Taiwan.15–,17 Of the brain abscesses caused by Streptococcus species, viridans streptococci, the most prevalent members of the flora in the upper respiratory tract, gastrointestinal tract, and female genital tract, accounted for 95% of strains. This study also shows that viridans streptococci are the prevalent pathogens with haematogenous spread secondary to cardiogenic origin or from infection of paranasal sinusitis; this finding is consistent with those of other reports.18,,19 In this study, no deaths due to viridans streptococci occurred, which may well be due to the low virulence of viridans streptococci.18,,20 In the brain abscesses of otogenic origin, Proteus species were the most commonly implicated pathogens among the brain abscesses with polymicrobial infection, and this phenomenon was also noted in other studies.13,,21 This study also showed that S. aureus, viridans streptococci, and K. pneumoniae were the most prevalent pathogens in patients with post‐neurosurgical states.

Regarding the portal of entry, brain abscess is almost always secondary to a focus of suppuration elsewhere in the body, and may develop either by a contiguous focus of infection, head trauma, or haematogenous spread from a distant focus. Although head trauma or neurosurgical procedures were once unusual as causes of brain abscesses, they were important in this study. As a previous report12 demonstrated and our study confirms, when haematogenous spread is the source of the brain abscess, this infection is frequently community‐acquired, and is commonly associated with underlying diseases and with more fulminant courses, as evidenced by higher rates for shock and death. By contrast, postneurosurgical causes and a contiguous focus of infection are associated with a favourable outcome.

In our 123 cases with brain abscesses; fever, headache, altered consciousness, and hemiparesis were the most common manifestations, occurring in 58%, 55%, 50%, and 44% of our patients, respectively. The presence of focal neurological findings (e.g. hemiparesis, hemisensory deficits, and aphasia) depends on the location of the abscess, and these signs were seen in approximately 63% (78/123) of our cases, while 37% (45/123) had no localized signs. The presence of septic shock significantly influenced the outcome of our patients. However, the presence of fever, headache, seizure, and focal neurological signs showed no statistical significance on the prognosis in our study. The interval between onset of symptom and signs and diagnosis was of no statistical significance in our study, as in a previous study.12

Metastatic abscesses are commonly located in the parietal, frontal, or temporal lobes,12,,14 and multiple brain abscesses are usually the result of metastatic spread from remote primary foci, accounting for 6% to 22% of cases.12,,14 In this study, the most common cause of brain abscesses was haematogenous spread, accounting for 26% (32/123) of the episodes, a figure higher than the study from China.14 Almost all abscesses were found in the frontal, temporal, and parietal lobes, and multiple abscesses were present in 19.5% (24/123). Although brain abscesses associated with neurosurgical procedures were once considered uncommon,2,,14 they accounted for 12% of the episodes in this study. The increasing frequency of neurosurgical procedures may be due to the increasing number of neurosurgeons, the improvement of neurosurgical facilities, and the large number of patients with head injury from motorcycle accidents. Unlike the data shown in one study from China,14 where the majority of the abscesses were temporal and cerebellar, suggesting otogenic spread, our patients had more frontal disease. This may reflect the trend towards better and more aggressive management of otogenic disorders between the 1970s and the 1990s.

Treatment of brain abscess requires a combination of antimicrobials, surgical intervention, and eradication of primary infected foci. In Taiwan, initial empirical antibiotics with third‐generation cephalosporins in combination with metronidazole, should be considered for the majority of abscesses cases resulting from infection with Gram‐negative bacilli and streptococcal species. Different therapeutic regimens show no significant statistic differences in prognosis in other reports,12 and were also of no statistical significance in this study.

In summary, the clinical presentations, underlying conditions and therapeutic outcomes vary greatly in brain abscesses, according to the different pathogens. In recent years, head trauma and/or post‐neurosurgical states have become important predisposing factors of brain abscess, and nosocomial infections are also important. Despite the availability of new antibiotics and the development of better neurosurgical techniques, therapeutic outcomes of brain abscess have not shown a statistically significant change, and only the presence of septic shock is a prognostic factor.


  • Address correspondence to Dr W.‐N. Chang, Department of Neurology, Chang Gung Memorial Hospital, 123 Ta Pei Road, Niao Sung Hsiang, Kaohsiung Hsien, Taiwan. e‐mail: chlu99{at}ms44.url.com.tw


View Abstract