Computed tomography (CT) was first introduced in South Africa at the neurosurgical unit at Wentworth Hospital, Durban in 1975, and despite it’s increasing availability, lumbar puncture (LP) appears still to be commonly performed in our region as a first diagnostic procedure in patients with brain abscess or subdural empyema. In a 15-year review of patients with brain abscess and subdural empyema treated at our institution between 1983 and 1997, nearly a third had undergone a diagnostic lumbar puncture prior to CT. We therefore evaluated the diagnostic role of lumbar puncture in these patients as well as it’s impact on patient outcome.
PATIENTS & METHODS
During the 15-year period, January 1983 to December 1997, patients with brain abscess and subdural empyema admitted to our neurosurgical unit were evaluated. The neurosurgical unit at Wentworth Hospital in Durban is the sole referral centre for the Province of KwaZulu-Natal and half of the Eastern Cape Province.
The diagnosis of brain abscess or subdural empyema was made on conventional CT criteria in all cases and definitively at surgery in almost all (97.8%) [1, 23]. The investigative procedures which these patients underwent were retrospectively reviewed. Patients undergoing diagnostic lumbar puncture prior to CT were identified and their case notes were carefully analysed with respect to the contribution of cerebrospinal fluid (CSF) analysis to diagnosis and to the impact of lumbar puncture on outcome.
During the 15-year period, a total of 4623 patients with all forms of intracranial infection were admitted to our neurosurgical unit at Wentworth Hospital in Durban, South Africa. Of these, 1411 patients were diagnosed as harbouring purulent infective intracranial mass lesions, in particular 712 with brain abscess and 699 with subdural empyema.
One hundred and forty-two of these 712 patients with brain abscess (19.9%) and 280 of the 699 patients with subdural empyema (40.1%) had undergone diagnostic lumbar punctue prior to CT and, importantly, prior to referral to our unit. Overall, 422 patients (29.9%) were subjected to lumbar puncture as the first diagnostic procedure, prior to CT. [Table 1]
CSF analysis from lumbar puncture revealed a normal CSF in 66 patients (15.6%), bacterial meningitis in a minority 73 (17.3%) and a pleocytosis in 283 (67.1%). In the latter case bacterial meningitis could not be proven and an organism could not be cultured. Typically, the CSF in such a situation revealed white cell counts < 500/cm3 with a predominance of polymorphs, an elevated protein level and, either normal or moderately depressed CSF glucose levels. Overall, therefore the CSF examination was either normal or non-diagnostic in 349 patients (82.7%). An organism was cultured in 42 of the 422 patients (10.0%) and this was predominantly in the group of infant patients with subdural empyema secondary to bacterial meningitis (83.3%). Of great concern and of significance the CSF pressure was only measured in 25 patients (5.9%) and when measured was raised (>20cm) in 15 (60%). [Table 2] As might have been expected 272 patients (64.5%) experienced clinical deterioration (drop in Glasgow Coma Scale (GCS) or development of a new focal sign) at some time following lumbar puncture. However, only in 81 patients (19.2%) could the deterioration predominantly be attributable to lumbar puncture. Twenty of the 81 patients died (4.7%). The fatalities predominantly occurred in patients with abscesses (hemispheric in 10 and cerebellar in 7). In the case of subdural empyema only three cases existed. [Table 3]
Many authors, including one of the present group  have strongly cautioned against the performance of lumbar puncture in patients with suspected or likely infective intracranial mass lesions due to the dubious value of the CSF analysis so obtained, and due to the inherent danger of clinical deteriorationprecipitated by a pressure cone [6-10,12,13,20-22].
The diagnostic value of CSF findings from lumbar puncture have proved to be limited. Carey et al found that approximately one-third of patients with proven brain abscess did not show any significant CSF pleocytosis, two-thirds had elevated protein levels, and glucose levels were lowered in one-quarter . Gregory et al noted that in three-quarters of brain abscesses, the CSF glucose level was normal, while Yang in an authoritative series of 400 brain abscesses reported that the CSF white cell count was not necessarily elevated, being < 10/mm3 in 21% of cases [21, 22]. Kratimenos et al in a series of 14 patients with multiple brain abscesses noted that the CSF obtained by lumbar puncture did not yield any positive cultures . Galbraith et al and Kaufman et al reported similar findings regarding CSF analysis in patients with subdural empyema [9,13]. It has been proposed that the arachnoid is a significant, hardy layer that protects the CSF from the subdural, extra-arachnoidal collection of pus in patients with subdural empyema. In patients presenting in a delayed fashion, the CSF may however exhibit an equivocal neighbourhood pattern due to prolonged contact of the pus with the arachnoid leading to arachnoiditis with resultant CSF changes .[9, 13]
In our series of 422 patients, CSF examination was normal or non-contributory in over 80% of cases. It has been our alarming experience that a normal or equivocal CSF examination often lulls the referring physician into complacency, who then treats the patient as one with viral meningitis or partially treated bacterial meningitis, leading to a delay in diagnosis and appropriate treatment. The dangers of lumbar puncture in patients with infective intracranial mass lesions have been well documented by many [8,9,10,12,13,18]. Gregory (1967), Duffy (1969) and Garfield (1969) have all described clinical deterioration following lumbar puncture[8,10,12]. Garfield described deterioration in the level of consciousness in the ensuing 48 hours in 41 of 140 patients who underwent lumbar puncture . Carey could attribute the deaths of 5 patients (5%) to lumbar puncture.  Chun et al described the death of 4 of 27 patients (14.8%) who died within 24 hours of undergoing lumbar puncture . Large series of patients with brain abscess or subdural empyema undergoing diagnostic lumbar puncture have previously been reported. In 1960, Bonnal et al reported 208 cases and, more recently, Yang reported 173 cases [21,22]. Both authors cautioned that lumbar puncture was of limited value and was hazardous. In our series, 20 deaths could be directly attributed to lumbar puncture (4.7%). Seven of the deaths were in patients harbouring cerebellar abscesses where supratentorial hydrocephalus has been documented as a concomitant adverse prognostic factor.  One of the 14 patients with infratentorial subdural empyema also died. Associated supratentorial hydrocephalus probably also being a contributory factor in the precipitation of the pressure cone . In addition to pressure cone, lumbar puncture may rarely also precipitate intracerebral or subdural haemorrhage [17,19].
We support the view of Ciarallo et al who cautioned against injudicious lumbar puncture in patients with periorbital cellulitis . We also concur with Garfield who advised that a lumbar puncture should not be performed in patients with meningeal irritation when a convulsion has occurred, or if papilloedema, hemisphere or cerebellar signs are present . Gower et al have recently described contra-indications to lumbar puncture as defined by CT, which would support the clinical view .
In addition to Garfield’s contra-indications to performance of lumbar puncture, we would recommend that a patient with meningeal signs and who also exhibits evidence of trauma, sinusitis or mastoiditis not undergo lumbar puncture but should rather be firstly investigated by CT.
In the absence of readily available CT facilities, we strongly recommend that such a patient be commenced on empirical, highdose, intravenous antibiotics until such time that a CTis obtained. CTis becoming an increasingly accessible modality in our region, with 8 public sector CTscanners already installed in the Province of KwaZulu-Natal, with 6 teleradiologically linked to Wentworth Hospital. Our report of 422 cases, which also represents the largest series reported to date, supports the view that lumbar puncture is of limited use in diagnosis of brain abscess and subdural empyema and, more over, is inherently dangerous and therefore students and practitioners should be advised, and taught, on the dangers. It is hoped that with the ever increasing availability of CT, the iatrogenic conversion of a patient with an eminently treatable brain abscess or subdural empyema into one with secondary irreversible brainstem damage from pressure cone could be avoided .
TABLE 2 : CSF ANALYSIS IN 422 PATIENTS UNDERGOING LUMBAR PUNCTURE
TABLE 3 : CLINICAL DETERIORATION FOLLOWING LUMBAR PUNCTURE IN 422 PATIENTS
|DETERIORATION ASSOCIATED WITH LUMBAR PUNCTURE||272||64.5|
|DETERIORATION DUE TO LUMBAR PUNCTURE||81||19.2|
|FATALITIES ATTRIBUTED TO LUMBAR PUNCTURE||20||4.7|