Cerebrospinal fluid leakage complicating skull base fractures: analysis of 81 cases

The aim of this study was to evaluate the results of conservative and surgical management options for traumatic cerebrospinal fluid (CSF) leakage complicating skull base fractures. The subjects were 81 patients who were treated between 1996 and 2003 for CSF leaks that had persisted for 24 h or longer after head injury. For each case the medical records were reviewed, and the data collected were as follows: demographic features, clinical and radiological findings, management options, complications and outcome scores. Analysis was done with patients grouped according to Glasgow coma scale (GCS) score at admission (<= 8 vs >8), and findings for three treatment methods (conservative, CSF drainage, surgery) were evaluated. In 32 cases (39.5%), the CSF leakage resolved spontaneously, and the mean hospital stay for these patients was 14 +/- 11 days. Twenty-four patients (29.6%) were treated by CSF drainage, and seven of these individuals ultimately required surgery to close the leak. Hospital stay was 17 +/- 7 days. Twenty-five patients (30.9%) underwent surgery as the initial treatment step, and the mean hospital stay for these individuals was 15 +/- 9 days. The large majority (74.2%) of patients with admission GCS scores <= 8 had poor outcomes. Compared with this group, a greater proportion of the CSF leaks in the patients with admission GCS scores >8 resolved spontaneously. The factors that had a critical influence on outcome in this series were level of consciousness on admission and presence of additional intracranial pathology associated with CSF leakage within cases of traumatic CSF fistulae due to skull base fractures. Treatment decisions should be dictated by the severity of neurological decline during the emergency period and the presence/absence of associated intracranial lesions. The timing for surgery and CSF drainage procedures must be decided with great care and with a clear strategy. The authors offer a treatment algorithm.