Objective: To determine the incidence of traumatic lumbar puncture (LP).
Methods: A retrospective study was conducted at an urban, university tertiary care referral center with 50000 annual emergency department (ED) visits. The study population included all patients who had cerebrospinal fluid (CSF) samples sent to the laboratory between August 15, 2000, and August 14, 2001. The numbers of red blood cells (RBCs) recorded in the first and last CSF tubes, the location where the LP was performed, and the discharge summary and the discharge diagnoses from the particular visit were obtained. All patients with intracranial pathology and CSF obtained via neurosurgical procedure or fluoroscopic guidance were excluded from the study group. Given no clear definition of traumatic LP in the literature, the incidence of traumatic LP was calculated using a cutoff of greater than 400 RBCs (visual threshold for bloody fluid) and 1000 RBCs (arbitrary threshold selected by other authors) in CSF tube 1. Proportions were compared using chi-square statistics.
Results: Seven hundred eighty-six CSF samples were recorded over one year. Twenty-four samples were obtained from patients with intracranial pathology or were obtained via a neurosurgical procedure. Of the remaining 762 CSF samples in the study population, 119 (15.6%) were traumatic using a cutoff of 400 RBCs, and 80 (10.5%) were traumatic, using a cutoff of 1000 RBCs in tube 1. Five hundred three LPs were done in the ED and 259 were attributed to all other locations in the hospital. Using a cutoff of 400 RBCs, the incidence of traumatic LP in the ED was 13.3%, compared with 20% in the rest of the hospital (p < 0.025). Similarly, using a cutoff of 1000 RBCs, the incidence of traumatic LP in the ED was 8.9%, compared with 13.5% in the rest of the hospital (p = 0.1). The incidence of "champagne taps" (defined as zero RBCs in the first and last tubes) in the ED was 34.4%, compared with 24.3% in the rest of the hospital (p < 0.01).
Conclusions: The incidence of traumatic lumbar puncture is approximately 15% using a cutoff of 400 RBCs and 10% using a cutoff of 1000 RBCs. In this study, the rate of traumatic lumbar puncture was significantly less (with a cutoff of 400 RBCs) and the rate of champagne tap was significantly greater for LPs done in the ED compared with the rest of the hospital.