Subarachnoid haemorrhage or traumatic lumbar puncture. Differentiation by cerebrospinal fluid parameters in a multivariable approach.

Lumbar puncture (LP) is recommended in patients with thunderclap headache and negative computed tomography to rule out spontaneous subarachnoid haemorrhage (SAH). Blood contamination of cerebrospinal fluid (CSF) due to traumatic LP poses a diagnostic dilemma. Therefore, routine CSF parameters were investigated to distinguish between SAH and a traumatic LP. CSF red blood cell (RBC), white blood cell (WBC) count, total protein, CSF colour and supernatant were used for group comparisons of patients with SAH and 'symptomatic controls'. Due to variable time intervals between bleeding onset and LP in SAH patients in contrast to patients with traumatic LP, where blood contamination of CSF occurs at the time of LP, CSF variables were adjusted for decay in time to allow comparability. Logistic regression analysis identified bloody CSF [odds ratio (OR) 32.6], xanthochromic supernatant [OR 15.5] and WBCadjusted [OR 4.5 (per increase of 100/µl)] as predictors of SAH, while age, sex and CSF total proteinadjusted were no predictors. Optimal cut-point of RBCadjusted (determined at day 1 after bleeding) was > 3667/µl to identify SAH patients with a 97% sensitivity and 94% specificity. Combination of low RBC and clear CSF supernatant was found in none of SAH patients. Combined CSF RBC count and CSF supernatant reliably distinguished traumatic LP from SAH.