Previous studies suggest that approximately 10 percent of subarachnoid hemorrhages are missed by noncontrast computed tomography (CT). The Stroke Council of the American Heart Association strongly recommends that lumbar puncture be performed if CT scans are negative in patients who present with suspected subarachnoid hemorrhage. Although cerebrospinal fluid analysis may provide important information, the discomfort, time and expense of lumbar puncture may inhibit its use. Morgenstern and associates assessed the adequacy of CT imaging in excluding subarachnoid hemorrhage and the value of various cerebrospinal fluid measurements in the diagnosis of subarachnoid hemorrhage.

This prospective study included 107 patients who presented with “the worst headache of their life.” All of the patients underwent CT scan of the head, and evidence of subarachnoid hemorrhage was noted in 18 patients (16.8 percent). If CT scans were negative, lumbar puncture was performed. Lumbar puncture was performed in 79 of the 89 patients in whom CT scanning did not show evidence of hemorrhage. It was not performed in 10 patients who refused the procedure. Subarachnoid blood was found in two of the 79 patients (2.5 percent) who underwent lumbar puncture.

Patients were considered to have subarachnoid hemorrhage on the basis of spinal fluid results if the sample contained more than 1,000 red blood cells and no decrement in the count of more than 25 percent was noted from the first to the last tubes. In addition, analysis of cerebral spinal fluid had to meet the following criterion: positive results on spectrophotometry, visual xanthochromia or d-dimer assay.

The authors conclude that modern CT imaging will exclude 97.5 percent of subarachnoid hemorrhages in patients who present with the worst headache they have ever experienced. Subarachnoid hemorrhage will be detected by analysis of cerebrospinal fluid in approximately 2 percent of patients who present with “worst headache” symptoms and have negative findings on CT scan. Because of the small sample in this study, the calculated 95 percent confidence interval for missing subarachnoid hemorrhage on CT scan was 0.3 to 8.8 percent. This number is lower than that reported with earlier-generation CT scanners. The authors recommend that patients with the worst headache of their lives undergo lumbar puncture if noncontrast CT scanning fails to show evidence of subarachnoid hemorrhage.