Around one of every ten patients presenting with a sudden severe headache in general practice has a subarachnoid haemorrhage (SAH) [1]. Computed tomography (CT) scanning of the brain can confirm the diagnosis in most but not all instances. In particular, in patients who present late, CT can be non-diagnostic, because the sensitivity of CT for subarachnoid blood decreases over time from nearly 100% in the initial hours [24] to 50% after 1 week [4]. To exclude SAH in patients with a negative CT scan, cerebrospinal fluid (CSF) is investigated to detect bilirubin pigments. The added value of CSF examination in patients with a negative CT is, however, a matter of debate because of the small likelihood of SAH in these patients [5]. In addition to the small likelihood of SAH, the likelihood of a ruptured aneurysm may be even smaller. Patients with nonaneurysmal subarachnoid haemorrhage, such as patients with perimesencephalic SAH, have usually headache only and no focal deficits or disturbed consciousness, and may present later than patients with more severe clinical symptoms. Patients with nonaneurysmal subarachnoid haemorrhage may therefore be overrepresented in late-arriving patients with a negative CT scan. We studied clinical characteristics and the chance of harbouring an aneurysm in this group of patients.


Patients were retrieved from the SAH database from the University Hospital Utrecht and from the files of the Laboratory of Clinical Chemistry, including CSF spectrophotometric investigations in the period 2002–2007. We included patients who had a history of sudden severe headache and a negative CT scan of the brain at initial assessment, but who had extinction of bilirubin >0.05 (at wavelength 458 nm) in their CSF, according to the guidelines for analysis of cerebrospinal fluid for bilirubin in suspected subarachnoid haemorrhage [6].

All patients with the diagnosis of SAH based on a positive CSF examination were investigated with CT angiography (CTA), magnetic resonance imaging (MRA) or conventional angiography to confirm or exclude an aneurysm as the cause of the bleeding.

The CSF from patients inside our hospital was protected from (day)light by wrapping them in aluminium foil, and centrifuged at 1,500 rpm during 10 min. The supernatant was stored at 4°C until analysis. CSF investigations were performed using a Beckman DU 650 spectrophotometer (Beckman Coulter, The Netherlands).

For the included patients, we describe clinical characteristics and the results of the radiological investigations. Clinical condition of the patients was assessed according to the World Federation of Neurological Surgeons (WFNS) scale on admission [7]. We analysed data according to the interval of symptoms and presentation in the hospital: (1) early presentation (0–3 days), (2) intermediate presentation (4–7 days) and (3) late presentation (8–14 days). The modified Rankin scale was used to describe outcome in all patients; we dichotomized grade 0–3 as good outcome and grade 4 and 5 as poor outcome [8].


During the study period 30 patients had been admitted with sudden severe headache or neck pain and a brain CT scan without subarachnoid blood but CSF stained with bilirubin. In total, 775 patients with SAH were admitted during this period in our hospital. Sixteen patients were men, and 14 were women. Median age was 50.9 years (range 27.3–72.9 years). On admission, WFNS score was 1 in all but one patient (who had WFNS score 2 and a slight hemiparesis). Of the total group, two patients (7%) presented early, within 3 days. Further investigation did not show any aneurysms in this group. The cause of the delay in the other patients was due to the patient (in 8), the doctor (in 11), or both (in 2) or unknown (in 7). Eighteen patients presented within 4–7 days, and eight of these (44%) had an aneurysm. Ten patients presented late (8–14 days), and five of them (50%) proved to have an aneurysm. Thus, in the total group of 30 patients, an aneurysm was found in 13 (43%) patients (Table 1).

Table 1 Chance of aneurysm and outcome for included patients with sudden headache

The CT scans of patients from the outside hospitals who were referred to our hospital were judged as normal by the radiologist from the outside hospital, but after revision by the neuroradiologist in our hospital (a tertiary referral hospital) slight abnormalities in 4 of 30 included patients were detected: 1 patient had a small hyperdensity in the left cerebellum, caused by an underlying arteriovenous malformation that was confirmed by angiography. This finding was not appreciated on the initial non-contrast CT scan by the radiologist from the outside hospital. One other patient had some sedimentation of blood in the posterior horns. Two other patients had ambiguous findings: suspicion of small amount of blood in the pentagon in two patients.

One other patient had on further follow-up with MRI a small hyperdensity in the parieto-occipital gyrus (not visible on contrast CT scan). Further investigations revealed an right subcortical aberrant vein in this patient draining to the sagittal sinus, without aneurysm.

All 16 patients without an aneurysm or arteriovenous malformation of the cause of their headache were investigated with CTA, MRI or MRA or additional four-vessel angiography (11 patients). In none of these 11 patients was focal or diffuse segmental arterial vasoconstriction shown.

Haematocrit was normal in 29 patients and not known in 1 patient. Serum bilirubin was measured in only one patient (who was pregnant).

Treatment of the aneurysm was done by coiling in nine patients and clipping in two. Two patients were not treated: one patient who developed ischaemia the day after admission and was in poor clinical condition, and one other patient who refused further angiography after MRA. The arteriovenous malformation in the cerebellum in one other patient was embolized successfully.

Eleven patients with an aneurysm were alive after 3 months and two died, because of rebleeding (1 patient) and secondary ischaemia (1 patient). One patient had a poor outcome with major neurological deficits including paresis and aphasia (because of secondary ischaemia). All patients without an aneurysm detected were alive after 2–7 years of follow-up and had good outcome without further episodes of SAH (Table 1).


In our series of patients with sudden headache with a negative brain CT scan but a positive lumbar puncture with bilirubin pigments, nearly half had a ruptured aneurysm. The group of patients presenting within 3 days who did not show an aneurysm was very small. We think that, in general, the chance of an aneurysm is small in patients who present early after the onset of headache and who have a negative CT scan and a positive lumbar puncture because of the overall small chance of very early reabsorption of blood on CT. In patients who present later, the chance of reabsorption is larger, with blood being no longer detectable on CT. For patients presenting later, the results of our study show a high yield in terms of aneurysms and underscore the importance of examining CSF in patients with acute headache but a ‘negative’ brain CT scan.

The outcome of patients with a negative CT but a positive lumbar puncture and an aneurysm detected was comparable to that of patients with SAH detected on CT scan admitted in a good clinical condition [911]. The occurrence of rebleeding, secondary ischaemia and poor outcome in our series shows that the aneurysms detected were truly ruptured aneurysms and not incidental aneurysms found after a false-positive CSF exam.

In our study, 2 of the 30 CT scans that were read as normal in a local hospital were positive for SAH when reviewed in our hospital, and 2 other CT scans were assessed as ambiguous. In a recent paper on examination of patients with acute headache in tertiary care emergency departments in Canada, negative predictive value of CT scanning within 6 h after onset of headache was 100% (99.5–100%) [12]. The data from our study show that the negative predictive value of CT scanning is less in local hospitals, at least in The Netherlands, because several CT scans that were read as negative in local hospitals were positive or at least ambiguous after revision in our hospital. These results reflect clinical practice. In referring hospitals and during out-of-office hours, sensitivity of CT readers is apparently less optimal than during office hours in a highly specialized centre [4]. Another reason for false-negative CT scans is a low haematocrit (<30%) [4]. In our patients haematocrit was normal. A false-positive CSF examination after a negative CT could also be explained by a high serum bilirubin, transferred through the blood–brain barrier, resulting in a high CSF bilirubin [13]. This chance is, however, very small in patients without clinically overt jaundice, which is correlated to a lower range of hyperbilirubinaemia than xanthochromia.

In conclusion, CSF analysis including spectrophotometry, according to the guidelines [6], is mandatory in patients with acute headache but a CT scan negative for subarachnoid blood to rule out an SAH and an underlying ruptured aneurysm. This holds true in particular in patients who present late after the onset of headache. In patients who present early after the onset of headache, CT scans may be falsely considered negative because of insufficient scan techniques or less experience of the radiologist. Thus, also in patients presenting early after onset of headache, CSF examination remains pivotal after a CT scan being read as negative. For this category of patients immediate referral to a tertiary care centre might be an alternative, although this will increase the workload in these centres considerably.