• cerebrospinal fluid
• guidelines
• oxyhaemoglobin
• subarachnoid haemorrhage

We note with interest the recent “Best Practice” article on cerebrospinal fluid (CSF) spectrophotometry in the diagnosis of subarachnoid haemorrhage (SAH) by Dr Cruickshank.¹ As a group that has produced a set of proposed national guidelines for the practice of spectrophotometry,² we wish to highlight several important differences between the two sets of recommendations.

Most importantly, Dr Cruickshank concludes that, as long as a CSF sample containing up to 40 000 × 10⁶ erythrocytes/litre is centrifuged within 15 minutes, no oxyhaemoglobin will be seen in the supernatant after centrifugation, and that within this cell count and time constraint, the presence of oxyhaemoglobin in CSF is supportive of SAH. This is entirely consistent with her in vitro data,³ although the only practical way of achieving CSF delivery within this time would appear to be by pneumatic tube, itself a cause of artefactual haemolysis.⁴ However, there are other in vitro data that allow for a longer time lapse before centrifugation. Thus, we know that variously 10 000 × 10⁶ erythrocytes/litre can be left for up to 30 minutes⁵ and 4000 × 10⁶ erythrocytes/litre can be left for up to 24 hours⁶ without oxyhaemoglobin appearing in the supernatant.

It is when we come to in vivo work that the data are conflicting. Again, Dr Cruickshank's conclusions are consistent with her data from patients undergoing spinal anaesthesia—that red blood cell counts from < 5 to 2215 × 10⁶/litre in CSF samples centrifuged within 40 minutes of puncture resulted in no detectable oxyhaemoglobin³—and also with data from Barrows et al.⁷ Such data are nevertheless at variance with those of Fahie-Wilson and Park, who found that red blood cell counts from 64 to 705 × 10⁶/litre in CSF samples centrifuged as soon as possible after receipt resulted in significant oxyhaemoglobin detectable in the supernatant.⁶ This set of experiments was performed because initial observations were that many CSF specimens taken for spectrophotometry in cases of suspected SAH showed the presence of oxyhaemoglobin totally out of keeping with cell counts or time lapse before centrifugation.

A survey of spectrophotometry findings against outcome in computed tomography negative suspected SAH from four participating centres (R Beetham et al, unpublished data, 2001) undertaken by our group has indicated that out of scans on 740 patients, 204 showed detectable oxyhaemoglobin without increased bilirubin. Thirty of these 204 patients proceeded to angiography and only two aneurysms were found. It has to be concluded that angiography in all 204 patients on the basis of the finding of oxyhaemoglobin alone would have been unwarranted when the required resource and known complication rate are considered. It is on this evidence, more in keeping with the in vitro data of Fahie-Wilson and Park than that of Cruikshank, that we make our recommendation that oxyhaemoglobin is discounted as evidence to support SAH.

On a second matter, we differ on the assertion that the occurrence of bilirubin alone without oxyhaemoglobin is a rare occurrence in computed tomography negative, angiographically confirmed aneurysms. In the above series, there were 11 patients with increased bilirubin who showed aneurysms on angiography and proceeded to surgery. Two of these were negative for oxyhaemoglobin and a third demonstrated only a trace of oxyhaemoglobin (absorbance above baseline of 0.012 AU). We agree however that an increased bilirubin without oxyhaemoglobin is found more frequently in association with an increased serum bilirubin than with SAH.

Finally, although we welcome the recommendation that non-haemorrhagic bilirubin is taken into account whenever bilirubin is detected, no mention is made of a reference range for bilirubin against which a value can be judged to be normal or abnormal. Quantitation is only of use if such a reference range is provided. We recommend the net bilirubin absorbance at 476 nm, as advocated by Chalmers,⁸ and provide a reference range based on angiographic outcome. It is also our experience that the correction for non-haemorrhagic bilirubin works well when the cause of the increased CSF bilirubin is an increased serum bilirubin, and less well when the serum bilirubin is normal and the cause of the increased CSF bilirubin is an increase in CSF protein.

Our proposed guidelines are available through UKNEQAS for Immunochemistry.² They can be accessed through www.immqas.org.uk and we recommend them to your readers.