We are pleased to see the evolution of the UK guidelines in this area (Ann Clin Biochem 2008;45:238–244), although contest the view that analysis of bilirubin in CSF using diazo methods has not been adequately validated in this context and should not be used.
1
We developed such an approach to diagnosis,
2,3
now used routinely in our laboratory to identify approximately 90% of samples that do not need to be submitted to spectrophotometry. Our approach employs the measurement of CSF bilirubin on an automated instrument, using the Jendrassik and Gróf method, calibrated to measure lower concentrations.
2
We found that at a CSF bilirubin cut-off of 359 nmol/L, based on our upper reference interval in CSF, there was 100% negative predictive value
2
for a net bilirubin absorbance greater than 0.007, i.e. positive initial scan. In a follow-up study,
3
the validity of this cut-off was confirmed, with no evidence from clinical records to suggest that any case of SAH had been missed. The accompanying editorial,
4
acknowledged that it could thus act as an initial screening method to eliminate those samples that would subsequently prove to be negative spectrophotometrically, thereby allowing the majority of samples to be screened out by a procedure potentially available on a 24-hour basis. An added advantage is that small volumes of sample, around 100 μL (compared with 0.5 mL needed for disposable cuvettes for absorbance scans) are sufficient for the CSF bilirubin determination.
3
Preliminary data suggest some negative interference of haemoglobin on CSF bilirubin, which requires further study and quantitation, although it also affects NBA derived by spectrophotometry. For operational purposes, CSF samples are also submitted to spectrophotometry if our analyser haemolytic index is greater than 2.0 mg/dL.
We believe that this methodology should be easily transferable to similar analytical platforms, although it is imperative that each laboratory should meticulously validate its own analytical performance. It offers a robust test and may be particularly useful in peripheral laboratories as a screen to identify those samples that need to be confirmed by spectrophotometry.
3
