Agreement between lactate measurements by blood-gas-analyzer and routine-CSF-measurement was high as indicated by the Pearson-correlation index of 0.94. Formally, the resulting Bland-Altman diagram does indicate that the methods compared are not fully interchangeable. However, even though no complete agreement between both methods was found with single values minimally outside the predefined boundaries of agreement (Fig. 3) unlike previously published results our values interestingly demonstrate high agreement for the full range of lactate values i.e. the correlation did not decrease with increasing lactate values. Hence, this method appears robust as indicator of CSF-alterations for the clinically relevant range of CSF-lactate values. Reasons for differences between measurements may include different times between taking CSF and its analysis which was negligible (around 2 minutes) in case of measurements by blood-gas-analyzers but could take several hours due to internal transport- and processing-times within the hospital. Due to its instability at room temperature prompt measurement of lactate (within 60 minutes after sampling) generally is preferable. In fact, processing time related decay may affect CSF-lactate values and indeed, the mean of CSF-lactate measurements by the standard analyzer including the aforementioned delay was lower (4.13 mmol/l) as compared to the results obtained by the blood-gas-analyzers (4.19 mmol/l) even though this difference was not significant. In order to address this source of possible bias lactate should have been measured simultaneously by blood-gas-analyzer and the reference method which this study did not account for. But obviously, immediacy is a systematic advantage of point-of-care-testing. Moreover, the usefulness of timely and frequent lactate-measurements as offered by blood-gas-analyzers becomes particularly apparent in light of those three patients who have had received a drainage for non-infectious maladies of the central nervous system and who later developed a catheter-associated meningitis. Rapid bed-side lactate-testing detected increases in lactate prior to our reference method which led to more immediate and accelerated anti-infective treatment as compared to the standard procedure.
Importantly, CSF lactate is an unspecific metabolite and may be elevated in a range of diseases including bacterial and fungal meningitis as well as Meningeosis neoplastica [11, 12]. The predominant source of CSF-lactate even in bacterial meningitis is the host organism, i.e. neuronal and immune cells, as studies differentiating D-lactate (prokaryotic) and L-lactate (eukaryotic) in CSF have shown [13] and which is supported by the positive correlation between leucocyte count and lactate levels in patients with meningitis [14]. Still, a cut-off for CSF-lactate of >3.5 – 4.2 mmol/l has demonstrated a high reliability in predicting a non-viral meningitis as confirmed by a recent study [3]. On the other hand, CSF-lactate values alone turned out to be of relatively low predictive value with respect to the development of a postsurgical meningitis in neurosurgical patients [7, 8]. This has been confirmed by a recent study of neurosurgical pediatric patients and retrospective analysis of 215 CSF-samples. Authors stated especially that the “added value of LCSF for diagnosing CSF infections in children with a history of neurosurgical procedures is unclear and may be influenced by the extent of blood in the CSF” [15]. Indeed, a general limitation of CSF-lactate as a predictor of CSF-infection is possible contamination of CSF by blood-derived lactate. Almost half of our patients had suffered from subarachnoid hemorrhage often introducing a high amount of blood into CSF. While only mild effects of blood-contamination on CSF-concentrations of amino-acids and a group of vitamins were found [16] blood contamination influences CSF protein diagnostics [17]. Still, in a more experimental setting the addition of different amounts of blood to otherwise normal CSF of 33 adults did not influence the lactate level significantly, but led to higher glucose measurements [18]. In general, single CSF-lactate-measurements in post-neurosurgical patients with external CSF-drainage especially in case of major blood-contamination may be less reliable in predicting CSF-infection as compared to otherwise non-contaminated CSF. Nonetheless, by providing a longitudinal view regular postsurgical measurements of CSF-lactate may help to readily detect inflammatory events in the CSF. Additionally, point-of-care blood-gas-analyzers are able to simultaneously deliver glucose measurements in CSF. As measurements of blood-glucose by blood-gas-analyzers is part of the routine for nearly any ICU-patient a pair of CSF- and blood-glucose is easily generated. This is important since CSF-glucose on its own has a rather poor while a low ratio of CSF-to-blood-glucose of <0.4 has a well-established predictive value for detecting a non-viral meningitis [19]. Importantly however and in contrast to the CSF-lactate-measurements we did not control for the accuracy of the blood-gas-analyzer derived CSF-glucose-measurements by comparing them to results generated by an established CSF-glucose-analyzer. Hence, these values so far remain not validated. Still, a recent study demonstrated that point-of-care-glucometers can reliably measure CSF-glucose and help to detect meningitis with a sensitivity of 94% and a specificity of 91% when using a cut-off for the CSF/blood glucose ratio of 0.46 [20]. As outlined before the major advantage of the lactate measurements by a blood-gas-analyzer is the immediate availability of results. Additionally though, broad availability of blood-gas-analyzers would facilitate diagnostic capabilities in resource limited situations as has been previously suggested and is currently under investigation for point-of-care CSF-glucometry [21, 22].