In this issue of Acta Neurochirurgica, Bhargava et al. [1]report on the efficacy of external ventricular drainage ofcerebrospinal fluid (CSF) for the management of raisedintracranial pressure (ICP) in patients with traumatic braininjury (TBI) refractory to conventional medical manage-ment. In contrast to clinical practice in many centres, in thisstudy CSF drainage via external ventricular drain (EVD)was used as a third-tier therapy rather than as one of thefirst approaches to the management of raised ICP. Only 2out of 16 patients treated in this manner required additionalthird-tier therapies (decompressive craniectomy or barbitu-rates therapy). In patients with raised ICP refractory tomedical management primarily treated by either decompres-sion or barbiturates therapy, two patients also required ad-ditional placement of an EVD. Thus, external CSF drainage,when used as third-tier therapy, would appear to be equallyeffective in controlling raised ICP as barbiturates or adecompressive craniectomy. This would seem to make astrong case for advocating use of a ventricular catheter forICP monitoring and when required external drainage of CSF.Previous reports have likewise provided evidence in sup-port of CSF drainage for controlling ICP. In a relatively smallcohort study on 20 TBI patients, Lescot et al. [2]showedthatcontinuous CSF drainage was effective not only in loweringICP levels but also in decreasing treatment intensity. Shore etal. [3] reported that continuous drainage was more efficientthan intermittent CSF drainage in paediatric TBI.The finding that external ventricular drainage of CSFconstitutes a useful ICP-lowering manoeuvre in patientswith raised ICP refractory to other measures is not new.Timofeev et al. [4] reported similar findings in 2008 andfurther demonstrated significant improvements in cerebralperfusion pressure, brain tissue oxygen and energy metabo-lism as evidenced by lactate/pyruvate ratios observed inmicrodialysis studies.The study by Bhargava et al. [1] now lends furthersupport for CSF drainage in TBI. Nevertheless, thestudy has its limitations and interpretation should bemade with caution. Besides its retrospective nature,and the lack of randomisation, there are two aspectsrelated to this study which are of particular concern.First, the authors focus on 33 patients with TBI andraised ICP, refractory to conventional medical therapy.All these patients had been diagnosed with severe TBIbut only one death is reported in this cohort, whichwould appear to be unbelievably low for a populationof severe TBI with raised ICP. The only possible expla-nations are either that the institution is incredibly goodat treating TBI or that in some way the patient popula-tion is selected. The latter would appear more likely.Secondly, and most importantly, the authors state thatpatients with sustained ICP increases were consideredfor the RESCUEicp trial (www.rescueicp.com). Ascanberead clearly from Fig. 1, patients treated by additional CSFdrainage were excluded from participation in the RESCUEicpstudy. Consequently the local procedures imply a selection inpatientsconsideredforrandomisationintoRESCUEicp.Intro-ducing a selection bias in this way jeopardises the principle ofequipoise underpinning RESCUEicp. Furthermore, this selec-tion illustrates how approaches to ICP monitoring and itstreatment may confound consistent use of enrolment criteria,not only across study centres but even within a single centre.Although the study reported by Bhargava et al. [1]isinteresting and provides evidence in support of CSFdrainage, these two concerns are such that they leaveme with an uncomfortable feeling, as in my opinionthey are illustrative of possible risks and consequencesof selection bias. It is highly likely that selection biasoccurred within the cohort study and local study