Editor,Regardless of possible biases related to post hoc testing, a sub-analysis can be appropriate when the primary point of interest has been reached and one wants to inspect whether sub-groups might have reacted to treatment differently than the main group. 1 The data presented in the recent "sub-analysis" by Aasvang et al 2 derive from a remarkable trial 3 addressing the question whether discharge from the post-anaesthesia care unit might be advisable and safe without assessment of lower limb motor function after total hip arthroplasty (THA) or total knee arthroplasty (TKA) performed under spinal anaesthesia. As the frequency of a successful fast-track recovery was the primary endpoint, 3 a sub-analysis of this parameter might be suitable, but not of failed spinal anaesthesia (FSA). 2 From the fairly large number of analysed patients (n = 1451), 57 patients (3.9%) fulfilled the criterion for FSA, ie, a patient required general anaesthesia at any time during surgery. 2 Supplemental analgesia or sedation with remifentanil or propofol per se was not considered as FSA, unless laryngeal mask placement or tracheal intubation was deemed necessary. As clinicians, we appreciate this definition of FSA. However, between the 5 participating centres, 3 the failure rate varied significantly from 2.2% to 7.1%. 2 Could the handling of intraoperative pain and problems have differed substantially from centre to centre and thus influenced the rate of FSA? A third of FSA occurred in TKA and two-thirds in THA. How should the reader interpret this difference without more details about the spinal technique? Only the rather small amounts of bupivacaine (5 mg/mL) were presented (2.2 [1.8-2.5] mL [mean [95% confidence intervals]]), but combined for TKA and THA. These 2 procedures were put under one umbrella without mentioning the distribution between plain and hyperbaric bupivacaine, the positions in which the spinal anaesthetics were injected, and for how long the patients were kept, eg, in the lateral position before return to the supine position. All these factors could have affected the efficacy of spinal anaesthesia.There are no data about the time when FSA was diagnosed after the start of surgery. 2 By reviewing the 57 FSA anaesthesia charts for the starting time of general anaesthesia and by using this parameter as a surrogate, it might be possible to discriminate between cases where the spinal turned out ineffective soon after incision and those where the spinal wore off gradually.Failed spinal anaesthesia was associated with no other factors but longer procedure times and younger age. 2 This corroborates what has been known before; namely, that the duration of bupivacaine spinal anaesthesia increases with increasing dose 4 and growing age. 5,6 The findings 2 do not allow the generalisation that intraoperative FSA is a relatively frequent outcome in TKA and THA patients. We are not sure how to interpret the authors' inference that pre-operative identification of risk factors is not feasible. When dealing with younger patients,...