To the Editor Robertson et al 1 conducted a crossover randomized clinical trial of gabapentin (GBP) vs pregabalin (PBG) in chronic sciatica (CS). We congratulate the authors for their effort toward improving CS management. However, some methodological issues hinder the conclusions in this trial.First, we believe that a crossover design may be inappropriate to answer the question of interest. Crossover trials assume that the disease is stable and the effects of an intervention do not carry over from one period to the next. Chronic sciatica intensity fluctuates over time and the 1-week washout period may be insufficient, meaning that the observed effects may be attributable to a longer clinical effect or the disease course and not exclusively to the differences between GBP and PBG. More important than surpassing a given number of half-lives, a preliminary test for differential carryover effects between the drugs should be made and the sequence and period effects should be accounted for to ensure washout period adequacy. Otherwise, the carryover effect is not excluded and only the first parallel phase of the study can be interpreted.Second, the choice of comparator is also important. Although there are some data supporting the efficacy of GBP and PBG in CS, we believe a placebo comparator is still mandatory given the current low level of evidence for these 2 drugs and the high placebo response in pain trials. 2,3 Additionally, safety analyses of randomized clinical trials may be also confounded by the nocebo effect, 4 which is nonnegligible in neuropathic pain trials. 5 In this trial, 12 participants (67%) reported 38 adverse events, with significantly more adverse events reported with PBG vs GBP (31 vs 7; P = .002). This can be partly a consequence of the 20% difference in group size. Hence, a common placebo comparison would also benefit the safety assessment of the drugs.Finally, the small number of participants ( 18) from a single center limits the generalizability of any findings. Despite the significant visual analog pain intensity scale reduction, a simple post hoc sample size calculation (SD, 1.35) shows that 48 participants would be required to demonstrate the same effect size. We believe that placebo-controlled, parallel-group multicentered trials are still needed for firmer conclusions on the benefit of new therapeutic interventions in CS.