Understanding outcomes after neuraxial anaesthesia: time to turn the page.

For most of the last century, anaesthetists have debated the advantages and disadvantages of neuraxial block for patients undergoing major surgical procedures. While major clinical trials comparing outcomes with and without such techniques havenot been carried out since the early 2000s, 3 the anaesthesia literature has seen a proliferation of interest in this subject over the past 10 yr, as epitomized by a flurry of non-randomized studies comparing outcomes among patients undergoing surgical procedures with and without neuraxial block. In this issue of the BJA, Kate Leslie and colleagues present an insightful new analysis that adds to the growing knowledge base regarding potential associations between the use of neuraxial techniques and patient outcomes, while also highlighting the vital need for high-quality randomized controlled trials to move this dialogue forward. Using data from the POISE-2 trial, a large multicentre study that randomized high-risk surgical patients to aspirin or placebo and to clonidine or placebo, Leslie and colleagues compared outcomes among study patients who did and did not receive neuraxial blocks as part of their anaesthesia care. Leslie and colleagues found no statistically significant association between the receipt of any intraoperative neuraxial block and the odds of death, myocardial infarction, or stroke at 30 days, although a subgroup analysis did find spinal anaesthesia alone to be associated with lower odds of death or non-fatal myocardial infarction than general anaesthesia alone. A smaller accompanying analysis found no statistically significant association between postoperative epidural analgesia and death, myocardial infarction, or stroke. The work of Leslie and colleagues took advantage of rich research data collected prospectively across a large sample of patients. As such, the investigators were able to adjust for potential confounding as a result of certain clinical factors, such as baseline co-morbidities and medication use, beyond the degree typically permitted by administrative or registry-based data sources. The authors used rigorous propensity-score weighting methods to balance observable patient factors for patients who did and did not receive neuraxial techniques. The results presented, most notably the authors’ findings on outcomes with spinal vs general anaesthesia, align with the findings of several other recent observational studies. 13 14 16 Nevertheless, the paper demands careful interpretation. As with any non-randomized study, the authors’ findings are limited by the possibility of residual confounding as a result of variables not included in their risk-adjustment models. In the present instance, residual confounding could exist because of differences between groups in terms of the specific surgical procedures received, the severity of patients’ chronic illnesses, or variations in the quality of care delivered at hospitals that tended to use neuraxial anaesthesia more or less frequently. As the authors acknowledge, such limitations preclude a causal interpretation of their results and demand that any application of these findings to clinical practice take place with recognition of the potential biases inherent in non-randomized studies. The overall findings of Leslie and colleagues also diverge in important ways from the group’s previous analysis of data from the POISE-1 trial, which found intraoperative neuraxial block to be associated with increased odds of 30 day cardiovascular death, myocardial infarction, or cardiac arrest. While such differences between studies may have occurred because of random error as the authors point out, they could also relate to observed or unobserved differences between studies in terms of the overall health of patients who were and were not treated with neuraxial anaesthesia, or in terms of the specific surgeries they received. Likewise, such divergent findings could also stem from differences between studies in the proportion of patients who received