Association between anaesthetic technique and unplanned admission to intensive care after thoracic lung resection surgery: a reply

We are grateful to Stone and Anwar 1 for their thoughtful response to our recent article 2. As they describe, in 7431 thoracic surgical cases drawn from 16 UK centres over a 2-year period, we demonstrated that both thoracic epidural blockade and a total intravenous anaesthetic (TIVA) technique appeared to have a risk-modifying effect when adjusted for age, sex, resection type, resection side and surgical approach (thoracotomy vs. VATS). Throughout our narrative, however, we have gone to some lengths to highlight that while ‘thought provoking’, our findings are of association, not causation, and we must therefore be careful not to over-interpret the results. As Licker describes in his accompanying editorial 3, ‘the Achilles heel of observational studies is the presence of confounding factors related to population characteristics, as well as to treatment and outcomes’. We quite agree with both commentators; there is no doubt that there are many potential confounders which might have influenced our findings. Both Stone and Anwar, and Licker highlight the potential confounding effect of anaesthetist experience. In a previous UK Association of Cardiothoracic Anaesthesia and Critical Care national audit, Papachristofi et al. demonstrated negligible influence of the anaesthetist on outcome in patients undergoing cardiac surgery 4, while in patients undergoing orthotopic liver transplantation, Hofer et al. 5 found that ‘an anesthesiologist's level of experience has a significant effect on outcomes’. To our knowledge, no such data exist for the thoracic surgical population, but it is intriguing to speculate that the regular thoracic anaesthetist (who may in turn be arguably more likely to use both TIVA and thoracic epidurals) may influence outcome by his or her close attention to lung isolation, protective ventilation techniques, fluid management and analgesia. Stone and Anwar also highlight their concerns that anaesthetists may be more likely to seek elective ICU admission in patients with a thoracic epidural in situ. We think this is unlikely to have confounded our results; firstly, if this was the case, this would only serve to magnify the apparent risk-modifying effect of epidural analgesia and secondly, as described in our paper, we defined ‘unplanned intensive care admission’ for the purposes of the study as the hard endpoint of ‘need for unplanned mechanical ventilation or renal replacement therapy’, with the explicit intention of excluding elective, precautionary or bed availability-related ICU admissions. Rather than focusing simply on a search for confounding influences to explain our findings, we must also be open minded to the potential that the association observed is causative; as discussed in the manuscript, there is clear biological plausibility for both TIVA and thoracic epidurals to reduce risk. To this end, we wholeheartedly agree with Stone and Anwar, as well as Licker that further well-conducted multicentre randomised controlled trials are required to answer these age old questions. We are at an exciting juncture for thoracic anaesthesia research. A National Institute of Health Research-funded trial 6 is currently underway in the UK and will randomly assign over 1000 patients undergoing elective thoracotomy to either thoracic epidural or PVB. Although the incidence of chronic pain is the primary outcome of this study, it contains a robust complications dataset and will be significantly larger than the cumulative population included in Yeung et al.'s Cochrane review comparing thoracic epidurals and PVB 7. Meanwhile, the Peri-operative Medicine Clinical Trials Network is actively seeking funding for a large multicentre randomised controlled trial of TIVA vs. volatile in patients undergoing major elective surgery, a trial in which it is hoped thoracic surgical patients will constitute a major part. As Licker concludes, our results should be ‘considered as hypothesis generating rather than definitive’ 3 and we eagerly await the results of these large trials to further inform this important clinical conundrum.