Time of day of cardiac surgery and postoperative outcomes: exploratory findings require caution

We read with interest the article by Kitchen et al., which examines the association between timing of planned cardiac surgery and clinical outcomes 1. It should be praised for highlighting the usefulness of big data in identifying patterns that may inform clinical practice. However, we wish to offer several clinical and methodological considerations to aid interpretation. The authors claim a modest increase in cardiovascular mortality with surgical ‘knife-to-skin’ time in the late morning and suggest peri-operative, organisational or circadian explanations. Knife-to-skin time represents an endpoint of a long and variable pre-operative sequence (operating theatre start time, setup, team brief, sign-in, anaesthesia, echocardiography, surgical preparation and time-out). Theatre start times vary across institutions and list progress depends on factors including postoperative bed availability and infrastructure. The study period spanned the H1N1 influenza pandemic, which increased pressure on the critical care services mandated for inclusion and predated the current era of enhanced recovery after cardiac surgery. Anaesthesia varies and anticipated or unanticipated issues can delay knife-to-skin time independently of surgical complexity or comorbidity. This limits the scope for adjusting the time of incision. Case distribution is deliberately non-homogeneous throughout the day. Complex procedures are often scheduled earlier after considering human and organisational factors. This is reflected in the increased EuroSCORE 2 values, cross-clamp times and proportions of non-isolated and ‘redo’ procedures in the late morning group. Skill and experience may affect confidence listing between surgeons. Additional confounders, such as cancellations, previous on-call commitments and surgeon-specific listing preferences, are difficult to capture but likely to be impactful. Afternoon cases are rescheduled more frequently, further complicating interpretation. The ‘early’ vs. ‘late’ morning distinction warrants scrutiny. Complexity affects preparation times for the surgical, anaesthetic and perfusion teams, and many institutions schedule standalone ‘all-day’ cases. A later knife-to-skin time may represent appropriate planning rather than delay. The pragmatic epochs include knife-to-skin times that are uncommon in planned cardiac surgery at many centres, raising questions about their representation of the clinical day. Scheduled urgent procedures were not excluded, although their pathophysiology (e.g. infective endocarditis) may differ substantially from that of elective surgeries; it would therefore be informative to know whether findings persist if these were omitted. Furthermore, when considering inflammatory or circadian mechanisms, factors such as prolonged fasting, medication withholding and peri-operative anxiety may also contribute, all of which are influenced by anaesthesia and perfusion-related factors and may not involve cardiopulmonary bypass. From a statistical perspective, the reported association is small and fragile. The key finding lies at the threshold of significance, with an adjusted hazard ratio for death due to cardiovascular disease of 1.18 (95%CI 1.00–1.39) in the late morning group. All other clinically relevant outcomes (e.g. major adverse cardiovascular event) show no meaningful difference, and an explanation is warranted for how cardiovascular death can differ when causative conditions do not. Multiple outcomes were analysed without an explicit multiplicity-control strategy, increasing the risk that a single borderline result represents a false-positive finding. The adjustment using EuroSCORE 2 and cross-clamp time may introduce collider bias, compounded by the limitations of EuroSCORE 2 itself 2. Results would be strengthened by comparing first with second cases and utilising negative-control outcomes, all-cause mortality as the primary outcome, centre-level adjustment and models without cross-clamp time. We remain unconvinced that this retrospective observational study should supersede previous contradictory, high-quality evidence 3-5. However, it does illustrate how patients with higher risk tend to undergo surgery later in the morning. While the authors state appropriately that the findings are hypothesis-generating and require further investigation, their interpretation is suggestive and surpasses description without acknowledging a potentially false-positive result. It is difficult to ignore the reciprocal relationship that anticipated postoperative outcomes influence the timing of surgery.