Background: Postoperative death measured 30 days after surgery is a conventional quality metric, whereas intervals up to 90 days are increasingly used, although data-driven time windows have scarcely been investigated.

Methods: The Swedish Colorectal Cancer Registry was used to identify all patients subjected resection for colorectal cancer between 2007 and 2020. All patients were followed up until 180 days after surgery. A join-point statistical hazard model was used to model a declining hazard to a transition point, followed by a stable death rate. This method was subsequently applied to describe postoperative deaths for the entire cohort and subgroups according to tumour location (colon and rectum).

Results: Some 56 096 patients electively operated on for colorectal cancer during the study interval were included, with a 30-day and 90-day fatality of 805 (1.43%) and 1458 (2.60%) patients respectively. The derived postoperative fatality window, after which the death rate transitioned to a stable rate, was 23.8 (95% c.i. 21.5 to 28.2) days after surgery. There was no significant difference in the time window between rectal cancer (22.9 days; 95% c.i. 15.1 to 28.4) and colon cancer (27.3 days; 95% c.i. 21.4 to 31.8) patients (P = 0.455). However, postoperative fatality time windows were extended in patients aged at least 80 years and with American Society of Anesthesiologists’ grade III or IV.

Conclusion: The traditional postoperative time window of 30 days was confirmed to be an appropriate metric in elective colorectal cancer surgery when evaluated with a hazards-based statistical framework. Importantly, this time window is influenced by older age and advanced co-morbidity, which could prompt increased vigilance for these patient groups.