BackgroundEarly diagnosis of anastomotic leak is challenging due to its heterogeneous clinical presentation. Continuous vital sign monitoring compared to periodic monitoring has potential to expedite anastomotic leak diagnosis. The aim is to assess this potential after esophageal and pancreatic cancer resection. MethodsIn this data modelling cohort study vital signs were continuously monitored and periodically recorded according to EWS protocol. We retrospectively calculated EWSs every one minute using continuous vital sign data. The percentage of patients having an EWS-alarm in 24 hours preceding an anastomotic leak, the elapsed time between the first alarm and diagnosis were compared between regular periodic nurse EWSs and the continuously calculated EWSs. These data were also compared to those of patients with other infectious complications and controls without complications. ResultsOne hundred ninety patients were included, twenty-six with an anastomotic leak (14%), thirty-six with an infectious complication (19%) and one hundred twenty-eight controls (67%). The time between the first continuous EWS-alarm and anastomotic leak was 23 hours [IQR 19-24] compared to 3hrs [IQR 2-13] with periodic alarms. Sixty-two percent of leak patients had at least one continuous alarm in the 24 hours before the event, compared to 15% for periodic EWS (p<.001). More patients with infections had a continuous alarm compared to anastomotic leak patients (86% vs. 62%, p=.026). Postoperative alarms in both groups were more frequent compared to controls. ConclusionAfter esophageal and pancreatic resections, continuous monitoring alarms earlier and more frequently than periodic monitoring before anastomotic leak diagnosis. However, combination with several other diagnostic modalities seems required for the definite diagnosis.