Associations of postoperative mortality with the time of day, week and year

Abstract

Studies that have investigated circadian, weekday and seasonal variation in postoperative mortality have been relatively small or have been for scheduled surgery. We retrospectively tested a large mixed surgical cohort from a German tertiary care university hospital for the presence of cyclical variation in all-cause in-hospital mortality after operations performed between 2006 and 2013. We analysed mortality rates after 247,475 operations, adjusted for age, sex, comorbidities, location, urgency and duration of the surgery, and intra-operative blood transfusions. The mortality odds ratio (95%CI) after operations started in the morning (08:00–11:00) were lowest, 0.73 (0.66–0.80), p < 0.001 and highest for operations started in the afternoon (13:00–17:00), 1.29 (1.18–1.40), p < 0.001. Mortality at the weekend was the same as during the week. There was no seasonal variation in mortality, p = 0.12. However, the interference of four-yearly and ten-monthly cycle amplitudes resulted in higher mortality odds ratio (95%CI) in winter 2008–2009, 1.41 (1.18–1.69), p < 0.001, and lower mortality in spring 2011 and 2012, 0.70 (0.56–0.85) and 0.67 (0.53–0.85), p < 0.001 and p = 0.001, respectively. The ability to predict cyclical phenomena would facilitate the design of interventional studies, aimed at reducing mortality following surgery in the afternoon and when cycles interfere constructively.