Objective  o optimize the system conditions and improve the patients' perioperative surgical safety through feedback learning for adverse events.

  Methods  We retrospectively analyzed the adverse events (AEs) and risks of patients' safety reported by the Department of Anesthesiology, Peking Union Medical College Hospital from January 2010 to December 2014. Fish bone diagram was used to analyze and summarize the root causes, timing, avoidability and injury from adverse events.

  Results  A total of 370 cases of AEs were collected. The root causes of AEs were mainly human factors (43.08%, 224/520), followed by systemic factors (30.96%, 161/520), and patient factors were the lowest (25.96%, 135/520). Forty-four cases (11.89%, 44/370) were emergent operations; 37 cases (10.00%, 37/370) occurred at night; 243 cases (65.68%, 243/370) could be avoided. According to the different occurrence time, the top four incidence rates were intraoperative (28.11%, 104/370), before anesthesia induction (17.30%, 64/370), during anesthesia induction (13.78%, 51/370), and at the end of surgery (12.97%, 48/370). Eighty AEs (21.62%, 80/370) did not cause injury; 262 cases (70.81%, 262/370) caused temporary minor injury; 15 cases (4.05%, 15/370) caused permanent injury; 13 cases (3.51%, 13/270) caused non-anesthesia related death intraoperatively or within 24 hours after the surgery. Compared with avoidable AEs, the proportion of patient factors (60.26% vs. 11.26%, P < 0.001) and that of non-anesthesia related death intraoperatively or within 24 hours after the surgery (8.66% vs. 0.82%, P < 0.001) in the unavoidable group were higher.

  Conclusions  The root causes of perioperative AEs are mainly human factors, most of which can be avoided and cause temporary mild injury to patients. Strengthening feedback learning of AE, improving personnel training and optimizing system conditions might reduce the incidence of AE and promote perioperative safety and quality.