Background: Analyses of technical complications in surgery have rarely been able to convincingly determine the causes. When newer concepts of human error analysis are used, however, insights of practical importance are more readily obtained.
Methods: Seeking ideas for prevention, we analyzed 200 laparoscopic bile duct injuries within the framework of human error analysis. The determinations included the following: 1) input data (visual and/or haptic); 2) decision mode (e.g, schema-based; rule-based; or knowledge-based); and 3) action (i.e, the operation). There were 157 women and 43 men whose average age was 46 years. The injury distribution was Class I 9%, Class II 22%, Class III 55%, Class IV 10%, bile leaks 4%.
Results: The primary cause of the error in all cases was a visual perceptual illusion. The laparoscopic environment contributed to 68% of injuries, and poor visibility to 32%. 18% of injuries were missed despite converting the operation to open. 51 injuries (26%) were recognized at the index operation, but only 16 (8%) involved using rule-based processing to an extent that limited the injury. In the remainder (184 pts), processing errors involved persistent schema-based action in spite of irregular input data (21%); persistent schema-based action despite recognized abnormalities (abnormal anatomy, poor visibility) (24%); faulty rule-based processing (32%) [e.g, misinterpretation of operative cholangiograms (49% of cholangiograms); identifying a hepatic duct as an accessory duct (12%); failure to identify an injury after opening (18%)]; and correct rule-based processing too late to limit the injury (15%).
Conclusions: These data show that errors leading to laparoscopic bile duct injuries stem principally from misperception, not knowledge deficits or lack of skill. In 60% the surgeon did not recognize the existence of a problem or initiated rule-based processing too late. Feedback via self-correcting rules did not prevent these injuries, because rules were often not applied or they were applied inappropriately. An understanding of these cognitive factors would highten vigilence (an attention factor), and the enuciation of specific learnable rules would improve performance. Nevertheless, explicitly defining the schema-based method of performing the operation (i.e, task analysis) to incorporate important safety measures should have the greatest preventive effect. This information is at hand.
