Introduction: It has been alleged that human error is impossible to eliminate in surgery because of the complex, multifactorial nature of clinical care. However, the level of detail that describes error types is limited and studies analyzing large numbers of cases are few. We developed a standardized error template and gathered data over five years to profile patterns of error in surgical cases with complications.

Methods: Data was collected at an accredited general surgery residency program utilizing a previously described complication reporting system. The system allows for click-based, electronic entry of complications, case summaries, and a standardized critique algorithm across multiple domains including diagnostic, technical, judgment, therapy/prophylaxis, communication, professionalism and system error. Data was collected on cases with complications and reviewed at weekly validation meetings by a committee of attendings (including the site Chief of Surgery, Program Director, divisional leadership, senior and junior residents, nursing and quality). The process was prospective and comprehensive; presented by residents responsible for care; designed to validate complications and identify error. Data was analyzed for the incidence of error across the seven domains, thirty-seven subtypes, and for specific procedures including bariatrics, cholecystectomy, colectomy, hepatic procedures, hiatal hernia, ileostomy closure, pancreatectomy and ventral hernia/component cases.

Results: The study population included 25,857 general surgery/vascular cases, 4,005 complications, 1,473 cases with complications, and 3,268 errors or 2.2 errors per complication case. The number and incidence of error by domain category was technical (831, 25.4%), therapy/prophylaxis (598, 18.3%), judgment (509, 15.6%), diagnosis (378, 11.6%), system issues (379, 11.3%), communication (356, 10.9%), and professionalism (226, 6.9%). The most common subtypes for each domain were 'error technical performance'(54.2%), 'post procedure care MD Team'(79.4%), 'error medical judgment'(50.5%), 'delay diagnosis'(11.6%), 'pre/post procedure care Nursing Team'(59.7%), 'error in handoff team members'(10.9%), and 'identified/accepted responsibility'(62%), respectively. All of the operative procedures evaluated mimicked the error profile of the domain and subtype categories with some differences in emphasis. We calculated the proportion of operations with complications that had errors. The median sub-specialty had a rate of .83, with a range of only .77 to .89. Thus, a large proportion of operations with complications had at least one error, and this did not vary much across sub-specialty.

Conclusion: These data suggest that patterns of error can be accurately described utilizing a standardized error profile template to analyze postoperative complications. Operation types with more errors had more complications. By understanding the distribution of error types, providers and colleagues may be informed of their individual propensity for error, and systems can be more focused in quality improvement initiatives based on patterns specific to procedure type.