The investigation and analysis of critical incidents and adverse events in healthcare

Objectives: To carry out a review of published and unpublished work on the analysis on methods of accident investigation in high-risk industries, and of critical incidents in healthcare. To develop and pilot guidelines for the analysis of critical incidents in healthcare for the hospital sector, mental health and primary care. Data sources: Literature already available in the Clinical Risk Unit, University College London. Work by known experts in the field of accident investigation and analysis. Electronic databases including PsycINFO and MEDLINE. Websites for accident investigation reports. Review methods: Twelve techniques from other highrisk industries were reviewed in detail using criteria developed for the purpose. This review provided a conceptual framework for the healthcare review and appraisal process, as well as providing a critical assessment of the industry techniques. Rigorous searching and screening identified 138 papers for formal appraisal and a further 114 were designated as providing potentially useful background information. A formal appraisal instrument was designed, piloted and modified until acceptable reliability was achieved. From the 138 papers, six techniques were identified as representing clearly definable approaches to incident investigation and analysis. All relevant papers were reviewed for each of the six techniques: Australian Incident Monitoring System, the Critical Incident Technique, Significant Event Auditing, Root Cause Analysis, Organisational Accident Causation Model and Comparison with Standards approach. Results: All healthcare techniques had the potential of being applied in any specialty or discipline related to healthcare. While a few studies looked solely at death as an outcome, most used a variety of outcomes including near misses. Most techniques used interviewing and primary document review to investigate incidents. All techniques included papers that identified clinical issues and some attempt to assess underlying errors, causes and contributory factors. However the extent and sophistication of the various attempts varied widely. Only a third of papers referred to an established model of accident causation. In most studies examined there was little or no information on the training of investigators, how the data was extracted or any information on quality assurance for data collection and analysis. There was some variation in the level of expertise and training required but to undertake the investigation to an acceptable depth all required some expertise. In most papers there was little or no discussion of implementation of any changes as a result of the investigations. A quarter of publications gave some description of the implementation of changes, though few addressed evaluation of changes. Conclusions: The reviews demonstrate that, while much valuable work has been accomplished, there is considerable potential for further development of techniques, the utilisation of a wider range of techniques and a need for validation and evaluation of existing methods which would make incident investigation more versatile and use limited resources more effectively. Further exploration of techniques used in high-risk industries, with interviews and observation of actual investigations should prove valuable. Existing healthcare techniques would benefit from formal evaluation of their outcomes and effectiveness. Studies should examine depth of investigation and analysis, adequacy and feasibility of recommendations and cost effectiveness. Examining implementation of recommendations is a key issue. Health Technology Assessment 2005; Vol. 9: No. 19 ii