Assessing the validity of prospective hazard analysis methods: a comparison of two techniques
1 Centre for Health Informatics and Multiprofessional Education, Institute of Epidemiology & Health Care, University College London, London, UK
2 Florence Nightingale School of Nursing and Midwifery, King’s College London, London, UK
3 Division of Neonatology, University of Virginia, Charlottesville, USA
4 Green Street Berman Ltd., London, UK
5 CHIME, 3rd floor, Wolfson House, 4 Stephenson Way, London NW1 2HE, UK
BMC Health Services Research 2014, 14:41 doi:10.1186/1472-6963-14-41Published: 27 January 2014
Prospective Hazard Analysis techniques such as Healthcare Failure Modes and Effects Analysis (HFMEA) and Structured What If Technique (SWIFT) have the potential to increase safety by identifying risks before an adverse event occurs. Published accounts of their application in healthcare have identified benefits, but the reliability of some methods has been found to be low. The aim of this study was to examine the validity of SWIFT and HFMEA by comparing their outputs in the process of risk assessment, and comparing the results with risks identified by retrospective methods.
The setting was a community-based anticoagulation clinic, in which risk assessment activities had been previously performed and were available. A SWIFT and an HFMEA workshop were conducted consecutively on the same day by experienced experts. Participants were a mixture of pharmacists, administrative staff and software developers. Both methods produced lists of risks scored according to the method’s procedure. Participants’ views about the value of the workshops were elicited with a questionnaire.
SWIFT identified 61 risks and HFMEA identified 72 risks. For both methods less than half the hazards were identified by the other method. There was also little overlap between the results of the workshops and risks identified by prior root cause analysis, staff interviews or clinical governance board discussions. Participants’ feedback indicated that the workshops were viewed as useful.
Although there was limited overlap, both methods raised important hazards. Scoping the problem area had a considerable influence on the outputs. The opportunity for teams to discuss their work from a risk perspective is valuable, but these methods cannot be relied upon in isolation to provide a comprehensive description. Multiple methods for identifying hazards should be used and data from different sources should be integrated to give a comprehensive view of risk in a system.