Adverse events in hospitals are a world-wide concern. To develop a tool for quantitative risk assessment of medical error, human reliability analysis was applied to incident reports from a prospective cohort study in a hospital. 374 nurses in a tertiary-care hospital of 670 beds joined this study. Numbers of tasks, classified into 170 items, were actually measured by nurses during three weekdays and one weekend. The total number of tasks per year was found to be about 13.2 million. 1, 030 incidents were reported over a year. The gross failure rate for tasks was estimated to be −10-4. The relative risk for failure rates in midnight and night shifts was 2-4 times greater than in dayshifts. When the failure rate for a patient category of ages 15 and 64 yr old was taken as a standard, the relative risk of failure was about 200 times greater in a patient category of problematic behavior and 90 times greater in patient categories of clouding of consciousness and postsurgical conditions. Failure rates for individual tasks were in the range of 10-5 to 10-3: the tasks with the highest failure rate were prevention of problematic behavior (2.32×10-3), safeguards against falls (1.47×10-3), intramuscular injection (1.30×10-3) and subcutaneous injection of insulin (9.82×10-4). The present study revealed two potential risks: occasional tasks with high failure rates and patient categories with small numbers of patients with high failure rates. The present protocol enabled us to conduct quantitative risk assessment on occupational health for medical workers.
2002 by the Japan Society for Occupational Health