Risk factors preventing immediate fall detection: Study using zero-inflated negative binomial regression

Abstract Purpose Falls are the most common accidents in healthcare facilities, and timely intervention can have a positive effect on the hazards and trauma experienced by patients after a fall. This study determined the factors affecting the time taken to detect a fall. Methods A total of 3,470 cases of falls reported through the Korea Patient Safety Reporting and Learning System were included in the analysis. A zero-inflated negative binomial regression method was used for this retrospective secondary data analysis study. Results There were 537 patients whose falls were not detected immediately; the count model was used to predict risk factors that delayed fall detection. Women aged 60–69 years—compared to those below 60 years and an evening nursing shift, compared to a day shift—were identified as significant factors. The fall detection time of about 2,933 patients was zero; therefore, the logit model was applied to predict a patient’s possibility of belonging to the group whose fall was detected immediately. Comparisons of tertiary hospitals with general hospitals and hospitals, of the evening shift with the day shift, and of the day shift with the night shift indicated significant influencing factors. Conclusions These findings can assist nurses in recognizing patient and hospital characteristics related to delayed fall detection. Strategies to improve patient safety in healthcare facilities that focus on patient characteristics such as age can be recommended. Furthermore, nurse staffing requires improvement to detect fall incidents immediately.