Medical accidents are typically classified into two types: accidents caused by healthcare staff while providing services, and accidents caused by patients. Among the latter type, patient falls account for a significant percentage, and have a significant impact on patients. The goal of our research here is to establish a methodology to prevent patient falls by identifying situations that are dangerous for patients and formulating concrete countermeasures based on the results of an assessment to prevent such situations from arising.
Kato et al. (2013b) evaluated risk factors, those are included in the existing assessment sheet, through Cox regression analysis by using data from approximately 1,000 cases at Iizuka Hospital, located in Fukuoka Prefecture. However, the number of this dataset was insufficient for a detailed analysis. It was also needed to arrange the data format and consult models for recurrent event analysis. Furthermore, the statistical model and covariates need to be considered in detail.
In this paper, we tried to statistically model patient falls based on multivariate analysis by using a logistic model, the Cox model, and models for recurrent events. We first discussed the statistical model as well as the covariates to be included in it. We then developed a multiple scoring system based on the results of multivariate analysis by using data from Iizuka Hospital from 2009 to 2011. Finally, we evaluated each scoring system by calculating the correlation between scores and probability of events.
View full abstract