Prediction factors at hospital admission of fall incidents

Preliminary study for a fall prevention assessment tool

Abstract

A sufficient prevention program against accidental falls is a necessity in hospitals. However, resources are limited due to insufficient numbers of nurses and caregivers, and poor medical funds. In the present study, we investigated the estimated number of foreseeable accidental falls on hospital admission in order to establish an efficient prevention program against falls.

The study included 2,258 patients admitted to the Kameda General Hospital between November 2008 and January 2009. Nurses conducted surveys using our new original checklist of 17 questions to predict a fall on admission. The 17 questions include age, physical functions (3 questions), mental functions (5 questions), compound factors (3 questions) and 5 others. For this checklist we had added some factors reported in previous studies to our original conventional checklist.

Among 2,258 patients, 55 were admitted as a result of a fall. We carried out a multivariate analysis using logistic regression analysis, and were able to extract the following 5 factors to foresee accidental falls: history of fall, trying to move alone without help, poor sitting balance, poor standing balance, and anemia. When applying these prediction factors to the patients, the area under the receiver operating characteristic curve was 0.776. Sensitivity was 0.709, specificity was 0.799.

Using this original checklist as screening tool, inpatient accidental falls could be predicted to some extent. Future studies should concentrate on a more precise and simple prediction tool.