Pragmatic development of an evidence-based intensive care unit-specific falls risk assessment tool: The Tyndall...

Falls may result in significant patient harm. A recommended strategy to prevent falls is the use of a falls risk assessment tool, but these tools are often specific for older people. Evidence suggests context-specific tools are more effective. Although a rare event in the intensive care unit (ICU), patients in the ICU are at high risk of falling. The primary trigger for the current study was an increase in falls in the study ICU.

The objective of this study was to develop and implement a valid and reliable ICU-specific falls risk assessment tool, with the aim of reducing falls.

Retrospective incident-monitoring database audits were performed. Prospective validity and reliability testing of an ICU-specific tool (Tyndall Bailey Falls Risk Assessment Tool [TB FRAT]) and the existing method for assessing falls risk (ONTARIO Modified Stratify [Sydney Scoring] Falls Risk Screen) was conducted. Seven raters (nurse clinicians) independently performed falls risk assessment using both tools on two occasions for six patients.

Correlation for risk stratification categories between the two tools was moderate (r = 0.60, P < .001). Intrarater reliability (correlation) for individual rater's scores was strong (r = 0.86, P < .001). Interrater reliability for the TB FRAT was moderate to excellent (interclass correlations = 0.76 [95% confidence interval: 0.54-0.94]), and internal consistency was excellent (Cronbach's alpha, 0.97). Falls resulting in serious injury reduced from 3.35 per 1000 separations 12 months before implementing the specific ICU tool to 0.85 per 1000 in the 12 months after implementation.

The TB FRAT provided a more reliable falls risk assessment than the existing method of assessing falls risk in this single-room occupancy ICU. This TB FRAT could be a valuable addition to quality improvement initiatives aimed at improving patient safety related to falls; however, adaptation to the local context should be considered.