Temporal Trends in Fall Rates with the Implementation of a Multifaceted Fall Prevention Program: Persistence Pays Off

Abstract

Background: Most fall prevention programs are only modestly effective, and their sustainability is unknown. An academic medical center implemented a series of fall prevention interventions from 2001 to 2014.

Methods: The medical center's series of fall prevention interventions were as follows: reorganized the Falls Committee (2001), started flagging high-risk patients (2001), improved fall reporting (2002), increased scrutiny of falls (2005), instituted hourly nursing rounds (2006), reorganized leadership systems (2007), standardized fall prevention equipment (2008), adapted to a move to a new hospital building (2008), routinely investigated root causes (2009), mitigated fall risk during hourly nursing rounds (2009), educated patients about falls (2011), and taught nurses to think critically about risk (2012). To evaluate temporal trends in falls and injury falls, piecewise negative binomial regression with study unit-level random effects was used to analyze structured validated data sets available since 2003.

Results: From July 2003 through December 2014, the crude fall rate declined from 3.07 to 2.22 per 1,000 patient days, and injury falls declined from 0.77 to 0.65 per 1,000 patient days. Nonsignificant increases in falls occurred after nurses started rounding hourly and after the move to the new hospital. On the basis of regression models, significant declines occurred after nurses began to mitigate fall risk during hourly rounds (p = 0.009).

Conclusion: Instituting incremental changes for more than a decade was associated with a meaningful (about 28%) and sustained decline in falls, although the rate of decline varied over time. Hospitals interested in reducing falls but concerned about competing clinical and financial priorities may find an incremental approach to be effective.

Figure 1.

The figure displays boxplots (median, 25^th and 75^th percentile [interquartile range], upper and lower whiskers) of observed fall rates across study nursing units at each time point. Red line represents predicted fall rates over time, based on negative binomial regression model with study nursing unit-level random effects. Observed and Predicted Fall Rates from July 2003 Through December 2014 (Event Indicator)