GRAF-PIF Fall Risk Assessment Tool

Predictive Accuracy in a Children's Hospital

Denise Sackinger, PhD, MN, RN, CPHP-PC, CPN; Kristen Carlin, MPH; Brenda Zierler, PhD, RN, FAAN

Disclosures

Pediatr Nurs. 2021;47(4):189-197, 180. 

In This Article

Abstract and Introduction

Abstract

Predictive qualities of a pediatric fall risk tool (Generalized Risk Assessment for Pediatric Inpatient Falls [GRAF-PIF]) were evaluated. Pediatric fall risk tools' predictive abilities vary when applied to different populations and settings. This observational study used retrospective review of GRAF-PIF scores, demographic characteristics, and fall incident reports for the period of January 2017 to December 2018 in a 407-bed pediatric hospital. One hundred thirty-six fallers were age-matched with 272 non-fallers (N = 408). GRAF-PIF sensitivity, specificity, odds ratio, and estimated area under the receiver operating characteristic curve were calculated. Odds ratios of falling were calculated across sex and diagnoses. The GRAF-PIF sensitivity in this patient population was 61%, and specificity was 58%. Results yielded an estimated receiver operating characteristic (ROC) curve with an area under the curve (AUC) of 0.59. For children with high GRAF-PIF scores, odds of falling were 2.08 times that of children with lower scores. Longer length of stay and cardiac and neurologic diagnoses were associated with higher odds of falling. Musculoskeletal diagnoses were associated with lower odds of falling. Although the sensitivity, specificity, and ROC AUC were not optimal, a GRAF-PIF score of greater than or equal to 2 points was associated with higher odds of falling. Weakness of tools to predict hospital falls may be due to prevention interventions implemented for those with high scores. Fall risk tools can be used to raise awareness of patient characteristics associated with falling. The recommendation is to continue to use the GRAF-PIF tool at this pediatric hospital. However, tool utilization should be accompanied with a critical evaluation of other fall risk factors (environmental, system, staff, and caregiver).

Introduction

A fall is "an unintentional descent…that results in the patient coming to rest" at a lower position (National Database of Nursing Quality Indicators, 2016, p. 2). Falls are the most common cause of hospital accidents in children, accounting for about 42% of inpatient accidents (Alemdaroglu et al., 2017; Da Rin Della Mora, Bagnasco et al., 2012; Fujita et al., 2013; Lee et al., 2013). Pediatric inpatient fall prevalence ranges from 0.4 to 3.8 falls per 1000 patient days (Almis et al., 2017; Fujita et al., 2013; Jamerson et al., 2014; Kim et al., 2019; Pauley et al., 2014; Schaffer et al., 2012). It is estimated that one-third to almost half (approximately 48%) of pediatric inpatient falls are preventable (AlSowailmi et al., 2018; Jamerson et al., 2014). Injuries from these falls can range from complaints of pain, skin redness/bruising, to broken bones, or damage to prior surgical repairs. These injuries can increase the length of stay (LOS) in the hospital and cost of care, and decrease parent satisfaction (AlSowailmi et al., 2018; Razmus et al., 2006).

Da Rin Della Mora, Calza and colleagues (2012) found that 51.7% of pediatric patients who fell in the hospital received additional interventions that were not previously part of their planned care. Although most children recovered from fall related injuries in two to three days, the recovery time ranged from 0 to 20 days (Da Rin Della Mora, Calza et al., 2012). Increased interventions and LOS associated with falls translates into increased health care costs related to hospital falls. Increased costs are concerning for health care organizations because many payers no longer reimburse organizations for care related to events that should "never" happen, such as falls resulting in injury (Garrard et al., 2016; Hagan & Jones, 2015; Inouye et al., 2009; Opsahl et al., 2017).

The current approach to fall prevention focuses on the use of risk assessment tools to identify children at higher risk of falling based on characteristics, such as age, sex, LOS, presence of an intravenous (IV) catheter, and cognitive/physical capabilities (Bagnasco et al., 2010; Franck et al., 2017; Graf, 2005; Hill-Rodriguez et al., 2009; Jamerson et al., 2014; Morse et al., 1989; Pauley et al., 2014; Razmus & Davis, 2012; Schaffer et al., 2012). As a result, fall risk tools are intended to predict anticipated physiologic and developmental falls, but not unanticipated physiologic or accidental falls. Many commonly used pediatric fall risk assessment tools lack adequate sensitivity and specificity to reliably predict children who will fall in the hospital (Almis et al., 2017; DiGerolamo & Davis, 2017; Harvey et al., 2010; Ryan-Wengeret al., 2012). Thus, even with the use of fall risk assessment tools, falls are difficult to predict and prevent (The Joint Commission, 2015).

processing....