Pediatric Hospital Falls: Patient Risk Assessment and Associated Parent Characteristics

Abstract

Background: Falls account for approximately 42% of adverse events reported in hospitalized children. These falls can result in patient injury which can increase hospital length of stay (LOS) and costs of care and decrease satisfaction with care. It is estimated that one-third to approximately one-half of these falls may be preventable. As a result, regulatory bodies, including The Joint Commission, require that patients be evaluated for fall risk during hospitalization. The current approach to risk assessment focuses on the use of fall risk assessment tools in which nursing staff rate individual patients against characteristics identified as increasing risk of falling in the hospital. Many pediatric fall risk assessment tools were developed based on characteristics specific to the patient population at the organization where the tool was developed. It is not clear if these fall risk assessment tools will be similarly predictive in other pediatric patient populations. Thus, organizations should validate fall risk tool performance in their patient care settings and populations. Even with the use of validated fall risk assessment tools, hospital falls are difficult to predict and prevent. Parents are present 60 – 83% of the times when children fall in the hospital. It has been hypothesized that parent characteristics such as anxiety, fatigue, and stress may contribute to the risk of pediatric hospital falls by causing parents to be distracted or less vigilant. However, this has not yet been tested. Objectives: This dissertation assesses an existing tool for predicting pediatric hospital falls and evaluates the feasibility of conducting a future study to describe associations of parent characteristics with falls in hospitalized children. These objectives are addressed in three papers. Paper one describes the predictive qualities of the Generalized Risk Assessment for Pediatric Inpatient Falls (GRAF-PIF) risk assessment tool in pediatric inpatient falls over a two-year period at a pediatric hospital. Paper two describes pediatric hospital fall risks identified in employee reports of pediatric falls over a two-year period and from a parent focus group discussion. Paper three evaluates the feasibility of conducting a future study to describe parental demographic and psychophysical characteristics associated with pediatric hospital falls. This paper identifies the sample size needed for an adequately powered future study. Finally, this paper provides recommendations for conducting a future study aimed at testing the association of parent demographic and psychophysical characteristics, specifically, anxiety; depression; fatigue; and sleep disturbance, and pediatric hospital falls. Research Design: A mixed-method study was undertaken to address the objectives. An observational case-control study using a retrospective review of GRAF-PIF scores and fall incident reports over a two-year period was used to describe predictive qualities of the fall risk assessment tool (paper one). A descriptive qualitative study was used to describe risks for pediatric hospital falls (paper two). A pilot study using an observational case-control design was used to evaluate the feasibility of and identify recommendations for conducting a future adequately powered study to test associations of parent characteristics with pediatric hospital falls (paper three). Results: The sensitivity of GRAF-PIF scores in the study sample (136 fallers and 272 non-fallers) was 61% and specificity was 58%. Results yielded an estimated Receiver Operator Characteristic (ROC) curve with an Area Under the Curve (AUC) of 0.59. For children with high GRAF-PIF scores > 2 points, the odds of falling was 2.08 times that of children with lower scores. The Interdisciplinary Momentary Confluence of Events Model (IMCEM) provided the framework for categorizing risks associated with child, environmental human, parent, or caregiver characteristics from fall incident reports and the parent focus group discussion. Nineteen risk factors were identified in fall incident reports, most of which were patient-level factors. Identified risks from the incident reports fell into more than one risk category thus providing support for the multifactorial nature of fall causation as proposed in the IMCEM. Responses from parent focus group discussions were categorized into 5 groups: 1) changing conditions of the child and environment, 2) distractions, 3) parents functioning as a different part of themselves, 4) lack of knowledge of risks, and 5) education considerations. The category of lack of knowledge of risk had three subthemes of “newness”, reliance on healthcare providers and changing conditions, and unexpected responses. For the pilot study, only four (28.6%) eligible faller dyads were recruited. Parental stress scores were correlated with anxiety and depression scores. Power calculations indicated a need for 392 fallers and 1,176 non-fallers for a future hypothesis testing study. Conclusions: Sensitivity and specificity of fall risk assessment tools may vary in different populations. The GRAF-PIF was only slightly better than chance at identifying fallers versus non-fallers. However, a score of two or more points was associated with a higher odds of falling. Thus, fall risk assessment tools can be used as part of the assessment of patient fall risk assessment. However, the use of these tools must be accompanied with a critical evaluation of other factors which may be associated with pediatric hospital falls. Parents are often with children when they fall in the hospital. It is important to identify and describe parent characteristics associated with the child’s fall risk. If parent characteristics are found to be associated with pediatric hospital fall risk, the healthcare team can partner with parents to identify and develop strategies to mitigate these characteristics to improve safety of the hospitalized child. A study to understand these associations may be limited by the sample size needed for an adequately powered study. A multi-site study should be undertaken to improve the likelihood of recruiting a sufficient sample. This study also provides recommendations for recruiting parent subjects and tool refinement to decrease concerns of multicollinearity among the data.