Learning Objective #1: Describe process used to validate a predictor model for pediatric in-patient falls

Learning Objective #2: Identify significant risk factors for pediatric in-patient falls

---

Assessment of In-patient Fall Risk was identified as one of the 2005 Patient Safety Goals by the Joint Commission on Accreditation of Healthcare Organization, requiring hospitals to review patient fall data to determine root causes and implement risk reduction strategies. Fall-risk assessment scales have shown predictive validity in identifying patients who fall from those who do not fall, and are used with adult and geriatric populations to determine care plan options based on potential fall-risk. Fall prevention programs based on an assessment of risk and implementation of risk-based fall protocols have shown sustained improvements in the reduction of inpatient falls for adults and the elderly. Unfortunately, these scales have not been validated for use with children. Therefore, a retrospective case/control study of 200 patients admitted, between 1998-2003, to a Children's Hospital: 100 children who fell while hospitalized and 100 matched controls who did not fall, was undertaken to identify predictor variables associated with pediatric in-patient falls. Potential risk factors were identified through a review of the fall literature. Falls were classified using the conceptual model developed by Morse (1997): accidental, anticipated physical/physiological or unanticipated physical/physiological.

Data were analyzed using descriptive statistics and univariate relative risks statistics. Principal component cluster analysis identified highly correlated or collinear variables. Within each cluster, the variables with the strongest association with the outcome variable were chosen. Logistic regression was used to develop a multivariate risk factor model. Significant risk factors were length of stay, orthopedic diagnosis, physical therapy/occupational therapy, seizure medication, and being IV/Heparin Lock free. The model correctly predicted 83.4% of the children who fell. This model may be used to further develop a fall risk assessment tool, which can guide clinical care planning for those children with the highest fall risk.