Comparative Analysis of Two Fall Risk Assessment Tools in the Obstetric Population.

Abstract

Objective: To examine the relative accuracy of the Morse Fall Scale (MFS) and the Obstetric Fall Risk Assessment System (OFRAS) in predicting obstetric patients' fall risk.

Design: Retrospective comparative analysis of the MFS and the OFRAS in obstetric inpatients.

Setting: A 560-bed urban teaching hospital in Hawaii.

Participants: Eighty-five records of people hospitalized for childbirth.

Methods: Adequate power modeling and statistical analyses were completed using the programs R packages Version 4.0.1 and SAS Version 9.4. Subsequently, a ratio of 17 fall records to 68 nonfall records (1:4) with similar dates of admission were reviewed. Investigators collected the MFS score/risk level as documented and the required data points to obtain the OFRAS fall risk score/level. Logistic regression models were fit using the MFS and OFRAS as predictors of falls. Results are expressed as odds ratios with 95% confidence intervals and p values to test for statistical significance. Receiver operating characteristic (ROC) curves were derived from logistic regression results and graphed to compare the instruments. Areas under ROC curve (AUROCs) were calculated to display the specificity and sensitivity of the risk assessment tools.

Results: Data for 85 pregnant or postpartum people were included in the sample. Analysis of AUROCs demonstrated that the OFRAS is more sensitive and specific for obstetric patients than the MFS. The OFRAS showed significance (p < .001) in predicting falls compared to the MFS (p = .40). Associations between fall scores and falls were examined in separate conditional logistic regression models.

Conclusion: The OFRAS demonstrated higher sensitivity and specificity in fall risk prediction. The MFS performed similarly to random chance regarding obstetric fall risk prediction. The potential exists to better anticipate patient falls, protect staff from injury related to patient fall, and decrease organizational risk using a population-specific tool.