Development and Preliminary Validation of a Fall Risk Assessment Tool for Psychiatric Inpatients: An Exploratory Sequential Mixed-Methods Study

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Falls are a major patient-safety concern in psychiatric inpatient settings, where conventional risk factors intersect with psychiatric symptoms, adverse effects of psychotropic medication, fluctuating cognition, and urgent toileting needs. We used an exploratory sequential mixed-methods design to develop and preliminarily validate a fall risk assessment tool for adult psychiatric inpatients. Interviews with 10 psychiatric nurses and five inpatients informed the initial item pool. The tool was then refined through two Delphi rounds involving 15 experts and weighted with a group-consensus analytic hierarchy process (AHP) matrix. Preliminary validation assessed inter-rater agreement, content validity, convergence with the Morse Fall Scale, and exploratory 1-month risk discrimination. The final tool comprised nine risk domains and 36 ordered scoring descriptors. Both Delphi rounds had 100% response rates. In round 2, absolute endorsement was high across the scoring cells (mean range, 4.53–5.00; CV range, .00–.14), although the tie-corrected 45-cell Friedman analysis showed statistically detectable but weak rank-order concordance (Kendall's W = .106, χ²(44) = 70.06, p = .007), consistent with ceiling effects and numerous tied ratings. The group AHP matrix showed acceptable consistency (CR = .024), and overall content validity was acceptable (S-CVI/Ave = .92). Among 20 inter-rater cases, ICC(2,1) was .986, risk-category agreement was 80.0%, and linearly weighted κ was .762. Among 80 convergent-validity cases, correlations with the Morse Fall Scale were strong (Pearson r = .848; Spearman ρ = .777). In the 50-patient prospective sample, the AUC was .876 (95% CI, .673–1.000). The Youden cut point was ≥12; at the exploratory clinical high-risk cut point of ≥14, sensitivity was .600 and specificity was .800. The tool demonstrated preliminary content relevance, inter-rater agreement, and convergence with the Morse Fall Scale. Its risk-discrimination performance requires confirmation in a larger multicentre external validation study.