Fractures After Inpatient Falls in a Tertiary Rehabilitation Hospital: Associated Factors, Performance of Fall-Risk Scores, and Orthopaedic Outcomes

Background/Objectives: Inpatient falls are an important patient-safety concern in rehabilitation hospitals; however, factors determining whether a fall results in fracture remain poorly understood. This study investigated clinical factors associated with radiographically confirmed fracture after inpatient falls, evaluated the fracture-discrimination performance of the Morse Fall Scale and Hendrich II Fall Risk Model, and characterized the resulting orthopaedic burden. Methods: This single-center retrospective cohort study included adults who experienced a documented fall while hospitalized in a tertiary physical medicine and rehabilitation hospital. Patients with and without radiographically confirmed fractures were compared regarding demographic characteristics, comorbidity burden, bone health, continence, mobility, medication exposure, laboratory findings, and fall circumstances. Factors associated with fracture were evaluated using univariable analyses and a parsimonious Firth penalized logistic regression model. Discrimination of the Morse and Hendrich II scores was assessed using receiver operating characteristic analysis. Results: Among 91 adult inpatient fall events, 51 (56.0%) resulted in fracture. In unadjusted analyses, patients with fractures had lower lumbar spine T-scores and higher frequencies of urinary and fecal incontinence, whereas age, sex, femoral T-score, mobility status, fall location, and fall mechanism did not differ significantly between groups. In the adjusted Firth model comprising 87 complete cases and 49 fracture events, no variable reached independent statistical significance; lower lumbar T-score (adjusted odds ratio [aOR], 0.74; 95% confidence interval [CI], 0.52–1.05; p = 0.073) and urinary incontinence (aOR, 2.30; 95% CI, 0.87–6.12; p = 0.082) showed the strongest associations. The Morse and Hendrich II scores demonstrated poor fracture discrimination, with AUCs of 0.372 (95% CI, 0.255–0.489) and 0.411 (95% CI, 0.295–0.526), respectively, without a significant difference between them (p = 0.500). Among patients with fractures, 41.2% sustained major fractures, 27.5% underwent surgery, and 11.8% developed complications. Conclusions: Among documented inpatient falls in this rehabilitation cohort, fractures were common and imposed a clinically important orthopaedic burden. Conventional fall-risk scores did not distinguish falls resulting in fracture, suggesting that prediction of fall occurrence and prediction of orthopaedic injury represent distinct clinical objectives. Fracture-prevention strategies in rehabilitation settings should therefore complement general fall-risk assessment with targeted evaluation of skeletal and functional vulnerability.