Baseline characteristics.

Purpose
We aimed to develop the Korean Hospital Frailty Risk Score (K-HFRS) by applying the International Classification of Diseases-10 codes to community-dwelling older adults’ medical data.
Methods
We selected data from 2,761 people with no missing main variable values from the Korean Frailty and Aging Cohort Data (KFACD) and National Health Insurance Database (NHID) for analysis. Frailty was determined based on modified Fried’s phenotype [MFP] and Korean Frailty Index for Primary Care [KFI-PC] in the KFACD. A previously established method calculated the K-HFRS, verified by the area under the receiver operating characteristic (ROC) curve. The calculated cutoff value predicted the medical use.
Results
The respective K-HFRSs of the frailty group using the MFP and KFI-PC criteria ranged from 3.64 (±3.03) to 8.15 (±5.72) and 4.07 (±3.42) to 9.10 (±6.28), with 7.67 (±5.40) and 8.59 (±6.03) when four diagnoses were included. The K-HFRS of the frailty group using the KFI-PC criteria was higher than that using the MFP criteria. With four diagnoses included using the MFP criteria, the adjusted odds ratio (OR) for medical expenditures in the frailty group compared to the non-frailty group was 3.01 (95% confidence interval [CI] 2.52–3.60, p < .001); for the number of emergency room (ER) visits was 2.19 (95% CI 1.77–2.70, p < .001); for inpatient days was 2.48 (95% CI 2.08–2.96, p < .001). With four diagnoses included using the KFI-PC criteria, the adjusted OR value for medical expenditures was 2.77 (95% CI 2.35–3.27, p < .001); for the number of ER visits was 1.87 (95% CI 1.51–2.32, p < .001); for inpatient days was 2.07 (95% CI 1.75–2.45, p < .001).
Conclusion
This study substantiated that the K-HFRS can measure frailty efficiently at a lower cost. Follow-up studies are needed for additional validity.