Psychometric evaluation of the Healthcare Improvement Scotland ‘Think Frailty’ tool in unscheduled care settings

INTRODUCTION: The Healthcare Improvement Scotland 'Think Frailty' tool (HIS-Frailty) was developed to allow frailty triage in unscheduled care. Development was iterative, with a change around assessment of support needs to improve completion. We evaluated the psychometric properties of HIS-Frailty. METHODS: We used data from prospective, national audits of frailty screening. Consecutive adults aged ≥65 who presented as unscheduled admissions to hospital were assessed with the tool across three phases of data collection (2017-2023). Frontline healthcare workers completed HIS-Frailty and a reference standard of the Clinical Frailty Scale. We assessed concurrent validity (Pearson's correlations), discrimination (area under curve; AUC), feasibility (percentage of missing data), predictive validity for 30-day mortality and duration of hospital stay (regression), factor structure (maximum likelihood factor analysis), intra-rater reliability and inter-rater reliability (adjusted kappa). RESULTS: A total of 2447 older adults were included. HIS-Frailty scores were significantly correlated with Clinical Frailty Scale (r = 0.78, P < .001), discrimination was excellent (AUC: 0.93, 95% CI = 0.92-0.95). All domains of HIS-Frailty had <1.5% missing data. HIS-Frailty scores were associated with 30-day mortality (OR: 1.49, 95% CI = 1.32-1.69) and being in hospital at 30-days (OR: 1.73, 95% CI = 1.52-1.96). Factor analysis indicated a single underlying factor accounting for 43.4% variance. Intra-rater agreement was good (Kappa = 0.71, 95% CI: 0.53-0.88). Inter-rater agreement between specialist and non-specialist assessors was substantial (Kappa: 0.83, 95% CI: 0.68-0.97). CONCLUSION: HIS-Frailty has good psychometric properties for 'front door' frailty screening and can be used consistently and effectively by non-specialist assessors.