Sukruth A Shashikumar1, Kristine Huang1, R Tamara Konetzka2, Karen E Joynt Maddox1,3. 1. Department of Medicine, Cardiovascular Division, Washington University School of Medicine, St. Louis, MO. 2. Department of Public Health Sciences, University of Chicago, Chicago, IL. 3. Center for Health Economics and Policy, Institute for Public Health at Washington University, St. Louis, MO.
Abstract
OBJECTIVE: The objective of this study was to evaluate claims-based frailty indices (CFIs) used to assess frailty on a population-based level. BACKGROUND: Frailty is a key determinant of patient outcomes, independent of demographics and comorbidities. Measuring frailty in large populations has implications for targeted interventions, public reporting, and risk adjustment. Frailty indices based on administrative data in health insurance claims allow such population-level assessments of frailty. METHODS: We used PubMed to search for studies that: (1) were development or validation studies of a CFI that predicted frailty; and (2) used only diagnosis codes within administrative claims or health services claims. We evaluated the CFIs on 6 axes: databases used to build the CFIs; variables used to designate frailty; methods used to build the CFIs; model performance for predicting frailty; model relationship to clinical outcomes; and model limitations. RESULTS: We included 17 studies. They showed variation in the claims codes used to designate frailty, although themes like limited mobility and neurological and psychiatric impairment were common to most. C-statistics demonstrated an overall strong ability to predict patient frailty and adverse clinical outcomes. All CFIs demonstrated strong associations between frailty and poor outcomes. CONCLUSIONS: While each CFI has unique strengths and limitations, they also all had striking similarities. Some CFIs have been more broadly used and validated than others. The major takeaway from this review is that frailty is a clinically relevant, highly predictive syndrome that should be incorporated into clinical risk prediction when feasible.
OBJECTIVE: The objective of this study was to evaluate claims-based frailty indices (CFIs) used to assess frailty on a population-based level. BACKGROUND: Frailty is a key determinant of patient outcomes, independent of demographics and comorbidities. Measuring frailty in large populations has implications for targeted interventions, public reporting, and risk adjustment. Frailty indices based on administrative data in health insurance claims allow such population-level assessments of frailty. METHODS: We used PubMed to search for studies that: (1) were development or validation studies of a CFI that predicted frailty; and (2) used only diagnosis codes within administrative claims or health services claims. We evaluated the CFIs on 6 axes: databases used to build the CFIs; variables used to designate frailty; methods used to build the CFIs; model performance for predicting frailty; model relationship to clinical outcomes; and model limitations. RESULTS: We included 17 studies. They showed variation in the claims codes used to designate frailty, although themes like limited mobility and neurological and psychiatric impairment were common to most. C-statistics demonstrated an overall strong ability to predict patient frailty and adverse clinical outcomes. All CFIs demonstrated strong associations between frailty and poor outcomes. CONCLUSIONS: While each CFI has unique strengths and limitations, they also all had striking similarities. Some CFIs have been more broadly used and validated than others. The major takeaway from this review is that frailty is a clinically relevant, highly predictive syndrome that should be incorporated into clinical risk prediction when feasible.