Comparing Frailty Assessment Methods and Their Ability to Predict Adverse Outcomes in Patients with Advanced CKD

Key Points Most measures of frailty, regardless of definition used, were associated with higher risk of mortality, hospitalizations, and emergency room visits. Claims-based definitions had poor agreement when compared with objective and subjective measures of frailty in the advanced CKD population. Administrative definitions require further development to accurately identify frail patients in the advanced CKD population. Background Frailty is common in patients with CKD, and those affected by both are at increased risk of adverse outcomes including disability, hospitalization, and death. Collecting data on frailty as part of clinical care could enhance care by identifying patients at risk of adverse events. However, clinical assessment of frailty requires time and resources. Frailty definitions based on administrative data might provide an efficient alternative. The primary objective was to compare agreement between administrative claims-based definitions versus objectively measured frailty in adults with advanced, nondialysis CKD and to examine their associations with adverse outcomes. Methods The cohort consisted of Manitoba participants from the Canadian Frailty Observation and Interventions Trial. This multicenter cohort study followed adults with advanced CKD longitudinally. Every visit, assessments were conducted to determine frailty status using the Fried Frailty Phenotype, Short Physical Performance Battery, and health care providers' impression. The Canadian Frailty Observation and Interventions Trial database was linked to administrative databases at the Manitoba Centre for Health Policy to calculate two claims-based frailty indicators, the Segal and modified preoperative frailty indices, which have been validated in the non-CKD literature. Results Of the 442 participants included, the mean age was 66±14 years and 58% were male; 88% had hypertension, 61% dyslipidemia, and 58% diabetes. The prevalence of frailty varied from 19% to 70% depending on definition. Agreement between frailty definitions was poor ( κ 0.09–0.33); however, individuals considered frail using both administrative or measured definitions had a higher risk of all-cause mortality and hospitalization, except for those identified by the Segal Frailty Indicator. Conclusions This study suggests that those identified as frail by nearly all measures were at higher risk of adverse outcomes. Thus, most frailty models in this study can be used to identify high risk advanced nondialysis CKD populations, allowing us to target individuals for interventions that aim to improve outcomes.