Frailty Assessment Tools in Chronic Kidney Disease: A Systematic Review and Meta-analysis

Abstract

Rationale & Objective

Frailty represents a loss of physiologic reserve across multiple biological systems, confers a higher risk of adverse health outcomes, and is highly prevalent among people with chronic kidney disease (CKD). We evaluated the measurement properties of frailty tools used in CKD and summarized the association of frailty with death and hospitalization.

Study Design

Systematic review and meta-analysis.

Setting & Study Populations

Studies assessing multidimensional frailty tools in adults at any stage of CKD and evaluating a measurement property of interest as per the Consensus-based Standards for the Selection of Health Measurement Instruments taxonomy.

Selection Criteria for Studies

Observational studies and randomized trials.

Data Extraction

Risk and precision measurements; measurement properties.

Analytical Approach

The Comprehensive Geriatric Assessment was the clinical standard for frailty identification. We pooled data using random effects models or summarized with narrative synthesis when data were too heterogenous to pool.

Results

We included 105 studies with data for at least one of the following: discriminative (n = 84; 80%), convergent (n = 20; 19%), and criterion validity (n = 2; 2%); responsiveness (n = 9; 9%) and reliability (n = 1; 0.1%). For the Fried Frailty Phenotype (FFP), the pooled adjusted HR (aHR) for mortality was 2.01 (95% confidence intervals [CI], 1.35-2.98; P = 0.001; I² = 58%) and 1.89 (95% CI, 1.25-2.85; P = 0.002; I² = 0%) for hospitalization in kidney failure (KF) populations. The pooled aHR for the Clinical Frailty Scale for mortality in pre-frail versus non-frail was 1.75 (95% CI, 1.17-2.60; I² = 26%) and 2.20 (95% CI, 1.00-4.80; I² = 66%) in frail versus non-frail. The Fatigue, Resistance, Ambulation, Illness, and Loss of weight scale showed consistent discriminative validity for higher mortality in non-dialysis CKD. The modified FFP (self-reported) showed acceptable discriminative validity and agreement with the FFP in patients with KF. In CKD and KF populations, agreement between clinicians’ subjective impression of frailty and frailty tools was low.

Limitations

Few studies compared the accuracy of frailty tools to the Comprehensive Geriatric Assessment. Only 1 study reported reliability. Studies were of overall low-moderate quality.

Conclusions

The FFP and Clinical Frailty Scale showed acceptable discriminant validity for clinical outcomes, and the modified FFP is an alternative tool to use if direct measurements are not feasible. The evidence does not support the use of clinicians’ subjective impression to identify frailty.

Plain-Language Summary

Frailty is a medical condition characterized by the loss of physiological reserve across multiple domains or an increased vulnerability to stress. Frailty is common among people with chronic kidney disease and is associated with poor health outcomes. There are numerous tools to assess frailty but the measurement properties of these tools, either for frailty identification, prognostication, or measuring changes in response to frailty interventions have not been identified in people with CKD. This information is important as frailty in CKD may be confounded by factors, such as those associated with uremia. By conducting this systematic review and meta-analysis, we found that frailty status, as measured by the Fried Frailty Phenotype and the Clinical Frailty Scale provided important prognostic information beyond age and clinical factors on the risk of mortality and hospitalization, with an approximate doubling in the hazard for these events among people with kidney failure. We also found that in both the kidney failure and non-dialysis CKD populations, the agreement between clinicians’ subjective impression of frailty and the FFP was low. There were limitations across studies, including heterogeneous follow-up period and covariate adjustment that may have influenced the results. In order to make recommendations for frailty tools across measurement domains, future studies should compare the diagnostic accuracy to the clinical standard, geriatric assessment, and examine responsiveness to change.