Evaluating concordance between International Myeloma Working Group (IMWG) frailty score and simplified frailty scale among older adults with multiple myeloma.

INTRODUCTION

Several frailty assessment tools exist for classifying older adults with multiple myeloma (MM) by their frailty status, such as the International Myeloma Working Group (IMWG) frailty score and the simplified frailty scale. The level of agreement between the IMWG frailty score and the simplified frailty scale remains unknown.

MATERIALS AND METHODS

In a cross-sectional analysis of a prospective cohort study, we identified adults ≥50y initiating a new treatment regimen for MM who underwent a baseline geriatric assessment (GA). Using data from the GA and electronic health records, we measured IMWG frailty score and the simplified frailty scale, and classified patients by frailty status. We merged the fit and intermediate-fit categories of IMWG frailty score to create a binary category (frail, non-frail) for comparison with simplified frailty scale and measured their agreement using Cohen's Kappa statistic. We tested the diagnostic utility of simplified frailty scale as a screening tool using IMWG frailty score as the gold standard, using sensitivity, specificity, and decision curve analysis (DCA).

RESULTS

Three hundred older adults were included with a median age at diagnosis of 64y; 56 % were male and 63 % were non-Hispanic White. By IMWG frailty score, 41 % were fit, 38 % intermediate-fit, and 21 % frail, while simplified frailty scale indicated 22 % frail and 78 % non-frail patients. The agreement between IMWG frailty score and simplified frailty scale was moderate (κ = 0.43); 19 % of the patients were misclassified. Despite discordance, when testing simplified frailty scale as a screening tool, we found a sensitivity of 56 % and specificity of 87 % to diagnose frailty. Substituting patient-reported performance status (PS) instead of physician reported ECOG PS led to a sensitivity of 91 % and specificity of 61 %. DCA showed that using simplified frailty scale (with patient reported PS) as a screening tool led to a 43-44 % reduction in the number of unnecessary GAs across reasonable threshold probabilities.

DISCUSSION

IMWG frailty score and simplified frailty scale have limited agreement with each other. This creates a possibility of misclassification bias and poses difficulty in comparing existing literature on frail patients with MM. Despite discordance, simplified frailty scale may have a potential role as a screening tool, when using patient-reported PS.