Abstract
Introduction: Older adults account for the majority of cancer diagnoses, yet oncology care systems remain insufficiently structured to address the complexity of this population. Oncology care prioritizes survival, while older adults often define treatment success in terms of quality of life (QOL), creating a mismatch between clinical endpoints and lived priorities. Frailty has emerged as a key construct in cancer care but is often applied within a biomedical structure that does not fully capture outcomes that matter to older adults. As a multidimensional, patient-informed measure, frailty may provide a framework to better align care with these outcomes.
Purpose: The purpose of this dissertation was to (1) examine the relationship between frailty and QOL; (2) evaluate concordance between patient- and physician-rated functional status; and (3) explore differences in treatment regret among older adults with cancer based on frailty status.
Methods: These multi-methods secondary analyses used data from the Geriatric Assessment–driven Intervention (GAIN) trial and the Hurria Older Patients with Breast Cancer (HOPE) study. Frailty was operationalized using a Deficit Accumulation Index (DAI), incorporating both clinical indicators and patient-reported domains reflecting symptoms, function, and daily life. In Aim 1, we examined the relationship between frailty and baseline QOL as well as within-person change over time using regression and linear mixed-effects models to identify frailty-stratified patterns of change; domain-specific analyses were conducted to identify drivers of these changes. In Aim 2, we characterized discordance in physician- vs. patient-rated performance status, including presence, magnitude, and direction, and examined how these patterns varied by frailty status. In Aim 3, we examined the association between frailty and treatment regret using regression modeling and conducted conventional content analysis of open-ended responses to further characterize how patients with varying frailty status evaluated and made meaning of their treatment decisions.
Results: Greater frailty burden was associated with worse baseline QOL, with variability within frailty groups. Longitudinally, QOL declined among fit patients following treatment completion, while pre-frail and frail patients demonstrated improvement. Discordance between patient- and physician-rated functional status was present in approximately two-thirds of patients and increased with frailty severity, with the largest increase between fit and pre-frail groups; clinicians more often rated functional status higher than patients. Greater frailty burden was also associated with higher treatment regret. Qualitative findings showed that fit participants were more likely to frame their decisions in terms of personal agency, whereas frail participants described treatment as necessary and expressed a need for more information.
Significance: Building on prior literature linking frailty to clinical outcomes, we found that frailty had associations with patient-centered outcomes: frailty status at diagnosis differentiated patient experience across the cancer care continuum, stratifying patterns of QOL change, revealing discordance in functional status at decision-critical thresholds, and demonstrating an independent association with treatment regret. These findings demonstrate that frailty is not a uniform marker of decline but captures heterogeneity in how patients experience treatment. Assessing frailty at the time of diagnosis provides important context at the point of decision-making. Integrating the measure into clinical workflows may better align clinical assessment with patient values, inform treatment and resource allocation, and shape how care is experienced and evaluated. It may also create an opportunity to identify population-level patterns across frailty groups, allowing these patterns to inform care at the time of diagnosis and over time.
