br When specifically examining the covariates used in the modeling

When specifically examining the covariates used in the modeling, our results suggested that Ionomycin and education were the stronger corre-lates of satisfaction across domains. Adjusted for the other covariates, non-whites reported, on average, lower satisfaction than whites, and patients with lower levels of education reported lower satisfaction. These incidental findings may represent disparities in health care, in-cluding perceived access to care, communication, mistrust, and/or liter-acy issues [2,3,34,35]. Comorbidities had a positive association with satisfaction, however of small to medium magnitude. This incidental finding is in line with a large national study of all-aged adults that re-ported an association between high patient satisfaction and higher mor-tality rates [36], suggesting that patient satisfaction with care may be an indicator of illness as these patients rely more on physician and healthcare provider support.
Table 4
Adjusted and unadjusted predicted satisfaction domain scores by the HRQoL sample mean and standard deviation change (N = 637).
Unadjusted satisfaction scores
Adjusted satisfaction scores
Unadjusted satisfaction scores
Adjusted satisfaction scores
At Mean At Mean At Mean At Mean At Mean At Mean
At Mean At Mean At Mean
At Mean At Mean At Mean
Communication
SD, standard deviation; HRQoL, health related quality of life.
Predicted satisfaction scores from fractional logit models. Satisfaction scores range from to 1.
Scores adjusted for cancer type, race, education, phase of care, and comorbidity score.
There are a number of limitations in this study. First, we cannot make conclusions about causal effects because of the cross-sectional study design. Second, these data were derived from a cohort of older patients in the Southeastern U.S. and therefore may reflect regional demographic, supplemental insurer, and institutional-related circum-stances that may not be generalizable to all older adults with cancer in other regions of the United States or elsewhere. Only those covari-ates that were available in our data were considered in the analysis; other factors may also influence both patient satisfaction and HRQoL. Lastly, use of the top-box method to estimate patient satisfaction scores may result in loss of information, however, this is the method used for reporting of satisfaction-related items in the CAHPS Clinician & Group Survey.
In our sample of older patients with cancer, we did not find a rele-vant association between HRQoL and patient satisfaction domains. Therefore, our results indicate that for health care policy determina-tions, these two constructs should not be assumed to correlate. We identified three domains as potential areas for improvement of satisfac-tion with oncology care among older patients, including how patients access care, coordinate care, and engage within the healthcare system. Incidental findings suggested that satisfaction with care differs by race and education, which warrants further examination. Finally, patient sat-isfaction assessment across age groups may contribute to inform oncol-ogy care providers on ways in which patients perceive the quality of care received, which ultimately affect healthcare organizations' accred-itation, ranking, and reimbursement.
Contributors
Conception and Design: GR, CW
Data Collection, Analysis and Interpretation of Data: CW, AA
Manuscript Writing: All authors
Approval of Final Article: All authors
Conflicts of Interest
None.
Funding/Support
This study was supported by grant from Genentech (G-55600) and from the Centers for Medicare & Medicaid Services (1C1CMS331023). The re-search was conducted by the awardee with input from Genentech per-sonnel. CMS had no role in any aspect of this study.