ASPPH logo


Member Research & Reports

Member Research & Reports

Brown Researchers Assess Differences Between Proxy and Patient Assessments of Cancer Care

There is a growing interest, in the United States and abroad, in evaluating patient experience and perceived quality of care. In July 2016, the Centers for Medicare and Medicaid Services initiated the Oncology Care Model, a new payment model for physician practices administering chemotherapy that will include financial and performance accountability, with performance based in part on patient experience surveys. In major national health and health care surveys, including the Consumer Assessment of Health Plans Survey (CAHPS), proxy respondents are allowed if patients are unable to respond. Proxy use can minimize missing data and improve representativeness, because proxies are typically required for sicker patients. However, using proxy responses raises concerns about whether they affect estimates, an important concern when reports and ratings are used for payment and planning purposes.

The purpose of this study, led by Jessica Roydhouse, doctoral student in the Department of Health Services, Policy and Practice, was to assess the impact of proxy survey responses on cancer care experience reports and quality ratings, using data from a large, clinically and demographically representative survey of patients with incident lung or colorectal cancer or their proxies.

The researchers found that, of the 6,471 respondents included in the analysis, 1,011 (16%) were proxies. The proportion of proxy responses varied from 6% to 28% across included settings. Adjusted proxy scores were modestly higher for medical care experiences, but lower for nursing care and care coordination experiences. There were no significant differences between adjusted patient and proxy ratings of quality.

The results of this study suggest that proxy reports of care experiences and ratings of care quality have small but statistically significant differences from patient reports. While proxy-reported health status differed from patient-reported heath status, likely reflecting a sicker population, this did not affect the outcomes of interest. Using proxy reports of experience and quality outcomes as well as patients’ health status covariates is unlikely to shift estimates substantially, even in situations of high rates of proxy use.

This study was published in Health Services Research, 2017 (ahead of print).

For more information: