Association for Public Policy Analysis & Management

Research Objective: To obtain a complete understanding of patient care experiences in a high mortality population (e.g., cancer patients), it is important to collect information from patients who survive and from the proxy respondents of patients who do not. We tested whether a tailored survey approach for proxy respondents of deceased cancer patients would increase response rates above those achieved without tailoring.

Study Design: We are surveying patients treated by the nearly 200 oncology physician group practices participating in Medicare’s Oncology Care Model (OCM). We conduct repeated waves of sampling/surveying to measure patients’ experiences in order to monitor changes over time. The OCM program also uses care experiences as measured by the survey to adjust performance-based payments to participating practices.

We sampled patients receiving chemotherapy in the spring/summer of 2016 (survey wave 1) and in the fall of 2016 (survey wave 2). In wave 1, the survey we mailed was the same for all recipients, whether or not the patient had died by the time of survey mailing (untailored approach). In wave 2, we sent one survey instrument to patients who were alive at the time of mailing and sent an alternative survey instrument intended for the family of patients who had died (tailored approach). Although mailed to the decedent’s address, the alternative survey and tailored approach included a more personalized cover letter that recognized the death/loss and survey items reworded specifically for the proxy of the decedent. We compared proxy responses for deceased patients with and without the tailored approach.

Population Studied: We used Medicare claims to identify cancer patients who met the OCM episode definition and drew a proportionate sample of patients treated by each oncology practice participating in OCM, stratified by patient age, race and cancer type.

Principal Findings: The overall response rate for proxies of deceased patients increased from 17.6 percent in wave 1 to 39.0 percent in wave 2 (P<0.001). This increase in response rate was true for all patient subgroups defined by age, sex, race, dual eligibility and cancer type. We also examined response rates for patients who were alive at the time of the survey mailing and saw an insignificant change from wave 1 to wave 2 (46.0% vs. 48.3%), suggesting that temporal trends do not completely account for our findings.

Conclusions: The response rate for proxy respondents of deceased patients was substantially higher with the tailored approach compared with the untailored approach. Mail survey response is a multi-part process, requiring respondents to open the mail, decide to respond, and complete a response. In addressing each of these steps, the tailored approach likely improved proxy respondents’ progression through this multi-part process.

Implication for Policy or Practice: Simple tailoring strategies can significantly enhance proxy response when surveying patients in a high mortality population. Such strategies may be especially relevant when measuring experiences of the entire patient population served by a physician group (or other provider entity).

Funding: Section 3021, Affordable Care Act