Assessing Representativeness of Seriously Ill Patient Survey Responders in a Pragmatic Clinical Trial

Pragmatic trials often implement an intervention across a population of patients but require information unavailable at the population level that must be reported by a subset of patients. In this pragmatic clinical trial, we compared characteristics of seriously ill patients with those who completed a survey evaluating advance care planning across 3 academic health systems.

A deliberate process including health system and external stakeholders and patients was used to design materials for and the approach to seriously ill patients. We developed a survey and conducted a multistep process to identify seriously ill primary care patients. We evaluated the relationships of age, gender, race and ethnicity, and vulnerability using the social vulnerability index in this population, and explored the representativeness of survey respondents compared with the underlying seriously ill population in terms of age, race and ethnicity, and vulnerability measured.

About 5% (8707 patients) of the primary care population was classified as seriously ill, 5351 were mailed a survey and 1100 provided survey responses. Hispanic and Black patients were younger than White patients, and Black and Hispanic patients were more vulnerable than White and Asian patients and patients of other races. Representativeness was high across age and race and ethnicity, although White and Hispanic patients were more likely to respond than Black and Asian patients and patients of other races. Vulnerability in the surveyed sample was nearly identical to the population.

A tailored survey and recruitment strategy yielded a representative sample of seriously ill, largely older, primary care respondents in the context of a pragmatic clinical trial.