A comparison of participants and non-participants in a serum sample survey in Texas City, Texas.

Lack of participation and response can affect the validity and generalizability of epidemiologic surveys. Participation has decreased in recent decades, and thus there is a need to determine reasons for and implications of this change. Usually respondents have better health, habits, and socioeconomic status and are interested in their health, being described as a “worried healthy” group. It is unclear if this explanation applies to minorities. The Environmental Risk, Coping, and Hispanic Health Project of the Center for Population Health and Health Disparities of UTMB, a population-based survey conducted in Texas City, Texas, included an interview of the randomly selected participants and an invitation to donate blood for biological measures. The study represented an opportunity to compare the characteristics of participants in the initial survey who, subsequently, provided blood sample and those who declined. In this Capstone the hypothesis that disadvantaged and less healthy groups are less likely to participate in serum sample collection as part of a sero-epidemiologic study was tested. The specific aims included: to determine the percentage of respondents declining to donate blood; to compare the characteristics of those who donated blood and those who declined, for the whole sample and across ethnic groups; and to examine the association with stress. We found that participation was lower than in similar surveys, with different response rates by ethnicity. We did not find great differences between respondents and nonrespondents, other than age, smoking habit, and perceived poor health. The only factors that were maintained in a multivariate analysis were age, family income, and smoking habit. We also found interaction between age and ethnicity, with advanced age being related to less likelihood of nonresponse in all ethnicities except in Non-Hispanic Blacks, where the direction was the opposite. A comparison of respondents and nonrespondents across ethnicities showed no additional factors, with the exception of results within US-born Hispanics. In addition to age, being a smoker and reporting poor health were associated with nonresponse, while low income was associated with less likelihood of being nonrespondent. For all participants and across ethnicities perceived stress did not have an association with response