Self-selection bias in an epidemiological study of respiratory health of a rural population.

A common source of bias in epidemiological studies is self-selection or volunteer bias. The self-selection into cohorts for further investigation or for participation in a substudy may be an important source of selection bias. The objectives of this paper were to identify differences in selected demographic characteristics, individual and contextual factors, and variables related to respiratory health between study participants who were willing to participate in subsequent clinical and allergy testing and those who were not willing to participate. One hundred and fourteen households (207 individuals) from a small town and 54 households (99 individuals) from a nearby rural municipality participated in this study. One key informant from each household provided information about each adult in that household. The question being studied was: "We wish to find out more about respiratory health of rural people. We would like to invite you to perform breathing and allergy tests. Would you be willing to be contacted for breathing and allergy tests in a nearby location?" One hundred and four participants said "Yes" they would participate in the clinical studies, 144 said "No" they would not participate in the clinical studies, and 52 said "Would like more information" about the clinical studies. More than one half (53.8%) of male participants and 46.2% of female participants indicated that they would like to participate in breathing and allergy tests. A higher proportion of study participants (26.1% males, 30.8% females) in the lowest income category requested more information compared to those answering either "Yes" (15.7% for males, 20.5% for females) or "No" (18.5% males, 23.3% females) to the question being studied. Study participants who were willing to participate in further breathing and allergy tests had a higher proportion of self-reported chronic phlegm and ever had allergic reaction to things eaten than those who either said "No" or "Would like more information." Among male study participants who said "Yes" to further participation, a higher proportion was exposed to one of the occupational exposures of interest compared to those who said either "No" or "Would like more information." This pattern was not observed for females.