[Response bias in a population postal survey on Health-Related Quality of Life (sf-12v2)]

the bias associated with the low response rate may limit the economic advantage of population surveys by mail. The factors associated with non-response were estimated with an emphasis on Health-Related Quality of Life (HRQoL).

people who had answered to the Health Survey Murcia-2007 2007 (≥ 18 years), realized by telephone, were sent another postal questionnaire four months later. Both evaluations included information about HRQoL: SF-12v2 (by phone) and EQ-5D (postal). A logistic regression was realized to identify baseline factors (sociodemographic factors, health-related behaviors and HRQoL-physical component summary (PCS) and mental component summary (MCS) of the SF-12v2) associated with non-response to the postal survey.

2,078 individuals (61.5%) did not answer. Multivariate analysis: non-response was associated with marital status, more likely not to respond if it was not married: widower OR: 2,24; IC95% 0,61/0,88; separate/divorced 1,69; 1,10/2,59-; single 1,53; 1,23/1,90 ) and inversely with age (25-34 years OR: 0,95; IC95% 0,69/1,29; 35-44 years 0,60; 0,43/0,85; 45-54 years 0,42; 0,29/0,60; 55-64 years 0,29; 0,20/0,43; 65-74 years 0,17; 0,11/0,26-;≥75 years 0,15; 0,09/0,23) and educational level (low-OR: 0,65; IC95% 0,48/0,86; intermediate 0,41; 0,30/0,57; high-0,22; 0,16/0,30). It was also associated with HRQoL, but differentially for PCS (those with intermediate scores have less probability of not answering OR: 0.73; 95% CI: 0.61/0.88) than for the MCS (the persons in the highest tertile were more likely to not answering 1.47; 1.22/1.78.

the HRQoL influences the non- response to the postal surveys, leading to an overrepresentation of individuals with middle physical health and low mental health, which it adds small magnitude bias in the estimation of population HRQoL.