Abstract
BACKGROUND: A common weakness of patient satisfaction surveys is a suboptimal participation rate. Some patients may be unable to participate, because of language barriers, physical limitations, or mental problems. As the role of these barriers is poorly understood, we aimed to identify patient characteristics that are associated with non-participation in a patient satisfaction survey. METHODOLOGY: At the University Hospitals of Geneva, Switzerland, a patient satisfaction survey is regularly conducted among all adult patients hospitalized for >24 hours on a one-month period in the departments of internal medicine, geriatrics, surgery, neurosciences, psychiatry, and gynaecology-obstetrics. In order to assess the factors associated with non-participation to the patient satisfaction survey, a case-control study was conducted among patients selected for the 2005 survey. Cases (non respondents, n = 195) and controls (respondents, n = 205) were randomly selected from the satisfaction survey, and information about potential barriers to participation was abstracted in a blinded fashion from the patients' medical and nursing charts. PRINCIPAL FINDINGS: Non-participation in the satisfaction survey was independently associated with the presence of a language barrier (odds ratio [OR] 4.53, 95% confidence interval [CI95%]: 2.14-9.59), substance abuse (OR 3.75, CI95%: 1.97-7.14), cognitive limitations (OR 3.72, CI95%: 1.64-8.42), a psychiatric diagnosis (OR 1.99, CI95%: 1.23-3.23) and a sight deficiency (OR 2.07, CI95%: 0.98-4.36). The odds ratio for non-participation increased gradually with the number of predictors. CONCLUSIONS: Five barriers to non-participation in a mail survey were identified. Gathering patient feedback through mailed surveys may lead to an under-representation of some patient subgroups.