Figure 2.

Description of sampling and response rate for baseline and follow-up. * wrong contact data, although the telephone numbers of the insurants were updated using accessible public databases. † not accessible = not personally accessible despite 10 attempts with defined time intervals of up to 7 days. ^# 138 further patients were in the baseline sample but could not be assigned to any group when more detailed enrollment data were available in the course of data analysis. 57 of these patients were interviewed at baseline and 39 reinterviewed at follow-up. Their interviews were not included into the analyses. ^§ according to patient; among enrolled patients: 14 (1,7%) no diabetes, 5 (0,6%) diabetes type 1; among not enrolled patients: 307 (16,9%) no diabetes, 51 (2,8%) diabetes type 1.