Sense of coherence and attrition during four-year follow-up in cohorts of permanent and non-permanent Finnish employees

Background: We studied whether health resources, measured as sense of coherence (SOC), are associated with participation in a follow-up survey among permanent and non-permanent employees who responded at baseline. Methods: Of a cohort of 5,981 permanent employees, those who after four years were still in the service of the same employer were asked to participate in a follow-up survey. Another cohort consisted of 2,194 fixed-term and 682 subsidised employees; among these the follow-up survey was posted to those whose addresses were found in the population register. Non-participation was divided into loss to follow-up (i.e., failure to locate the individual, death and, among permanent employees, turnover or exit from labour market) and non-response to the follow-up survey. Logistic regression analyses were used to examine whether the respondents differed from the non-respondents with respect to SOC and other characteristics at baseline. Results: Among permanent employees the follow-up survey yielded 3,998 respondents, 1,051 were lost, and 932 did not reply. Among non-permanent employees the follow-up survey yielded 1,563 respondents on initially fixed-term and 467 on subsidised contracts, the corresponding figures for those lost were 145 and 38, and for the non-respondents 486 and 177. Low SOC was associated with lower response rate among fixed-term but not among permanent or subsidised employees. No association was found between SOC and loss to follow-up. Conclusion: SOC is a potential source of non-random sample attrition and should be taken into account for when estimating bias due to non-participation in occupational cohorts that include fixed-term employees.