Are spouses' socio-economic classifications interchangeable? Examining the consequences of a commonly used practice in studies on social inequalities in health.

Abstract

Indicators of socio-economic position are not always available for all subjects. To avoid losses of large parts of study populations, missing data are replaced by spouses' information. Despite this commonly practiced solution, systematic analyses of the consequences on substantive results of studies are rare. We examined the consequences of assigning the educational position of subjects to their partners. German statutory health insurance data from 2005 (N=1,801,744) and 2011 (N=1,987,707) were used. Diagnoses of type 2 diabetes were used as outcome. Effects were examined in terms of differences in diabetes prevalence and by the reproduction of social gradients in women and men as compared to their partners. Social gradients were reproduced for subjects and for their partners, but diabetes prevalences were higher in partners. From a pragmatic point of view the practice of replacing missing information by spouses' information turned out as viable. However, the usefulness of this solution has to be examined in every case anew, because it may not be suitable for every health-related outcome.