The Norwegian Institute of Public Health twin study of mental health: examining recruitment and attrition bias.

All Norwegian twin pairs born 1967-1974 and still living in Norway in 1992 were invited to a health questionnaire study (Q1). 2,570 pairs (65%) participated. These cohorts and the twin cohorts born 1967-1979 were invited to a new questionnaire study (Q2) in 1998. This time 3,334 pairs (53%) participated. Almost all pairs having participated in the 1998 study were invited to an interview study of mental health (MHS), taking place 1999-2004. 1,391 complete pairs (44%) participated. The questionnaire studies included extensive data on somatic health with fewer items on mental health and demography. Health-related and demographic information available from the Medical Birth Registry on all invited twins was applied to predict participation to the first study. A few registry variables indicating poor health predicted nonparticipation in Q1. Health information and demography from Q1 were tested as predictors of participation in the follow-up study (Q2). Monozygosity, female sex, being unmarried, having no children, and high education predicted participation, whereas few indicators of poor mental and somatic health and unhealthy lifestyle moderately predicted nonparticipation in Q2. No health indicators reported in Q2 predicted further participation. Standard genetic twin analyses of indicators of various mental disorders from Q2, validated by diagnostic data from the MHS, did not indicate differences in genetic/environmental covariance structures between participants and nonparticipants in MHS. In general the results show a moderate selection towards good mental and somatic health. Attrition from Q2 to the MHS does not appear to affect twin analyses of mental health related variables.