Response rates and selection problems, with emphasis on mental health variables and DNA sampling, in large population-based, cross-sectional and longitudinal studies of adolescents in Norway

Table 4 Baseline health characteristics in 2001, prevalence and relative risk of lost to follow-up in 2004 in Oslo and Hedmark combined

Variables collected	Base-line	Lost to follow-up		Relative risk of lost to follow-up
In 2001	N	n	%	RR_crude ¹	95% CI		RR_adj ^1,2	95% CI
Self-reported health
Good/very good	4951	2057	42	Ref			Ref
Bad/not so good	706	341	48	1.16	1.07	1.26	1.09	0.98	1.20
Total	5657	2398	42
Mental distress/HSCL above cutt-off
No	4695	1961	42	Ref			Ref
Yes	971	415	43	1.02	0.94	1.11	1.06	0.96	1.16
Total	5666	2376	42
Externalized symptoms
No	4985	1990	40	Ref			Ref
Yes	702	399	57	1.42	1.32	1.53	1.18	1.08	1.28
Total	5687	2389	42
Physical activity
3+ h	3556	1426	40	Ref			Ref
0-2 h	2041	913	45	1.12	1.05	1.19	1.03	0.95	1.10
Total	5597	2339	42
Smoking
Never or stopped	3962	1538	39	Ref			Ref
Occasionally or daily	1749	871	50	1.28	1.21	1.36	1.19	1.11	1.28
Total	5711	2409	42

¹When substituting "lost to follow-up" with "failure to provide DNA" as dependent variable, the RRs were similar
²Adjusted for invitation group, gender, ethnicity, family economy, parental marital status, educational plans, father's education and income. The number included in adjusted analyses is lower than in crude analyses due to missing data for some of the variables. For example, in analyses of self-reported health there are 5657 in crude analyses and 4527 in adjusted analyses. In particular, we lack data on father's education and income, which are obtained from Statistics Norway. In reanalyses of RR_crude, including only the number available in adjusted analyses, the associations were slightly stronger (except for physical activity), but did not change confidence intervals to significantly/non-significantly for either of the associations.

ISSN: 1471-2458