Environmental clustering of drug abuse in households and communities: multi-level modeling of a national Swedish sample.

Abstract

Both epidemiological and genetically informative studies indicate that shared environmental influences contribute to resemblance in siblings for drug abuse (DA). To what degree do these influences arise from living in the same household versus residing in the same community? We performed a cross-classified multi-level logistic regression on all individuals born in Sweden 1975-1990 (N = 1558,654). We assessed the proportion of the total population variation in DA that was due to household versus community effects controlling for genetic resemblance. DA was assessed from medical, criminal and pharmacy records. Expressed as an intraclass correlation (ICC), the combined household/community effects accounted for ~8 % of the total population variation in DA. The variance attributed to the community was greater than that seen for household (4.5 versus 3.4 %). In males, the variance components were slightly larger and nearly equal at the community (5.3 %) and household level (5.1 %). In females, household effects (4.8 %) were stronger than those arising from the community (3.2 %). In the total population and among males, community effects on DA were somewhat more potent than household effects. However, in females, household effects on DA were stronger than community effects. In Sweden, shared environmental effects for DA arise both at the household and at the community level. Community effects on DA are more potent in males than in females.