Hazard ratios from unadjusted and adjusted Cox analyses are shown in Table 3 for AA twins. Effects of parental alcoholism and/or parental separation were nonsignificant, with one exception: AA twins from alcoholic separated families were at 1.88 times increased likelihood of smoking onset through age 14, compared to twins from nonalcoholic intact families. In adjusted models, significant protective effects were observed. Parental alcoholism was associated with 75% decreased likelihood of regular smoking from age 15 onwards if parents were separated, and across adolescence for twins whose parents remained together. In nonalcoholic AA families where parents separated, AA twins were at 92% decreased likelihood of regular smoking from age 18 onwards. Parental alcoholism predicted 96% decreased likelihood of non-cannabis illicit drug use from age 15 onwards if parents were separated, and 97% decreased likelihood across risk period(s) for twins whose parents remained together; in nonalcoholic families where parents separated, twins were at 93% decreased likelihood of other illicit drug use through age 14.
