Results of multinomial logistic regression using 1,919 offspring at varying levels of genetic vulnerability for ND suggested ever smoking was associated with increasing offspring age, white race, high maternal pressure to succeed in school, sibling drug use, and friend smoking, alcohol and drug use. Offspring regular smoking was associated with these same factors with additional contribution from maternal ND. Offspring ND was associated with increasing offspring age, male gender, biological parents divorce, high genetic risk from father and mother ND, maternal problem drinking, maternal rule inconsistency and sibling drug use, and friend smoking, alcohol and drug use. To our knowledge these findings are novel in that family and non-family measured environmental factors contribute to offspring smoking outcomes after controlling for familial vulnerability.
For all outcomes the largest magnitude of influence came from friend smoking, but we are unable to determine if this is due to peer influence or peer selection (Madden et al. 2002
). In a parallel study of offspring of drug dependent fathers, we found the largest risk for cannabis dependence was perceived friend cannabis use but peer smoking was not associated with cannabis abuse/dependence. Taken together, these two studies suggest some specificity for the environmental contribution to offspring substance use. Further research is warranted to determine the relative magnitude of peer drinking on offspring alcoholism.
Many of our observations are consistent with other large cohort studies. Increasing age is associated with increasing smoking and ND (Hu, Davies & Kandel 2004
) in adolescents and young adults. And our finding for an association between white race and regular smoking is consistent with previous studies in the United States (Park, Weaver & Romer 2009
). Previous studies have demonstrated peer smoking is significantly associated ever smoking (Hu, Davies & Kandel 2004
), transitions to daily smoking (Bricker et al. 2007
) and with developing ND (Hu, Davies & Kandel 2004
; Kandel et al. 2007
). Our study is unique in demonstrating the significance of peer influences after accounting for familial vulnerability for ND and measured environmental influences. Similarly unique is the finding that sibling substance use remained significantly associated with all offspring outcomes. This is consistent with other population based cohorts (Boyle et al. 2001
) but we believe our analysis is the first to demonstrate a contribution from sibling smoking even after accounting for degree of genetic and environmental risk for ND from the parent generation.
We found paternal alcoholism and illicit drug dependence did not predict offspring smoking initiation or regular smoking, but high familial risk for ND was associated with offspring ND. This finding extends our previous analyses demonstrating the specificity of the inter-generational transmission of alcohol dependence and nicotine dependence in a subset of offspring of alcoholic twins (Volk et al. 2007
). Our current results are not completely consistent with classical twin models of smoking outcomes in which genetic factors have been shown to contribute to tobacco use, regular smoking and ND. For example, Maes et al. (2004) reported high heritability for initiation of tobacco use, regular smoking, and ND (75%, 80%, and 62%, respectively) and substantial genetic correlations between phenotypes, but found that over one-third of genetic liability was specific to ND (Maes et al. 2004
). Investigations of common heritable influences on initiation and persistence of smoking (Madden et al. 2006
; True et al. 1997
; True et al. 1999
) and on initiation and ND (Fagerstrom 1978
) are mostly consistent in that genetic factors for regular smoking are not completely accounted for by the genetic factors contributing to smoking initiation (True et al. 1997
; Heath & Madden 1995
). We speculate that the variance attributed to genetic factors in the classical twin design are partly associated with sibling substance use and peer substance use which are modeled in the present paper but not accounted for in classical twin studies.
The absence of a significant association between parental closeness and smoking is consistent with other multivariate analysis of parent and sibling influences on smoking (Picotte et al. 2006
A principal strength of the present work is the use of the twin-offspring design which permits a sensitive test for measured environmental influences while accounting for genetic vulnerability due to parental ND. Additional strengths include the large sample size and non-clinical sample that enhance generalizability while avoiding bias inherent to clinical samples. The structured method of data collection reduces interviewer bias. Last the large age range of the offspring permitted modeling the major smoking milestones in adolescents and young adults. We computed post-hoc sensitivity analysis using data only on offspring 18 years and older and can conclude that point estimates were not significantly different before and after limiting the cohort to these older aged offspring.
Sample size limitations may have reduced our statistical power to detect differences in the risk for smoking outcomes between Group 2 (HG-HE) and Group 3 (MG-LE) effects. Power to detect differences between these groups was 30%. However, we are able to conclude that high genetic and high environment as compared to low genetic and low environmental risk is associated with offspring ND, and we can conclude that environmental factors contribute to offspring smoking outcomes after accounting for familial vulnerability (i.e. Group 4 vs. 1).
It is not possible to measure all environmental influences on offspring. We lacked data on offspring perception of peer smoking and lacked self reported measures from siblings on smoking, alcohol and drug use. Expansion of the shared environment assessment may reveal key parent, sibling and peer level variables that have not been adequately measured. Longitudinal data will help clarify the direction of effect for peer substance use. Follow-up is necessary to collect more detailed assessment of the parent-offspring environment, to expand upon sibling substance use and drug problems and the impact of older as compared to younger siblings on offspring smoking and to model trajectories related to life events associated with adulthood.
Because DSM-III-R criteria ND is weakly correlated with FTND it is possible that our results have underestimated the genetic contribution from paternal ND to offspring ND and overestimated the environmental contribution to offspring ND. Since offspring who smoke are more likely to perceive peers smoke it is possible that we overestimate the environmental contribution from peer smoking to offspring smoking.
Last, because of the ubiquitous exposure to tobacco in the military the present cohort may be more exposed to tobacco than those found in offspring of civilian populations.
Parent, sibling and peer level variables contribute to offspring ever smoking, regular smoking and ND. Even after controlling for familial vulnerability to ND, environmental contributions to smoking remain significant.