Differential implications of persistent, remitted, and late-onset ADHD symptoms for substance abuse in women and men: A twin study from ages 11 to 24

Abstract

Background: Persistence and emergence of ADHD in adulthood are associated with substance problems. We investigate differential implications of ADHD course for tobacco, alcohol, or marijuana problems by sex, then whether substance misuse results from ADHD or contributes to it, through a twin differences design.

Methods: A population-based cohort of 998 twins (61 % monozygotic; 52 % female), born in Minnesota from 1988 to 1994, was prospectively assessed from ages 11-24. Childhood ADHD was oversampled. At age 24, 255 had a history of childhood-onset ADHD (160 persistent, 95 remitted); 93 had late-onset ADHD symptoms identified in late-adolescence/adulthood. Persistent, remitted, and late-onset groups were compared to those without ADHD (N = 459) on childhood characteristics and age-24 substance problems.

Results: Persistent and late-onset groups differed in childhood; twin concordances suggested greater genetic etiology for persistent ADHD. As adolescents, however, both groups were high in conduct problems; by adulthood, they were comparably high in substance problems. In particular, women whose ADHD persisted were 5 times more likely to develop tobacco use disorder than women without ADHD. Remitted ADHD was associated with less-increased risk, except for alcohol problems among women. Consistent with possible causality, monozygotic female twins with more age-17 ADHD symptoms than co-twins had more age-24 tobacco symptoms; a similar association was found for alcohol.

Conclusions: Presence or emergence of ADHD in early adulthood increases substance problems to a greater degree for women than men. While effects of substances on later ADHD were not statistically significant, detection was limited by the relative rarity of late-adolescent substance symptoms.

Keywords: Alcohol; Attention deficit hyperactivity disorder; Discordant twin design; Gender differences; Longitudinal; Marijuana; Tobacco.

Figure 1.

Mean DSM-IV Conduct Disorder Symptoms at Each Adolescent Wave and Adult Antisocial Behavior by Age 24, By Course of ADHD Mean symptoms reported at ages 14, 17, and 24 are shown; error bars represent standard errors. Both Childhood-Onset groups (in green) were significantly higher in early CD (between ages 11-14) compared to those who never had ADHD (both p<.0001) but did not differ from each other. However, Persistent and Remitted cases diverged later on CD (1.1 vs. 0.4 symptoms at age 17) and adult antisocial behavior (both p=.001). Conversely, while Persistent and Late-Onset groups (in red) differed on CD early, they were very similar in late-adolescent CD and adult antisocial behavior.

Figures 2.

Cross-lagged Monozygotic Twin Difference Models of DSM-5 Symptoms of ADHD and Substance Use Disorders at Ages 17 and 24: (a) Tobacco Use Disorder; Females Only (N=125 pairs), (b) Tobacco Use Disorder; Males Only (N=98 pairs), (c) Alcohol Use Disorder (N=223 pairs), and (d) Marijuana Use Disorder (N=217 pairs). TOB = Tobacco Use Disorder; ALC = Alcohol Use Disorder; MAR = Marijuana Use Disorder. Significant standardized path estimates are shown in bold (with standard errors in parentheses); paths that were not significant are represented by dotted lines. The lack of stability in substance symptoms within pairs across time, as evident for alcohol (β = −.01; Figure 2c), is not surprising, as many had not yet engaged in heavy substance use prior to age 17 (see section 4.2.). Within-age correlations (curved paths) function as residuals at age 24. p < 0.05* p < 0.01** p < 0.001***

Figures 2.

Cross-lagged Monozygotic Twin Difference Models of DSM-5 Symptoms of ADHD and Substance Use Disorders at Ages 17 and 24: (a) Tobacco Use Disorder; Females Only (N=125 pairs), (b) Tobacco Use Disorder; Males Only (N=98 pairs), (c) Alcohol Use Disorder (N=223 pairs), and (d) Marijuana Use Disorder (N=217 pairs). TOB = Tobacco Use Disorder; ALC = Alcohol Use Disorder; MAR = Marijuana Use Disorder. Significant standardized path estimates are shown in bold (with standard errors in parentheses); paths that were not significant are represented by dotted lines. The lack of stability in substance symptoms within pairs across time, as evident for alcohol (β = −.01; Figure 2c), is not surprising, as many had not yet engaged in heavy substance use prior to age 17 (see section 4.2.). Within-age correlations (curved paths) function as residuals at age 24. p < 0.05* p < 0.01** p < 0.001***

Figures 2.

Cross-lagged Monozygotic Twin Difference Models of DSM-5 Symptoms of ADHD and Substance Use Disorders at Ages 17 and 24: (a) Tobacco Use Disorder; Females Only (N=125 pairs), (b) Tobacco Use Disorder; Males Only (N=98 pairs), (c) Alcohol Use Disorder (N=223 pairs), and (d) Marijuana Use Disorder (N=217 pairs). TOB = Tobacco Use Disorder; ALC = Alcohol Use Disorder; MAR = Marijuana Use Disorder. Significant standardized path estimates are shown in bold (with standard errors in parentheses); paths that were not significant are represented by dotted lines. The lack of stability in substance symptoms within pairs across time, as evident for alcohol (β = −.01; Figure 2c), is not surprising, as many had not yet engaged in heavy substance use prior to age 17 (see section 4.2.). Within-age correlations (curved paths) function as residuals at age 24. p < 0.05* p < 0.01** p < 0.001***