Attention Dysfunction Predicts Substance Involvement in Community Youths

Abstract

Objective:

Cognitive impairments influence alcohol and drug treatment outcomes, though little is known about how neurocognition affects the development of harmful substance use patterns. This study examined the influence of adolescent attention functioning on the development of substance use problems in 66 high-risk youths over an 8-year period.

Method:

Participants were community youths who were free from any history of substance use disorders, neurological illness, and mood, anxiety, or psychotic disorders at project intake and were administered neuropsychological tests and substance involvement interviews from ages 15 through 23 on average. Substance involvement was assessed by self-report, resource person reports, and randomly sampled toxicology screens.

Results:

Attention/executive functioning scores obtained at the intake neuropsychological assessment significantly predicted substance use and dependence symptoms 8 years later, even after controlling for intake substance involvement, gender, education, conduct disorder, family history of substance use disorders, and learning disabilities.

Conclusions:

These results suggest that adolescents with limited attentional abilities, but not necessarily attention-deficit/hyperactivity disorder diagnoses, may be at risk for developing more problematic alcohol and drug involvement. Thus prevention and treatment efforts should target youths with attentional difficulties by using programs that are effective for those with compromised concentration and processing abilities.

Protracted heavy alcohol or drug use may cause neurocognitive dysfunction in young people (Brown et al., 2000a; Tapert et al., 2001; Tapert and Brown, 1999). At the same time, cognitive deficits may be risk factors for the development of chronic substance use or substance use disorders (SUD). This study prospectively examined the influence of adolescent attentional functioning on the development and maintenance of alcohol and drug use and related problems.

Attention has been described in terms of three major functions: alerting, orienting, and executive control (Berger and Posner, 2000; Posner and Raichle, 1995). Concentration, mental tracking, divided attention, and working memory all require the ongoing allocation of cognitive resources to a determined stimuli or set of stimuli (Robertson et al., 1991). Impairments in these ability areas may be diagnosed as attention-deficit/hyperactivity disorder (ADHD) if functioning is significantly disrupted in more than one setting. ADHD is estimated to occur in 3% to 6% of school-age youths (Goldman et al., 1998), and persists into adolescence for 70% of those who receive the diagnosis as children (Barkley et al., 1990; Biederman et al., 1996).

ADHD is common among adolescents in treatment for SUD; estimates range from 20% to 50% (Brown et al., 1990; DeMilio, 1989; Horner and Scheibe, 1997). Whether or not ADHD directly or indirectly increases the risk of SUD is a topic of debate. Prospectively, adolescents with ADHD show no increased risk of alcohol or drug problems (Biederman et al., 1997), while retrospectively, adults meeting ADHD criteria appear twice as likely to have a SUD (Biederman et al., 1998). Sensation seeking, poor academic achievement, and early disruptive behavior are substance involvement risk factors associated with ADHD (Riggs and Whitmore, 1999; Zuckerman, 1994). Inattention is symptomatic of conditions linked to increased substance involvement risk, such as prenatal alcohol exposure (Streissguth et al., 1991), family history of SUD (Hill et al., 1999, 2000; Tarter et al., 1989a), and depression (Tramontana et al., 1988). Children with ADHD have an elevated risk for developing conduct problems (Spencer et al., 1999), and a higher prevalence of ADHD is found in delinquent samples than in the general population (27% versus 5%, respectively; Moffitt and Silva, 1988).

Comorbid conduct disorder (CD) and ADHD have been associated with a notably poor behavioral prognosis (Moffitt, 1990), including substance-related problems (Barkley et al., 1990; Crowley and Riggs, 1995; Pihl and Peterson, 1991; Thompson et al., 1996; Wilens et al., 1996). Across multiple studies, CD has mediated or moderated the relationship between ADHD and SUD (Barkley et al., 1990; Biederman et al., 1997; Crowley and Riggs, 1995; Wilens et al., 1996). However, substance-involved youths may engage in CD behaviors solely as a function of their alcohol or drug use (Brown et al., 1996; Myers et al., 1995). Consequently, when evaluating the role of CD in the relationship between attentional deficits and substance involvement, it is important to distinguish substance use–provoked from independent CD behaviors. Once substance-involved, adolescents with ADHD tend to transition more rapidly from substance abuse to dependence (Wilens et al., 1999), consume and meet dependence criteria for more substances (Thompson et al., 1996), and experience more severe substance dependence symptoms (Horner and Scheibe, 1997) than non-ADHD adolescents. Adults with ADHD show elevated rates of alcohol abuse/dependence (12%–33%) and drug abuse/dependence (5%–21%) (Downey et al., 1997; Robins et al., 1984).

Attention appears to be involved in the persistence of alcohol and drug problems after treatment. Recovering teenagers with limited coping repertoires who performed poorly on tests of attention show greater difficulty avoiding a relapse than those with adequate coping or intact attention scores (Tapert et al., 1999). This may be due to compromised learning during treatment, limitations in detecting cues that signal potential relapse situations, or failure to appraise situations as requiring proactive behaviors to prevent a relapse (Myers and Brown, 1995; Myers et al., 1993; Tapert et al., 1999). Social information-processing models (Coie and Dodge, 1998) describe how proximal cognitive and emotional states lead to drinking and drug use within the social context of peer use. Distal biological, cultural, and familial factors are linked with proximal peer, availability, reinforcement, and mood cues in intentional and automatic decisions regarding substance use. Neuromaturational processes (e.g., prefrontal synaptic refinement and myelination) may dictate the extent to which immediate versus long-term consequences are incorporated into social cognitions (Tapert et al., 1999).

The current study examined the influence of adolescent neuropsychological functioning on the subsequent development of substance involvement in young adulthood. It was hypothesized that poor functioning on neuropsychological tests of attention during adolescence would predict greater levels of substance use and substance dependence symptoms in young adulthood.

METHOD

Participants

Participants were adolescents and their parents who were solicited through community advertisements to serve as a nonabusing comparison sample in a longitudinal study of adolescent SUD (Brown et al., 1987, 1989, 1994, 2000a; Tapert and Brown, 1999, 2000). To serve this purpose, our comparison sample had the same socioeconomic, genetic, and family background characteristics as adolescents with SUD and is thus a high-risk group. Exclusion criteria for adolescents were (1) history of alcohol/drug problems, (2) no collateral (parent) available, and (3) possible history of DSM-III-R mood, anxiety, or psychotic disorder as determined by Diagnostic Interview Schedule for Children (Shaffer et al., 1993) screening questions administered independently to teenagers and parents. Sixty-six youths were included in the present study (Table 1). Of these, 61% reported having had at least one outpatient mental health visit in their lifetime, including 4 who received medication evaluations for ADHD and 29 who received family therapy. Informed consent, approved by the University of California San Diego institutional review board, was obtained from all participants.

TABLE 1.

Characteristics of Community Youths Followed for 8 Years (N = 66)

	Intake	8-Year Follow-up	Time Effect (p Value)
Female (%)	59
White (%)	79
Years of age (intake)	15.05 (1.79)	23.42 (1.81)	.000
Years of education (intake)	8.88 (1.17)	14.08 (2.22)	.000
Learning disability placement (%)	15	27	.016
Attended college (%)	0	75	.000
Hollingshead SES (intake)a	30.55 (13.31)	45.31 (15.06)	.000
Grades completed by mother	14.12 (2.81)
Grades completed by father	14.61 (2.58)
Family history of alcohol dependence (%)	61
Family history of drug dependence (%)	22
Met diagnostic criteria for CDb (%)	18
Had one or more CD symptomsb (%)	44
Alcohol use per month	2.32 (5.75)	5.27 (4.52)	.001
Marijuana use per month	0.74 (3.82)	2.17 (7.05)	NS
Stimulant use per month	0.23 (0.99)	0.39 (3.08)	NS
Total substance use days per month	3.39 (7.75)	5.85 (7.66)	.04
Substance dependence symptomsc	0.00 (0.00)	1.55 (3.16)	.001

Note: Values represent mean (SD) or percent. SES = socioeconomic status; CD = conduct disorder; NS = not significant.

^aHigher scores represent lower SES, range 11–77.

^bDSM-IV CD criteria met independent of alcohol or other drug involvement.

^cDSM-IV substance dependence criteria for alcohol and participant’s three most frequently used drugs (range: 0–28).

Measures

Structured Clinical Interview.

A 90-minute structured interview was given to participants and collateral informants at each time point (Brown et al., 1989), assessing demographic information, academic functioning, mental health care, and family history of substance use disorders (FH).

Neuropsychological Test Battery.

A 2-hour neuropsychological test battery was administered at intake and follow-ups. For the present study, five measures of attention and executive functioning were selected on the basis of principal components analysis (Tapert and Brown, 2000): WISC-R (Wechsler, 1974) Digits Forward, Digits Backward, Arithmetic, and Coding, and the Halstead-Reitan Battery Trail Making Test (Reitan and Wolfson, 1993) Part B time.

Customary Drinking and Drug Use Record.

The Customary Drinking and Drug Use Record (Brown et al., 1998), given at each time point, assesses alcohol and other drug use, related life problems, withdrawal symptoms, and DSM-III-R and DSM-IV (American Psychiatric Association, 1987, 1994) SUD criteria. Good internal consistency, test-retest and interrater reliability, and convergent, discriminant, and criterion validity of this instrument have been demonstrated with community and clinical adolescent and young adult samples (e.g., Brown et al., 1998; Stewart and Brown, 1995). Outcome variables were calculated from the 8-year follow-up: days on which any substance use occurred and number of DSM-IV substance dependence symptoms met, both for the previous 2 years. To follow significant results concerning frequency, days on which the most widely used substances (alcohol, marijuana, and stimulants) each occurred were also calculated. Dependence symptoms were targeted because they are easier to discriminate in youths than are abuse symptoms (Pollock et al., 2000).

Conduct Disorder Questionnaire.

The Conduct Disorder Questionnaire (CDQ) (Brown et al., 1996), administered at project intake, is a structured interview that assesses CD behaviors according to DSM-III-R and DSM-IV criteria. For each behavior endorsed, the relation to substance use is assessed. Behaviors that occurred solely in an attempt to acquire alcohol or drugs, while under the influence, or during acute withdrawal are considered related to substance use. When a behavior had ever occurred without such relation to substance involvement, the behavior was considered to be independent from substance use. The CDQ has shown good psychometric characteristics, and scores of independent CD behavior predict progression to antisocial personality disorder (Brown et al., 1996; Myers et al., 1995, 1998; Stice et al., 1998).

Child Behavior Checklist.

The Child Behavior Checklist (CBCL) (Achenbach, 1991) was administered to parents of participants at the 4-year follow-up to assess child behaviors and characteristics during childhood prior to substance involvement. The CBCL Attention Problems factor scale was used in selected analyses.

Conners Parent Rating Scale.

The Conners Parent Rating Scale (CPRS) (Conners, 1994) was completed by participants’ parents at the 2-year follow-up. This measure has shown 93% accuracy in diagnosing ADHD (Conners, 1999; Conners et al., 1998), and interrater reliability (Conners, 1973) and concurrent validity are well established (Achenbach and Edelbrock, 1983; Arnold et al., 1981; Campbell and Steinert, 1978; Mash and Johnston, 1983). The CPRS Hyperactivity-Impulsivity Index was used for some analyses.

Procedures

Youths were assessed at project intake and at 6-month, 1-year, 2-year, 4-year, 6-year, and 8-year follow-ups. The present study uses data collected at the intake and 8-year follow-up time points. Ninety-three percent of subjects who had complete neuropsychological information at project intake were successfully followed 8 years later. At each follow-up, 10% of participants were randomly selected to provide urine samples for toxicology screening. Assessments were completed at the research facility or participant residences. Interviewers were bachelor’s- and master’s-level research assistants trained to criterion. Resource persons (typically the biological mother) were separately assessed by a different interviewer for corroboration of life history, FH, CD, behavioral, and substance use information. In instances of discrepant information between sources, data were obtained from additional resource people or medical or legal records and data were coded with the report indicating the poorer level of functioning. Toxicology results were consistent with self-report in each instance.

RESULTS

T scores were computed using intake score means and standard deviations of youths with no substance problems throughout the 8 years of follow-up (n = 34). Trail Making Test Part B time was inverted so high scores reflected good performance on all tests. The attention/executive composite T score was calculated according to the principal components analysis factor structure (Tapert and Brown, 2000). Although our measure of attention is neurocognitive rather than diagnostic, attentional test performances were associated with parent reports of attention and hyperactivity problems. The attention/executive score correlated with CBCL Attention Problems (r = −0.29, p < .01), and Trail Making Test Part B time (inverted) correlated with CBCL Attention Problems (r = −0.27, p < .05) and CPRS Hyperactivity (r = −0.23, p < .05).

On average, substance involvement increased as youths transitioned into the legal drinking age and young adulthood (Table 1), with 17% meeting diagnostic criteria for substance dependence (typically alcohol, marijuana, or amphetamines) at the 8-year follow-up, commensurate with the high-risk nature of this sample. Male gender and baseline substance use were significantly associated with more frequent use and more substance dependence criteria in young adulthood, and baseline educational attainment correlated with the attention/executive score, so these variables were controlled in analyses. Maternal drinking during pregnancy was not correlated with the attention/executive score or follow-up substance use.

Hierarchical regressions evaluated the influence of attention abilities on 8-year follow-up substance use frequency and dependence symptoms. Attention/executive scores at intake predicted substance use days per month (F_4,60 = 4.31, p < .005; R²Δ = 12%, β = −.39, p < .005) and marijuana use days per month (F_4,60 = 3.97, p < .01; R²Δ = 5%, β = −.25, p < .05) at the 8-year follow-up above and beyond the effects of gender, grades completed, and the corresponding intake substance use variable (Fig. 1). Substance dependence symptoms at the 8-year follow-up were not predicted by the set of independent variables (F_5,58 = 1.95, p < .09), but intake attention/executive scores appeared notable (R²Δ = 8%, β = −.33, p < .05). In contrast, follow-up alcohol and stimulant use frequency were not predicted by baseline attention/executive scores.

Fig. 1.

Attention functioning in adolescence predicts substance use frequency 8 years later in a community sample (N = 66). ● = any substance use (days per month averaged over the past 2 years; y = 28.18–0.46x, p < .005; regression line is solid); ○ = marijuana use (days per month averaged over the past 3 months; y = 16.95–0.30x, p < .05; regression line is dashed).

To examine how intake neuropsychological test scores differed for youths who met at least one of the seven DSM-IV criteria for substance dependence over the 8-year follow-up period (n = 27) and those who did not (n = 39), two independent sample t tests were conducted. Estimated IQ scores based on intake WISC-R Vocabulary and Block Design scores (Sattler, 1992) were virtually identical for youths who later met substance dependence criteria and those who did not (104.78 ± 10.18 versus 105.75 ± 12.45, respectively). However, intake attention/executive T scores were significantly lower among those who later met at least one dependence criterion by the 8-year follow-up compared with youths who did not (51.04 ± 8.22 versus 45.93 ± 5.25, p < .01). Post hoc t tests revealed significantly poorer scores on intake WISC-R Arithmetic, WISC-R Coding, and Trail Making Test Part B time for youths who later met one or more substance dependence criteria by the 8-year follow-up.

To understand the relationship between attention and substance involvement, we examined the influence of factors related to inattention: CD, FH, and learning problems. Intake attention/executive scores predicted 15% of the variance in substance use days at the 8-year follow-up over and above effects accounted for by gender, education, intake substance use, and the number of CD symptoms at intake (F_6,57 = 3.47, p < .005; β = −.46, p < .001). Similar results were obtained in predicting 8-year substance dependence symptoms and marijuana use frequency, but baseline attention/executive scores did not significantly predict 8-year alcohol or stimulant use frequency. Substance-independent CD behaviors at intake did not predict subsequent substance involvement and thus did not mediate the relationship between attention/executive scores and substance use or dependence.

We also tested a moderation model to determine whether the attention/executive–substance use relationship might vary as a function of conduct severity. The centered attention/executive factor score and centered independent CD symptom count at intake were multiplied to create an interaction term that was entered as the final step in a hierarchical regression with covariates as above on the first step and main effects on the second step. The interaction term significantly predicted 8-year substance use days (F_7,56 = 3.75, p < .005; R²Δ = 5%, β = .28, p < .05) above and beyond preceding steps. For youths who had one or more substance-independent CD symptom at intake (n = 29), attention/executive scores did not predict later substance use frequency (β = −.28, not significant), but attention/executive functioning significantly predicted substance use frequency for subjects (n = 37) who had not met any criteria for lifetime conduct problems (β = −.42, p < .01). These results were replicated using a full count of CD behaviors regardless of independence to alcohol or drugs.

FH was not significantly associated with the frequency of substance use at the 8-year follow-up. Furthermore, FH did not moderate or mediate the relationship between attention/executive scores and subsequent substance use. While the number of biological relatives with SUD was associated with poorer attention/executive scores (r = −0.26, p < .05), intake attention/executive scores continued to predict 15% of the variability in substance use days at the 8-year follow-up above and beyond effects accounted for by FH.

History of placement in learning disability educational services was examined for potential mediating and moderating effects on the attention/executive–substance use relationship in hierarchical regressions. Youths who had been placed in learning disability services had lower attention/executive scores (p < .001) and more substance use days at 8-year follow-up (p < .05), but such a history did not account for incremental variance beyond that accounted for by attention/executive scores in predicting 8-year substance involvement. Rather, attention/executive scores significantly predicted 8-year substance use above and beyond variance accounted for by covariates or learning disability placement (F_5,56 = 4.96, p < .001; R²Δ = 7%; attention/executive T score β = −.36, p < .025).

DISCUSSION

This study explored the influence of adolescent neurocognitive functioning on the development of substance use and dependence in young adulthood. The hypothesis that poorer attention/executive functioning would predict greater substance involvement was supported in this sample of community adolescents above and beyond effects accounted for by preexisting substance use levels, gender, education, CD symptoms, FH, and school placement for learning disabilities. Lower scores on neuropsychological tests of attention were prospectively associated with greater substance use frequency and more marijuana use in particular. Marijuana, used by more than a quarter of the sample at follow-up, has been suggested as a gateway substance in the initiation (Kandel and Yamaguchi, 1993) and resumption (Brown et al., 2000b) of alcohol and drug problems. In addition, youths who went on to meet one or more substance dependence criteria in young adulthood had significantly poorer attentional performance during adolescence. Other researchers have also documented a substance involvement risk for youths with attention difficulties (Biederman et al., 1998). These results suggest a trajectory of progressively deteriorating attentional abilities in which limited skills result in protracted alcohol and drug involvement that further compromises these functions (Tapert and Brown, 1999).

It is interesting that in contrast to previous research, CD behaviors did not mediate the relationship between attention functioning and substance involvement. Scores on neuropsychological tests of attention predicted subsequent substance use above and beyond the effects of CD symptoms, suggesting that factors other than problem behavior may contribute to escalating substance involvement. There may be several explanations for this finding. First, whereas previous studies investigated ADHD and CD diagnoses, we did not assess disorders categorically but instead used dimensional measures of attention and CD behaviors and only considered CD behaviors that could not be accounted for solely as a function of substance involvement. This included categorical analyses based on meeting one or more CD symptoms, not necessarily full diagnostic criteria. Second, we examined the relationship between attention and substance involvement across a very long time span, whereas most studies have examined this association cross-sectionally or over briefer intervals. Finally, this was a sample at elevated risk for SUD rather than a sample from the general population. We did, however, find that CD symptoms moderated the relationship between attention and substance use such that the association was stronger for youths who had not met any criteria for CD symptoms. Thus it may be that inattention or impulsivity places youths at risk for substance use initiation or escalation. Furthermore, factors other than adolescent problem behavior may contribute to the development of SUD. Youths with alcohol use disorders are more likely to have been victims of physical abuse, sexual abuse, or other forms of violence, and they report substantially more negative life events than control peers (Clark et al., 1997).

FH was not associated with acceleration into young adult substance involvement but was linked to attention in this predominantly FH-positive sample. Youths with family histories of alcohol dependence have been found to be at an increased risk for neuropsychological deficits, predominantly frontal lobe dysfunction (Harden and Pihl, 1995; Tarter et al., 1985, 1989b), and are more likely to develop SUD during their lifetime than youths without such family histories (Cloninger et al., 1981; Goodwin, 1979; Schuckit, 1985). Consistent with a previous study (Tapert and Brown, 2000), FH was associated with poorer attention scores at intake but not higher levels of young adult substance use. Other authors (Hill et al., 1999, 2000) have proposed that youths with family histories of alcohol dependence may have subtle developmental delays in brain maturation and cognition that result in high-risk trajectories but which are not directly apparent by young adulthood.

Limitations

Some limitations of the present study warrant consideration. First, clinical diagnoses of ADHD were not available in the current study. Therefore, the relationship found between attention and substance involvement is limited to attentional abilities as measured by neuropsychological tests rather than a clinical diagnosis. These results do not necessarily generalize to youths who meet diagnostic criteria for ADHD. On the other hand, neuropsychological measures of attention correlated significantly with well-validated behavioral measures of attention problems. A second limitation was the fairly small sample size; thus findings should be replicated in a larger sample of youths.

Clinical Implications

This study provides several clinical implications. First, because adolescents with poorer neurocognitive skills are at greater risk for developing substance use problems, prevention and treatment messages should be designed with attentional impairments in mind. One suggestion is to package intervention information as brief entertainment in multimedia, interactive formats (Epstein and McGaha, 1999). Second, teaching specific cognitive, behavioral, and abstinence-focused coping strategies may be particularly important for teenagers with limited attentional abilities because such coping skills may be critical for sustaining abstinence (Tapert et al., 1999). The identification of specific cognitive risk factors for substance use problems may guide more targeted prevention and intervention programs at earlier stages of problem development.