Does Stimulant Therapy of Attention-Deficit/Hyperactivity Disorder Beget Later

Substance Abuse? A Meta-analytic Review of the Literature

Timothy E. Wilens, Stephen V. Faraone, Joseph Biederman and Samantha
Gunawardene
Pediatrics 2003;111;179
DOI: 10.1542/peds.111.1.179

The online version of this article, along with updated information and services, is
located on the World Wide Web at:
http://pediatrics.aappublications.org/content/111/1/179.full.html

PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly

publication, it has been published continuously since 1948. PEDIATRICS is owned,
published, and trademarked by the American Academy of Pediatrics, 141 Northwest Point
Boulevard, Elk Grove Village, Illinois, 60007. Copyright © 2003 by the American Academy
of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.

Downloaded from pediatrics.aappublications.org at Bibliotheque de l'Universite Laval on July 14, 2014

 Does Stimulant Therapy of Attention-Deficit/Hyperactivity Disorder
Beget Later Substance Abuse? A Meta-analytic Review of the Literature

Timothy E. Wilens, MD*‡; Stephen V. Faraone, PhD*‡; Joseph Biederman, MD*‡; and
Samantha Gunawardene, BS*

ABSTRACT. Objective. Concerns exist that stimulant risk for subsequent drug and alcohol use disorders.
therapy of youths with attention-deficit/hyperactivity Pediatrics 2003;111:179 –185; attention-deficit/hyperactiv-
disorder (ADHD) may result in an increased risk for ity disorder, substance use, pharmacotherapy.
subsequent substance use disorders (SUD). We investi-
gated all long-term studies in which pharmacologically
treated and untreated youths with ADHD were examined ABBREVIATIONS. ADHD, attention-deficit/hyperactivity disor-
for later SUD outcomes. der; SUD, substance use disorders; OR, odds ratio; POR, precision
of the odds ratio; SN, standard normal deviate; CI, confidence
Methods. A search of all available prospective and
interval;
retrospective studies of children, adolescents, and adults
with ADHD that had information relating childhood ex-

A
posure to stimulant therapy and later SUD outcome in ttention-deficit/hyperactivity disorder (ADHD)
adolescence or adulthood was conducted through is the most common neurobehavioral disorder
PubMed supplemented with data from scientific presen-
that is presented for treatment. It is estimated to
tations. Meta-analysis was used to evaluate the relation-
ship between stimulant therapy and subsequent SUD in affect from 4% to 9% of youths.1–3 Pharmacotherapy in
youths with ADHD in general while addressing specifi- general and stimulants in particular remain a mainstay of
cally differential effects on alcohol use disorders or drug treatment for ADHD.3–7 Data from ⬎200 randomized
use disorders and the potential effects of covariates. clinical trials have consistently documented that stimulant
Results. Six studies—2 with follow-up in adolescence drugs are highly effective in the treatment of youths and
and 4 in young adulthood—were included and com- adults with ADHD.4–7 A recently published large multi-
prised 674 medicated subjects and 360 unmedicated sub- site and randomized study documented the essential role
jects who were followed at least 4 years. The pooled that medication treatment plays in the long-term treat-
estimate of the odds ratio indicated a 1.9-fold reduction
ment of children with ADHD.8
in risk for SUD in youths who were treated with stimu-
lants compared with youths who did not receive phar- Despite stimulants’ well-documented efficacy in
macotherapy for ADHD (z ⴝ 2.1; 95% confidence interval the treatment of ADHD, concerns remain as to
for odds ratio [OR]: 1.1–3.6). We found similar reductions whether their use in youths with ADHD could in-
in risk for later drug and alcohol use disorders (z ⴝ 1.1). crease the risk for substance use disorders (SUD;
Studies that reported follow-up into adolescence showed denoting drug or alcohol abuse or dependence).9 –13
a greater protective effect on the development of SUD Although a recent report by our group showed that
(OR: 5.8) than studies that followed subjects into adult- anti-ADHD pharmacotherapy protected youths with
hood (OR: 1.4). Additional analyses showed that the re- ADHD from later SUD,14 another study reported just
sults could not be accounted for by any single study or by
publication bias.
the opposite: cocaine and nicotine abuse were asso-
Conclusion. Our results suggest that stimulant ther- ciated with previous stimulant treatment.15 These
apy in childhood is associated with a reduction in the contradictory findings call for additional efforts to
help resolve this critical issue.
Whether pharmacotherapy for ADHD in general
From the *Clinical Research Program in Pediatric Psychopharmacology,
and stimulant treatment in particular leads to SUD in
Massachusetts General Hospital, Boston, Massachusetts; and ‡Harvard
Medical School, Boston, Massachusetts. children with ADHD has serious clinical implica-
Dr Joseph Biederman receives research support from the following sources: tions given that medications are fundamental in the
Shire Laboratories, Eli Lilly & Company, Wyeth Ayerst, Pfizer Pharmaceu- treatment plan of individuals with ADHD.3,8 If stim-
tical, Cephalon Pharmaceutical, Janssen Pharmaceutical, Noven Pharma- ulant therapy for ADHD leads to SUD, then clini-
ceutical, Stanley Foundation, National Institute of Mental Health, National
Institute of Child Health and Human Development, and National Institute cians, patients, and families would need to weigh
on Drug Abuse; is a speaker for the following speakers’ bureaus: Glaxo- carefully the risk of SUD against its therapeutic ben-
Smith Kline, Eli Lilly & Company, Pfizer Pharmaceutical, Wyeth Ayerst, efits. If, however, stimulant treatment does not lead
Shire Laboratories, Alza Pharmaceutical, and Cephalon Pharmaceutical; to SUD, then clinicians, patients, and families could
and is on the advisory board of the following pharmaceutical companies: Eli
Lilly & Company, Cell Tech and Shire Laboratories, Noven Pharmaceutical,
approach pharmacological treatment of youths with
and Alza/McNeil Pharmaceuticals. ADHD without ungrounded fears of addiction-re-
Received for publication Jan 17, 2002; accepted Jun 18, 2002. lated complications. Furthermore, if stimulant treat-
Reprint requests to (T.E.W.) Pediatric Psychopharmacology Clinic, ACC ment for ADHD protects against SUD in youths with
725, Massachusetts General Hospital, Boston, MA 02114. E-mail:
[email protected]
ADHD, then pharmacotherapy would serve as a pre-
PEDIATRICS (ISSN 0031 4005). Copyright © 2003 by the American Acad- ventive approach for SUD risk in youths with
emy of Pediatrics. ADHD.

PEDIATRICS Vol. 111 No. 1 January 2003 179

Downloaded from pediatrics.aappublications.org at Bibliotheque de l'Universite Laval on July 14, 2014
 One approach for reconciling conflicting findings precision have large standard errors and therefore should have
among studies is meta-analysis. This method evalu- small SNDs; large samples have higher precision and smaller
standard errors and should have large SNDs). Egger’s publication
ates whether the aggregate evidence across all avail- bias statistic is the intercept of the regression, which will be
able studies provides evidence for statistical signifi- significantly greater than 0 in the presence of publication bias.
cance. Thus, to examine the putative association When the intercept is ⬎0, the smaller studies are finding larger
between SUD and previous exposure to stimulants, SNDs (and hence larger ORs) than expected. This would occur, for
example, if other negative smaller studies had not been published
we applied meta-analysis to all long-term studies in or if they had preferentially published outcomes with large ORs.
which pharmacologically treated and untreated All analyses used Stata 6.0.20
groups of individuals with ADHD were examined
for SUD outcomes.
We tested 3 competing hypotheses. The first was RESULTS
the null hypothesis that stimulant therapy would Our literature search revealed a total of 6 studies
have no demonstrable effect on the development of from the United States and Germany (Table 1). There
SUD in children with ADHD. The second was that were 5 prospective longitudinal studies: 2 involving
exposure to stimulants would predict a higher risk children who were followed at least 4 years14,21,22
for SUD in general and stimulant and other sympa- and 3 involving children who were followed into
thomimetic abuse in particular. The third posited young adulthood (Barkley RA et al, unpublished
that stimulant management of ADHD would dimin- observations).15,23–25 One retrospective report was
ish the later risk for SUD. This hypothesis derives available from a study of adults with ADHD.26 Over-
from the idea that SUD in children and adolescents all, these studies comprised 674 medicated subjects
with ADHD may be secondary to ADHD, because of and 360 unmedicated subjects. Four of 6 studies
attempts at self-medication16 or to direct effects of demonstrated similar levels of severity and psychi-
symptoms (eg, impulsivity) and their correlates (eg, atric comorbidity between the medicated and un-
poor self-esteem). medicated ADHD groups at baseline. For the vast
majority of medicated subjects (97%), stimulants
METHODS (methylphenidate or amphetamine) were used. In 1
We conducted a systematic literature search of all available study, because no overall rate of drug abuse was
prospective and retrospective studies of children, adolescents, and reported, rates of cocaine abuse in treated and un-
adults with ADHD that had information on childhood exposure to treated groups (the only statistically significant find-
stimulant pharmacotherapy and data on SUD outcome in adoles- ing in substance abuse rates reported in that study)
cence or adulthood. Overall rates of any nonnicotine drug and
alcohol use disorder were used. We searched journal articles were used as a proxy of overall drug abuse.15
through PubMed at the National Library of Medicine using For each of the 6 studies, Table 2 gives the ORs that
ADHD, pharmacotherapy, stimulants, and SUD as key words. index the protective effect of pharmacotherapy on
This search was supplemented with additional data from scientific drug abuse or dependence and their 95% confidence
presentations at national and international scientific meetings.
We used meta-analyses to evaluate the direction and strength intervals (CIs) and provides the results for alcohol
of the overall association, differential effects on drug or alcohol abuse or dependence. The ORs indicate the increased
use disorders, and the potential effects of covariates. For the odds of not having an SUD for youths who were
analysis, each study provided the 2 ⫻ 2 table classifying subjects treated previously with medication. ORs ⬎1 indicate
by treatment status (pharmacotherapy [stimulants] or not) and the a protective effect; those ⬍1 suggest that stimulant
subsequent development of SUD (present or not) using the odds
ratio (OR). For these studies, the OR estimates the increase in the therapy increases the risk for the SUD outcome. An
odds of not developing SUD among individuals who were previ- individual OR is statistically significant at the 0.05
ously treated pharmacologically compared with individuals with level if its 95% CI does not include 1.0.
ADHD who were not treated pharmacologically. Thus, ORs ⬎1 As Table 2 shows, 7 of the ORs (from 4 studies) are
indicate a protective effect of stimulant therapy on SUD.
Naturalistic studies of psychiatric disorders can lead to the
⬎1.0, suggesting a protective effect of stimulants.
paradoxical result in which greater treatment intensity predicts Five of these ORs are statistically significant. Four of
worse outcome, even when the treatment is known to be effica- the ORs (from 2 studies) are ⬍1, suggesting an ad-
cious from randomized trials.17 This paradox occurs because of 2 verse impact of stimulants, but these are not statisti-
naturalistic correlations: 1) for many psychiatric disorders, people cally significant. The pooled estimate of the OR from
with more severe disorders are usually given more intense treat-
ments (ie, higher doses or longer duration of treatment); and 2) for the meta-analysis was 1.9 and was statistically sig-
many disorders, increasing severity predicts worse outcome. nificant (z ⫽ 2.1; P ⫽ .037; 95% CI for OR: 1.1–3.6).
Thus, we assessed studies for evidence of baseline severity differ- This OR indicates an almost 2-fold reduction in risk
ences between the treated and untreated ADHD groups. for SUD in youths who were treated pharmacologi-
We used a random-effects meta-analysis to analyze the ORs
using the method of Carlin.18 To determine whether the results of cally compared with youths who did not receive
the meta-analysis were unduly influenced by any 1 study, we pharmacotherapy for their ADHD. There was statis-
recomputed the meta-analysis statistic after deleting each study 1 tically significant evidence for heterogeneity of the
at a time. Although the meta-analysis accounts for sample size by ORs in Table 2 (␹2 ⫽ 57.3; df ⫽ 10; P ⬍ .001).
weighting studies according to their sample size, it does not Table 3 presents a sensitivity analysis in which the
determine whether the set of published studies shows evidence
for the biased publication of positive studies. We addressed this combined estimate of the OR was computed after
issue using the method of Egger et al.19 This method is based on omitting 1 data point at a time. This analysis shows
the fact that the precision of the OR (POR) increases with larger whether the significance of the combined estimate
sample sizes. The methods of Egger et al regresses the standard can be attributed to a single datum. Table 3 shows
normal deviate (SND) of the OR (the OR divided by its standard
error) against the POR (the inverse of its standard error). In the that the estimates of the combined OR range from 2.4
absence of bias, Egger et al showed that the regression of SND on to 3.2, suggesting that no 1 study is heavily influenc-
POR should run through the origin (ie, small samples with low ing the combined estimate. Moreover, the confidence

180 STIMULANT TREATMENT OF ADHD AND LATER SUBSTANCE ABUSE

Downloaded from pediatrics.aappublications.org at Bibliotheque de l'Universite Laval on July 14, 2014
 TABLE 1. Description of Studies of ADHD Pharmacotherapy and Later SUD
Study ADHD Ctrl (⫹)Med (⫺)Med Age at Years Alc Use Alc Use Drug Use Disorder Drug Use Disorder Baseline Comments
(N) (N) ADHD ADHD F/U of Disorder Disorder (⫹)Med (N [%]) (⫺)Med (N [%]) Severity
(N) (N) F/U (⫹)Med (⫺)Med Similar?
(N [%]) (N [%])
24, 25 219 0 182 37 22 15 49 (27%) 21 (56%) 31 (17%) 7 (19%) Yes (⫺)Med with more drug use and more likely to
have disorders related to drugs; responders
with less SUD than nonresponders; 84% of
medicated group Tx ⬎ 1 y
15 174 175 93 81 Adult* ⬎10 41 (44%) 26 (32%) Marij: 29 (31%); Marij: 18 (22.5%); No No overall SUD rates reported; no differences
cocaine: 25 (27%); cocaine: 12 (15%); in alcohol, marijuana, stimulants between
stims: 23 (25%) stims: 17 (21%) groups; significant difference in
cocaine/tobacco abuse between groups; trend
to higher rates of SUD with increased
duration of exposure to stimulant
14, 22 190 311 145 45 15.5 4–5 31 (21%) 31 (68%) 23 (16%) 19 (42%) Yes Tx duration not assessed; family history of SUD
similar between (⫺)med and (⫹)med groups;
marijuana most common drug of abuse;
(⫹)med with similar risk for SUD as non-
ADHD
21 138 53 73 15 5 2 (4%) 15 (21%) 4 (8%) 20 (28%) Yes Medication status in adolescence; previous
stimulant response did not predict SUD;
(⫺)med 1 y older than (⫹)meds; marijuana
most common drug of abuse
27 206 0 103 103 21 12 NA NA 10 (11%) 20 (29%) Yes Psychotherapy had no effect on drug abuse;
medication tx reduced marijuana abuse most
robustly; (⫺)med with earlier onset of alcohol
and nicotine use
26 119 98 21 21 15 33 (33%) 7 (33%) 32 (32%) 6 (29%) No Only alcohol and cannabis rates reported; no
differences in sedatives, stimulants, opiates,
and hallucinogens between groups; baseline
difference controlled statistically
Tx indicates treatment; (⫹) meds, youths with ADHD treated pharmacologically; (⫺) meds, youths with ADHD not treated pharmacologically; F/U, follow-up; alc, alcohol; ctrl, control; stims,
stimulants, marij, marijuana; NA, not applicable.
Tx refers to any exposure to agents used in ADHD.
* Age at follow-up not specified.

Downloaded from pediatrics.aappublications.org at Bibliotheque de l'Universite Laval on July 14, 2014

181 REVIEW ARTICLE
TABLE 2. Studies That Examined the Impact of ADHD Phar- The age effect showed that studies that reported
macotherapy on Later Substance Use Disorders follow-up into adolescence14,21,22 showed a greater
Study Protective Effect protective effect (OR: 5.8) than studies that followed
(OR) subjects into adulthood (OR: 1.4) (Barkley RA et al,
OR 95% CI unpublished observations).15,24 –26 The age effect was
still significant after removing the 4 data points from
Meta-analysis of drug studies
Lambert15 0.47 0.22–1.0
2 studies (Barkley RA et al, unpublished observa-
Biederman14 3.9 1.8–8.1 tions)15 with different baseline severity between
Huss26 2.2 0.99–5.1 treatment groups (OR: 5.8 vs 2.3, z ⫽ 2.7, P ⫽ .008).
Loney25 1.1 0.46–2.8 It is possible that, because of publication bias, the
Molina21 4.6 1.5–14.5 group of studies that controlled for baseline severity
Barkley 0.83 0.29–2.3
Meta-analysis of alcohol studies overestimates the protective effect of stimulants.
Lambert15 0.6 0.32–1.1 However, our analyses found no evidence of such
Biederman14 8.1 3.9–17.2 bias. The publication bias statistic was not significant
Loney25 3.6 1.7–7.4 (t ⫽ 0.02, P ⫽ .99). Moreover, its value (0.05) was
Molina21 6.6 1.4–30.2
Barkley 0.98 0.36–2.7
very close to the expected value under the hypothe-
sis that the studies are not biased (0.0).
The OR measures the increase in the odds of not having an SUD
outcome between medicated and unmedicated youths with DISCUSSION
ADHD. ORs ⬎1 indicate a protective effect of pharmacotherapy
on SUD outcome. The larger the OR, the greater the protective The results of this meta-analysis using data from 6
effect of pharmacotherapy on SUD outcome. studies that examined the impact of early medication
treatment for ADHD in childhood on subsequent
TABLE 3. Sensitivity Analysis of Studies SUD outcome in adolescent and young adult years
show that treatment for ADHD significantly de-
Study Omitted-SUD Combined Estimate
of Protective Effect creases the risk for subsequent SUD. These results
(OR) After Omission provide compelling evidence that, contrary to asser-
of Study tions in the popular media, pharmacotherapy with
OR 95% CI stimulants for ADHD does not lead to SUD but
15
instead seems to have protective effects for adverse
Lambert-Alcohol 3.2 1.6–4.9 SUD outcomes in youths with ADHD.
Lambert-Drug15 3.2 1.6–4.9
Biederman-Alcohol14 2.4 1.3–3.5 Examination of individual findings from the 6
Biederman-Drug14 2.9 1.2–4.6 studies used in this meta-analysis reveal that 4 (Bark-
Huss-Drug26 3.1 1.3–4.8 ley RA et al, unpublished observations)14,21,22,24 –26 of
Loney-Alcohol25 2.9 1.2–4.7 the 6 available studies identified striking protective
Loney-Drug25 3.2 1.5–4.8
Molina-Alcohol21 2.6 1.1–4.2
effects of stimulant medications for ADHD on sub-
Molina-Drug21 2.8 1.2–4.5 sequent SUD outcome. Two14,21,22 of these studies
Barkley-Alcohol 3.2 1.5–4.8 that used a comprehensive SUD assessments showed
Barkley-Drug 3.2 1.5–4.9 significantly reduced SUD risk in adolescence. Mo-
The OR measures the increase in the odds of not having an SUD lina and Pelham21 as part of a comprehensive longi-
outcome between medicated and unmedicated youths with tudinal follow-up of children who previously partic-
ADHD. ORs ⬎1 indicate a protective effect of pharmacotherapy ipated in a summer camp for youths with ADHD
on SUD outcome. The larger the OR, the greater the protective
effect of pharmacotherapy on SUD outcome.
showed that stimulant treatment in childhood was
associated with a reduction in risk for both drug and
alcohol use disorders in mid-adolescence. Using data
intervals in Table 3 show that the combined OR from a longitudinal study of boys with ADHD, our
retains statistical significance regardless of which group similarly documented that youths with
study is deleted. ADHD who were treated with stimulants (⬎90% of
We further evaluated factors that potentially influ- cases) and other medicines for ADHD in childhood
ence the SUD outcome. A meta-analysis regression had a 3-fold decreased risk for adverse SUD outcome
found no effect of type of substance (drug versus 4 years later in mid-adolescence compared with
alcohol; z ⫽ 1.1, P ⫽ .3), but there were significant youths with ADHD who were not treated pharma-
effects of study design (z ⫽ 2.9, P ⫽ .004) and age at cologically.14,22 Moreover, the risk for SUD did not
follow-up (z ⫽ ⫺4.7, P ⬍ .001). The study design differ between medicated youths with ADHD and
effect indicated that studies in which groups of non-ADHD controls. In both the Molina and Bieder-
treated and untreated youths with ADHD had sim- man studies, comparable reductions in the risk for
ilar baseline severity found larger ORs than studies stimulant, cocaine, and other substances was found
that had dissimilar baseline severity. As a group, the in the pharmacologically treated compared with un-
data from studies that had similar baseline severity treated youths with ADHD.
showed a statistically significant protective effect Loney et al24,25 found significant reductions in the
(OR: 3.5 [2.2, 5.8]). The 4 data points from the 2 risk for alcohol use disorders in treated youths with
studies that did not have similar baseline severity ADHD as young adults but failed to identify a sim-
between treatment groups (Barkley RA et al, unpub- ilar effect for drug use disorders. The severity of
lished observations)15 both suggest that stimulants ADHD and rates of comorbidity of this sample were
increased the risk for SUD outcomes. similar between medicated and unmedicated sub-

182 STIMULANT TREATMENT OF ADHD AND LATER SUBSTANCE ABUSE

Downloaded from pediatrics.aappublications.org at Bibliotheque de l'Universite Laval on July 14, 2014
jects at baseline assessment in childhood. In addition, lication bias, the group of studies that controlled for
in an earlier report by this group of investigators baseline severity overestimates the protective effect
based in part on the same cohort, a positive response of stimulants, our analyses found no evidence of
to treatment was associated with a lower risk for such bias.
later SUD.27,28 Likewise, in a systematic retrospective Despite15 findings, this meta-analysis rejects the
study, Huss26 reported a clinically and statistically idea that stimulant therapy of ADHD increases the
significant reduction in the risk for drug use disor- risk for neither SUD in general nor specific type of
ders—in particular, marijuana—in young adults who alcohol or drug use disorder. Although some preclin-
were treated previously with methylphenidate. The ical animal models suggest that SUD-related behav-
authors found a linear relationship between risk re- iors (eg, preference for sympathomimetic com-
duction of SUD and duration of exposure to ADHD pounds) are associated with early stimulant
pharmacotherapy. administration,9 –12 the route and dose of administra-
Very recently, Barkley et al (unpublished observa- tion of stimulants used in these models may not be
tions) evaluated the outcome for cigarette and sub- applicable to human data.34,35 For example, in pre-
stance use as well as SUD. In a longitudinal study of clinical studies with rats, methylphenidate was often
147 children with ADHD and 73 controls without administered intraperitoneally at supratherapeutic
ADHD who were followed into adolescence (5 years) human dosing equivalence.9,10 Given that parental
and young adulthood (15 years), the authors re- administration of methylphenidate exceeds oral dos-
ported no differences in SUD between groups in ing,36,37 the dosing in animals is in excess of the
adolescence or young adulthood. Similarly, the au- upper limit of therapeutic dosing recommended in
thors found no significant differences in the use of humans.4,6
any specific drugs with the exception of cocaine use Our finding of a less robust protective effect of
in adulthood, which was mediated entirely by con- ADHD pharmacotherapy in reducing SUD in adult-
duct disorder. A linear effect between risk reduction hood (OR: 1.4) relative to adolescence (OR: 5.8) is
of hallucinogen and cocaine use disorders and dura- noteworthy. Although data on duration of exposure
tion of exposure to stimulant treatment was noted, to pharmacotherapy were not available, it is possible
although the effect was mitigated for cocaine abuse that the adult samples— because of dated recom-
when controlling for conduct disorder. mendations to discontinue treatment in adoles-
In contrast to these findings that reported protec- cence38— had experienced more years without treat-
tive effects of stimulant treatment for SUD outcomes ment than the adolescent samples. If so, then it may
in adolescents and adults, Lambert et al15 found that be that lack of medication coverage in adulthood
stimulant treatment of ADHD was a risk factor for reduced the overall protective effect of earlier stim-
subsequent drug use disorder in young adults. In ulant treatment. Alternatively, enhanced parental
particular, exposure to earlier stimulant treatment monitoring of youths who receive medications may
was linearly related to nicotine and cocaine abuse have a preferential effect in adolescents compared
with notable similar trends to other substance and with young adults. It may also be that adolescents
alcohol abuse. This study, however, had significant have not fully passed through the age of risk to
differences on baseline characteristics between med- develop SUD given that retrospectively derived data
icated and unmedicated youths that may have influ- from adults indicate that the mean onset of SUD is at
enced the outcome. For example, conduct disorder is 19 years in individuals with ADHD.39 Clearly, more
reported in approximately 10% of youths with work to disentangle these issues is warranted.
ADHD.29 Prospective studies in youths with ADHD ADHD has been shown to be a risk factor for
have consistently demonstrated that conduct disor- cigarette smoking in children40,41 and adults.42 Al-
der is a major risk factor for the development of though the current meta-analysis lacked adequate
early-onset SUD.30 –33 In the study by Lambert et al,15 power to evaluate stimulant exposure and cigarette
conduct disorder was overrepresented in the medi- smoking, notable trends emerged in studies that ex-
cated group.23 Hence, it remains unclear whether the amined this issue. One study in adults by Lambert et
higher rates of cocaine and nicotine use in the med- al15 found a linear increase in smoking related to
icated group were a result of the conduct disorder, stimulant exposure in individuals with ADHD. In
stimulant treatment, or other variable related to the contrast, 1 study in adolescents21 and 1 study in
severity of illness at baseline or follow-up. adults24,25 reported that the risk for stable tobacco
Because all of the reviewed studies were natural- use was higher in individuals with ADHD who were
istic and, hence, not randomized at baseline to med- not receiving stimulant treatment. No effect of stim-
ication, attempts to disentangle positive or deleteri- ulant treatment of ADHD on later risk for cigarette
ous effects of treatment from the severity of the smoking was reported in 1 study of adolescents14,22
underlying condition(s) are potentially confound- and 1 in adults (Barkley RA et al, unpublished ob-
ed.17 In all studies in which treated and untreated servations). Hence, reminiscent of findings on SUD,
youths with ADHD had a similar severity at base- the aggregate literature seems to support that stim-
line, examination of individual studies14,21,22,24 –26 re- ulant pharmacotherapy of ADHD is not related to an
veals a reduction in the risk for SUD (Table 2). In increased risk for subsequent cigarette or tobacco
contrast, in the 1 study in which SUD was associated abuse.
with earlier stimulant exposure, severity at baseline The mechanism by which ADHD stimulant phar-
was asymmetrically represented in the treated macotherapy protects against SUD remains unclear.
group.15 Although it is possible that, because of pub- It may be that the reduction of ADHD symptoms,

REVIEW ARTICLE 183

Downloaded from pediatrics.aappublications.org at Bibliotheque de l'Universite Laval on July 14, 2014
demoralization, poor self-esteem, and academic or research in this area must rely on naturalistic de-
occupational failure associated with ADHD43– 45— signs.
factors associated independently with SUD risk46 – 49— Adolescents with ADHD may not have passed
results in reduced SUD. It may also be that by their through the full risk for developing SUD. Hence, our
pharmacological efficacy in diminishing conduct findings may have been biased by the more robust
symptoms,50 the stimulants may have indirectly re- reduction in SUD associated with earlier stimulant
duced the risk for SUD by reducing the risk that treatment that was observed in the adolescent stud-
conduct imparts on SUD. Families who seek medi- ies (OR: 5.8) relative to the adult studies (OR: 1.4).
cation treatment for their youths may be more intact The majority of youths with ADHD on whom data
or of higher socioeconomic status, more invested in were available were male, limiting the generalization
their children’s education success, or more involved of these findings to female individuals with ADHD.
in their parenting. Alternatively, the close monitor- Although the vast majority of youths with ADHD
ing of youths who receive medications may directly were treated with stimulants, a small minority (3%)
influence SUD risk, independent of the actual med- of the medicated group received nonstimulant med-
ication effect, as has been purported elsewhere.40 ications for their ADHD. SUD outcome relied on self-
Additional research to evaluate the connection be- or parental report, and the criteria used to denote
tween symptom expression, self-esteem, and func- abuse or dependence of substances varied between
tional status and later SUD in individuals with studies. Most studies did not elaborate on the rela-
ADHD is necessary. tionship of SUD to the duration of medication expo-
Our finding that ADHD pharmacotherapy pro- sure or adequacy of treatment. Because stimulants
tects against later SUD is of high clinical and public were the most common class of medications used for
health relevance. Clinically, the absence of evidence ADHD, the effects of other classes of medication on
linking SUD with stimulant medication should reas- SUD outcome remain unclear.
sure clinicians and families when discussing the risks Despite these limitations, a meta-analysis of the
and benefits of medication intervention for ADHD. available literature indicates that stimulant therapy
The apparent effect of stimulant treatment of ADHD of ADHD does not increase the risk for subsequent
reducing SUD is among the most robust findings in SUD but seems to have a protective effect. Addi-
pediatric mental health indicating a protective effect tional studies to investigate the long-term SUD out-
come and putative mechanism(s) of reduced SUD
of treatment on lessening SUD risk. The mechanism
risk in youths of both genders with ADHD treated
of treatment in general and medication management
pharmacologically are necessary.
in particular as a protective factor against SUD in
ADHD could serve as a template for other mental
health disorders. From the public health perspective, REFERENCES
given the high prevalence of ADHD in youths and 1. Anderson JC, Williams S, McGee R, Silva PA. DSM-III disorders in
their high risk of developing SUD,44,51–56 the identi- preadolescent children. Prevalence in a large sample from the general
population. Arch Gen Psychiatry. 1987;44:69 –76
fication and treatment of youths with ADHD may 2. Bird HR, Gould MS, Staghezza BM. Patterns of psychiatric comorbidity
affect a large segment of the adolescent and young in a community sample of children aged 9 through 16 years. J Am Acad
adult population culpable to SUD. Child Adolesc Psychiatry. 1993;32:361–368
The results from this meta-analysis need to be 3. Goldman L, Genel M, Bezman R, Slanetz P. Diagnosis and treatment of
attention-deficit/hyperactivity disorder in children and adolescents.
tempered against their limitations. In general, there JAMA. 1998;279:1100 –1107
was a paucity of research data available for review 4. Greenhill L, Osman B. Ritalin: Theory and Practice. New York, NY: Mary
(N ⫽ 6 studies). However, meta-analytic techniques Ann Liebert; 1999
derive statistical power to detect group differences 5. Swanson J, McBurnett K, Christian D, Wigal T. Stimulant medications
and the treatment of children with ADHD. In: Ollendick T, Prinz R, eds.
from the size of the individual studies in the analysis.
Advances in Clinical Psychology. New York, NY: Plenum Press; 1995:
The total number of treated and untreated subjects 265–322
with ADHD across all studies was substantial (N ⫽ 6. Wilens T, Spencer T. The stimulants revisited. Child Adolesc Psychiatric
1034), suggesting that our findings are not a result of Clin North Am. 2000;9:573– 603
low power. Although our finding of no publication 7. Spencer T, Biederman J, Wilens T, et al. Pharmacotherapy of attention
deficit disorder across the life cycle. J Am Acad Child Adolesc Psychiatry.
bias could be attributable to low power, the value of 1996;35:409 – 432
the publication bias statistic was so close to 0 (the no 8. Group MTS. A 14-month randomized clinical trial of treatment strate-
bias value) that it is reasonable to assert that bias gies for attention-deficit/hyperactivity disorder. The MTA Cooperative
may not have been an issue.19 Group. Multimodal Treatment Study of Children with ADHD. Arch Gen
Psychiatry. 1999;56:1073–1086
The naturalistic nature of the studies may have 9. Drug Enforcement Administration. Methylphenidate Review Document.
created confounds (eg, severity of illness, family his- Washington, DC: Office of Diversion Control, Drug and Chemical Eval-
tory of SUD) that may have independently affected uation Section; 1995
outcome. For example, families with a history of 10. Vitiello B. Long-term effects of stimulant medications on the brain:
possible relevance to the treatment of attention deficit hyperactivity
SUD may be less inclined to place their children on disorder. J Child Adolesc Psychopharmacol. 2001;11:25–34
medication, providing a potential bias for SUD in the 11. Kollins SH, MacDonald EK, Cush CR. Assessing the abuse potential of
untreated groups. In contrast, more severe cases of methylphenidate in nonhuman and human subjects: a review. Pharma-
ADHD may be referred for pharmacotherapy, bias- col Biochem Behav. 2001;68:611– 627
12. Schenk S, Davidson E. Stimulant preexposure sensitizes rats and hu-
ing the treated groups to SUD. Ideally, such biases mans to the rewarding effects of cocaine. NIDA Res Monogr. 1998;169:
would be handled with long-term, randomized, con- 56 – 82
trolled trials, but because such trials are not ethical, 13. Robinson TE, Berridge KC. The neural basis of drug craving: an incen-

184 STIMULANT TREATMENT OF ADHD AND LATER SUBSTANCE ABUSE

Downloaded from pediatrics.aappublications.org at Bibliotheque de l'Universite Laval on July 14, 2014
 tive-sensitization theory of addiction. Brain Res Brain Res Rev 1993;18: Arch Gen Psychiatry. 1995;52:456 – 463
247–291 35. Volkow ND, Wang GJ, Fowler JS, et al. Therapeutic doses of oral
14. Biederman J, Wilens T, Mick E, Spencer T, Faraone SV. Pharmacother- methylphenidate significantly increase extracellular dopamine in hu-
apy of attention-deficit/hyperactivity disorder reduces risk for sub- man brain. J Neurosci. 2001;21:RC121
stance use disorder. Pediatrics. 1999;104(2). Available at: 36. Gerasimov M, Franceschi M, Volkow ND, et al. Comparison between
www.pediatrics.org/cgi/content/full/104/2/e20 intraperitoneal and oral methylphenidate administration: a microdialy-
15. Lambert NM, Hartsough CS. Prospective study of tobacco smoking and sis and locomotor activity study. J Pharmacol Exp Ther. 2000;295:51–57
substance dependencies among samples of ADHD and non-ADHD 37. Dedrick RL, Flessner MF. Pharmacokinetic problems in peritoneal drug
participants. J Learn Disabil. 1998;31:533–544 administration: tissue penetration and surface exposure. J Natl Cancer
16. Khantzian EJ. The self-medication hypothesis of substance use Inst. 1997;89:480 – 487
disorders: a reconsideration and recent applications. Harv Rev Psychia- 38. Laufer MW, Denhoff E. Hyperkinetic behavior syndrome in children.
try. 1997;4:231–244 J Pediatr. 1957;50:463– 474
17. Faraone SV, Simpson JC, Brown WA. Mathematical models of complex 39. Wilens TE, Biederman J, Mick E, Faraone SV, Spencer T. Attention
dose-response relationships: implications for experimental design in deficit hyperactivity disorder (ADHD) is associated with early onset
psychopharmacologic research. Stat Med. 1992;11:685–702 substance use disorders. J Nerv Ment Dis. 1997;185:475– 482
18. Carlin JB. Meta-analysis for 2 ⫻ 2 tables: a Bayesian approach. Stat Med. 40. Chilcoat HD, Breslau N. Pathways from ADHD to early drug use. J Am
1992;11:141–158 Acad Child Adolesc Psychiatry. 1999;38:1347–1354
19. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis 41. Milberger S, Biederman J, Faraone S, Chen L, Jones J. ADHD is associ-
detected by a simple, graphical test. BMJ. 1997;315:629 – 634 ated with early initiation of cigarette smoking in children and adoles-
20. Stata Corporation. Stata Reference Manual: Release 6.0. College Station, cents. J Am Acad Child Adolesc Psychiatry. 1997;36:37– 43
TX: Stata Corporation; 1999 42. Pomerleau O, Downey K, Stelson F, Pomerleau C. Cigarette smoking in
21. Molina B, Pelham W, Roth J. Stimulant medication and substance use by adult patients diagnosed with attention deficit hyperactivity disorder. J
adolescents with a childhood history of ADHD. Poster at the Biennial Subst Abuse. 1995;7:373–378
Meeting of the International Society for Research in Child and Adoles- 43. Weiss G. Attention-Deficit Hyperactivity Disorder. Philadelphia, PA: WB
cent Psychopathology; Barcelona, Spain; June 1999 Saunders Company; 1992
22. Wilens T. Attention deficit hyperactivity disorder and substance use 44. Mannuzza S, Klein RG, Bessler A, Malloy P, LaPadula M. Adult out-
disorders—the nature of the relationship, subtypes at risk, and treat- come of hyperactive boys: educational achievement, occupational rank,
ment issues. In: Jensen PS, Cooper J, eds. Attention Deficit Hyperactivity and psychiatric status. Arch Gen Psychiatry. 1993;50:565–576
Disorder: State of the Science; Best Practices. Kingston, NJ: Civic Research 45. Biederman J, Faraone SV, Spencer T, et al. Patterns of psychiatric
Institute; 2002:1–17 comorbidity, cognition, and psychosocial functioning in adults with
23. Lambert N. Stimulant treatment as a risk factor for nicotine and sub- attention deficit hyperactivity disorder. Am J Psychiatry. 1993;150:
stance abuse. In: Jensen PS, Cooper J, eds. Attention Deficit Hyperactivity 1792–1798
Disorder: State of the Science; Best Practices. Kingston, NJ: Civic Research 46. Yamaguchi K, Kandel DB. Patterns of drug use from adolescence to
Institute; 2002:1–24 young adulthood: III. Predictors of progression. Am J Public Health.
24. Paternite CE, Loney J, Salisbury H, Whaley MA. Childhood inattention- 1984;74:673– 681
overactivity, aggression, and stimulant medication history as predictors 47. Kandel DB, Logan JA. Patterns of drug use from adolescence to young
of young adult outcomes. J Child Adolesc Psychopharmacol. 1999;9: adulthood: I. Periods of risk for initiation, continued use, and discon-
169 –184 tinuation. Am J Public Health. 1984;74:660 – 666
25. Loney J, Kramer JR, Salisbury H. Medicated versus unmedicated 48. Crum RM, Bucholz KK, Helzer JE, Anthony JC. The risk of alcohol
ADHD children: adult involvement with legal and illegal drugs. In: abuse and dependence in adulthood: the association with educational
Jensen PS, Cooper J, eds. Attention Deficit Hyperactivity Disorder: State of level. Am J Epidemiol. 1992;135:989 –999
the Science; Best Practices. Kingston, NJ: Civic Research Institute; 2002: 49. Brook JS, Whiteman M, Cohen P, Shapiro J, Balka E. Longitudinally
1–16 predicting late adolescent and young adult drug use: childhood and
26. Huss M. ADHD and substance abuse. In: IX Annual European Congress adolescent precursors. J Am Acad Child Adolesc Psychiatry. 1995;34:
of Psychiatry. Hamburg, Germany; 1999 1230 –1238
27. Loney J, Klahn M, Kosier T, Conboy J. Hyperactive boys and their 50. Klein R, Abikoff H, Klass E, et al. Clinical efficacy of methylphenidate
brothers at 21: predictors of aggressive and antisocial outcomes. In: van in conduct disorder with and without attention deficit hyperactivity
Dused KT, Mednick SA, eds. Prospective Studies of Crime and Delinquency. disorder. Arch Gen Psychiatry. 1997;54:1073–1080
Hingham, MA: Kluwere/Nijhoff; 1982:181–207 51. Weiss G, Hechtman L, Milroy T, Perlman T. Psychiatric status of hy-
28. Kramer J, Loney J, Whaley-Klahn M. The role of prescribed medication peractives as adults: a controlled prospective 15 year followup of 63
in hyperactive youths’ substance use. Poster presented at the American hyperactive children. J Am Acad Child Adolesc Psychiatry. 1985;24:
Psychological Association; Los Angeles, CA; August 1981 211–220
29. Biederman J, Newcorn J, Sprich S. Comorbidity of attention deficit 52. Biederman J, Wilens TE, Mick E, et al. Psychoactive substance use
hyperactivity disorder with conduct, depressive, anxiety, and other disorders in adults with attention deficit hyperactivity disorder
disorders. Am J Psychiatry. 1991;148:564 –577 (ADHD): effects of ADHD and psychiatric comorbidity. Am J Psychiatry.
30. Mannuzza S, Gittelman-Klein R, Konig PH, Giampino TL. Hyperactive 1995;152:1652–1658
boys almost grown up: IV. Criminality and its relationship to psychi- 53. Disney ER, Elkins IJ, McGue M, Iacono WG. Effects of ADHD, conduct
atric status. Arch Gen Psychiatry. 1989;46:1073–1079 disorder, and gender on substance use and abuse in adolescence. Am J
31. Mannuzza S, Klein RG, Bonagura N, et al. Hyperactive boys almost Psychiatry. 1999;156:1515–1521
grown up. V. Replication of psychiatric status. Arch Gen Psychiatry. 54. Faraone SV, Biederman J, Mick E, et al. Family study of girls with
1991;48:77– 83 attention deficit hyperactivity disorder. Am J Psychiatry. 2000;157:
32. Hechtman L, Weiss G. Controlled prospective fifteen year follow-up of 1077–1083
hyperactives as adults: non-medical drug and alcohol use and anti- 55. Levin FR, Kleber HD. Attention-deficit hyperactivity disorder and sub-
social behaviour. Can J Psychiatry. 1986;31:557–567 stance abuse: relationships and implications for treatment. Harv Rev
33. Biederman J, Wilens T, Mick E, et al. Is ADHD a risk for psychoactive Psychiatry. 1995;2:246 –258
substance use disorder? Findings from a four year follow-up study. 56. Schubiner H, Tzelepis A, Milberger S, et al. Prevalence of attention-
J Am Acad Child Adolesc Psychiatry. 1997;36:21–29 deficit/hyperactivity disorder and conduct disorder among substance
34. Volkow ND, Ding Y, Fowler JS, et al. Is methylphenidate like cocaine? abusers. J Clin Psychiatry. 2000;61:244 –251

REVIEW ARTICLE 185

Downloaded from pediatrics.aappublications.org at Bibliotheque de l'Universite Laval on July 14, 2014
Does Stimulant Therapy of Attention-Deficit/Hyperactivity Disorder Beget Later
Substance Abuse? A Meta-analytic Review of the Literature
Timothy E. Wilens, Stephen V. Faraone, Joseph Biederman and Samantha
Gunawardene
Pediatrics 2003;111;179
DOI: 10.1542/peds.111.1.179
Updated Information & including high resolution figures, can be found at:
Services http://pediatrics.aappublications.org/content/111/1/179.full.ht
ml
References This article cites 43 articles, 6 of which can be accessed free
at:
http://pediatrics.aappublications.org/content/111/1/179.full.ht
ml#ref-list-1
Citations This article has been cited by 47 HighWire-hosted articles:
http://pediatrics.aappublications.org/content/111/1/179.full.ht
ml#related-urls
Subspecialty Collections This article, along with others on similar topics, appears in
the following collection(s):
Pharmacology
http://pediatrics.aappublications.org/cgi/collection/pharmacol
ogy_sub
Therapeutics
http://pediatrics.aappublications.org/cgi/collection/therapeutic
s_sub
Toxicology
http://pediatrics.aappublications.org/cgi/collection/toxicology
_sub
Permissions & Licensing Information about reproducing this article in parts (figures,
tables) or in its entirety can be found online at:
http://pediatrics.aappublications.org/site/misc/Permissions.xht
ml
Reprints Information about ordering reprints can be found online:
http://pediatrics.aappublications.org/site/misc/reprints.xhtml

PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly

publication, it has been published continuously since 1948. PEDIATRICS is owned, published,
and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk
Grove Village, Illinois, 60007. Copyright © 2003 by the American Academy of Pediatrics. All
rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.

Downloaded from pediatrics.aappublications.org at Bibliotheque de l'Universite Laval on July 14, 2014