Causal Factors of Increased Smoking in ADHD: A Systematic Review

ABSTRACT

Background: ADHD is a highly prevalent disorder and poses a risk for a variety of mental disorders and functional impairments into adulthood. One of the most striking comorbidities of ADHD is nicotine dependence. Youth diagnosed with ADHD are 2–3 times more likely to smoke than their peers without ADHD, initiate smoking earlier in life and progress more quickly and more frequently to regular use and dependence. Possible explanations for these increased risks are: (a) self-medication of ADHD symptoms with the stimulant nicotine; (b) ADHD symptoms like inattention and hyperactivity/impulsivity predispose for smoking initiation and impede smoking cessation; (c) peer pressure; and/or (d) common genetic or environmental determinants for ADHD and smoking. Objective: Identify the most probable causes of the high prevalence of smoking and nicotine dependence in subjects with ADHD. Methods: A systematic literature review was performed and the causality of the observed relations was ranked using the Bradford Hill criteria. Findings: ADHD medication reduces early smoking initiation and alleviates smoking withdrawal. Nicotine patches, bupropion and (probably) varenicline ameliorate ADHD symptoms. Imitation of and interaction with peers and genetic and environmental determinants may contribute to the comorbidity, but seem to contribute less than self-medication. Conclusion: Smoking is probably best explained by a combination of imitation, peer pressure and typical traits of ADHD. In contrast, the positive relation between ADHD and nicotine dependence is currently best explained by the self-medication hypothesis. This hypothesis has a clear pharmacological rationale and is supported by ample evidence, but awaits confirmation from longitudinal naturalistic studies.

KEYWORDS:

• ADHD
• comorbidity
• tobacco
• smoking
• nicotine dependence

Introduction

Attention-deficit/hyperactivity disorder (ADHD), with a worldwide prevalence in children of 5.3 percent and the male-to-female ratio of at least 3:1 (Wilens et al., 2008b; Ramtekkar, Reiersen, Todorov, & Todd, 2010), is one of the most common child psychiatric disorders. ADHD may resolve in adulthood, but many patients (50–70%) remain to suffer from ADHD during their entire lifespan (Kessler et al., 2006). ADHD is a serious burden to the patient, the family, the health care system and society as a whole. For example, the costs of ADHD related absence from work and healthcare are estimated at 31.6 billion dollars per year in the United States alone (Biederman & Faraone, 2006). In addition, ADHD negatively affects health considering that ADHD patients are more anxious, depressed and more prone to develop a substance use disorder (SUD).

Compared with control subjects without ADHD, children with ADHD are at least 2.5 times more likely to develop a substance use disorder (SUD), overall (Lee et al., 2011). SUD also includes nicotine dependence as a result of tobacco smoking and ADHD has been identified as an important risk factor nicotine dependence (Charach, Yeung, Climans, & Lillie, 2011; Wilens et al., 2011). In a sample of ADHD diagnosed patients, 40% of adults with ADHD were current tobacco smokers compared to 28% in healthy controls (Pomerleau, Downey, Stelson, & Pomerleau, 1995) and a similar difference was observed in adolescents with ADHD (30% vs. 12%) (Molina & Pelham, Jr., 2003). Moreover, subjects with ADHD start smoking earlier in life, smoke more cigarettes per day and the withdrawal symptoms following smoking cessation are more severe, leading to a greater difficulty to quit smoking (see later).

Possible explanations for the much higher tobacco use in people with ADHD are: (a) self-medication with the stimulant nicotine resulting in a reduction of ADHD symptoms; (b) ADHD symptoms that predispose for initiation of smoking and impede smoking cessation; (c) imitation of smoking behavior of parents and friends; and/or (d) common genetic or environmental determinants for ADHD and nicotine dependence.

Given the high public health burden of smoking, factors that affect the development of tobacco use in youth with ADHD should be identified to improve preventive measures and treatment efforts. In this paper we review the currently available literature and used the data described to identify causes for the increased prevalence of nicotine dependence in subjects with ADHD. Knowing that regular smoking mostly starts with smoking in adolescence and young adulthood, we primarily focus on the initiation and maintenance of smoking by adolescents and young adults with childhood ADHD, in particular those with the most prevalent subtype 3 (inattention, impulsivity and hyperactivity combined). Comorbid disorders of ADHD, like Oppositional Defiant Disorder (ODD), major depression or anxiety disorders, as predictors of smoking are excluded from this review.

Methods

“Search strategy and selection criteria

We developed a systematic search strategy with PubMed controlled vocabulary medical subject headings (MeSH) in combination with free text search strings to include untagged PubMed references (e.g., recent publications) published until September 15, 2016, in English. Our search aimed to retrieve studies related to ADHD associated smoking. MeSH terms and cognate free text search terms included the following substance-related search terms: “smoking,” “tobacco use disorder,” “dependence, tobacco,” “dependence, nicotine,” nicotine/administration and dosage,” “attention deficit disorder with hyperactivity” and “methylphenidate.” These MeSH terms were searched in combination with free text terms applicable to ADHD i.e. “novelty seeking,” “inattention,” “hyperactivity” and “impulsivity,” and smoking, i.e., “smoking cessation,” “quitting,” “withdrawal,” and “peer pressure.” Our initial PubMed search retrieved 334 peer-reviewed journal articles since 1960. Recent, not yet indexed references (N = 22; ‘in process’) were retrieved by replacing “MeSH terms” by “[tiab].” Duplicates were avoided by using the extension “NOT medline[sb].” We also used the ancestry approach whereby we examined the references of relevant studies to find other articles that met criteria.

Qualitative evaluation of studies

The abstracts of retrieved journal articles were scanned for inclusion. To be included, the articles must have described some measures of smoking (e.g., starting age, number of cigarettes per day, smoking trajectories), measures of ADHD or ADHD symptoms, and the medication that was taken. In addition, the sample studied had to include young subjects (aged 15–25). Though we focused on young ADHD subjects, some relevant studies in ADHD adults were included, as well. Using this selection, 86 of the 356 articles used for the current review. The studies included show important differences in methodologies across the studies. Most studies have a retrospective or cross-sectional design, are based on self-report of participants, have been performed in small samples.

Ranking causality

To assess causality in the case of increased prevalence of smoking in subjects with ADHD we ranked these explanations using Bradford Hill's criteria. The nine Bradford Hill criteria, a way to examine causality in naturalistic studies, are described in (Table 1). Though more recent models for understanding causality methods have been developed, the Bradford Hill criteria are still the most frequently cited framework for causal inference. It should be noted, however, that these criteria are not simple requirement that must be fulfilled before an association can be judged as causal i.e. the criteria are not intended as checklist for causation but as considerations whether causation might be involved. Other models to assess causality, like Rothman's “component cause model” or “causal pie model,” Lewis's “counterfactual models” and Pearl's “Directed Acyclic Graphs” (DAGs) are variants of Hill's concept or less suitable for the present subject.

Table 1. Description of the nine Bradford Hill criteria to assess causality.

Number	Criterion	Description
1	Strength (effect size)	A strong association is more likely to have a causal component than a modest association.
2	Consistency (reproducibility)	A relationship is observed repeatedly by different persons in different samples.
3	Specificity	A factor influences specifically a particular outcome or population with no other likely explanation.
4	Temporality	The effect occurs after the cause which may include a delay if there is an expected delay between the cause and expected effect.
5	Biological gradient	The outcome increases monotonically with increasing dose of exposure or according to a function predicted by a substantive theory.
6	Plausibility	A plausible mechanism between cause and effect is helpful.
7	Coherence	A causal conclusion should not fundamentally contradict present substantive knowledge, like laboratory findings.
8	Experiment	Causation is more likely if evidence is based on randomised experiments.
9	Analogy	For analogous exposures and outcomes an effect has already been shown.

ADHD and smoking behavior

ADHD and smoking initiation

Mostly driven by curiosity, novelty seeking, experimenting behavior or peer pressure, people usually start smoking in adolescence (Dinn, Aycicegi, & Harris, 2004). Subsequently, they stop smoking or rapidly progress into daily smoking. Impulsivity has emerged as an important risk factor for smoking initiation, where neuroticism seems to be an important driver to continue smoking (Hakulinen et al., 2015; Kvaavik & Rise, 2012). Consistent with these findings, ADHD and the related high level of impulsivity seems to facilitate the initiation of smoking at very young age (12 years) (Milberger et al., 1997; Pomerleau et al., 2003). A 4-yr longitudinal study showed that boys with ADHD started smoking significantly earlier than healthy controls (mean age: 15.5 yrs vs. 17.3 yrs). Between the ages of 15 and 16, 46% of boys with ADHD started smoking compared to 18% of healthy controls, and smoking before the age of 15 was started by 25% of boys with ADHD compared to 9% of healthy controls (Milberger et al., 1997). These findings were confirmed by Pomerlau et al. (2003) who showed that the age of onset of experimentation was 12.8 yrs for ADHD and 14.6 yrs for non-ADHD youngsters, and regular smoking started at 15.6 yrs for ADHD vs. 17.0 yrs for non-ADHD youngsters. Interestingly, the age of smoking initiation was strongly and inversely correlated with the severity of ADHD symptoms (Kollins, McClernon, & Fuemmeler, 2005).

Please note that Kollins et al. (2005), Milberger et al. (1997) and Pomerleau et al. (2003) have assessed smoking initiation and progression retrospectively. Recently, it was confirmed that adolescents with, versus without, ADHD were more likely to start smoking at younger ages and to progress to regular smoking more quickly (Rhodes et al., 2016).

Smoking expectancies, including those about negative consequences, precede smoking initiation and they may vary among the different ADHD subtypes. It was shown that adolescents with inattention were more reluctant to accept the negative consequences of smoking. This contrasts with those with clinical type of hyperactivity/impulsivity who more appreciated the positive consequences, such as ‘expected benefit’ (OR = 5.3), ‘affect control’ (OR = 2.6), and ‘reduced boredom’ (OR = 3.1) (Foster, Racicot, & McGrath, 2012). Of further note is that in this study inattentive female adolescents appreciated well reduced weight as a consequence of smoking. This corroborates with the finding that males were more likely to start smoking (Hamilton et al., 2006). Furthermore, these findings indicate that smoking initiation may differ across ADHD subtypes, ADHD severity, and gender.

ADHD and severity of smoking

Although the smoking prevalence in ADHD is high, smoking severity as an important characteristic of smoking has been understudied. On the other hand, nicotine dependence is relatively independent of the daily consumption of tobacco (Donny, Griffin, Shiffman, & Sayette, 2008). Mixed results have been found in studies on the effects of ADHD on the intensity of smoking. In one study, the severity of ADHD symptoms was found to be correlated with the subjects' smoking severity score according to the Fagerström Test (Fuemmeler, Kollins, & McClernon, 2007) and in another study both inattentive and hyperactive-impulsive ADHD symptoms significantly predicted the number of cigarettes smoked per day (Kollins, McClernon, & Fuemmeler, 2005). Adolescents with ADHD were found to be 4–5 times more likely than controls to escalate to heavy cigarette use after trying it once (Sibley et al., 2014). Two other studies, however, found no relation between the severity of ADHD and cigarette consumption (Pomerleau et al., 2003; McClernon et al., 2008b). The latter observation was recently confirmed (Rhodes et al., 2016). The discrepancy between these results is not well understood, but may be due to differences in sampling (McClernon & Kollins, 2008; August, Realmuto, Crosby, & MacDonald, III, 1995). For example, childhood ADHD seems to be a risk factor for heavy smoking only in subjects with a high IQ (Molina & Pelham, 2001) and subjects with a relatively low IQ were overrepresented in the community-based studies. As outlined in the next section, ADHD symptoms obstruct smoking cessation extending their smoking career.

ADHD and smoking cessation

ADHD patients also show greater difficulty to stop smoking and to remain abstinent. Only 29% of 77 adult ADHD patients successfully quitted smoking compared to 49% of the smokers in the general population, though this difference was only apparent in male subjects (Pomerleau, Downey, Stelson, & Pomerleau, 1995; Rohde, Kahler, Lewinsohn, & Brown, 2004). Moreover, male subjects who had experienced ADHD symptoms only during childhood (and not during adulthood) also showed lower quit rates compared to non-ADHD controls (Humfleet et al., 2005).

A recent study among adolescents and young adults with and without childhood ADHD showed that smokers with ADHD reported more intense withdrawal and craving during periods of abstinence than non-ADHD smokers (Rhodes et al., 2016), but the two groups did not significantly differ in number of quit attempts. Typical withdrawal symptoms smokers experience are a variety of affective (e.g., irritability) cognitive (e.g., inattention and difficulty concentrating) symptoms, and craving (a strong desire to smoke). Adult smokers with ADHD recall more severe self-report and laboratory-based withdrawal symptoms than adult smokers without ADHD (McClernon et al., 2008b; Pomerleau et al., 2003). These findings suggest that smokers with ADHD will more often fail in quitting from smoking, because they experience more withdrawal symptoms, in particular inattention and irritability, as compared with non-ADHD smokers.

Summary ADHD and smoking behavior

ADHD is associated with various features of smoking behavior, i.e., ADHD adolescents start experimenting with smoking earlier in life, are more likely to become regular smokers and nicotine dependent, and also show lower quitting rates. In the following sections the relation between ADHD and smoking will be further explored to explain why tobacco smokers are over-represented among individuals with ADHD.

Causality between ADHD and smoking

In explaining the causality between ADHD and smoking, the initiation stage should distinguish from both continued smoking as well as smoking cessation. As will be outlined below, smoking initiation in people with ADHD is mainly driven by typical traits of ADHD and imitation, whereas continued smoking, nicotine dependence and low cessation rates in people with ADHD are mainly driven by shared genetic factors, typical ADHD traits, and self-medication.

Self-medication

Both nicotine dependence and ADHD are disorders that involve the striatal dopaminergic system. ADHD is hypothesized to be the result of an aberrant striatal dopaminergic system that results in disrupted dopaminergic transmission in cortico-striatal circuits and low dopaminergic tone primarily in the prefrontal cortex known to be critical for the regulation of behavior, attention and affect (Brennan & Arnsten, 2008; Posner & Petersen, 1990; Dougherty et al., 1999). The dopaminergic modulation of these pathways follows an inverted U-model, i.e., optimal stimulation of the prefrontal cortex at intermediate dopamine levels, and maladaptive inactivation of the prefrontal cortex at low and high dopamine levels (Brennan & Arnsten, 2008). Similarly, both higher (Madras, Miller, & Fischman, 2005; Dresel et al., 2000; Cheon et al., 2003) and lower (Jucaite et al., 2005; van Dyck et al., 2002) densities of the dopamine transporter (DAT) have been found in the ventral striatum of ADHD patients (Spencer et al., 2005). Furthermore, it has been suggested that patients with ADHD may have a lower tonic dopamine tone (Dougherty et al., 1999) which in turn leads to an over-reactive phasic dopamine response (Grace, 2000; Grace, 2001).

By increasing the availability of dopamine, nicotine may normalize the low dopamine tone in the striatum of ADHD patients resulting in less pronounced ADHD symptoms (Mihailescu & Drucker-Colin, 2000). On the other hand, the over-reactive dopamine system in ADHD patients could result in a much larger rewarding dopamine boost resulting in a higher risk of nicotine dependence (McClernon & Kollins, 2008). Therefore, due to the higher responsiveness of the dopamine system in ADHD subjects, smoking may be more rewarding, explaining the relatively high attractiveness of smoking in ADHD subjects. Indeed, the reinforcing effects of cigarette smoking in adult regular smokers with ADHD were significantly larger than in non-ADHD smokers (Kollins et al., 2013).

Nicotine use may be reduced in ADHD by reducing its characteristic symptoms. A first argument for nicotine use as a form of self-medication might be that—like the typical ADHD psychostimulant medication methylphenidate—nicotine acts as an indirect dopamine agonist (De & Dani, 2011; Volkow et al., 2009) and improves attention and arousal and reduces impulsivity in subjects with ADHD. However, only four human studies have examined (in small samples) whether smoking ADHD patients show fewer or less severe ADHD symptoms compared to their non-smoking peers, and they showed relatively small beneficial effects of nicotine on attention, impulsivity and ADHD symptoms (Conners et al., 1996; Levin et al., 1996; McClernon et al., 2008a; Shytle, Silver, Wilkinson, & Sanberg, 2002). However, a prospective study performed in smoking and non-smoking twins showed that smoking twin developed more attention problems in childhood and adolescence than non-smoking co-twins and these attention problems lasted into adulthood (Treur et al., 2015).

The self-medication hypothesis of smoking in ADHD was supported by studies in which treatment of ADHD patients with nicotine patches alleviated attention problems during cognitive tasks (Conners et al., 1996; Levin et al., 1996; Shytle, Silver, Wilkinson, & Sanberg, 2002), reduced self-reported ADHD symptoms and negative mood (Gehricke et al., 2009), and improved performance both in stop signal reaction time and delayed aversion tasks (Potter & Newhouse, 2004). Similarly, in the Conner's Continuous Performance Test, adult non-smokers with low attentiveness treated with a 7 mg nicotine patch made significantly fewer errors of commission and showed improved stimulus detectability and fewer perseverations as compared to placebo (Poltavski & Petros, 2006). ADHD patients may also smoke tobacco to alleviate unpleasant ADHD-related symptoms, like sleeping problems, restlessness and irritability. Indeed, subjects with ADHD considered tobacco a therapeutic aid, as they reported positive effects on “inner tension” and cognitive function in addition to the social benefits of smoking (Liebrenz et al., 2014). However, improvement of ADHD-related sleep and mood problems related to nicotine use was not consistently found (Wilens et al., 2007).

The self-medication hypothesis is also endorsed by the results obtained with the nicotine analogues bupropion and varenicline (Potter, Schaubhut, & Shipman, 2014; Ferris & Beaman, 1983; Mihalak, Carroll, & Luetje, 2006; Rhodes et al., 2012) which have been successfully applied as smoking cessation aid (Hurt et al., 1997; Aubin et al., 2008). Bupropion showed beneficial effect during treatment of ADHD. In a small sample of 38 adults with ADHD, bupropion improved ADHD symptoms in 76% of the ADHD cases (placebo 37%), and “much improvement” was reported in 52% of bupropion treated ADHD subjects (placebo 11%) (Wilens et al., 2001). In three other studies, using small samples (N = 13–30) of substance-abusing adolescents with ADHD and comorbid psychopathology, bupropion (200–400 mg/day) also significantly reduced ADHD symptoms at 2 weeks to 6 months post-treatment (Solhkhah et al., 2005; Riggs, Leon, Mikulich, & Pottle, 1998; Wilens et al., 2003). In an 8-week RCT, where 162 adult ADHD patients were treated with up to 450 mg/day of bupropion, the proportion of ADHD responders (defined as at least 30% reduction in the investigator-rated ADHD Rating Scale score at week 8) was significantly higher in the bupropion compared to the placebo group (53% vs. 31%) (Wilens et al., 2005). A recent study confirmed the effect of bupropion to reduce ADHD symptoms though the effect was smaller (mean difference d = −0.32) as compared to the effect of methylphenidate and atomoxetine (d = – 0.68 and d = −0.75, respectively) (Stuhec, Munda, Svab, & Locatelli, 2015). Though one study showed that varenicline clearly improved overall attention (Rhodes et al., 2012), no studies have yet been published on the use of this drug in the treatment of ADHD. Finally, two other nicotine analogues, i.e., ABT-894 (Bain et al., 2013) and ispronicline (Potter et al., 2014) have been reported to significantly improve primary and secondary outcomes in adults with ADHD.

As dopaminergic medications can act as a substitute for nicotine, it is conceivable that ADHD treatment with stimulants would prevent smoking and nicotine dependence in ADHD subjects, endorsing the self-medication hypothesis. On the other hand, the sensitization hypothesis predicts that CNS stimulant treatment is a risk factor for subsequent use of stimulants like tobacco (and cocaine). These two contrasting options may explain the controversial results obtained in studies evaluating the effect of ADHD-medication on smoking which either support or refute the self-medication hypothesis as outlined below.

Stimulant treatment of ADHD patients was reported to be associated with (1) a delayed onset of regular smoking in adolescence and to protect them from becoming a daily smoker in adulthood (Lambert, 2002), (2) less smoking in treated adult ADHD subjects (Wilens et al., 2008a; Whalen et al., 2003; Hammerness et al., 2013) and (3) lower tobacco use during the 28-years follow-up, which went up when the medication was stopped (Lambert, 2005). A recent meta-analysis of 14 studies showed that treatment of ADHD children with methylphenidate was associated with a two-fold lower tobacco use in adulthood (Schoenfelder, Faraone, & Kollins, 2014). On the other hand, a small meta-analysis of six studies (Humphreys, Eng, & Lee, 2013) showed that overall ADHD medication did not affect the risk for nicotine dependence (OR: 1.34; 95% CI 0.90–1.99) with 5 out of 6 studies reporting no significant association and one study reporting a significant increase of the risk for nicotine dependence (Lambert & Hartsough, 1998). Such an increased risk for smoking was supported by an experimental studies in smoking controls and smoking ADHD patients which showed that methylphenidate dose-dependently increased total number of cigarettes smoked and number of puffs) (Rush et al., 2005; Vansickel et al., 2011; Vansickel, Stoops, Glaser, & Rush, 2007). According to Schoenfelder et al. (2014) the variation in outcome of epidemiological studies may be explained by the larger effect sizes that are obtained in studies which used clinical samples, included relatively more females and measured smoking in adolescence rather than adulthood.

Others see self-medication more^{Table 2} broadly and have incorporated mood regulation and improvement of sleeping hygiene in the hypothesis (Glass & Flory, 2010). For instance, in a longitudinal study of children with ADHD, Wilens et al. (Wilens et al., 2007) asked participants whether smoking helped them to improve sleep or mood. Of all ADHD subjects, 3% and 34% reported that smoking improved sleep and mood, respectively. Gehricke et al. proposed that subjects smoke to regulate their affect (Gehricke et al., 2007). Using a small sample of 22 subjects with ADHD and 22 controls who self-reported smoking motivation and sensory effects of smoking, van Voorhees et al. (2012) showed that smokers with ADHD reported smoking as more calming, enjoyable and satisfying than smoking controls.

In an attempt to estimate the causal role of self-medication in the increased prevalence of smoking in subjects with ADHD, we applied the nine criteria of Hill (1965), depicted in (Table 1), to summarize the findings presented so far (cf. Table 2). From this summary, it seems that self-medication is very likely to play a causal role since there is experimental evidence, the finding are quite consistent, the relation is mechanistically plausible, and experimental nicotine administration studies showed not only temporal relations but the findings were also consistent with epidemiological findings. On the other hand, effect sizes and specificity were moderate at best leaving some room for alternative explanations (see later). It should be noted, however, that the self-medication hypothesis awaits confirmation from longitudinal naturalistic studies which show a true and prolonged amelioration of ADHD-symptoms following smoking initiation. Moreover, the negative results obtained by Humphreys et al. (2013), implying that treatment of ADHD with stimulants and the related reduction in ADHD symptoms do not diminish smoking and nicotine dependence, argue against a role of ADHD symptoms as risk factor of smoking in ADHD.

Table 2. Validation of two explanatory models for the causal relation between ADHD and smoking using Hill's criteria (Hill, 1965).

	Self-medication		ADHD-traits/symptoms
Hill criterion	Score	Arguments	Score	Arguments
Strength	+	Moderate effect sizes	+	Moderate effect sizes
Consistency	++	Consistent findings reported	++	Consistent findings reported
Specificity	±	Not very specific for smoking and nicotine dependence; also applicable for other substances	±	Not very specific for smoking and nicotine Dependence; also applicable for other substances. For specificity, smoking initiation should be distinguished from smoking continuation and dependence
Temporality	++	Experimental studies show temporal relationships and predictors have statistical significance	+	Predictors have statistical significance
Biological gradient	—	Not clear.	—	Not clear
Plausibility	++	Plausible mechanism	++	Plausible mechanism
Coherence	++	Experimental data consistent with epidemiological evidence	±	Consistency of laboratory findings with epidemiological evidence is impeded by lack of experimental studies
Experiment	++	Experimental data partly confirm the outcome of epidemiological results	±	Low
Analogy	+	Other attributive factors cannot be excluded	±	Traits collectively play a role, but their relative importance is unclear

++: High; + moderate; ±: doubtful or low; −: not clear.

ADHD symptoms predispose for smoking initiation and impede smoking cessation

Self-medication may be important in smoking continuation and reduced smoking cessation, but it does not seem to explain frequent, early smoking initiation in youngsters with ADHD. ADHD in childhood and adolescence is associated with a variety of self-management deficits and impairments in cognitive and social functioning in later life (Barkley, Fischer, Edelbrock, & Smallish, 1990; Biederman et al., 1996). Negative coping (become helpless and angry), an example of poor self-management, is a strong risk factor of early smoking initiation (Siqueira, Diab, Bodian, & Rolnitzky, 2000), whereas positive coping (trying to solve the problem and to see the problem differently) has been associated with a lower vulnerability to nicotine dependence (Wills & Hirky, 1996). As such, the higher risk for nicotine dependence in ADHD is partly explained by poor cognitive and behavioral coping skills (Molina, Marshal, Pelham, Jr., & Wirth, 2005).

Other typical traits of ADHD resulting in behavioral inhibition, like inattention, novelty seeking, hyperactivity and impulsivity have also been identified as predictors of substance use disorder (SUD), including smoking and nicotine dependence (Burke et al., 2007; Fuemmeler, Kollins, & McClernon, 2007; Pingault et al., 2013; Rodriguez & Bohlin, 2005). However, it is not clear which of the mentioned ADHD traits only drive smoking initiation or are also involved in smoking progression.

In addition, nicotine withdrawal symptoms are similar to ADHD symptoms (Shiffman & Jarvik, 1976). As such, ADHD patients who stop smoking will experience an increase in ADHD severity making it more difficult to quit smoking and creating a higher risk of relapse, resulting in a higher current prevalence of smoking in people with ADHD. Indeed, higher self-reported withdrawal symptom severity was reported by former (female) ADHD smokers compared to non-ADHD former smokers (McClernon et al., 2008b). Though ADHD medication did not diminish smoking and nicotine dependence (cf. 4.1), nicotine patches alleviated ADHD symptoms by 8% and negative moods by 9% in abstinent smokers and non-smokers with ADHD (Gehricke et al., 2009). Similarly, ADHD-medication reduced withdrawal during overnight abstinence and improved difficulty concentrating during no smoking events (Gehricke et al., 2011).

Again we applied the Hill criteria to summarize the potential causal role of pre-existing ADHD symptoms in the increased initiation of smoking and the higher prevalence of nicotine dependence in ADHD (Hill, 1965). In this summary (cf. Table 1), moderate effect sizes go together with a plausible mechanism and consistent findings, although the number of studies is rather limited. Moreover, the fact that smoking withdrawal leads to ADHD-like symptoms may result in less successful quit attempts in ADHD.

Imitation and peer pressure

Social interaction and peer pressure are known determinants ^{Table 2}of smoking behavior. ADHD adolescents have more often friends that smoke, live more often with a smoker or have more often a parent that smokes (Tercyak & Audrain-McGovern, 2003; Kalyva, 2007) suggesting that ADHD individuals are more vulnerable for social smoking. In addition, smoking behavior in ADHD may be facilitated by poor parent-child communication and low parental support (Molina, Marshal, Pelham, Jr., & Wirth, 2005; Waa et al., 2011; Burke, Loeber, & Lahey, 2001). It is not known whether parents of children with ADHD fail to set less rules about smoking. This is important, because parental rule setting is known to postpone or prevent smoking initiation (Waa et al., 2011). Based on Hill's criteria it is concluded that there is not enough data to establish the causal role of imitation and peer pressure in the relation between ADHD and smoking.

Common genetic or environmental determinants

The dysregulation of dopaminergic and nicotinic pathways are likely to arise, at least in part, from genetic and environmental variations. With rates of 60%–80% for ADHD and 56% for smoking, both disorders show considerable genetic influence (Li, Cheng, Ma, & Swan, 2003; Faraone et al., 2005; Li et al., 2014). Moreover, ADHD and smoking share a number of candidate genes, suggesting shared aberrations in some neurobiological pathways (Faraone & Mick, 2010; Gizer, Ficks, & Waldman, 2009; Tanner, Chenoweth, & Tyndale, 2015; Ware & Munafo, 2015). Promising targets of genetic overlap between ADHD and smoking are the DRD2, the DRD4 and DAT genes, exhibiting associations with both phenotypes (Ware & Munafo, 2015; Faraone et al., 2005; McClernon et al., 2008a; Bidwell et al., 2012) and genes regulating nicotinic receptor functioning, like CHRNA3 (Polina et al., 2014), CHRNA4 (Lee et al., 2013; Kamens et al., 2013; Faraone & Mick, 2010; Alemany et al., 2015). For review, we refer to (McClernon & Kollins, 2008; Thapar, Cooper, Eyre, & Langley, 2013).

Of the environmental determinants, maternal smoking during pregnancy (SDP) is an important risk factor in the etiology of ADHD. Even after adjustment for measured confounders, various epidemiological studies have reported robust associations between SDP and ADHD in offspring (Banerjee, Middleton, & Faraone, 2007; Rodriguez & Bohlin, 2005) though others have expressed great skepticism (Thapar & Rutter, 2009), because of bias by unmeasured familial factors. Indeed, in a Swedish nationwide population-based cohort (n = 813,030) study the predicted estimates of ADHD in offspring (hazard ratio: 1.9 to 2.5) gradually attenuated toward the null after adjustment for a variety of confounders, including unmeasured confounders shared within the extended family (i.e., cousin comparison) and the nuclear family (i.e., sibling comparison) (hazard ratio: 0.8 to 0.9) (Skoglund et al., 2014).

Also low socioeconomic status (SES) is associated with both ADHD and tobacco smoking. Of 42 studies that were included in a systematic review, 35 found a significant univariate association between low SES and ADHD; children in families of low SES are on average 1.85–2.21 times more likely to have ADHD than their peers in high SES families (Russell, Ford, Williams, & Russell, 2016). Similarly, in the lower SES social networks smoking is more prevalent (Hitchman et al., 2014; Hiscock et al., 2012).

Again using Hill's criteria, we conclude that genetic overlap between ADHD and smoking and social disadvantage may both contribute to the higher risk for smoking in ADHD, but it should be noted that the effect sizes are relatively small.

Treatment options

Understanding the link between ADHD and smoking is important to develop successful interventions. Parents seem to be crucial because ADHD is often diagnosed at an early age (<7 years), well before children start experimenting with smoking. For instance, children of long-term smokers were at a higher risk of initiating smoking than children of parents who were not long-term smokers (Melchior et al., 2010). Treatments aimed at improvement of the patient's positive coping abilities could be promising in preventing them from smoking. In addition, the direct social environment of ADHD subjects is important since parents and friends who smoke will stimulate ADHD subjects to start smoking and thus prevention should be aimed at the direct social environment of ADHD subjects. Given the potential for ADHD symptoms to interfere with smoking cessation success, it is advocated that—at least in severe ADHD—the first priority is to stabilize ADHD using long-acting psychostimulants. Subsequently, cognitive behavioral therapy or motivational interviewing (NICE, 2008), either or not in combination with smoking cessation aids, like NRT, bupropion or varenicline can be used for further treatment (Karpinski, Timpe, & Lubsch, 2010; Stead, Koilpillai, Fanshawe, & Lancaster, 2016). Indeed, bupropion and perhaps also varenicline proved to be promising agents to reduce smoking in individuals with ADHD, though a post-hoc analysis showed a divergent treatment response in smokers with inattention and those with hyperactivity/inattention symptoms (Covey et al., 2008). Secondly, withdrawal symptoms in smokers with severe ADHD were successfully treated with osmotic-release oral system (OROS) methylphenidate + NRT (Gray & Upadhyaya, 2009). A post-hoc subgroup analysis further showed that this treatment also increased abstinence in the smokers with severe ADHD but was counterproductive among smokers with lower severity of ADHD (Nunes et al., 2013).

Discussion

Tobacco use is the leading cause of preventable deaths in Western societies (Danaei et al., 2009). Given the high morbidity and mortality associated with tobacco smoking, the much higher rate of smoking and nicotine dependence among subjects with ADHD is a significant public health issue.

Methylphenidate and d-amphetamine reduce ADHD symptoms by increasing synaptic dopamine concentrations in the striatum, including the nucleus accumbens, the brain's reward center via presynaptic transporters. Considering that nicotine i.e. smoking also increases dopamine levels in a similar way, self-medication is an attractive hypothesis to explain the relation between ADHD and smoking. Further support for the self-medication hypothesis comes from studies showing that cognitive and attentional deficits, common to ADHD are improved by smoking (Levin et al., 1998; Lawrence, Ross, & Stein, 2002), nicotine patches (Potter et al., 2014; Rhodes et al., 2012; Wilens et al., 2001) as well as by bupropion and varenicline (Potter & Newhouse, 2004; Potter & Newhouse, 2008; Poltavski & Petros, 2006). Two studies, however, failed to confirm the principle of self-medication (Wilens et al., 2007; Dinn, Aycicegi, & Harris, 2004), presumably because both studies used a definition of self-medication based on ADHD-related symptoms rather than ADHD-symptoms and because both studies relied on self-report of the symptoms. Finally, one meta-analysis indicated that methylphenidate treatment of childhood ADHD reduces smoking as an adult (Schoenfelder, Faraone, & Kollins, 2014). However, this conclusion was not confirmed by another meta-analysis, showing that overall stimulant medication in children with ADHD does not affect the risk for nicotine dependence in adulthood (Humphreys, Eng, & Lee, 2013). In summary: though ample evidence is available in favor of the self-medication theory, others have presented data that are not consistent with this hypothesis. Therefore, this hypothesis still awaits confirmation from longitudinal naturalistic studies. It should be noted, however, that ADHD severity, presence of comorbid conduct disorder and stimulant treatment duration were not taken into account in these meta-analyses.

The hypothesis that typical traits of ADHD predispose for early smoking initiation and nicotine dependence (Foster, Racicot, & McGrath, 2012; Hakulinen et al., 2015; Kvaavik & Rise, 2012) and make it difficult to quit smoking may partly explain the high prevalence of tobacco use among ADHD subjects. One would therefore expect that treatment of ADHD with methylphenidate would subside or dampen these predisposing effects. However, laboratory studies show that methylphenidate increases smoking in ADHD subjects and that early stimulant treatment of ADHD patients does not prevent nicotine dependence.

The imitation hypothesis is supported by the finding that ADHD adolescents reported to have significantly more smoking friends and family members (Kalyva, 2007; Wilens et al., 2008b), leading to social pressure to start smoking. However, there are very few prospective studies that examine this hypothesis more specifically.

Finally, ADHD and nicotine dependence seem to share genetic factors resulting in considerable overlap in genetic predisposition constituting a shared risk to develop both ADHD and nicotine dependence. However, specific SNPs at four loci have been found to be associated with a range of psychiatric disorders, such as bipolar disorder, autism spectrum disorder, ADHD, major depressive disorder, and schizophrenia (CDGPGC, 2013). It can therefore not be excluded that the overlap that was found between nicotine dependence and ADHD genes reflects in fact an overlap between nicotine dependence and psychiatric illness in a broader sense, including ADHD comorbid psychiatric illness (pleiotropy).

In conclusion, the self-medication hypothesis at least partly explains why ADHD subjects have an increased risk to smoke, because they seem to find relief for their ADHD symptoms via indirect dopaminergic stimulation However, this hypothesis does not explain the earlier initiation of smoking in young ADHD subjects. Here, social-environmental factors (e.g. smoking peers and relatives), shared genetic risk factors and the presence of specific ADHD traits may be more prominent, although to which extend remains unclear. The use of smoking cessation aids retaining dopamine stimulating potency such as bupropion and perhaps also varenicline seems promising to reduce smoking in individuals with ADHD, whereas extended release methylphenidate in combination with nicotine patches may be helpful in particular to increase abstinence in smokers with severe ADHD.

This study has some limitations. First, the results described are not always comparable because of important differences in methodologies across the studies which impede the generalizability of the findings. For instance, some studies exclusively relied on self-report of ADHD symptoms, which can be an underestimation of severity. Second, most studies have been performed in small samples. Third, most studies were retrospective or cross-sectional thus limiting causal interpretation, and only very few studies took a prospective approach. Fourth, studies on nicotine dependence and tobacco consumption in relation to ADHD have largely focused on adult smokers, were based on adult-diagnosed ADHD and reported mixed findings e.g. some found higher tobacco consumption and dependence while others found no differences. Fifth, diagnostic misclassification could produce spurious evidence of overlap between risk factors related to ADHD versus those of simple inattention or conduct disorder. Finally, knowing that some ADHD symptoms resemble those of nicotine withdrawal, one has to be prudent in drawing conclusions about a causal relationship between ADHD and smoking. Future research should focus on longitudinal studies that allow us to establish true causality with regard to the following questions: “does smoking result in impaired attention, is pre-existing inattention self-medicated by smoking, and does smoking cessation increase the severity of ADHD symptoms in the long-term?”

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Abbreviations

ADHD	=	Attention-deficit/hyperactivity disorder
IQ	=	intelligence quotient
OR	=	odds ratio
CI	=	confidence interval
DAT	=	dopamine transporter
NRT	=	nicotine replacement therapy
OROS	=	osmotic-release oral system
SES	=	socioeconomic status