Introduction

Attention deficit hyperactivity disorder is a common condition affecting in the region of 7% of children and adolescents and 5% of adults (Willcutt 2012). It is characterised by the core psychopathologies of attention difficulties, impulsivity and hyperactivity (American Psychiatric Association 2013). In children, the condition may be diagnosed as a primarily hyperactive/impulsive subtype, a primarily inattentive subtype, or a combined subtype (Willcut 2012). In adults, the condition is described mostly as a combined type (Bell 2011). ADHD is associated with poorer psychosocial outcomes (Loe and Feldman 2007) and shows a high frequency of comorbidity with other psychiatric and psychological disorders (Matthews et al. 2014). The aetiology of ADHD is complex and incompletely understood, although there appears to be a high level of heritability of the condition, indicating a significant genetic component (Matthews et al. 2014).

A common finding in ADHD in both adults and children is that there are significant sleep disturbances associated with the condition (Hvolby 2015). For example, chronic insomnia has been reported in 27% of adults with ADHD (Schredl et al. 2007). Intriguingly, many of the core cognitive and behavioural symptoms of ADHD can also be contributed to by poor sleep quality and/or quantity (Babkoff et al. 1991), and as such sleep disturbance may be important in the aetiology of ADHD. In terms of the regulation of sleep, the classic two-process model (Borbély et al. 2016) proposes that sleep/wake behaviour is driven by overlapping homeostatic processes (accumulating drive to sleep with increasing cumulative hours of wakefulness) and circadian drive to wakefulness (normally strong during the day, declining moving towards the later evening). The circadian component of this process is now known to be determined by an endogenous circadian timekeeping mechanism which manifests itself in a host of rhythms in behavioural, physiological and biochemical rhythmic outputs recurring with periods of approximately 24 h (Buttgereit et al. 2015). Such outputs include diurnal rhythms in cortisol resulting from strong circadian control of the hypothalamic–pituitary–adrenal axis (Spiga et al. 2014) and nocturnal secretion of the pineal hormone melatonin gated through the suprachiasmatic nucleus (Erren and Reiter 2015).

The clock itself is encoded at a genetic level by a number of clock genes, whose expression leads to the formation of a series of interlocking feedback/feedforward transcriptional loops which result in the near 24-h expression of clock gene products, but also the rhythmic expression of a large set of clock-controlled genes across the genome (Zhang et al. 2014). From an anatomical perspective, the master circadian clock resides in the suprachiasmatic nuclei of the hypothalamus, with other clock distributed widely throughout the brain and periphery (Mohawk et al. 2012). Given this widespread distribution of components of the circadian timing system, it is somewhat unsurprising that the operation of the internal clock imposes a temporal governance on a wide range of physiological and molecular processes (Buijs et al. 2016) and as such are also implicated in a wide range of diseases (Smolensky et al. 2016). Amongst these are involvements in common psychiatric conditions such as major depression, bipolar disorder and schizophrenia (Coogan 2013; Tam et al. 2015; Melo et al. 2016). Given this, it is being increasingly recognised that circadian function in these conditions may be an appropriate target for psychological, behavioural and/or pharmacological therapies (Dallaspezia and Benedetti 2015). Such therapies may take advantage of fundamental knowledge of the neurobiology of the circadian system. For example, the main photoreceptive system for the circadian clock is not the visual system, but rather a system utilising the photopigment melanopsin in a subset of light-sensitive retinal ganglion cells (Lucas et al. 2014). It is long known that light is the most profound influence on circadian function and circadian clock are normally entrained to environmental light/dark cycles (Mohawk et al. 2012). As some psychiatric conditions, such as bipolar disorder and seasonal affective disorder, have been described to be associated with alterations in circadian phase reflective of altered entrainment of the clock to the environment, the use of timed light therapy has been shown to be of promise in resynchronising rhythms and improving clinical ratings (Dallaspezia and Benedetti 2015; Lewy 2007).

Given the prevalence of sleep problems in ADHD, recent attention has begun to focus on the role that circadian dysfunction may play in such sleep problems, and more generally how circadian dysfunction might contribute to the aetiology and/or symptomatology of ADHD. We have recently reviewed the mechanisms that could link ADHD and circadian rhythm abnormalities (Coogan et al. 2016). However, to date there has been no systematic review of the relevant literature linking circadian rhythms and ADHD, and it is to this gap that the present study addresses itself.

Methods

Search strategy

A structured search of PubMed, the Cochrane Library and Clinical Trials.gov was conducted in July and September 2016 using the terms “circadian” OR “chronotype” OR “diurnal preference” OR “biological rhythm” OR “sleep” OR “melatonin” OR “light” AND “attention deficit hyperactivity disorder” OR “ADHD”. Publication date and language were not restricted. Articles that focussed solely on sleep parameters, with no measures of sleep timing or sleep/wake cycle descriptions, or circadian or diurnal rhythms description, were excluded, as the focus of this review is on circadian function in ADHD. Only primary studies were included, so reviews and meta-analyses were excluded. Studies of neurodevelopmental disorders that did not explicitly delineate ADHD populations were excluded. Animal studies were also excluded. Genetic studies were included once they included descriptions of links between ADHD and core circadian clock genes. Interventional studies assessing impact on circadian function were included. Studies assessing the impact of melatonin, whether used as a somnolent or as a chronobiotic, were also included.

Measures of circadian rhythmicity

There are a number of different measures of circadian function that are appropriate for the assessment of circadian function in clinical populations. An area that is relatively accessible for examination is inter-individual differences in sleep/wake behaviour (chronotype) and diurnal preference for sleep/wake behaviour. These parameters can be examined using psychometric instruments for the assessment of diurnal preference, such as the morning–eveningness questionnaire (MEQ) and the composite scale of morningness (CSM). These instruments may be described as being indicators of psychological preference for the timing of sleep/wake behaviour, and as such may not be fully indicative of underlying circadian phase (Levandovski et al. 2013; Adan et al. 2012). The Munich Chronotype Questionnaire (MCTQ) is another instrument which records actual timing of sleep/wake behaviour, and as such may be a truer indicator for underlying endogenous circadian phase (Roenneberg et al. 2003). The timing of mid-sleep on “free” days is taken as the key measure from the MCTQ. The MCTQ also differentiates between “free” and “work” days, and as such allows for calculation of the discrepancy between the timing of sleep on both work days, a construct referred to as social jetlag (Wittmann et al. 2006). The MEQ, CSM and MCTQ are well validated and have been deployed in numerous studies interrogating the epidemiology of circadian rhythmicity in humans. The MCTQ and MEQ measures are correlated with melatonin secretion profiles (Kantermann et al. 2015), and there is a moderate to strong correlation between the MEQ and MCTQ (Zavada et al. 2005).

Objective measures of sleep/wake behaviour and circadian rhythms in locomotor activity include the use actigraphy. This approach utilises accelerometers, usually wrist-worn, to gather data on activity patterns and from these derive measures of circadian function as well as measures of sleep timing and quality (Ancoli-Israel et al. 2003). Measures of circadian rhythms often involve nonparametric analysis of circadian rhythms, yielding measures such as relative amplitude of the rhythm (higher values indicating clearer demarcation between consolidated bouts of rest and activity), L5 and time of L5 onset (where L5 is the activity in the least active 5 h), M10 and M10 onset (amount of activity in the most active 10 h), and measures of rhythm precision (interdaily stability or ‘IS’) and fragmentation (intradaily variability or ‘IV’; van Someren et al. 1997). Sleep parameters of interest for this review that can be derived from actigraphy include time of bedtime, time of sleep onset and time of wakening.

As many endocrine systems operate under profound circadian control, measures of daily variability salivary or serum levels of hormones such as cortisol and melatonin (or melatonin’s metabolite 6-sulfatoxymelatonin in urine) are important indicators of circadian function. For melatonin, an important measure is the dim light melatonin onset (DLMO), which is taken as an important indicator of underlying circadian phase (Pandi-Perumal et al. 2006). On a molecular level, the circadian clock is known to be driven by a series of interlinked transcriptional feedback and feedforward loops involving a panel of circadian clock genes. Expression patterns of such genes can be followed in peripheral tissues such as oral epithelia or peripheral blood mononuclear cells to allow insight into the molecular clockwork. Polymorphisms in clock genes have also been implicated in various neuropsychiatric disorders (Landgraf et al. 2014), and as such genetic association studies between ADHD (and ADHD symptoms) and clock gene polymorphisms are included in this review.

Chronotherapeutic studies

There are a number of pharmacological and non-pharmacological interventions that may be utilised to alter circadian function, and such approaches are broadly termed chronotherapy. Such treatments may include melatonin [in either its application at higher doses as a somnolent, or in smaller, time-controlled doses as a chronobiotic to alter circadian phase (Lewy 2007)], drugs that specifically target the circadian system [such as the antidepressant agomelatine (San and Arranz 2008)], or environmental or behavioural manipulations, such a light therapy, which alters circadian phase and/or amplitude (Terman 2007). We included in our review any study seeking to manipulate the circadian system as a primary intervention.

Results and discussion

A total of 1719 articles were initially identified through database searches. Figure 1 shows the process for article selection. A final total of sixty-two articles were included for review, with the oldest article included being from 2000. Overall, studies involving 33,212 participants were included in this review. Of these, a total of 4462 ADHD patients were studied. Of the studies included for review, 25 studies were in adult populations whilst 37 studied children or adolescents.

Fig. 1
figure 1

Graphic representation of the study selection process

Full size image

Chronotype and diurnal preference in ADHD

The strongest and most consistent finding in the analysed literature is for association of later chronotype/evening preference with ADHD and ADHD symptoms. In total, 9 papers examining chronotype/diurnal preference were reviewed (Table 1). Assessment of chronotype/diurnal preference is made through the morningness/eveningness questionnaire (MEQ; 6 studies), the Munich Chronotype Questionnaire (MCTQ; 4 studies), the composite scale of morningness (CSM; 3 studies) and derivative questionnaires based on the MEQ and/or MCTQ (3 studies), with some studies utilising overlapping measures. In studies of ADHD adult patients, there is a remarkable agreement of findings of later chronotype/increased eveningness in ADHD groups compared to controls. Rybak et al. (2007) report increased eveningness associated with increased ADHD symptoms and poorer sustained attention on a continuous performance task in 29 adults with ADHD. A similar report of increased eveningness is reported in 13 adults with ADHD by Baird et al. (2012), whilst Bijlenga et al. (2013b) report that ADHD was associated with later mid-sleep on free days measured by the MCTQ in 202 adults with ADHD. Vogel et al. (2015) report that ADHD was associated with later chronotype and with more prevalent delayed sleep phase disorder and unstable timing of meals in 202 adults with ADHD. Kooij and Bijlenga (2014) report in a sample of 543 adults that more extreme evening typology associates with ADHD. Similarly, Gruber et al. (2012) report greater eveningness in 26 children with ADHD, and this later diurnal preference was associated with later sleep onset.

Table 1 Summary of studies included for review which examined chronotype/diurnal preference in ADHD
Full size table

A number of studies have examined the link between diurnal preference/chronotype and ADHD symptoms too using ADHD screening instruments applied to non-clinical samples. In a cross-sectional study of 205 adults, Caci et al. (2009) report that eveningness rated on the CSM was associated with inattentive symptoms, more than impulsiveness and hyperactivity. Voinescu et al. (2012) also used the CSM in a cross-sectional study of 301 adults and report that evening preference is associated with likely-ADHD status and inattentive symptoms. McGowan et al. (2016) report, using the MCTQ in a sample of 396 adults, that social jetlag, rather than chronotype, is a predictor of ADHD symptoms and impulsivity possibly indicating that circadian desynchrony is an important factor mediating the putative relationship between chronotype/circadian typology and ADHD.

Sleep-onset insomnia and delayed sleep phase disorder in ADHD

Delayed sleep phase disorder (DSPD) is believed to be driven by circadian factors (Micic et al. 2016). Likewise, sleep-onset insomnia (SOI) may have a significant circadian component in its aetiology (Lack and Wright 2007). Sixteen studies were included in this review examining either DSPD or SOI in ADHD (Table 2). Van der Heijden et al. (2005) report that SOI in ADHD was associated with a phase delay of the dim light melatonin onset (DLMO) compared to subjects with ADHD but without SOI. In this study, 87 of 110 children with ADHD had SOI, indicating that it is prevalent in ADHD. Van der Heijden et al. (2006) report that sleep hygiene measures do not account for the differences in the ADHD + SOI group and the ADHD alone group. Hvolby et al. (2008) report that 45 children with ADHD had significantly longer sleep latencies, as assessed by actigraphy, compared to both control and children with other psychiatric diagnoses. Hoebert et al. (2009) examined the use and discontinuation of melatonin for SOI in ADHD in 94 children and report that melatonin discontinuation was associated with relapse of SOI in 92% of cases. Similar to the finding of van der Heijden et al. (2005) in children, Van Veen et al. (2010) report that of 40 adults with ADHD studied, 31 reported SOI that was associated with attenuated circadian rhythm amplitude of activity and delayed DLMO. Fargason et al. (2013) on the other hand report that in a group of 105 adults with ADHD, circadian delay was not associated with ADHD, although later dosing time with stimulants was associated with later sleep times.

Table 2 Summary of studies included for review which examined sleep-onset insomnia and delayed sleep phase disorder in ADHD
Full size table

The role of ADHD medication in sleep-onset problems in ADHD is indicated by the findings of Boonstra et al. (2007) who report that in a double-blind crossover trial in 31 adults with ADHD, methylphenidate treatment was associated with later bed times and greater sleep latency. Likewise, Ironside et al. (2010) report that methylphenidate treatment of 16 children with ADHD was associated with a phase delay of the sleep/wake cycle and that methylphenidate treatment was associated with an increase in motor activity during the sleep latency period. Interestingly, Gruber et al. (2000) report that in an actigraphic study of 32 children with ADHD that sleep onset times are unstable, perhaps indicating a weakened circadian component of sleep/wake regulation.

Regarding DSPD in ADHD, there are case reports of the condition co-occurring in ADHD (Khurshid and Khan 2009), including a striking example of seemingly complete inversion of the sleep/wake cycle in one patient (Fargason et al. 2013). Chiang et al. (2010) report that in a cross-sectional study of 325 children with ADHD, the combined subtype of ADHD was associated with circadian rhythm sleep disorders. Sivertsen et al. (2015) report that in a retrospective cohort of 9338 adolescents there was a moderate association of DSPD with inattentive and hyperactive symptoms, whilst Hysing et al. (2016) report in a cohort of 9816 adolescents that likely-ADHD status is associated with DSPD and later bedtimes. Bron et al. (2016) report in another retrospective cohort of adults at risk for anxiety or depression there is an increased risk of late chronotype and delayed sleep phase in those scoring highly on the ADHD screener. An increased risk for delayed sleep was also noted in a cohort study by Wynchank et al. (2016).

Functional assessment of circadian rhythms in ADHD

Twenty-two studies included in this review examined functional circadian rhythms in ADHD (Table 3). These rhythms include actigraphy recorded activity/rest cycles, diurnal sampling of the hormones melatonin and cortisol, assessment of DLMO as a circadian phase marker, patterns of clock gene expression in accessible tissues, cognitive assessments and physiological measures such as heart rate variability. In a case–control study of 42 children with ADHD, Dane et al. (2000) report that actigraphy indicates greater afternoon, but not morning, activity in ADHD and that there is no subtype difference between inattentive and combined types. As noted earlier, Van der Heijden et al. (2005, 2006) indicate that ADHD is associated with a significant delay of DLMO. Van Veen et al. (2010) report that in 40 adults with ADHD that there was a delay of the diurnal cycle of activity/wake and a delayed DLMO. Baird et al. (2012) report that in 13 adults with ADHD there was loss of rhythmic expression of the clock genes PERIOD 2 (PER2) and BMAL1 in oral mucosa, and the period of the actigraphic rest/activity rhythm differed significantly from 24 h. Gamble et al. report (2013) in 24 adults with ADHD that the rest/activity cycle was delayed and that this delay was correlated positively to the severity of the ADHD symptoms. A case–control study of 12 adults with ADHD reported that ADHD was associated with phase delays in DLMO, sleep onset and core and skin temperature to similar extents (Bijlenga et al. 2013a). Faedda et al. (2016) report that actigraphy can reveal differences in circadian parameters diurnal skew, L5 and relative circadian amplitude that can be used to discriminate ADHD from bipolar disorder and control in children, indicating that changes in rest/activity patterns associated in ADHD may be different from those associated with other psychiatric populations.

Table 3 Summary of studies included for review which involved functional assessment of circadian rhythms in ADHD
Full size table

In studies which examined the daily profile in melatonin and/or cortisol, there are more mixed findings. Daily rhythms in salivary cortisol in 28 adults with ADHD were examined in a case–control study, and no difference was reported in ADHD compared to control (although there was not a complete diurnal sampling procedure utilised in this study, with cortisol assessment taking place at awakening, 30 min after awakening, between 17:00 and 18:00 and at 23:00; Hirvikoski et al. 2009). Personen et al. (2011) also do not report ADHD-symptom-related changes in salivary cortisol profiles assessed across the day in a cross-sectional study of 322 children with ADHD. Imeraj et al. (2012a, b) showed no differences in morning cortisol levels in 33 children with ADHD, but did note a steeper afternoon decline of cortisol in children with ADHD and oppositional defiant disorder. Baird et al. (2012) report a significant delay of the cortisol diurnal rhythm relative to wakening time in a sample of 13 adults with ADHD with 6 samples taken across 24 h. Isaksson et al. (2015) report that in a case–control study of 81 children with ADHD, wakening levels of cortisol were lower in the ADHD than the control group, and the area under the curve was also lower in the ADHD group for the daily profile compared to controls.

With regards to melatonin, as noted above a number of studies have reported delayed DLMO in adults and children with ADHD (van der Heijden et al. 2005, 2006; Van Veen et al. 2010; Bijlenga et al. 2013a). Paclt et al. (2011) report that ADHD score correlation with melatonin levels was dependent on symptoms of conduct disorder. In a case–control study of 34 children with ADHD, (Nováková et al. 2011) note that ADHD is not associated with alterations of the 24 h melatonin profile in the overall sample, although older children with ADHD showed narrower melatonin secretion profiles than controls. In their study, Baird et al. (2012) noted a blunted melatonin rhythm in adults with ADHD, but also noted more nocturnal light exposure in the ADHD group which may have contributed to acute suppression of melatonin secretion. A higher level of serum melatonin in children with hyperactive-impulsive ADHD with comorbid conduct disorder compared to inattentive ADHD was noted by (Cubero-Millán et al. 2014) in a study of 148 children. These authors also report that methylphenidate treatment was associated with a lower level of the melatonin metabolite 6-sulfatoxymelatonin. Molina-Carballo et al. (2013) report that in 226 children with ADHD treatment with methylphenidate ablates morning/evening differences in melatonin levels, possibly through raising morning levels of melatonin. An increase in urinary 6-sulfatoxymelatonin was noted in another study of 27 children and adolescents with ADHD (Büber et al. 2016). A potential involvement of pineal function and anatomy in mediating ADHD-related changes in melatonin is suggested by the study of Bumb et al. (2016) which reports that ADHD was associated with lower pineal volume in a sample of 74 adults. Further, in a case report Cunniffe et al. (2010) report that sleep/wake alterations in an adolescent with a pineal tumour had been masked clinically by the presence of an ADHD diagnosis, thus highlighting the overlap of sleep/wake symptoms resulting from ADHD or pineal dysfunction.

There are three studies included in this review which have examined rhythms in physiological and cognitive parameters. Imeraj et al. (2011) report that children with ADHD (N = 30) showed elevated heart rate, most pronounced in the afternoon and the night. Buchhorn et al. (2012) also examined diurnal changes in heart rate, and heart rate variability, in a case–control study of 31 children with ADHD, and report that ADHD was associated with a marked increase in heart rate in the afternoon and less pronounced diurnal variation in heart rate variability. Both of these studies indicate that altered circadian function may manifest in alterations of autonomic function relevant for ADHD (Buijs et al. 2016; Rash et al. 2012). Impaired sustained attention is a key cognitive dysfunction in ADHD, one that can be assessed using objective tests of attention and vigilance. Hunt and colleagues (2012) report that for one such cohort, the test of variables of attention, applied to 55 young adults with ADHD in the morning and the late afternoon, indicates that there is no time-of-day effect on performance on this test.

Intervention studies

Nine studies included in this review have examined the impact of various interventions that target the circadian system in ADHD (Table 4). The largest group of these studies examine the impact of melatonin administration on ADHD symptoms and other associated outcomes. Tjon Pian Gi et al. (2003) examined the efficacy of melatonin administration prior to bedtime in 24 children with ADHD and insomnia in a prospective open quasi-experimental study and report that such treatment lead to a rapid onset decrease in sleep latency, which was significantly sustained at three-month follow-up. Weiss et al. (2006) report that in a randomised, double-blind crossover trial in 27 children and adolescents with ADHD, melatonin reduced insomnia; the largest reduction in insomnia resulted from co-administration of melatonin and a sleep hygiene intervention. Van der Heijden et al. (2007) report results from a randomised double-blind placebo-controlled trial of evening melatonin for 105 children with ADHD, which indicate that melatonin treatment resulted in advanced sleep-onset time and DLMO, indicating an effect of melatonin on underlying delayed circadian phase. Mohammadi et al. (2012) conducted a randomised control trial (RCT) in 60 children with ADHD which examined the co-administration of methylphenidate with melatonin over a period of 8 weeks and report that melatonin treatment was associated with shorter sleep latency, but not with ADHD ratings. Similar decrease in sleep latency was described in a study by Mostafavi et al. (2012) using melatonin co-administered with methylphenidate, which also noted an increase in height and weight following melatonin treatment, with no effect reported on food intake or appetite. A long-term follow-up study by Hoebert et al. (2009) in 94 children with ADHD and sleep-onset insomnia indicated that melatonin discontinuation was associated with relapse of sleep-onset insomnia in 92% of cases. A register study in Sweden indicates that 50% of children prescribed melatonin were also using ADHD medications and that melatonin usage was higher in boys than girls (Furster and Hallerbäck 2015).

Table 4 Summary of studies included for review which examined chronotherapeutic interventions in ADHD
Full size table

There is little comprehensive evidence for other chronotherapeutic approaches in ADHD. Gruber et al. (2007) report a single case for the use of morning bright light therapy in a child with ADHD with resulting improvements in ADHD ratings and amelioration of delayed phase sleep disorder. Niederhofer (2012) report results of a small randomised placebo-controlled trial of the use of the chronobiotic antidepressant agomelatine in 10 adolescents with ADHD which indicate that agomelatine treatment was associated with decreased ADHD symptoms. Rybak et al (2006) report that morning light therapy in adults with ADHD results in alleviation of core ADHD symptoms, improvement of mood symptoms  and advances of circadian preference.

Genetic and environmental association studies

Five studies are included in this review that examined association of polymorphisms in circadian clock genes with ADHD (Table 5). Kissling et al. (2008) reported that the T3111C single-nucleotide polymorphism (SNP) in CLOCK is associated with ADHD symptoms in a forensic, adult male population of 143, with the T allele conferring risk. Xu et al. (2010) report similar results from adult ADHD, with the T allele of the T3111C SNP in CLOCK associated with ADHD. Jeong et al. (2014) report that the T3111C SNP in CLOCK is also associated with ADHD symptoms in males. Cao et al. (2012) also find that the T3111C SNP in CLOCK is associated with ADHD in children, although contrary to previous finings the authors found in their study that the C allele confers risk. Brookes et al. (2006) in a genome-wide study provide speculative evidence that polymorphisms in PER 2 may be associated with ADHD in children, but the association was not significant at the study-wide or genome-wide level, and so this evidence should be treated with caution. Chaste et al. (2011) report the presence of coding mutations in the melatonin biosynthesis pathway found in a sample of 101 adults with ADHD, but these mutations were not significantly enriched compared to the control population.

Table 5 Summary of studies included for review which examined circadian-related genetic and environmental associations with ADHD
Full size table

Three studies examining the association of circadian-relevant environmental factors with ADHD are included in this review (Table 5). Arns et al. (2013) report that in an analysis of records from ten countries that solar intensity is associated with ADHD prevalence and postulate that this may be in part due to enhanced early morning bright light which might advance and correct otherwise delayed circadian phase, which may be associated with ADHD symptoms. Arns et al. (2015) also suggest that an association noted by Huber et al. (2015) between altitude and ADHD prevalence may be mediated through solar intensity. Finally, in a naturalistic study in northern latitudes, Langevin and Ramdé (2012) find that the shortening of photoperiod in the winter is associated with more nocturnal agitation in ten children with ADHD studied with actigraphy.

Conclusions

This review of the literature highlights a number of striking findings with regards to circadian function and ADHD. First is the remarkable consensus in findings for delayed circadian phase and later chronotype/evening preference associated with ADHD. This level of consistency is in contrast with the mixed picture reported for phase alterations associated with different psychiatric conditions, where a mixture of phase delays, phase advances and no changes are reported (e.g. bipolar disorder; Melo et al. 2016). This circadian phase delay may of course be linked to the observed link between ADHD and sleep-onset insomnia and delayed sleep phase disorder. There appears to be little evidence for other circadian changes in ADHD, such as alterations in rhythm amplitude. This again is in contrast with circadian changes associated with other conditions such as dementia, where dampened circadian amplitude may be the major change that occurs (Coogan et al. 2013a, b). It is not clear why circadian rhythms show such delays in ADHD, although the observation by Kooij and Bijlenga (2014) of higher than expected levels of photophobia associated with ADHD may indicate that there is a change in the melanopsin photoreceptive system, as this is associated with both circadian entrainment and other non-visual responses to light including photophobia (La Morgia et al. 2011). Future studies might address themselves to the role of melanopsin function and circadian entrainment explicitly in ADHD.

The presence of consistent findings of delayed circadian rhythms in ADHD raise the possibility of chronotherapeutic approaches to “correct” underlying phase alterations with a view to increasing sleep quality/quantity and ultimately leading to a decrease in ADHD symptoms. Such an approach has been shown in other conditions to be of considerable promise (e.g. Lewy et al. 2006). However, the current evidence base for such approaches in ADHD is disappointingly weak, with only one small trial of agomelatine and one case report for light therapy assessed in the course of this review. Thus, there would appear to be considerable scope to develop this avenue of investigation in future work. There is stronger evidence of the use of melatonin in ADHD, with studies included in this review indicating beneficial effects on sleep but not on ADHD symptoms. It should be noted that most of the studies examined utilised melatonin as a somnolent, and not explicitly as a chronobiotic whose main effect would be to advance circadian phase. Such use of melatonin requires lower doses, but timing of dosing is vital and should be tailored to the underlying circadian phase misalignment (Lewy 2007). Further, the use of the chronobiotic antidepressant agomelatine, which is a melatoninergic agonist, should be further explored in ADHD. There is clearly a need here for well-designed, sufficiently powered RCTs for such interventions in ADHD, and it should be noted that recent results of chronotherapy from RCTs are providing promising results (e.g. light therapy in major depression; Lam et al. 2016).

From an aetiological perspective, as there is a lack of prospective and longitudinal studies of circadian function in ADHD, inferences about causality of circadian dysfunction and ADHD cannot be made. Further, whilst there are some indications that polymorphisms in clock genes may be associated with ADHD, evidence is not strong although there are four separate studies implicating the CLOCK gene. Another potentially important issue on which there appears to be too little information to make firm conclusions is the role ADHD medication plays in observed circadian changes in ADHD. Animal studies indicate that stimulant and non-stimulant medications can impact on circadian processes (e.g. Antle et al. 2012; Baird et al. 2013; O’Keeffe et al. 2012). There are also indications that ADHD medications can impact on sleep (e.g. Cortese et al. 2013). However, there are little explicit analyses of ADHD medication on circadian function, and future studies might address medicated and non-medicated ADHD populations for circadian function in either case–control or crossover studies.

Overall, results from this systematic review highlights a striking commonality of findings of circadian phase delays associated with ADHD, but also highlights the weakness of the literature surrounding chronotherapeutic interventions in ADHD and delineation of the aetiological significance of circadian dysfunction in ADHD.