Comorbid attention-deficit hyperactivity disorder and bipolar disorder

An initial psychiatric clinical assessment usually seeks to determine ‘what is the diagnosis’, so allowing a condition-specific management plan to be derived. In practice, the majority of patients present with multiple conditions, suggesting that a psychiatric interview should more seek to establish what are the diagnoses. When multiple diagnoses are made, the clinician’s task is generally to judge whether the conditions are independent or interdependent (and why), clarify primacy and then determine whether management should seek to address the diagnosed conditions sequentially or concurrently.

A not uncommon clinical scenario is that of patients having comorbid attention-deficit hyperactivity disorder (ADHD) (or attention-deficit disorder [ADD]) and a bipolar disorder (BD). Almost invariably such patients are referred for clarification about the existence of only one of those conditions, further supporting the need for a comprehensive interview to identify the presence (or the likelihood) of any other conditions. As ADHD and BD appear strikingly distinct conditions, their coterminous presence invites several questions, including whether their cooccurrence exceeds chance, and if both are present, should a sequencing or concurrent management model be implemented.

Many papers have reviewed and quantified their ‘comorbid’ status. While comorbid in this clinical situation strictly means cooccurring or coterminous conditions, for some authors and researchers, the term is used to capture non-concurrent lifetime episodes of differing disorders. Thus, while this review focuses on pursuing issues in relation to their coterminous presence, data references may include studies employing the comorbid term to the non-coterminous presence of both ADHD and BD over time. The relevant literature generally includes ADD (or attention deficit disorder) in the term ADHD but may refer to it as ‘inattentive’ subtype as against the ‘hyperactive-impulsive’ subtype of ADHD.

Turning to prevalence data, Schiweck et al. (2021) reported lifetime prevalence rates of 6.5% for ADHD and 1–2% for BD and then quantified the theoretical chance of their cooccurrence at 0.38%. Klassen et al. (2010) overviewed studies suggesting that children and teenagers with ADHD are at a 10-fold risk for BD (and thus comorbid status), and rates of ADHD comorbidity in BD adult patients ranged from 9% to 35%. Nierenberg et al. (2005) overviewed studies quantifying that in children and adolescents with a BD, rates of comorbid ADHD ranged from 60% to 90%, suggesting that such comorbidity is distinctly higher in children and adolescents as compared to adults. Such variable levels of associations may reflect differences between childhood and adult onset ADHD but any such age-related causal factor remains unestablished.

Schiweck et al. (2021) have recently published the first meta-analysis of some 71 studies evaluating the degree of comorbidity between the two conditions. Inclusion criteria required subjects with a BD to be 15 years or older and those with ADHD to have received that diagnosis before they were 12 years of age. They quantified that 8% of adults with ADHD had a BD and that 17% of those with BD had ADHD, while comorbidity rates were similar for those with a bipolar I as against those with a bipolar II condition. We judged that their quantified comorbidity rates were higher than might be expected by chance (8.9% vs 1–2% for BD, and 18% vs 6.5% for ADHD). A subset of 14 studies providing data on age of onset were consistent in quantifying an earlier age of onset of BD in those with ADHD (the mean being 4 years earlier). Such studies quantify that comorbidity rates are distinctly higher than might be expected by chance if the two are independent conditions.

A few studies have examined for clinical correlates. In their analysis of STEP-BD data, Nierenberg et al. (2005) reported that, of their 919 bipolar patients (minimum age = 15; mean age = 41 years), 87 (9.5%) met criteria for a lifetime diagnosis of ADHD (but with that diagnosis being current for only 62% of the 87). Those receiving a comorbid diagnosis were more likely to be male, had a younger age of BD onset (with comorbid status far less prevalent in those who had their BD occur at 18 years or older), were more likely to have a bipolar I BD sub-type and less likely to have a bipolar II BD sub-type, had a more severe mood disorder course, were more likely to have attempted suicide, more likely to evidence violent behaviour (and to have it commence at an earlier age) and more likely to meet criteria for social phobia, posttraumatic stress disorder (PTSD), alcohol/drug abuse or dependence and agoraphobia. In another clinical study involving 124 adults with a diagnosis of BD (Perroud et al., 2014), those with a comorbid ADHD diagnosis had a younger age at BD onset (and of initial depressive and hypomanic/manic states), had experienced more depressive episodes, were less likely to have a BD I sub-type, more likely to have an anxiety disorder, substance disorder, alcohol dependence, and borderline personality traits, and to be recipients of emotional abuse during childhood. In a third study of a cohort with a BD (Tamam et al., 2008), 16.3% had a comorbid diagnosis of ADHD. Those in the comorbid group had an earlier age of onset of their BD, had higher rates of axis I disorders, anxiety disorders and alcohol abuse or dependency, had an earlier age of onset, and had more depressive and hypomanic episodes. In a further study, State et al. (2004) reported that their BD subjects having ADHD in childhood were less likely to respond to mood stabilising medication.

In a recent paper, Nunez et al. (2022) compared three sub-sets of those with a BD, comprising (1) those with ADHD diagnosed in childhood, (2) those with adult-onset ADHD and (3) those without ADHD. Those in the first group had a greater number of comorbid conditions (e.g. anxiety, substance abuse) and showed a poorer response to mood stabilisers lithium and lamotrigine, the last findings being consistent with findings by State et al. (2004).

Sachs et al. (2000) offered four explanations for the high comorbidity: (1) chance, (2) overlapping criteria for the two disorders (especially of emotional dysregulation, distractibility and increased activity), (3) a common diathesis (e.g. shared genetic vulnerability) and (4) early onset ADH symptoms being a prepubertal expression of a BD not fully expressed. ‘Chance’ would account for less than 1% of instances (as noted earlier from several papers), and with comorbid prevalence data noted here rejecting that possibility as having much valency. A variant of ‘chance’ is worthy of noting. Many papers considering this nuanced topic make reference to the Berkson paradox or fallacy (Berkson, 1946) which argues that seemingly correlated events are not so in fact and, in this context, explainable as a consequence of studies being undertaken on formal patients rather than those in the general community. Its salience here is that those with such comorbid states would be more likely to seek services as a consequence of greater severity and/or impairment, so leading to clinicians observing higher comorbidity rates than observed in community subjects. While it may so contribute, the data presented earlier (reporting distinctly high comorbidity rates across differing samples) suggest that any such contribution would be slight.

The second (artefactual) explanation is a seemingly strong candidate explanation and has been evaluated. Marangoni et al. (2015) provided a lengthy set of symptom criteria (26 for BD and 26 for ADHD) for differentiating one condition from the other, and with few having any potential for diagnostic overlap. In addition, they showed that when overlapping symptoms were removed from diagnostic criteria sets for the two conditions, there was a reduction of nearly 50% in comorbidity rates but that the association between the two conditions remained significantly elevated. In a more formal analysis, Milberger et al. (1995) effectively removed overlapping symptoms for the two conditions and quantified that symptom overlap was not contributing distinctly to comorbid diagnoses.

Turning to a shared genetic contribution, Schiweck et al. (2021) reported data from family-based studies indicating that the risk of either one condition was doubled if the other condition was present. They noted that reported GWAS studies had returned quite varying genetic correlation estimates (ranging from 0.05 to 0.71) but that the largest data sample returned a genetic correlation of only 0.21. They interpreted the data to suggest that rare variants, main effects of environmental factors or gene/environmental interactions were more likely than shared hereditability. They also considered several environmental candidate factors (e.g. premature birth, low birth rate, maternal stress during pregnancy, maternal substance abuse and childhood maltreatment) but without there being substantive supportive studies for any such candidates, thus weighting the shared genetic effect explanation.

If genetic factors contribute, then the two comorbid conditions might be independent or interdependent outcome conditions. In an intriguing study, Klassen et al. (2010) analysed family study data which suggested that the two conditions are transmitted together and not independently (and thus form a distinct subtype of either BD or ADHD). Other data analyses indicated that their study members with comorbid ADHD/BD developed manic symptoms up to 5 years earlier than those with BD alone, and also experienced a more severe course (with the latter either being intrinsic or a reflection of that group being less compliant with medication).

In terms of the fourth explanation, ADHD might also be a prodromal state preceding formal onset of a BD (a developmental heterotypy model akin to the caterpillar becoming a butterfly). Such an explanation does not, however, account for the distinctly high comorbidity rates in adults, whereby both conditions are evident.

A fifth explanation (and one which challenges the co-morbid association) is that, as attentional and executive problems are common in adults with a BD (even when euthymic) as quantified in several studies (e.g. Martinez-Arán et al., 2004), there is a risk of falsely assigning an ADHD diagnosis for those with a bipolar condition.

Other explanations include overlapping risk factors, and while some candidates have been proposed, Youngstrom et al. (2010) have suggested that all remain nonspecific or nonreplicated. In addition, while one condition (e.g. ADHD) may increase the risk of the other (e.g. BD), but Schiweck et al. (2021) made reference to data from the Avon Longitudinal study that failed to support ADHD as a bipolar prodrome. It is also possible that early exposure to stimulant medication for the ADHD condition might increase the likelihood of developing a BD.

In summary, comorbid rates of BD and ADHD appear particularly high in children but still elevated above chance in adult samples. The elevated rates could reflect diagnostic error, with certain symptoms (e.g. emotional dysregulation, distractibility, attentional problems in those with BD) leading to false-positive diagnoses of one or both conditions, and while support for this possibility has been demonstrated, it does not appear to fully account for the high co-morbidity rate. If not due to such artefactual influences, it is possible that BD might manifest as a forme fruste sub-threshold condition akin to ADHD prepubertally before developing a more characteristic prototype in adulthood, and so suggest co-morbid status. The converse heterotypy explanation that ADHD might be a prodromal precursor state for BD is unlikely, when several studies (see Youngstrom et al., 2010) of ADHD cohorts have shown that few members develop a BD.

The suggestion that a comorbid sui generis sub-type may exist (and have an earlier age of onset than observed in those with pure BD, as well as higher rates of other co-morbid conditions and a poorer response to mood stabilisers) is intriguing. In terms of sub-type, studies variably suggest either BD I or BD II disorder to be over-represented, so no conclusion can be drawn on this nuance.

While seeking to determine the correct diagnosis is always a key assessment priority, for those presenting with co-morbid states the bar for accepting each diagnosis would seemingly need to be set high to avoid false-positive primary or co-morbid diagnoses, and it would appear important to determine whether any ADHD condition emerged in childhood or in adulthood in light of prognostic treatment implications. In relation to management, many authors (e.g. Nierenberg et al., 2005) note that clinicians are often hesitant to prescribe psychostimulants to those with such comorbid states, reflecting concern about exacerbating the course of the BD or inducing more ‘switching’ into hypo/manic states. Klassen et al. (2010) noted several treatment studies for those with the comorbid states. All were either anecdotal and/or involved small samples, but with differing treatment models evident in young and adult patients. In children, the model was generally to first use a mood stabiliser and, after stabilisation, to introduce a psychostimulant, while in adults the patients tended to be treated for their BD condition only. Pataki and Carlson (2013) overviewed four studies of children to conclude that stimulant treatment did not worsen mood symptoms in any of the studies, while Schiweck et al. (2021) judged that ‘First evidence’ suggests no general disadvantage in combining a mood stabiliser with methylphenidate for management of ADHD with comorbid BD. In line with other writers on the topic, Youngstrom et al. (2010) recommend treating the BD first, reflecting its greater severity and more concerning prognosis, as well as conceding the risk of any primary ADHD treatment involving psychostimulants worsening the illness course of the BD. Such a sequencing model is worthy of close clinical consideration. While seemingly preferable, it may well be that, for some individuals, settling their ADHD symptoms first may lead to greater success in managing their BD, whether by providing them with a clearer comprehension of management nuances (e.g. compliance) or assisting them to take up a well-being plan with its multiple component strategies.

Clinician awareness of the distinctive comorbidity phenomenon should be fostered while focussed research addressing possible determinants appears worthwhile. It will be important to clarify if those with the comorbid conditions differ in terms of their long-term trajectory and in response to condition-specific medications (either intrinsically, or as a consequence of onset age or the presence of other differing co-morbid states such as anxiety and drug dependence being present), an objective benefitting from clinician observation as well as from empirical studies.