Neuropsychological profiles of adult bipolar disorder patients with and without comorbid attention-deficit hyperactivity disorder

The prevalence of comorbid attention-deficit/hyperactivity disorder (ADHD) in adults with bipolar disorder (BD) ranges between 9.5 and 28% (Nierenberg et al. 2005; Rydén et al. 2009; Torres et al. 2015). Although the longitudinal course of illness differs between BD and ADHD, where BD is an episodic disorder and ADHD is a persistent condition, differentiating between ADHD and BD is sometimes a challenge (Brus et al. 2014). BD and ADHD feature partially overlapping symptoms that cross-sectionally might be interpreted as hypomanic symptoms, such as talkativeness, restlessness, impulsivity, distractibility, and affective lability.

The overlap of symptoms not only begets diagnostic challenges, it also raises the question if so-called ‘comorbid’ BD and ADHD condition might be an artefact produced by the diagnostic system (Wingo and Ghaemi 2007). The comorbidity concept is, however, supported by the fact that a majority of patients with comorbid BD and ADHD maintain both diagnoses after correction for overlapping symptoms (Milberger et al. 1995). Moreover, BD with comorbid ADHD presents a worse course of illness than BD without ADHD, including earlier onset of bipolar disorder (Nierenberg et al. 2005; Tamam et al. 2008; Rydén et al. 2009; Karaahmet et al. 2013; Perroud et al. 2014), as well as higher frequency of mood episodes (Nierenberg et al. 2005; Tamam et al. 2008; Rydén et al. 2009; Perroud et al. 2014; Torres et al. 2015), suicide attempts (Nierenberg et al. 2005; Torres et al. 2015; Harmanci et al. 2016), and interpersonal violence (Nierenberg et al. 2005; Rydén et al. 2009). This means that it is clinically important to identify BD patients with comorbid ADHD. It is noteworthy that the mere history of childhood ADHD, irrespective of adult ADHD status, is sufficient to generate a worse course of illness in BD (Rydén et al. 2009; Torres et al. 2015). This suggests that reviewing childhood history is important when assessing BD prognosis.

BD has been associated with general cognitive impairment across different cognitive domains (Mann-Wrobel et al. 2011; Sparding et al. 2015). Meta-analyses suggest that executive function, verbal memory, and attention/working memory are particularly affected (Robinson and Ferrier 2006; Torres et al. 2007; Arts et al. 2008; Bourne et al. 2013). However, more recent studies suggest that cognitive impairment may not be a general trait in BD, but limited to specific subgroups (Burdick et al. 2014; Bora et al. 2016; Sparding et al. 2017): one group of patients that perform within the normal range, and one or more subgroups that exhibit varying degrees of cognitive impairment. It remains unknown what accounts for this cognitive heterogeneity in BD, but a possible explanation is that the cognitively impaired group suffer from comorbid conditions, e.g., ADHD. Cognition in adults with ADHD shows general impairments across multiple cognitive domains, with possible emphasis on impaired attention, inhibition, and verbal memory (Hervey et al. 2004; Schoechlin and Engel 2005). Despite this, studies of cognitive function in BD have generally not accounted for potential comorbid ADHD.

Examining cognition in BD patients with and without comorbid ADHD might also serve to identify a BD subgroup with poor outcome in need for specific interventions. One study compared adults with BD, ADHD, BD + ADHD, and healthy controls across a broad spectrum of cognitive domains, but found no significant differences between BD and BD + ADHD (Torres et al. 2017). However, the assessment of comorbid ADHD did not include any informant report. The accuracy of retrospective self-reports of childhood symptoms has been questioned (Barkley et al. 2002; Mannuzza et al. 2002), and it is considered best practice to include multiple sources of information, e.g., a parental or next of kin interview (Davidson 2008). Furthermore, replication is needed given the limited sample size in the previous study (Torres et al. 2017).

The aim of the present study was to compare the neuropsychological profiles of adult BD patients with childhood ADHD (BD + cADHD) and without childhood ADHD (BD − cADHD) in a cohort with meticulous assessment of childhood as well as adult ADHD.

The participants in this study were recruited from a bipolar outpatient clinic at the Northern Stockholm psychiatric clinic in Sweden between October 2005 and April 2008, as a part of the St. Göran Bipolar project. The project and the clinical assessment process has been described in detail elsewhere (Rydén et al. 2009). In brief, the project is a prospective, longitudinal study providing assessment, treatment, and follow-up of patients with BD. For the present study, we included patients who were at least 18 years of age, met the DSM-IV criteria for BD type I or II, and consented orally and in writing to participate in the project. Exclusion criteria were a history of intellectual disability (IQ < 70), organic brain disorder, neurological disorder, and severe medical conditions. The study was approved by the ethical review board in Stockholm, Sweden.

This study compared the neuropsychological profiles of adult BD patients with and without comorbid childhood ADHD. We payed particular attention to childhood history and performed interviews with parents or next of kin to establish childhood ADHD diagnoses. The main finding was that the neuropsychological profiles of the groups were similar. The only exception was that the BD + cADHD group performed significantly worse on WAIS working memory than the BD − cADHD group, but that finding would not withstand correction for multiple testing (N = 10).

There are few previous studies in this area. Our finding of small overall cognitive differences between BD-cADHD and BD + cADHD is largely in agreement with the results of Torres and colleagues (Torres et al. 2017) who found no significant differences between BD and BD + ADHD. Another study (Silva et al. 2014) used only the Wisconsin Card Sorting Test and found that BD + ADHD showed set-shifting difficulties compared with ADHD and healthy controls. One study comparing adolescents with BD, ADHD, BD + ADHD, and healthy controls (Rucklidge 2006) found that the BD group in general performed similar to controls, whereas BD with comorbid ADHD was associated with cognitive impairment. The greatest deficits in the BD + ADHD group were found in verbal memory and inhibitory control. Working memory was impaired only in the ADHD group.

It is noteworthy that the few studies comparing BD and ADHD with respect to cognition suggest that differences were directly or indirectly attributable to working memory impairment in ADHD (Torralva et al. 2011; Baez et al. 2014). With the caveat that our results were not corrected for multiple comparisons, the finding of worse working memory of moderate effect size in the BD + cADHD group is interesting because conceptual models of ADHD place deficits in working memory as either a core feature of the disorder (Rapport et al. 2001), or as an area of secondary but fundamental impairment (Barkley 1997). In fact, a meta-analysis examining working memory in adults with ADHD found deficits of moderate effect size and concluded that working memory impairment in ADHD persist into adulthood (Matt Alderson et al. 2013).

Drilling deeper into the BD + cADHD group, we found no difference in working memory between those who met criteria for ADHD in childhood as well as adulthood, and those who only did so in childhood. This indicates that adults with BD with a history of childhood ADHD are likely to show working memory impairments regardless of current ADHD status.

Our findings have three important clinical implications. First, previous studies have found that BD with comorbid ADHD is associated with worse course of illness, calling for instruments to readily identify this group clinically (Rydén et al. 2009; Torres et al. 2015). Unfortunately, this study suggests that cognitive testing is not likely to be a valuable tool to differentiate between BD with and without ADHD given that we found quite similar neuropsychological profiles. This might be surprising since cognitive testing is considered to be an essential source of information in ADHD assessment. But there is in fact no agreed upon cognitive profile specific for ADHD (Davidson 2008), nor is there consensus on a testing battery that would identify cognitive impairments in BD [although there have been suggestions (Yatham et al. 2010; Sparding et al. 2015)]. This absence of disorder-specific cognitive profiles might be due to inherent limitations with current neuropsychological batteries or that cognitive impairments are orthogonal to psychiatric syndrome diagnoses.

Second, our results show that childhood ADHD was associated with impaired working memory in BD regardless of whether the diagnosis persisted into adulthood or not. This aligns with studies indicating that the mere history of childhood ADHD is sufficient to generate a worse course of bipolar illness (Rydén et al. 2009; Torres et al. 2015). Other studies have found cognitive impairment in BD to be associated with worse course of illness (Robinson and Ferrier 2006; Bourne et al. 2013)—with the highest correlation for working memory—and poorer everyday functioning (Depp et al. 2012). Hence, future research could investigate in more detail if cognitive testing focusing on working memory together with a thorough assessment of childhood symptoms of ADHD might be of value in identifying a group of BD patients at risk for poor outcomes.

Third, cognitive impairments observed in BD patients as a group have been attributed to one or more subgroups with more pronounced impairment (Burdick et al. 2014; Bora et al. 2016; Sparding et al. 2017). So far, no proposed characteristics of these cognitive subgroups have been able to explain the cognitive variability, i.e., bipolar subtype (Pålsson et al. 2013; Bora 2018), prior psychotic manifestations (Savitz et al. 2009; Bora 2018), number of affective episodes (Mann-Wrobel et al. 2011), residual mood symptoms, or medication (Bourne et al. 2013). Our findings suggest that nor is comorbid ADHD likely to explain the heterogeneous cognitive performance in BD beyond working memory deficits.

Some limitations of this study should be considered. First, although the sample size was on par with previous studies in the area, a limited number of study subjects might have affected the possibility to detect group differences with ensuing risk of type II errors. Second, we did not correct for multiple comparisons; doing so would revoke the group difference in working memory. Moreover, in the supplementary analyses where a healthy control group was added, working memory was no longer statistically significantly between the bipolar disorder groups. Third, ADHD is a childhood onset diagnosis that in this study was made in retrospect. Even though we used validated instruments to assess childhood symptoms, conducted a separate diagnostic interview focusing on ADHD, and even included an interview with next-of-kin when possible, this study design is susceptible to recall bias. Fourth, although not statistically significant, there were more BD II patients in the BD + cADHD group. We did not find any differences between BD I and BD II in a previous study (Sparding et al. 2015,), but a recent meta-analyses found that BD I performed significantly worse than BD II with respect to global cognition, verbal memory, processing speed, and executive functioning speed and accuracy (Bora 2018). Fifth, the use of mood stabilizers differed such that lithium was more common in the BD − cADHD group whereas other mood stabilizers (e.g., valproate, lamotrigine) were more common in the BD + cADHD group. It cannot be excluded that medication might affect cognitive performance.

Cognitive performance was similar in bipolar disorder patients with and without ADHD in childhood, except that the former group performed worse on working memory. This means that comorbid ADHD cannot explain the cognitive heterogeneity seen in bipolar disorder patients, and that cognitive testing is not likely to be useful for differentiating between BD with and without ADHD.