Frying pan to fire? Commentary on Practitioner Review: Definition, recognition and treatment challenges of irritability in young people

Abstract

The bipolar disorder diagnosis in prepubertal children became popular because it answered a clinical need to treat the explosive behavior component of irritability and the hope that anti-manic strategies would be helpful. Poor definition of episodes resulted in mixing chronic and episodic irritability in samples of children with bipolar disorder. The subsequent dramatic increase in neuroleptic use is a testimony to the importance of the problem of irritability and our need to better understand it. Insofar as our use of the term irritability conflates proneness to anger with the subsequent aggressive response, it will again not be clear who is being studied. We need to uncouple the mood and behavior aspects of irritability for further study or we will have traded the imprecision of “bipolar” for the imprecision of irritability.

Interest in irritability began with research suggesting there was a unique subtype of bipolar disorder beginning in prepubertal children. The intramural team at NIMH, including Dr. Stringaris, has been instrumental in producing or stimulating much of the research on irritability as well as several excellent reviews (Evans et al.,in press; Stringaris et al., in press; Wakschag et al, in press;). My commentary will address 5 points made in the Practitioner Review by Stringaris and colleagues in the current issue: the importance of the behavioral component of irritability, measurement issues, relationship between community and clinical samples, limitations in outcome studies and treatment implications.

What has made juvenile bipolar disorder so compelling over the past 25 years, at least in the United States, has been that it provides a diagnostic home for children with severe anger outbursts. Calling these outbursts “tantrums” normalizes them and diminishes their ferocity whereas mania conveys severity. The research group at Massachusetts General Hospital articulated this most clearly, reporting that 77% of children identified as manic in their psychopharmacology clinic had extreme, persistent, “super-angry” irritability (i.e. “explosive outbursts that are often violent, prolonged, and largely unprovoked”). This kind of irritability was said to be specific to mania and distinguished it from losing temper or simply being grouchy, as seen in oppositional or depressed youth (Mick et al., 2005). Indeed, Child Behavior Checklist (CBCL) T scores for aggression were at least 80 compared to T scores of less than 70 for children with other conditions. Their hypothesis met a need in that none of our diagnoses addressed explosive children. Conduct disorder has been bred to select proactive aggression. Emotion dysregulation, originally part of the conceptualization of hyperkinetic children, was removed as a primary criterion for ADHD in DSM III. All of the other conditions that evaluate irritability did and still do so by specifying the frequency of feeling irritable/ easily annoyed in the context of different symptoms (depression, anxiety).

Intermixing chronic and episodic irritability occurred in part because of poor episode definition in prior versions of DSM and the interviews developed from it (Carlson and Klein, 2014). DSM 5 shored up the episode definition and created disruptive mood dysregulation disorder (DMDD) to provide a better home for chronically irritable, explosive children. As others have noted, however, it is not without its own controversies (Evans et al., 2017) and we still have the problem that samples of bipolar youth include children with chronic and episodic irritability. We cannot separate them post hoc.

Although Stringaris et al (2018) said that irritability is easy to measure, it is not. It means different things to different people. Inadequate attention has been given to the tonic (“persistently angry, grumpy or grouchy mood”, i.e. touchy or easily annoyed, angry or resentful) versus phasic (excessive temper or upset with crying, stomping, name calling and/or verbal or physical aggression against persons or property) components (Copeland et al. 2015). In other words, how one feels versus what one does. Most interviews do not make those distinctions (Carlson et al., 2016). General rating scales have broad-band aggression scales though a few have a specific irritability factor. The Multidimensional Assessment of Preschoolers (Wakschlag et al., in press) provides an elegant breakdown of tantrum behaviors but only for preschool children and not, to my knowledge, for clinical use. The Affective Reactivity Index (ARI) measures angry mood (i.e. loss of temper, becoming and staying angry) but not what the child does when s/he gets angry (Stringaris et al, 2012). The concept of “reactive aggression” or “impulsive aggression” has been used to describe the behavior but the term was not tied specifically to irritable mood (Jensen et al., 2007).

We developed a simple, parent-completed measure of what a child actually does during an outburst (Carlson et al., 2016). We recently reported it correlates well with the ARI and temper loss. More importantly, though, the number of parent-reported behaviors during a severe outburst predicted the need for subsequent seclusion/restraint and the immediate need for oral or intramuscular injections (emergency interventions to manage severe outbursts in the hospital). “Often loses temper/easily annoyed” did not predict either one. In other words, at least in our sample, what the child actually did during an outburst had more predictive utility than frequency of temper m (Matthews et al., 2017).

Related to the mood versus behavior confusion is how irritability should be operationalized and quantified. There are 3 separate symptoms often considered synonymous: irritability, loss of temper and having a tantrum. They are not the same things. In the Stony Brook community sample (Carlson et al., 2016), as well as the Great Smoky Mountain (GSM) sample (Copeland et al., 2015), children could have tantrums without being irritable. Moreover, in the Stony Brook Community sample, about half the 6 year olds lost their tempers without having a tantrum. In the clinic sample, losing one’s temper culminated in a tantrum over 90% of the time but parents didn’t necessarily describe their child as irritable (Carlson et al, 2016). One solution would be to use the term “irritable” for grumpy and grouchy mood with subsequent loss of temper and use the term “reactive” or “impulsive” aggression separately for the actual tantrum behavior. There are considerable data for the latter (Jensen et al, 2007).

Another issue is how community samples relate to clinical samples. Do the “phasic”-irritable children in the GSM sample identify the same children seen in an outpatient clinic so that we can be assured that outcome and treatment implications can be extrapolated? The rate of “phasic irritability” [tantrums] in this non-clinical population is 51.4%, nearly identical to rates in clinical settings where severe outbursts are a reason for referral in 47.2% of outpatients and 56% of inpatients (Copeland et al, 2015; Carlson and Dyson, 2012; Margulies et al, 2012). Similarly, the rate of DMDD in the Stony Brook community sample was nearly identical to rate in the Stony Brook outpatient clinic (8.7% vs 10.9%) (Carlson et al., 2016). These are not the same children, however. In the community sample, median outburst duration was 7 minutes with outbursts occurring about once a month; 11.3% of those with “phasic events” were violent. In the clinic and hospital sample respectively, median outburst duration was 30 minutes and 50 minutes and occurred at least weekly and often daily in both groups; 44.5% and 100% respectively involved assault on property or people. In the 6 year olds from the Stony Brook community and clinic samples, the Global Assessment of Function (GAF) score was 61.9 (SD10). In the clinic sample the GAF score was 39.5 (SD 3.3). Only in the 90%ile of the community sample children, where outburst duration was 30 minutes and frequency was “more days than not”, did severity begin to approximate that of clinic children. It would be useful to know if their outcomes are the same as those less severe. Those are the children whose outcomes are of greatest interest to clinicians.

Irritability and tantrums differ by age and diagnosis. For instance, mild tantrums in normal preschool children are common. Pathological tantrums differ in terms of frequency, destructiveness, duration, and ease of provocation (Copeland et al., 2015; Wakschlag et al., in press). Interestingly, directly observed pathological tantrums in 6–12 year old psychiatric inpatients were found to have a structural and temporal organization similar to a preschooler’s temper tantrums and were not the rapid cycles of manic episodes (Potegal et al, 2009). It is a researchable question, then, how irritability/severe outbursts as a developmental delay in emotion regulation differ from the irritability that occurs in other psychiatric disorders. Irritability should certainly not be ascertained in one interview module and extrapolated to other conditions

How one elicits symptoms and behaviors of irritability in adults is equally relevant. Outcomes of oppositional defiant disorder (ODD), DMDD and reactive aggression in children consistently predict negative emotionality/depression and anxiety in adults. That may be in part because adult clinical interviews don’t ask about oppositional defiant disorder, or “tonic” and “phasic” irritability and if they do, it is in the depression section of the interview! We need to know explicitly the degree to which the oppositional or outburst behavior in childhood continues to affect the subject as an adult. Does ease of losing temper and outbursts attenuate with age (as one might expect of a developmental disorder) but still complicate the mood/anxiety disorder or dissipate? Etiologically, it is important to understand how irritability grown up compares with irritability in de novo or episodic disorders.

DMDD was hijacked from a research designation to prevent children from getting a bipolar diagnosis and treated for a lifetime disorder with medications whose side effects are significant. Keeping in mind, however, that the bipolar frenzy evolved because of the treatment problems posed by explosive children, the treatment perspective is important. Although not every child with DMDD has comorbid ADHD and ODD, many do. We have known since the Multimodal Treatment of ADHD (MTA) study that both stimulant medication and behavior modification are important in the treatment of children with ADHD with and without ODD. However, even after 14 months of optimal medication treatment for children in the medication-only and combined treatment groups, 44% of 267 children with initial impulsive aggression remained significantly symptomatic (Jensen et al. 2007). The 55 children in that study who had the CBCL dysregulation profile (i.e, CBCL T scores ≥70 on the aggression and anxious/depressed subscales and probably equivalent to DMDD), were even more impaired with T scores close to 80 (Galanter et al 2003).

Certainly there are a number of randomized controlled trials of non-medication behavior modification/parent management training and cognitive-behavioral treatment for children with anger and irritability (Sukodolsky et al., 2017). By extrapolation, there should be some efficacy in DMDD. However, even when such children get the best of all treatments, only 6 % normalize (Waxmonsky 2008) and when they improve, the relapse rate is high (Waxmonsky, 2016).

It may be useful to return to the tonic/phasic distinction of irritability in thinking about treatment. The tonic aspect is a target for psychological interventions, i.e. teaching parents and children coping skills. The average parent training program is several months long and the level of severity of the children rates an aggression T score of about 67 (Menting et al., 2013). The availability of knowledgeable care and the willingness to invest in it is very variable. Severe outbursts, however, are psychiatric emergencies. As noted, children in this category have aggression T scores between 75 and 80, about one standard deviation higher. We have written elsewhere (Carlson, 2012) that the reason doctors have used (and abused) neuroleptic medication is that we often need a fast, effective treatment for behaviors that are getting children carted off by the police in handcuffs and sent to emergency rooms with increasing frequency, hospitalized, suspended from school, and/or placed residentially. The problem with neuroleptics, besides their long-term side effects, frankly, is that they are not effective enough even in the short term, hence the inappropriate use of often 5–6 medications.

Irritable, explosive children have psychiatric, learning and language disorder challenges for which there is often not a simple solution. Treating the primary disorder makes sense as a start. However, it is important to realize that both short term and long term interventions will be needed. Neuroleptics may have a legitimate place in the acute treatment of explosive children though if they are started there needs to be an exit strategy. Etiology and outcome research need to account for severity. Policy makers need to recognize that children with developmental delays in emotion regulation need developmentally sensitive policies, not those meant for adults.

In conclusion, it would be a shame if we jumped from the frying pan of mislabeling children as “bipolar” because we had no accurate diagnosis to categorize their explosivity to the fire of conflating the mood and behavioral components of irritability.