During more than 18 months of data collection, 117 outbursts were recorded for 49 patients. Of the coded behaviors observed, angry behaviors (eg, yelling, screaming, cursing, violent threats, stamping, kicking, hitting, throwing objects) were the most common and occurred in 93% of the episodes; tearful/sad or anxious/fearful behaviors occurred in fewer than half (46%) the rages. An episode of rage generally lasted over 45 minutes, but length varied widely, with 19% of outbursts lasting less than 30 minutes and 19% lasting longer than 60 minutes.
Rage outbursts usually occurred early in the course of hospitalization. For instance, 44% of children with outbursts had their first/only episode within 2 days of admission. The remainder occurred 2 days after admission through discharge. Half of the children had an episode of rage immediately after admission; then, after “testing the waters,” they quickly desisted.
In addition to the question of whether outburst content and/or structure might vary depending on psychiatric conditions, we were mindful of frequent parental comments to the effect that their rage-prone child had “never outgrown the terrible twos.” This raised the hypothesis that rages might resemble childhood temper tantrums. A set of factor analyses of coded behaviors identified 5 groups: 3 were interpretable as progressive levels of anger intensity, 2 others as levels of distress (ie, sadness/anxiety).9
This model of outburst organization accounted for 54% of total variance. An independent cluster analysis of behavior slope revealed that anger behaviors peaked early and declined relatively rapidly; distress behaviors were more evenly distributed. Thus, rages closely resembled the observed tantrums of typically developing preschoolers with respect to behavior types and time course (early peaking anger, more evenly distributed distress).10,11 However, in terms of severity and duration, they were similar to parent-reported tantrums of preschoolers with depression and disruptive behavior disorders whose rages lasted more than 20 minutes.12 Inpatient rages lasted at least twice as long as temper trantrums of typically developing preschoolers, and because the patients were bigger, they were more destructive.
Diagnostic considerations
In determining how children with observed rage outbursts during hospitalization differed from their hospitalized peers, we found that younger children were marginally more likely to have rages (r = –0.155). Their IQ, history of abuse, and living status (with or without a parent) had no significant effect. Best-estimate diagnoses encompassing parental and school history, child mental status, and hospital course were made after discharge.
Considering single psychiatric diagnoses, children with outbursts were almost 5 times more likely to have a best-estimate diagnosis of ADHD than children without rages. They were over 5 times more likely to have a learning or language disorder but 3 times less likely to have an anxiety disorder. However, children with rages were almost 3 times more likely than those without rages to have 3 or more concurrent diagnoses. The most common combination was ADHD, oppositional defiant disorder, and learning/language disorder. This was seen in 70% of children who had rages compared with 34% of those who did not. The association between language disorder and outbursts is consistent with the more general connection between language and behavior13; however, it is almost never studied in research based solely on structured interviews.
Given the current interest in rages as a symptom of mania and severe mood dysregulation,2,14 we examined the frequency of those conditions based on referral information and subsequent best-estimate diagnoses for all hospitalizations (N = 151) because a child might have been admitted for mania on 1 admission but not another. In fact, hospitalized children with rages had been referred with a diagnosis of mania more often than those without rages (34.7% vs 15.7%, odds ratio [OR], 2.8; confidence interval [CI], 1.3, 6.3). However, of the 33 admissions where a referral diagnosis of mania had been given, mania was observed and confirmed in only 5 children. Of 44 children with rages, we observed only 9% with mania or manic symptoms (ie, bipolar not otherwise specified), compared with 4.7% of 86 children without mania, and always comorbid with other disorders. Mania, as defined by DSM-IV, did not account for most children whose rages occurred while hospitalized, nor were their outbursts examples of rapid cycles.
We also tried to determine how many children might have met criteria for severe mood dysregulation. Two-thirds of children with rages (65.9%) were defined as having this condition based on behavior at home, compared with only one-quarter of children without rages in the hospital (25.6%) (OR, 5.6; CI, 2.55, 12.39). Once hospitalized, 1 child had rage episodes 3 times a week. He was given a diagnosis of childhood-onset schizophrenia.
Treatment
The evidence-based treatment algorithm for ADHD and aggression consists of medication and behavior modification.15 If there is treatment resistance, an atypical antipsychotic is added, and then lithium(Drug information on lithium) or divalproex. While treatment makes some positive impact, the reality is that complete remission is rare in children with ADHD who have outbursts of rage. For example, in the Multimodal Treatment of ADHD study, rigorous treatment with stimulant medications with or without behavior modification significantly improved behavior in children with comorbid ADHD and in children with the Child Behavior Checklist bipolar phenotype (T scores greater than 67 on the ADHD, aggression, anxiety/depression scales).16 At the end of 14 months, however, these children were still considerably more impaired than those with uncomplicated ADHD.
