Age at Onset, Childhood Psychopathology, and 2-Year Outcome in Psychotic Bipolar Disorder

Early age at onset and/or poor premorbid functioning may be associated with poor course and outcome in bipolar disorder (1). Our previous research demonstrated that an earlier age at onset in bipolar disorder is associated with prior child psychopathology (2). We hypothesized that childhood psychopathology may, in fact, explain the association between young age at onset and poor outcome. In this report we address this issue by examining the independent and combined contributions of age at onset and childhood psychopathology to functional outcome and clinical course over 2 years of follow-up in a representative group of first-admission subjects with psychotic bipolar disorder.

Method

The Suffolk County Mental Health Project, described extensively elsewhere (3, 4), focuses on illness course in a countywide group of first admissions, aged 15–60, presenting with psychotic symptoms. The baseline interview, administered by master’s-level mental health professionals, usually occurred 1–3 weeks after admission. Written informed consent for the interview and use of records was obtained. Home follow-up interviews were conducted 6 and 24 months later. At 24 months, consensus DSM-IV diagnoses of the index episode were made by project psychiatrists using all sources of information, including the Structured Clinical Interview for DSM-III-R (SCID) (5). Of the 695 patients interviewed at baseline (72% of those targeted), 537 were assessed at 24 months, of whom 123 (22.9%) received a diagnosis of definite DSM-IV bipolar I disorder.

Age at onset and childhood psychopathology were determined from SCID histories, school and medical records, and interviews with significant others. Early-onset bipolar disorder was defined as a first affective episode before age 19. As described previously (2), three child psychopathology categories were formed: behavior disorder (persistent, impairing disorders of hyperactivity/impulsivity warranting referral, three or more conduct symptoms, or substance/alcohol abuse before age 16); other diagnostically ambiguous symptoms (fewer than three conduct symptoms, significant anxiety, suicide gesture without evidence of major depression, or emotional response to physical or sexual abuse); and no childhood psychopathology (kappa=0.88 for categorization of child psychopathology versus no psychopathology for 40 subjects).

Four outcome variables are presented: the interviewer-rated Global Assessment of Functioning for the best level of functioning achieved during the follow-up, dichotomized into poor (score<70, N=44) and good (score≥70, N=64) (not rated, N=15); an independently rated 4-point index of overall functioning modeled after Beiser et al. (6) (intraclass correlation=0.86) and stratified into fair/poor (score=1–2, N=53) and good (score=3–4, N=70); rehospitalization between the 6- and 24-month follow-ups (N=28 rehospitalized) (rapid rehospitalizations were usually due to misdiagnosis and mismanagement); and the World Health Organization’s classification of illness course (7), categorized into partial or no remission (N=23) versus complete remission regardless of the number of episodes (N=94) (not rated, N=6).

Logistic regression was used to assess the association of age at onset and childhood psychopathology to each outcome. Two indicator (dummy) variables were created to distinguish the three childhood psychopathology groups. Unadjusted odds ratios and 95% confidence intervals and odds ratios adjusting for the two predictors as well as for sex and education (≤high school or >high school) are presented.

Results

Of the 123 subjects, 21.1% (N=26) had a behavior disorder in childhood, 48.0% (N=59) had other comorbidity, and 30.9% (N=38) had no childhood psychopathology. Both the behavior disorder and other symptoms groups had significantly poorer course and outcome compared to those with no psychopathology (Table 1).

Twenty-seven subjects (22.0%) had an early age at onset, and 96 (78.0%) had a later age at onset. The patients with an early age at onset had significantly poorer results on measures of course and outcome, except for the Global Assessment of Functioning rating. For example, 63.0% of the early-onset group (N=17) versus 37.5% of the later-onset group (N=36) were rated as fair/poor on the Beiser index.

In the multivariate model (Table 1>, adjusted odds ratio), behavior disorder and other symptoms were associated with a significantly poorer functional outcome, compared with no childhood psychopathology. Subjects with behavior disorder had a higher likelihood of rehospitalization than subjects with no childhood psychopathology. Patients with an early age at onset were less likely to achieve full remission, but the odds ratios for the Beiser index and the likelihood of rehospitalization were no longer significant.

Discussion

Our data extend and clarify those of other studies on the relationship of age at onset to poorer functional outcome and episode relapse or recurrence in bipolar I disorder (1, 8, 9). Both early age at onset and childhood psychopathology were related to outcome. However, in the multivariate model, age at onset and childhood psychopathology had somewhat different effects, with childhood psychopathology (especially behavior disorders prior to the onset of mood disorder) related to poor functional outcome at 2 years and to rehospitalization, and early age at onset related to continued symptoms.

Our findings are limited to subjects with psychotic bipolar disorder first hospitalized at age 15 or older. Although some attrition occurred, there were no significant differences in the background characteristics of the completers and the subjects who refused or were lost to follow-up. Data on most child psychopathology features were based on parents’ and subjects’ recall. School records independently substantiated early behavior and academic problems.

Although several factors contribute to outcome, we conclude that it is important to separate the constructs of age at onset and psychopathology before onset of bipolar disorder, especially when considering clinical and functional outcome.

TABLE 1

Enlarge table