Outcomes of Nontransitioned Cases in a Sample at Ultra-High Risk for Psychosis

PACE is a specialist clinic for young people at ultra-high risk for psychosis in Melbourne, Australia. The current data are from a study aiming to reassess all research participants at PACE between 1993 and 2006 (N=416). Follow-up interviews were completed by 311 participants (74.8%), 85 of whom had developed psychotic disorder (14). The current sample consisted of 226 participants (125 female, 101 male) who completed a follow-up assessment but had not transitioned to psychosis. Figure 1 shows the composition of the current sample.

At baseline, the participants were ages 15 to 30 years and met criteria for being at ultra-high risk for psychotic disorder. These are 1) attenuated psychotic symptoms, 2) brief limited intermittent psychotic symptoms, and/or 3) trait vulnerability for psychotic illness (schizotypal personality disorder or history of psychosis in a first-degree relative) and deterioration in functioning or chronic low functioning (see reference 14 for a full description of determination of ultra-high risk status of this cohort). Exclusion criteria for entry to PACE are a previous psychotic episode, organic cause for presentation, and past antipsychotic exposure equivalent to a haloperidol dosage of more than 15 mg/day.

A previously developed tracking system (23) was used to relocate participants. If participants did not consent to face-to-face assessment, they were asked for a telephone interview or written assessment. This study was approved by the local research and ethics committee. All participants provided written informed consent.

Data were examined for the frequency of current attenuated psychotic symptoms, nonpsychotic DSM-IV disorders during the follow-up period (current or since baseline), and the course of disorders. Three disorder groups were examined: mood, anxiety, and substance use disorders, as well as the frequency and course of any disorder. Somatoform and eating disorders occurred rarely and were not included.

The course of disorders was examined for participants who had diagnostic assessments at baseline and follow-up (N=203 for mood/anxiety, N=192 for substance use). For each disorder, the participant’s status was classed as “never” if the disorder was not present at baseline or during follow-up, “persistent/recurrent” if the disorder was present both at baseline and during follow-up, “remission” if the disorder was present at baseline but absent during follow-up, or “incident” if the disorder was absent at baseline but present during follow-up (see Figure 2).

To investigate candidate predictors of the course of disorders, participants with incident disorder were compared with participants who never had the disorder. Participants with persistent or recurrent disorder were compared with those with remitted disorder. Candidate predictors were intake group; GAF score; BPRS scores for psychotic and affective symptoms; SANS total score; CAARMS scores for disorders of thought content, perceptual abnormalities, and conceptual disorganization; and verbal, performance, and full-scale IQ. For primary analyses, predictors with a univariate association at p<0.1 were entered together into a binary logistic regression to identify the strongest predictors. Age at baseline, gender, and length of the follow-up period were always included as predictors. Analyses were conducted for mood, anxiety, and substance use disorders separately and then for any disorder. Our study numbers gave us 80% power to detect (at significance level p<0.05) a reduction in persistence or recurrence of any disorder from 67% to 45% associated with removal of a common risk factor (e.g., gender) with 50% prevalence, and we had similar power to detect a reduction from 67% to 41% associated with removal of a rarer risk factor with 25% prevalence. For incidence of any disorder, we were similarly powered to detect a reduction from 84% to 57% associated with a common risk factor and 84% to 54% with a rarer risk factor.

Given the large variability in follow-up period, the cohort was divided into three subsamples on the basis of when they were identified as at risk: long follow-up period (1993–2000, N=82), medium follow-up (2001–2003, N=77), and short follow-up (2004–2006, N=67). Frequencies are presented for the entire cohort and each subsample. Given the volume of data, some analyses (including exploratory analyses of neurocognitive predictors) are presented in online supplementary data only.

More female than male participants completed follow-up (55.3% female; χ²=5.12, df=1, p=0.02). There were no other significant differences between participants who were followed up and those who were not. Their mean age was 18.6 years (SD=3.3) at baseline and 25.5 years (SD=4.8) at follow-up. Follow-up was conducted between 2.4 and 14.1 years after baseline (mean=6.9, SD=3.1, median=5.72). Eighty-two (36.3%) of the participants received trial treatment at PACE—cognitive-behavioral therapy, N=25; cognitive-behavioral therapy and low-dose antipsychotics, N=38; or low-dose lithium, N=19)—all for 12 months or less. There were no significant differences in disorder rates during follow-up between participants who received trial treatment and those who did not. Further characteristics for each subsample are presented in Table S1 in the data supplement accompanying the online version of this article.

The use of antipsychotics or any psychiatric medication in the 2 years prior to follow-up was documented for 184 participants (81.4% of the sample). Of these participants, five (2.7%) reported using antipsychotic medication and 70 (38.0%) had used any psychiatric medication “some or all of the time” in the past 2 years.

Diagnostic outcomes at follow-up are presented in Table 1. Of the entire cohort, 68.1% met criteria for at least one disorder during the follow-up period. Mood disorder was present during follow-up for 48.7%, anxiety disorder for 34.5%, and substance use disorder for 29.2%. Proportions were not notably different between subsamples.

For the entire cohort, both mood and anxiety disorders were present in 24.3%, mood and substance use disorders in 17.7%, anxiety and substance use disorders in 13.7%, and all three disorders in 10.2%. Patterns of comorbidity were similar in the 1993–2000 and 2004–2006 subsamples but lower in the 2001–2003 group (see Table 1).

The proportion of participants reporting attenuated psychotic symptoms at follow-up that were at or above the threshold for ultra-high risk was 28.3% for the entire cohort, 24.4% for the 1993–2000 subsample, 23.4% for the 2001–2003 subsample, and 41.9% for the 2004–2006 subsample (data were missing for 30 of the participants with telephone or written assessment at follow-up).

The co-occurrence of attenuated symptoms and disorders at follow-up is presented in Table 2. For the entire cohort, the presence of attenuated psychotic symptoms was significantly associated with mood disorder (χ²=7.81, df=1, p=0.005) and with any nonpsychotic disorder over the follow-up period (χ²=5.91, df=1, p=0.02), but not with anxiety and substance use disorders. Results were similar for the 2004–2006 subsample (mood disorders: χ²=9.14, df=1, p=0.003; any disorder: χ²=8.19, df=1, p=0.004).

Of the five participants using antipsychotics in the 2 years before follow-up, three had persistent attenuated psychotic symptoms and two had no attenuated symptoms. Participants who reported any psychiatric medication use were more likely to report current attenuated psychotic symptoms than those who did not (45.7% versus 26.3%; χ²=6.46, df=1, p=0.01).

Table 3 shows the frequencies of baseline disorder and of remission, incidence, persistence/recurrence, and absence of nonpsychotic disorders. In the following, we report results for the entire cohort. Of the participants who had a mood disorder at baseline (71.4%), 53.8% had persistent or recurrent disorder. In those without mood disorder at baseline, 32.8% developed one. Of those with anxiety disorder at baseline (39.9%), 40.7% experienced persistent or recurrent anxiety. Of those without anxiety disorder at baseline, 29.5% developed one. Substance use disorders were present at baseline for 21.9% (of the 192 with available baseline substance use diagnoses). Of them, over half (52.4%) showed persistent or recurrent substance use disorder over follow-up. Of those without substance use disorder at baseline, 22.0% developed a substance use disorder.

In terms of any disorder, 90.1% of the cohort had any nonpsychotic disorder at baseline. Over the follow-up period, 26.0% of the entire cohort had remission of a nonpsychotic disorder, 37.5% developed a new disorder, and 51.6% had a persistent or recurrent disorder. Only 7.3% never experienced any disorder.

For the most part, the courses of disorders did not notably differ between subsamples. An exception was seen for the 2004–2006 subsample, who had a lower rate of substance use disorders at baseline. However, the rate of incident substance use disorder in this subsample was comparable to the rates for the other groups.

Baseline symptoms, GAF score, IQ, and age were poor predictors of the course of disorder. Gender emerged as a significant predictor of specific disorders, although the overall models were not statistically significant. Being female was associated with persistent or recurrent mood disorder compared with remitted mood disorder (odds ratio=2.07, 95% confidence interval [CI] for odds ratio=1.02–4.23, p=0.05) and with incident anxiety disorder compared with never having an anxiety disorder (odds ratio=2.66, 95% CI for odds ratio=1.11–6.39, p=0.03).

The incidence of any disorder was associated with a higher baseline score on the SANS (odds ratio=1.14, 95% CI for odds ratio=1.01–1.29, p=0.03) compared with never having a disorder. The persistence or recurrence of any disorder, as opposed to remission from any disorder, was associated with being female (odds ratio=2.40, 95% CI for odds ratio=1.12–5.15, p=0.02). Meeting the criteria for brief limited intermittent psychotic symptoms at intake was associated with a lower chance of persistent or recurrent disorder (odds ratio=0.19, 95% CI for odds ratio=0.05–0.72, p=0.01). Despite its variability, the length of follow-up did not predict the course of disorder in the entire cohort.

Predictors of the course of disorders for each subsample and exploratory analyses of neurocognitive performance are presented in the online supplementary data.

Attenuated psychotic symptoms were reported by 28.3% of the current sample at follow-up assessment. Considered together with the individuals in the cohort who developed psychosis (14), half of those who met the criteria for ultra-high risk showed continued or recurrent positive psychotic symptoms (threshold or subthreshold). The presence of attenuated psychotic symptoms may reflect continued risk for psychosis in some individuals. This is possible since transitions occurred up to 10 years after identification of risk in this sample (14). Alternatively, attenuated symptoms may occur in the context of nonpsychotic disorders, which resolve with resolution of that disorder (3, 6, 7). This would be consistent with the idea of “incidental” psychotic symptoms (33). The fact that participants with the shortest follow-up periods showed the highest rates of attenuated symptoms and that their attenuated symptoms were associated with nonpsychotic disorders could support either of these possibilities.

Mood disorders were the most common diagnosis during follow-up. Major depressive disorder was especially common. This was followed by high rates of anxiety disorders, cannabis dependence, and alcohol abuse. These rates are higher than would be expected in the general population. A detailed comparison of our cohort with the Australian general population (34) is presented in online Table S3. Briefly, the rates of nonpsychotic disorders in this cohort were higher than the 12-month prevalences of these disorders for a similar age group in the general population, as well as higher than lifetime prevalences of adults of all ages. Notably, the prevalence of mood disorder over follow-up in our cohort was higher by a factor of five than the 12-month prevalence and higher by a factor of three than the lifetime prevalence in the general population.

This would be expected of a selected help-seeking sample. Indeed, many nonpsychotic disorders were already present at baseline. An important finding was that disorders persisted for approximately half of these young people who did not develop psychosis. In addition, for those without a nonpsychotic disorder at baseline, the incidence of new disorders was common. In fact, over a third of the sample developed an incident disorder over the follow-up period. Thus, the ultra-high risk criteria might also represent a useful system for identifying young people at risk for chronic and emerging nonpsychotic disorder, especially since they are already linked with youth mental health services. This highlights the need for further investigation to develop a better understanding of the risk factors associated with nonpsychotic disorder in this population.

We explored positive and negative psychotic symptoms, affective symptoms, functioning, IQ, gender, and age as predictors of course of disorder. Being female was associated with a higher risk of disorder than being male, a finding that is consistent with general population data (34). However, no other baseline variables were associated with the course of a specific disorder. It is interesting that higher SANS scores at baseline and not meeting the criteria for brief limited intermittent psychotic symptoms were associated with the incidence and the persistence or recurrence of any disorder, respectively. This could demonstrate the specificity of brief limited intermittent psychotic symptoms to psychotic disorder and, on the other hand, the nonspecificity of symptoms measured on the SANS. It may be that depressive symptoms were interpreted as negative symptoms and rated on the SANS. Alternatively, individuals with high negative symptom scores and depressive disorder may continue to be in the prodrome of a psychotic disorder, as both are known to occur during this phase in schizophrenia (35, 36).

Although highly variable, the length of the follow-up period was not strongly associated with the course of disorders. Notably, in the subsample with the shortest follow-up period, persistence or recurrence of any disorder was associated with a shorter follow-up period, consistent with the decrease in the rate of nonpsychotic disorders that was noted by Addington and colleagues (19). Together with the finding of considerably higher levels of attenuated psychotic symptoms in this subsample, our data suggest that the time for which participants are monitored may be important over the short term (first 2 to 4 years) but becomes less important over the longer term. This has implications for many studies, which typically track at-risk participants for 1 to 3 years.

The lack of strong predictors of nonpsychotic disorder is distinctly different from the prediction of psychotic illness in at-risk samples, where a number of baseline symptoms are consistently shown to predict onset of psychosis. This does not imply that the course of nonpsychotic illness cannot be predicted. Rather, it suggests that clinical variables that predict nonpsychotic disorder may be different from those that predict psychotic illness, highlighting the need to design studies with a focus on multiple outcomes at inception (37).

The strengths of this study are the large number of participants recruited from a single site, the long follow-up period, and high follow-up rates. The greatest limitation is the variable length of the follow-up period. Although we have presented data for the entire cohort as well as for subsamples of short-, medium-, and long-term follow-up, the subsamples did differ in some respects and the analyses were complicated by this. Moreover, the epochs used were arbitrary.

Another limitation is that we did not comprehensively document treatment over the entire follow-up period, limiting it as a potential predictor of the course of disorders. Additionally, follow-up diagnosis of 32 participants was made from telephone or written interviews. Finally, women were overrepresented at follow-up, which may bias toward higher levels of mood and anxiety disorders.

The current findings demonstrate significant psychopathology in nontransitioned cases 2 to 14 years after identification of risk. Persistent or recurrent nonpsychotic disorders were frequent even though these young people had previously been involved with youth mental health services, albeit in a time-limited manner. Clinically, the results suggest the need for at-risk clinics to include nonpsychotic outcomes in their treatment and follow-up plans.

We previously proposed a clinical staging model positing that severe mental disorders (e.g., schizophrenia, bipolar disorder, severe unipolar depression) develop from initial nonspecific symptoms, such as depressed mood, anxiety, and distress (38). Acquisition of new symptoms, including psychotic symptoms, and worsening of emotional dysregulation occur in some people, who might then meet the psychosis at-risk criteria. From this clinical picture, a number of trajectories and outcomes are possible, including the major mental disorders noted above, remission, or persistence of subthreshold syndromes. The ultra-high risk criteria were developed to detect incident psychotic disorders and have proved valid to that end. It is not surprising therefore that they identify high rates of schizophrenia (39). Future studies need to investigate the risk factors for chronic and incident nonpsychotic disorder by incorporating variables of interest to nonpsychotic outcomes in their designs—for example, axis II disorders, mood disturbance, cognitive biases, or family history of nonpsychotic disorders. Moreover, there is a need to investigate how functional outcome is associated with the presence of nonpsychotic disorders. This knowledge will increase the understanding of the factors associated with the onset, course, and outcome of disorders in this population and how they can best be treated.