Psychiatric Antecedents in Adolescents at Clinical High Risk for Psychosis: Insights from the “Parma At-Risk Mental States” Follow-up Program

While there’s sufficient evidence on psychiatric antecedents in psychosis and schizophrenia (e.g., anxiety, depressive symptoms, attenuated psychotic symptoms, disruptive behaviors, sleep problems, basic symptoms/schizotypy) (Rutter et al., 2006; Uher et al., 2024), very little is known about antecedents in adolescents at “Clinical High Risk for Psychosis” (CHR-P) (Pelizza et al., 2025a). Since psychoses (and schizophrenia, in particular) are now increasingly considered as neurodevelopmental disorders, identifying early subclinical/clinical symptoms and syndrome of CHR-P mental states is crucial, as they could suggest psychopathological continuity and may interact with other risk factors synergistically increasing the risk of later developing psychosis (Cannon et al., 2002; Welham et al., 2009). From this point of view, CHR-P period could be considered a relatively late phase of an earlier developmental derailment (Rapoport et al., 2012; Poletti et al., 2019). Unfortunately, a recent meta-analysis of prospective investigations indicated that the association of psychiatric antecedents with transition in CHR-P studies was numerically weaker (Solmi et al., 2023a). In particular, for the two more commonly reported psychopathological antecedents (i.e., anxiety and depressive symptoms), a meta-regression showed a reduced effect size of the predictor-outcome association, controlling for antecedent type (Uher et al., 2024).

Overall, there is a lack of knowledge about childhood-onset disorders in prodromal teens, even though many risk factors for psychosis and schizophrenia occur during infancy (Mazzoni et al., 2009). Specifically, to date, most clinical antecedents (such as depression and anxiety) appear to be trans-diagnostic (Solmi et al., 2023b) and presented relevant associations with the onsets of major depressive, bipolar, and psychotic disorders. Furthermore, given the low number of prospective studies, the CHR-P design appears to underestimate the strength of antecedent-onset relationships. However, biological and familial investigations reported interesting findings. In this respect, a neurobiological review showed specific structural abnormalities, especially in prefrontal cortex and hippocampus, in 7 years of age children with genetic burden for schizophrenia (Nanda et al., 2014). Moreover, polygenic risk scores seem to account for about 80% of the variation in the risk of psychosis (Sullivan et al., 2003). In offspring of individuals with schizophrenia, developmental abnormalities during childhood frequently precede the onset of symptoms, especially attentional and behavioral impairments, social difficulties, and motor abnormalities (Erlenmeyer-Kimling, 2001).

Furthermore, the importance of further elucidating on the premorbid phase in CHR-P mental states is also accentuated by recent empirical investigations suggesting that poor premorbid academic functioning, poor social adjustment, and the presence of neurodevelopmental disorders (e.g., childhood repetitive behavior and attention-deficit/hyperactivity disorder [ADHD]) may act as predictors of conversion to psychosis (Tarbox et al., 2013; Dannevang et al., 2018; Jutla et al., 2020). Specifically, in an interesting cohort study by Welham and colleagues (2009), maternal reports related to general psychopathology, social and attentional dysfunction, aggression or delinquency at years 5 and 14 showed significant association with non-affective psychosis in youths by year 21. Child trauma is also a crucial risk factor for psychosis worth to be investigated as antecedent in CHR-P subjects, showing a prevalence rate around 60% and a risk higher of 2.72 times than healthy individuals (Varese et al., 2012; Loewy et al., 2019).

Additionally, the “Jerusalem Infant Development Study” found a 25% prevalence rate of affective and anxiety disorders and a 37% prevalence rate of disruptive behavior disorders in the group at high genetic risk for psychosis (Hans et al., 2000). Other research examining diagnoses during prodromal phase of psychosis showed prevalence rates around 50% for mood disorders, 30% for anxiety disorders, 35% for substance use disorders (Meyer et al., 2005; Rosen et al., 2006), but with small attention to childhood onset disorders. Finally, in a case series of 9 teens (aged 13–17 years) identified as prodromal to psychosis (Mazzoni et al., 2009), childhood-onset diagnoses commonly endorsed ADHD (5/9), oppositional defiant disorder (5/9), and separation anxiety (3/9).

In summary, to our knowledge, very few studies systematically investigated psychiatric antecedents and past specialist contact in young people at CHR-P, especially from a prognostic point of view and as predictors of treatment response and long-term outcomes. In this neurodevelopmental framework, we believe that better knowing and correctly identifying the most common psychiatric antecedents of CHR-P mental states is very important to avoid their psychopathological underestimation by clinicians, to counteract care discontinuity over time, and to longitudinally monitor symptoms and syndromes that, although trans-diagnostic, may be predictive of severe mental illness (such as psychotic, bipolar, and major depressive disorders). Together with the “pluripotent” CHR-P features (Lo Buglio et al., 2024), this would allow us to illustrate an “early psychopathology” that requires great attention in the context of prevention/early intervention of serious mental disorders. Therefore, the specific aims of this examination were: (a) to investigate psychiatric antecedents in a young CHR-P population treated within a specialized CHR-P service, and (b) to determine whether these antecedents could be considered a predictor of poor clinical or functional outcome during a 2 years follow-up period.

A total of 170 CHR-P individuals were recruited in this research through the PARMS program. Of them, 95 (55.9%) had a previous specialist (mental healthcare) contact and were included in the CHR-P + subgroup. The remaining 75 subjects were grouped in the CHR-P- subsample (Fig. 1).

Afbeelding vergroten

In the CHR-P + group, 73 participants (76.8%) had a past contact within child/Adolescent Mental Health Services (CAMHSs), while only 22 (23.1%) within Adult Mental Health Services (AMHSs). Their age at presentation was 19.62 ± 6.42 years compared to an age at previous contact of 16.30 ± 4.57 years. The main DSM-5 diagnoses at the earlier contact were learning disorder (n = 26, 27.4%), anxiety disorder (n = 22, 23.1%), conduct disorder (n = 21, 22.1%), and depressive disorder (n = 14, 14.7%). The conduct disorder diagnostic entity also encompassed hyperkinetic disorders. Finally, only 14 individuals (14.7% of the CHR-P + subsample) were directly referred to the PARMS program by other mental health services within a clinical pathway of care continuity. In the remaining 81 (85.2%) subjects, their past specialist contact ended with their care retention in mental health services being terminated (care discontinuity).

Baseline clinical and sociodemographic comparisons between the two CHR-P subgroups are shown in the Table 1. Compared to CHR-P-, CHR-P participants with previous specialist contact had a longer DUPrS. No other inter-group differences in terms of socio-demography, psychopathology, functioning, treatment and clinical characteristics were found.

Notably, our Kaplan–Meier survival analysis results showed that CHR-P + participants had a lower 2-year incidence rate of service disengagement (38.9% vs. 53.3%) and a higher 2-year incidence rate of psychosis transition (16.5% vs. 7.2%) in comparison with CHR-P- individuals (Table 2; see also Fig. 2 for details on survival functions). No between-group differences in incidence rates of new hospitalization and new suicide attempt were observed across the follow-up period.

Afbeelding vergroten

Our binary logistic regression analysis results for 2-year not time-to-event outcome parameters in the two CHR-P subgroups showed no statistically relevant inter-group differences in terms of hazard ratio values for suicidal ideation, functional recovery, SOFAS and functional remission, PANSS symptomatic remission, CHR-P criteria persistence, and AP prescription rate along the follow-up period (Table 3). Conversely, our mixed-design ANOVA analysis results showed a statistically significant effect of time on all HoNOS, PANSS, and SOFAS scores, as well as on equivalent AP doses (Table 4). Specifically, we observed longitudinal improvements in outcome, psychopathological and functioning severity levels, as well as a reduction in AP dosage over time. However, statistically significant group effects for HoNOS “Social problems” subscores and HoNOS total scores were notably found. Specifically, compared to CHR-P-, CHR-P + individuals maintained higher scores in these HoNOS dimensions along the 2 years of follow-up (see also Supplementary Materials [Figure S1] for details on mixed-design ANOVA profile plots).

Exploring psychiatric antecedents in youths at CHR-P remains a crucial opportunity to better describe and define their clinical trajectories, especially in prognostic terms (Polari et al., 2018). Very few investigations systematically examining this topic and there’s an overall lack of knowledge about child/adolescent prodromal syndromes in CHR-P teens, even though it has been frequently reported that several clinical risk factors for psychosis occur during childhood and adolescence (Raballo et al., 2022). In fact, this knowledge gap prevents the development of truly effective EIP interventions for these young clinical populations, resulting in treatment delay and worsening prognosis (Pelizza et al., 2024b). Furthermore, neglecting any psychiatric history in early psychosis patients may also lead to “one-off” (often non-specialist) care, to an increase in the duration of untreated psychosis or DUPrs, and to the fragmentation/discontinuity in specialist treatments (Catalano et al., 2025).

In this examination, approximately 56% of CHR-P participants had a past contact with mental healthcare services before entering the PARMS protocol. Specifically, most of them (85.3%) showed psychiatric antecedents but without current retention in care within specialized centers (care discontinuity), i.e., presenting to the PARMS program as new specialist access. Only 15% of CHR-P participants were directly referred to our EIP service by other mental healthcare centers within a generalist-to-specialized care program transition (care continuity). These findings highlight a critical issue for clinical practice: a relevant portion of youths who will later develop CHR-P mental states lacks sufficient monitoring for effective psychosis prevention and early detection within help-seeking behaviors in a crucial age range (Signorini et al., 2017). In this sense, it is important to underline the need to train psychiatrists and psychologists working in general and “low-threshold” mental health services on EIP (as well as general practitioners and emergency physicians), in order to avoid underestimating patients’ requests for help and to build a solid knowledge of a specific and shared prodromal psychopathology that can precede the onset of psychosis and other severe mental illnesses (Gammino et al., 2025). Furthermore, it is important to emphasize the need to implement early care pathways aimed at reducing the duration of help-seeking behavior and promoting appropriate access to specialist EIP services, particularly through primary care, emergency departments, and generalist mental health centers (Skrobinska et al., 2024). In this respect, the recent paradigm shift from EIP to “Youth Mental Health” (YMH) services might help to overcome some potential diagnostic delay factors (such as misdiagnosis, diagnostic underestimation of APS, considering sub-threshold psychotic features as secondary to other mental problems, including substance abuse) and to promote timely CHR-P detection, to reduce the DUPrS, and to improve long-term outcomes (Uhlhaas et al., 2023).

Additionally, in a non-negligible portion (41.8%) of CHR-P + participants, care discontinuity interested the transition age and the care transition from CAMHS to AMHS. This is not surprising considering the low rate of service engagement as care continuity within CAMHS-to-AMHS transition reported in young populations with severe mental illness (Pelizza et al., 2025d). In this respect, previous investigation examining care continuity in clinical samples with mental disorder reported that young people referred for treatment from CAMHS to AMHS are less likely to receive a discharge diagnosis of hyperkinetic and pervasive developmental disorders (Maurice et al., 2022). Based on this evidence, some authors proposed the shift towards a neurodevelopmental perspective on the development of psychosis (Gur et al., 2025) and the potential role of neurodevelopmental disorders (especially autism spectrum disorders and ADHD) as predictors of psychosis conversion (Gering et al., 2021; Guerrera et al., 2024).

Furthermore, the European MILESTONE survey (Gerritsen et al., 2022) reported that the current organization of mental health services along the pediatric/adult care model irremediably fails to match the natural course of emerging severe mental illnesses in youths, especially psychosis. Indeed, individuals aged 12–25 years have the highest incidence of severe psychiatric disorders across the lifespan, while also having the worst service engagement with CAMHS and AMHS compared to other age groups, with obvious consequences in terms of care discontinuity, unmet needs, and under-treatment (Poletti et al., 2022b). Therefore, we can no longer postpone a radical review of the architecture and resourcing of health care for youths in transition from childhood to adulthood (Raballo et al., 2017). In particular, it is absolutely necessary to strengthen those health pathways that may facilitate the transition and continuity of care between adolescent and adult services (such as YMH centers, functional teams that provide joint support from both CAMHS and AMHS operators) (Hartmann et al., 2019; McGorry et al., 2024).

Compared to the few published data on this topic, our baseline prevalence rate of past specialist contact is positioned in the highest values observed in other comparable studies. In this regard, some authors reported rates of psychiatric antecedents ranging from 25 to 37% in samples at high genetic risk for psychosis (Hans et al., 2000), while others described a prevalence of about 50% for mood and neurodevelopmental disorders during prodromal phase of psychosis (Meyer et al., 2005; Rosen et al., 2006; Mazzoni et al., 2009). These epidemiological discrepancies could be due to differences in the mean age at presentation and severity levels of clinical features. In this respect, our CHR-P sample was older compared to those recruited in previous research and did not exclusively include minors. However, overall considered, our findings confirm the importance of carefully monitoring help-seeking behavior in youths manifested in their early 20, especially in terms of psychosis prevention (Scott et al., 2024).

In this examination, the main psychiatric antecedents before entering the PARMS program in the CHR-P + subsample were anxious-depressive disorders (37.8%), learning disorder (27.4%), and conduct disorder (22.1%, including hyperkinetic disorders), with the last two diagnoses formulated within a previous contact with CAMHS. Our high rates of past anxiety and conduct disorders are in line with what was reported in the case series described by Mazzoni and co-workers (2009), who identified separation anxiety and oppositional/defiant disorder as main childhood-onset prodromes to psychosis, together with a high proportion of neurodevelopmental disorders (especially ADHD). In this respect, only two of our CHR-P participants had ADHD as specific previous diagnosis, and one had an autism spectrum disorder. Therefore, we could not examine differences between these specific psychiatric antecedents because this would likely have compromised the statistical power of our analyses. Future studies exploring these specific diagnostic entities in larger CHR-P populations are recommended.

In this investigation, we also notably found a high prevalence of previous learning disturbances. Therefore, also these diagnoses could be considered as specific antecedents of psychotic risk to be monitored in the developmental age. In this regard, some researchers formerly conceptualized psychosis also as a learning disorder, specifically as a disease generated by alterations in cognition systems (Ivleva et al., 2012; Tamminga, 2013). Indeed, it was reported that the hippocampus is altered in patients with early psychosis, with associated impairment in associative memory and learning processes (McHugo et al., 2019). Therefore, professionals working with children manifesting learning impairment should be made aware of this potential clinical long-term connection and monitor these individuals over time, also in terms of psychosis prevention.As for clinical characteristics at presentation, our CHR-P + participants showed longer DUPrS compared to CHR-P- peers. This further suggests the crucial connection of psychological suffering between past and current mental disorder and how care discontinuity may represent a critical delay in early intervention in psychosis, especially in the transition from adolescence to adulthood. Moreover, DUPrS extension represents one of the most negative prognostic factors for EIP treatment outcomes, critically prolonging the patient’s suffering and reducing the chance of effectiveness of later specialized EIP interventions. In other words, longer DUPrS has been demonstrated to be a crucial predictor related to poorer prognosis in early psychosis (Murden et al., 2024) and implementing EIP services capable of decreasing it represents one of the most important objectives to promote favorable long-term outcomes (Salazar de Pablo et al., 2024).

In this investigation, having psychiatric antecedents was associated to a higher 2-year incidence rate of psychosis transition. This finding supports the role of previous mental disorder as adverse prognostic factor, potentially favoring conversion to psychosis. This negative prognostic meaning is also confirmed by our longitudinal evidence related to higher social problems (as measure with the HoNOS) and HoNOS total scores in the CHR-P + subgroup compared to CHR-P- subjects across the whole follow-up period. Specifically, the HoNOS results further suggest that CHR-P with past contact with mental healthcare services are more likely to presents poorer long-term outcome levels and to manifest prolonged and more severe social functioning over time, maybe as consequence of a putative less effective response to our specialized EIP treatments.

However, the presence of psychiatric antecedents was also associated with a lower 2-year incidence rate of service disengagement. In this perspective, the presence of psychiatric antecedents could be considered as an index of greater psychopathological severity and more severe clinical trajectories by mental health professional working within our EIP service. This may have led to stronger retention in care and more careful monitoring of these CHR-P + subgroup.

The findings of this study showed that most (55.9%) of CHR-P participants showed psychiatric antecedents and a past contact with mental healthcare services. Specifically, 53.7% of these diagnoses (i.e., learning disorder, conduct disorder, and anorexia nervosa) were formulated within CARMS. Moreover, in 85.2% of these cases, we documented a care discontinuity. Greater effort should thus be made to identify psychiatric antecedents of psychosis in both YMH services and primary care, as well as to favor and implement care continuity during child-adult transition, and to better manage the interfaces among specialist mental healthcare services, also in order to reduce the DUPrS. Indeed, CHR-P individuals with past contact showed to have higher risk of psychosis transition and poorer functioning and outcome scores over time. In this sense, previous specialist contact may be considered as a predictor of unfavorable prognosis and poor response to treatment.

Lorenzo Pelizza and Fabio Catalano: conceptualization and study design; Lorenzo Pelizza, Fabio Catalano, and Simona Pupo: literature search; Lorenzo Pelizza, Emanuela Leuci, and Emanuela Quattrone: data collection and curation; Lorenzo Pelizza and Fabio Catalano: Formal analysis; Lorenzo Pelizza and Fabio Catalano: writing – original draft; all authors: writing – review & editing.