Introduction
Early intervention programs for psychosis have been developed within the last 30 years. Intensive and specialized intervention in the early phase of psychosis improves symptomatic and functional outcomes, increases satisfaction and engagement in care, reduces the risk of suicide and relapse, and is superior to treatment as usual in various outcomes (Birchwood, Todd, & Jackson, Reference Birchwood, Todd and Jackson1998; Correll et al., Reference Correll, Galling, Pawar, Krivko, Bonetto, Ruggeri and Kane2018; Fusar-Poli, McGorry, & Kane, Reference Fusar-Poli, McGorry and Kane2017; Galletly et al., Reference Galletly, Castle, Dark, Humberstone, Jablensky, Killackey and Tran2016). Although specialized and integrated early intervention programs improve care in psychosis globally, there are still some challenges to resolve, such as further improving relapse prevention in the most difficult to treat patients, and the identification of patient sub-groups with specific treatment needs, in order to adapt early intervention strategies (Conus & McGorry, Reference Conus and McGorry2002; Dempster, Li, Sabesan, Norman, & Palaniyappan, Reference Dempster, Li, Sabesan, Norman and Palaniyappan2021; Ramain, Conus, & Golay, Reference Ramain, Conus and Golay2021b). In this regard, mood symptoms may prove crucial in improving our understanding of the early course of first-episode psychosis (FEP) as they are notably associated with poor outcome and could help differentiate sub-groups of FEP patients with specific needs (Arrasate et al., Reference Arrasate, Gonzalez-Ortega, Alberich, Gutierrez, Martinez-Cengotitabengoa, Mosquera and Gonzalez-Pinto2014; Bebbington, Reference Bebbington2015; Ramain et al., Reference Ramain, Conus and Golay2021b; Ramain, Conus, & Golay, Reference Ramain, Conus and Golay2022).
Mood symptoms are common in psychosis. The prevalence of depressive symptoms ranges between 14.15 and 44.80% after FEP. In addition, around 30% of FEP patients suffer from an affective psychosis (Coentre, Talina, Góis, & Figueira, Reference Coentre, Talina, Góis and Figueira2017; Kennedy et al., Reference Kennedy, Everitt, Boydell, Van Os, Jones and Murray2005; Ramain, Conus, & Golay, Reference Ramain, Conus and Golay2021a). Studies on early psychosis have highlighted the burden of manic and depressive dimensions. Indeed, patients with manic symptoms may hardly fully recover after a first episode (Conus et al., Reference Conus, Cotton, Abdel-Baki, Lambert, Berk and McGorry2006; Conus & McGorry, Reference Conus and McGorry2002; Marwaha et al., Reference Marwaha, Hett, Johnson, Fowler, Hodgekins, Freemantle and Birchwood2021), and depressive symptoms are associated with long-term functional impairment (Alameda et al., Reference Alameda, Golay, Baumann, Progin, Mebdouhi, Elowe and Conus2017; McGinty & Upthegrove, Reference McGinty and Upthegrove2020). Despite the burden of mood symptoms in early psychosis, their course over time and predictive values remain understudied (Arrasate et al., Reference Arrasate, Gonzalez-Ortega, Alberich, Gutierrez, Martinez-Cengotitabengoa, Mosquera and Gonzalez-Pinto2014; Ciompi, Reference Ciompi2015).
Furthermore, literature on early psychosis investigated depressive and manic symptoms independently, with a large focus on first-episode mania (Arrasate et al., Reference Arrasate, Gonzalez-Ortega, Alberich, Gutierrez, Martinez-Cengotitabengoa, Mosquera and Gonzalez-Pinto2014; Conus & McGorry, Reference Conus and McGorry2002; Ratheesh et al., Reference Ratheesh, Davey, Daglas, Macneil, Hasty, Filia and Cotton2017). Nevertheless, the possible alternation of manic and depressive episodes in sub-groups of psychosis suggests that manic and depressive dimensions can interact (Strakowski et al., Reference Strakowski, Keck, McElroy, West, Sax, Hawkins and Bourne1998). Depressive and manic symptoms may also co-occur, leading to poor quality of life and functioning (Bauer, Simon, Ludman, & Unützer, Reference Bauer, Simon, Ludman and Unützer2005). In sum, despite manic and depressive dimensions, and their possible co-occurrence, remain understudied in FEP, previous literature suggests that manic and depressive symptoms may be determinant in the course of early psychosis. There is a subsequent need for studies investigating the co-occurrence of mood dimensions and their impact in FEP.
Consequently, the aims of this study were to investigate (1) the co-occurrence of manic and depressive dimensions defining mood profiles in the early course of psychosis, (2) their trajectories over three years of treatment in early psychosis and, (3) their impact on functional and symptomatic outcomes in an early intervention in psychosis program.
Method
Sample and procedure
This is a prospective study on a cohort of FEP patients treated at the Treatment and Early Intervention in Psychosis Program (TIPP) that has been implemented by Lausanne University Hospital's Department of Psychiatry in 2004 (Baumann et al., Reference Baumann, Crespi, Marion-Veyron, Solida, Thonney, Favrod and Conus2013; Conus & Bonsack, Reference Conus and Bonsack2004). Patients entering the program are aged 18–35, reside in the Lausanne catchment area and have crossed the psychosis threshold according to the Comprehensive Assessment of At-Risk Mental States scale (CAARMS; Yung et al., Reference Yung, Yuen, McGorry, Phillips, Kelly, Dell'Olio and Buckby2005) based on the Psychosis Threshold subscale. Patients are referred to other programs if they have been on antipsychotic medication for more than six months, have an intoxication-induced or organic brain disease-induced psychosis, or have an intelligence quotient below 70. The TIPP favors a bio-psycho-social approach and provides 3 yearlong psychiatric and case management follow-up including psychotherapy, psychoeducation, family support and therapy, cognitive assessment and remediation, social support, supported employment, psychological interventions for cannabis use, and pharmacological treatment. In line with international guidelines, atypical antipsychotics are a first-line pharmacological treatment used in order to minimize side effects (Baumann et al., Reference Baumann, Crespi, Marion-Veyron, Solida, Thonney, Favrod and Conus2013). Case managers fill out a specifically designed questionnaire for the TIPP with every patient. This includes items on socio-demographic characteristics, medical history, exposure to traumatic life events, psychopathology and usual functioning. Follow-up assessments are carried out at 2, 6, 12, 18, 24, 30 and 36 months, by a psychologist and a case manager, to explore various aspects of treatment: changes in psychopathology and treatment, functional status, insight and treatment adherence, as well as intermittent exposure to trauma, suicide attempts, forensic events and substance use. The study was approved by the Human Research Ethics Committee of the Canton of Vaud (protocol #2020-00272). The data generated during follow-up were only used if patients provided written informed consent; only five of them refused to have their clinical data used for research, yielding a highly representative sample of early psychosis patients.
Diagnostic assessment
The diagnoses result from an expert consensus built from discussions held at 18 and 36 months, based on the DSM-IV criteria and using information from patients' medical records provided by their treating psychiatrists and their case manager. We used the latest consensus diagnosis available.
Depressive and manic symptoms
Depressive and manic symptoms were assessed at 2, 6, 12, 18, 24, 30, 36 months follow-up. We measured the severity of depressive symptoms using the overall score of the Montgomery-Asberg Depression Rating Scale (MADRS; Montgomery & Asberg, Reference Montgomery and Asberg1979), and manic symptoms with the overall score of the Young Mania Rating scale (YMRS; Young, Biggs, Ziegler, & Meyer, Reference Young, Biggs, Ziegler and Meyer1978).
Outcomes at discharge
We considered scores at 36 months follow-up or 30 if not available, for the outcome measures at discharge. We used 8 items of the PANSS (delusion, unusual thought content, hallucinatory behavior, conceptual disorganization, mannerisms, blunted affect, social withdrawal, lack of spontaneity; Andreasen et al., Reference Andreasen, Carpenter, Kane, Lasser, Marder and Weinberger2005)) following Andreasen's Criteria (score ⩽3) to determine symptomatic recovery. A PAS score equal or lower to the premorbid rating on four of the five PAS general scale's items defined functional recovery (Strakowski et al., Reference Strakowski, Keck, McElroy, West, Sax, Hawkins and Bourne1998). The assessment of independent living recovery (head of household/living alone, with partner, or with peers/living with family with minimal supervision) was carried out using the Modified Vocational Status Index and working recovery (paid or unpaid full- or part-time employment/being an active student in school or university/head of household with employed partner (homemaker/full or part-time volunteer) using the Modified Location Code Index Independent living (Tohen et al., Reference Tohen, Hennen, Zarate, Baldessarini, Strakowski, Stoll and Cohen2000). Insight recovery was defined as full insight at discharge. We assessed quality of life at discharge with the World Health Organization Quality Of Life scale (‘The World Health Organization Quality of Life assessment (WHOQOL): position paper from the World Health Organization,’, 1995). It measures satisfaction with life and self-esteem through 26 self-rated items with 5-point Likert scales ranging from 1 (low satisfaction) to 5 (high satisfaction).
Statistical analysis
First, we performed a latent profile analysis (LPA) for the beginning, middle and end of the program. Depressive and manic symptoms, respectively MADRS and YMRS scores, were used as indicators. To guarantee model statistical identification and deal with missing data, each LPA was estimated using depressive and manic symptom scores of two neighboring assessment (2 and 6 months, 12 and 18 months, 30 and 36 months) with measurement invariance imposed across the two depressive symptom respectively the two manic symptom indicators. The best solution was determined using the Bayesian information criterion (BIC) coefficient, which balances model fit and model complexity, i.e. the number of parameters (Schwarz, Reference Schwarz1978).
Second, we performed a latent transition analysis (LTA). Given the meaning of the classes was very close between each three LPA, we also imposed longitudinal measurement invariance in the LTA model. While reducing the total number of model parameters, it also ensured the meaning of the classes would stay the same between the beginning, middle and the end of the program, thus facilitating interpretation.
Finally, in order to compare scores from the different classes at discharge, we used a Bayesian model comparison approach. This represents an elegant alternative to the classic problem of multiple comparisons and enables evaluations to support the null hypothesis (Golay et al., Reference Golay, Laloyaux, Moga, Della Libera, Larøi and Bonsack2020; Golay, Morandi, Silva, Devas, & Bonsack, Reference Golay, Morandi, Silva, Devas and Bonsack2019; Noël, Reference Noël2015). The first model was the homogeneous model (1, 2, 3, 4), stating that the four groups did not differ and were issued from the same distribution. It corresponds to the null hypothesis in the classical statistical testing framework. Another model was the heterogeneous model: (1) (2) (3) (4) (i.e. all the groups were different from each other and were issued from four different distributions). All other possible combinations, which adds up to 15 – for instance (1, 2, 3), (4) or (1, 3), (2, 4) – were also estimated. For continuous variables, the best possible Gaussian model (μ, σ2) was determined by use of the BIC (Schwarz, Reference Schwarz1978). For nominal variables, the best multinomial model was determined using the exact likelihood with a uniform prior on all parameters (Noël, Reference Noël2015). An equal prior probability of 1/15 was assumed for all models so that no model was favored. The Bayes factor was also computed (Kass & Raftery, Reference Kass and Raftery1995) and provided a comparison between the best model and the homogenous model. A Bayes factor of 4 indicates that the best model was 4 times more likely to be true than the homogenous model. Values over 3 are generally considered sufficiently important to favor one model over another (Jeffreys, Reference Jeffreys1961; Wagenmakers, Wetzels, Borsboom, & Van Der Maas, Reference Wagenmakers, Wetzels, Borsboom and Van Der Maas2011).
All statistical analyses were performed using the Mplus statistical package, version 8.3, IBM SPSS, version 26, the AtelieR package for R (Noël, Reference Noël2013) and the Bayes R2STATS group models calculator (Noël, Reference Noël2018).
Results
Patient sample
The sample consisted of 313 patients (Mean age = 24.8; s.d. = 4.81), and included a majority of males (65.5%). Among these patients, 56.9% met diagnostic criteria for schizophrenia, 12.5% for schizophreniform or brief psychotic disorder, 9.9% % for schizoaffective disorder, 8.0% for bipolar disorder with psychotic features, 3.2% for major depressive disorder with psychotic features, and 9.6% for other psychotic disorders.
Characteristics of the different LPA solutions (Table 1)
A six-class solution was chosen for the first LPA (2–6 months corresponding to the beginning of the program) based on its lowest BIC and clinical interpretability. The first class consisted of patients with moderate depression (MADRS score 20–34). The second class consisted of patients without any mood disturbance (MADRS and YMRS scores <12). The third class consisted of patients with hypomania (YMRS score 12–19). The fourth class consisted of patients with co-occurring manic and depressive symptoms (MADRS score 20–34 and YMRS score 12–19). The fifth class consisted of patients with severe depression (MADRS score >33) and the sixth class with mania (YMRS score >20). For the LPA performed in the middle of the program (12–18 months), BIC indicated a six-class solution similar to the classes observed in the first LPA. The LPA performed at the end of the program (30–36 month) revealed a four-class solution with a similar interpretation, but no class including patients with severe depression or mania.
Table 1. Characteristics of the three latent class analysis solutions
Note. BIC, Bayesian information criterion; a, one class is empty. The best class was determined on the basis of the lowest BIC coefficient and is indicated in bold.
Depressive and manic dimension trajectories over the 36 months follow-up (Table 2)
Our results suggested that 38.3% of patients had not any mood disturbance from the beginning to the end of the program. Among patients who had not any mood disturbance at the end, 15.3% of patients had moderate depression, 8.6% had hypomania, 2.2% had severe depression, and 0.9% had severe mania at the beginning. 12.8% of patients remained moderately depressed from the beginning to the end of the program, while 4.8% transited from moderate depression to hypomania. Moreover, among patients who transited to hypomania at the end, 1.9% had co-occurring manic and depressive symptoms, 1.6% had not any mood disturbance, and 0.6% had severe depression at the beginning. 0.9% of patients had severe depression at the beginning but co-occurring manic and depressive symptoms at the end. 0.6% of patients had not any mood disturbance at the beginning but were moderately depressed at the end.
Table 2. Depressive and manic dimension trajectories over the 36 months follow-up (N = 313)
Note. The trajectories are presented according to their frequency of occurrence in the cohort.
Transition matrices between the beginning and the end of the program (Table 3)
All patients with mania at the beginning of the program had not any mood disturbance at the end of the program. A majority of patients (58.7%) with severe depression at the beginning of the program had not any mood disturbance at the end, but some of them developed hypomanic symptoms (16.7%), and others co-occurring manic and depressive symptoms (25%). Nearly all patients without any mood disturbance at the beginning of the program did not display mood symptoms at the end, only few of them had a moderate depression (1.6%) or hypomania (3.9%). While nearly half of patients with moderate depression at the beginning of the program had not any mood disturbance at the end, 38.8% of them remained moderately depressed and 14.6% presented hypomanic symptoms. Despite 64.3% of patients with hypomania at the beginning of the program had not any mood disturbance at the end, 35.7% of them remained hypomanic at the end. Three-quarters of patients with co-occurring manic and depressive symptoms at the beginning of the program had a moderate depression at the end of the program, the 23.1% remaining had hypomanic symptoms at the end.
Table 3. Transition matrixes between the beginning and the end of the program
Note. Cells in bold indicates stability.
Comparison of outcomes between groups of mood profiles at discharge (Table 4)
Globally, patients without any mood disturbance at discharge had a better outcome at the end of the program. Indeed, patients without any mood disturbance had better symptomatic and functional recovery, a better working recovery, as well as better quality of environment and social relationships than the other groups. Patients with hypomania at discharge had poorer insight recovery than the other groups. Patients with moderate depression at discharge had a lower rate of return to premorbid functional level. Patients with moderate depression or co-occurring manic and depressive symptoms had a poorer quality of physical and psychological health than patients without any mood disturbance or hypomania. All groups had a similar independent living recovery.
Table 4. Comparison of outcomes between groups of mood profiles at discharge
Note.
a Based on BIC coefficient.
b Bayes factor comparing the best model to the homogeneous model (1, 2, 3, 4).
c Compared to all possible models.
Discussion
In this study, we investigated the co-occurrence of mood dimensions and their impact on symptomatic and functional outcomes in a cohort of FEP patients. In this regard, we identified latent sub-groups based on depressive and manic dimensions and explored their evolution and outcomes at the end of the program. Results of LTA revealed six mood profiles at the beginning of the program: a subgroup of patients without any mood disturbance, one with moderate depressive symptoms, one with co-occurring manic and depressive symptoms, one with hypomanic symptoms, one with severe depressive symptoms, and one with severe manic symptoms. These profiles evolved at the end of the program into four different profiles: one without any mood disturbance, one with moderate depressive symptoms, one with co-occurring manic and depressive symptoms, and one with hypomanic symptoms. These results suggested that manic and depressive dimensions can co-occur at a moderate level (Co-occurrence profile), but not at a severe level (severe depressive or manic profile). They also showed that mood dimensions have an important impact in the early course of psychosis. Indeed, the subgroup without any mood disturbance at discharge had better outcomes, especially regarding psychotic symptomatic recovery, functional and working recovery, as well as quality of environment and social relationship at discharge.
In line with previous literature, our results suggested that either depressive or manic symptoms worsen prognosis in early psychosis (Bauer et al., Reference Bauer, Simon, Ludman and Unützer2005; Ciompi, Reference Ciompi2015; Conus et al., Reference Conus, Cotton, Abdel-Baki, Lambert, Berk and McGorry2006; Marwaha et al., Reference Marwaha, Hett, Johnson, Fowler, Hodgekins, Freemantle and Birchwood2021; Morrissette & Stahl, Reference Morrissette and Stahl2011). In addition, our results showed that 94.5% of patients without any mood disturbance at the beginning of the program had not any mood disturbance at the end, suggesting a stability of the absence of mood disturbance profile: on this basis, it seems possible to identify FEP patients with a risk of enduring mood symptoms early in treatment. Reciprocally, our results suggest that identification of mood symptoms at the beginning of treatment justifies their treatment and may even deserve the development of specific interventions.
However, treating mood dimensions at the beginning of early intervention program would require assessing and adjusting treatment of mood dimensions accurately, as previously suggested (Lambert, Conus, Lambert, & McGorry, Reference Lambert, Conus, Lambert and McGorry2003; Ramain et al., Reference Ramain, Conus and Golay2021a). Unfortunately, our results also revealed that, while all patients with manic symptoms, and a majority of patients with hypomanic (64.3%) or severe depressive symptoms (58.3%) at the beginning of the program did not have any mood disturbance at the end, none of the patients with co-occurring manic and depressive symptoms at the beginning were free of mood disturbance after 3 years. Indeed, they had moderate depressive (76.9%) or hypomanic (23.1%) symptoms at discharge. These results suggested that pure manic/hypomanic or severe depressive forms, contrary to mild mood symptoms, are probably well identified, and appropriately treated leading to the resolution of mood disturbance at the end of treatment. However, results also showed that all patients with co-occurrence profile at the beginning remained symptomatic at the end, suggesting the need to improve assessment and specific treatment for patients with co-occurring manic and depressive dimensions, even under threshold of a florid manic or severe depressive episode (Bauer et al., Reference Bauer, Simon, Ludman and Unützer2005; Berk, Reference Berk2007; Marneros, Röttig, Wenzel, Blöink, & Brieger, Reference Marneros, Röttig, Wenzel, Blöink and Brieger2004; Ramain et al., Reference Ramain, Conus and Golay2021a). Further studies on FEP investigating depressive and manic dimensions together, rather than independently, are also required.
Moreover, our results showed that patients with mild depressive symptoms at discharge had a lower rate of return to premorbid functional level than the other subgroups. They also showed that displaying a depressive component was associated with poorer quality of physical and psychological health. These results may confirm previous findings that a depressive dimension, even under depression diagnostic threshold, has a negative impact on outcome (Alameda et al., Reference Alameda, Golay, Baumann, Progin, Mebdouhi, Elowe and Conus2017; Fiedorowicz et al., Reference Fiedorowicz, Persons, Assari, Ostacher, Goes, Nurnberger and Coryell2021). It also supports the idea that some patients develop a poorer outcome via an affective pathway (Alameda et al., Reference Alameda, Rodriguez, Carr, Aas, Trotta, Marino and Spinazzola2020; Alameda, Conus, Ramain, Solida, & Golay, Reference Alameda, Conus, Ramain, Solida and Golay2022; van Os et al., Reference van Os, Pries, Ten Have, de Graaf, van Dorsselaer, Delespaul and Guloksuz2020). Contrary to previous studies (Alameda et al., Reference Alameda, Conus, Ramain, Solida and Golay2022; Calderon-Mediavilla et al., Reference Calderon-Mediavilla, Vila-Badia, Dolz, Butjosa, Barajas, Del Cacho and Ochoa2021), we did not observe a significant association between depressive symptoms and psychotic symptomatology. However, to our knowledge, this is the first study that investigated the co-occurrence of manic and depressive dimensions in early psychosis, considering the impact of the relative expression of both dimensions at the same time, rather than separately.
Our results should be considered with some degree of caution considering several limitations. First, due to too many missing data, we were not able to include analysis of medication. Second, the prospective realistic design of the follow-up did not allow to control for all the confounding factors possibly explaining for the appearance of mood disturbance and its variation (e.g. trauma, cannabis use). Third, the methodology used did not allow us to differentiate outcome of patients without any mood disturbance since the beginning of the program, from those of patients remitted from a mood disorder at the end. Fourth, because the size of several sub-groups at the beginning of the program was very small, we were not able to include robust results about the prediction of outcomes at discharge based on latent class profiles of the beginning of the program.
Conclusion
This study investigated the co-occurrence of manic and depressive dimensions in FEP and their impact on outcome. Based on transition LTA, we identified subgroups of patients with different mood profiles considering the relative expression of manic and depressive symptoms. Our results showed that the subgroup of FEP patients without any mood disturbance at the end of the program had better outcomes. They also revealed that the co-occurrence of manic and depressive dimension, as well as a depressive component under threshold, may require specific treatment adjustment in early intervention. Further studies investigating both manic and depressive dimensions together in FEP are required.
Acknowledgements
We would like to thank the case-managers of the TIPP program for collecting data over years.
Financial support
This study was supported by the Swiss National Science Foundation (320030_122419 to Philippe Conus), by the FNS ‘SYNAPSY—The Synaptic Bases of Mental Diseases’ (#320030 158776) and institutional funding.
Conflict of interest
The authors declare that there is no conflict of interest.
