Introduction

Schizophrenia spectrum disorders (SSD) is a severe mental health condition characterized by hallucinations, delusions, and disorganized thinking [1]. Biological sex-specific differences in psychosis have been consistently observed across clinical presentation, course of illness, and treatment response [2–4]. Males with schizophrenia tend to experience earlier onset, more negative symptoms, fewer affective symptoms, worse psychosocial functioning, including more social isolation, and higher rates of substance use. Conversely, females often present with more affective symptoms, such as depression and anxiety, resulting in higher rates of schizoaffective diagnoses, and generally demonstrate better premorbid and social functioning [2,5]. However, regarding cognitive and psychosocial functioning, sex differences are complex and heterogeneous: females have been shown to exhibit lower performance across several cognitive domains, including current and premorbid intellectual functioning, working memory, and non-verbal abstract reasoning, while demonstrating preserved or superior abilities in processing speed and verbal learning [6]. These differences are often particularly pronounced during the early stages of the disorder [3]. Consistent with these findings, a recent meta-analysis of first-episode psychosis reported that males exhibited more pronounced negative symptoms, whereas females had higher functioning and more depressive symptoms [4]. Such sex-related differences in clinical presentation may contribute to distinct recovery trajectories in males and females.

Beyond clinical presentation, sex differences also extend to treatment response. With respect to pharmacological interventions, females generally respond better to antipsychotics, requiring lower doses and experiencing fewer relapses, although they are more susceptible to side effects like hyperprolactinemia and weight gain [3,7]. Psychological interventions, including cognitive-behavioral and metacognitive therapies, have likewise been shown to display sex-related differences in effectiveness. For instance, a recent study found that females with first-episode psychosis (FEP) showed higher rates of functional remission and fewer hospitalizations over time compared to males, suggesting potential sex-related variability in response to early psychological treatments [8]. Similarly, some reports suggest females treated in early intervention services for psychosis were more likely to achieve symptom remission and better functioning than males after two years of treatment, responded more favourably to minimal antipsychotic dosages, and showed higher recovery rates during the first three years of early intervention, including psychological support [9,10]. Although premorbid functioning and other risk factors partially account for these differences, they nonetheless suggest the presence of sex-specific outcomes in psychological care [11]. Moreover, sex differences in predictors of psychological treatment response, such as cognitive insight, may underlie differential effectiveness of therapies including Cognitive Behavioural Therapy (CBT) and Metacognitive Training (MCT) [12].

MCT, a psychological intervention derived from CBT, and originally developed by Moritz and Woodward [13], is a promising intervention for individuals with psychotic disorders, particularly schizophrenia. Designed to target cognitive biases, MCT aims to reduce symptom severity without directly confronting delusional content [14]. Importantly, MCT uses normalizing language and examples to encourage participants to discuss and think about their symptoms and experiences with psychosis. Its efficacy has been demonstrated across various clinical settings, with improvements observed in positive, negative, and general symptoms of psychosis [15] and it is strongly recommended (WFSBP-grade 1) to improve positive symptoms [16]. Despite growing evidence supporting MCT’s effectiveness, relatively few studies have examined whether treatment outcomes differ by biological sex. To date, only one study has explicitly investigated sex differences in response to MCT [12]. This study found that females in the MCT group showed greater improvements in general symptoms, cognitive insight, and reduced personalizing bias and dependency-related beliefs, while males improved more in frustration tolerance and affective reasoning tasks, suggesting the existence of sex-specific mechanisms underlying response to MCT. However, this study was limited to a sample of participants with first episode psychosis. It therefore remains unclear whether these sex-related differences, particularly in symptom outcomes, generalize across MCT formats and different stages of the disorder.

The current study addresses this gap by analyzing harmonized retrospective data from the PERMEPSY project (www.permepsy.org), which compiled pre- and post-treatment assessments from international studies involving different versions of MCT on participants at various stages of the disease. Using psychotic symptoms as the primary outcome measure, we investigated whether biological sex moderates symptom trajectories following MCT. By leveraging a large, harmonized dataset and standardized analytic procedures, this study provides a robust examination of sex-specific treatment effects in psychosis. Specifically, the aims of this study are to provide further evidence of the effect of MCT at reducing psychotic symptoms and explore whether the recovery trajectory is sex-specific. We hypothesized a reduction of psychotic symptoms in both males and females, with a more pronounced reduction in general symptoms among females, in line with the findings reported by Salas-Sender et al. [12].

Methods

Results

A total of 573 individuals with SSD were included in the analysis. Participant age ranged from 16 to 68 years (M = 34.7, SD = 12.3), and participants had completed an average of 11.5 years of education (SD = 3.08). Sex distribution comprised 358 males (62.5%) and 215 females (37.5%). Living situation analysis revealed that most participants (81.8%) lived with family members or parents, while a small minority (3.6%) resided alone. Results indicated that 40.9% of patients were inactive or unemployed, 19.0% were actively employed, and 15.9% on temporary sick leave. Length of illness had a mean duration of 5.28 years (SD = 8.14, median = 2.0 years). Diagnostic distribution indicated that most participants were diagnosed with schizophrenia, accounting for 56.0% of the sample, while 9.2% were diagnosed with schizoaffective disorder, 8.9% with first episode psychosis, and 25.9% with other psychotic spectrum disorders Baseline comparisons between male and female participants revealed significant differences in age (p < .001), positive symptoms (p = .014), negative symptoms (p = .006), and diagnostic distribution (χ² = 21.24, p = .002, V = .19), with males being younger and presenting with more severe positive and negative symptoms at baseline. No significant differences were found for education, illness duration, or general psychopathology (all p > .10). Full baseline characteristics by sex are presented in Table 1. The distribution of PANSS scores at pre- and post-treatment is presented in Table 2.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Sociodemographic and clinical characteristics of the retrospective sample.

https://doi.org/10.1371/journal.pone.0356632.t001

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Descriptives of sample’s psychotic symptom distribution (pre- and post-treatment).

https://doi.org/10.1371/journal.pone.0356632.t002

Analysis of PANSS_T scores revealed a significant main effect of Time, indicating a reduction in overall symptom severity following treatment (F(1, 395) = 139.169, p < .001, η²p = .261, 90% CI [.202,.318]). The main effect of sex approached, but did not reach, statistical significance (F(1, 395) = 3.60, p = .058, η²p = .009, 90% CI [.000,.031]). The time × sex interaction was not significant (F(1, 395) = 0.514, p = .474, η²p = .001, 90% CI [.000,.014]), suggesting similar therapeutic gains for both male and female participants.

For positive symptoms (PANSS_P), a significant main effect of Time was observed (F(1, 476) = 170.347, p < .001, η²p = .264, 90% CI [.210,.316]), reflecting reduced positive symptom severity following MCT. The main effect of Sex showed a trend toward significance (F(1, 476) = 3.24, p = .073, η²p = .007, 90% CI [.000,.024]), but did not reach the conventional threshold for statistical significance. The time × sex interaction was not significant (F(1, 476) = 0.323, p = .570, η²p = .001, 90% CI [.000,.010]), indicating equivalent treatment-related changes across sexes.

Analysis of negative symptoms (PANSS_N) revealed a significant main effect of Time (F(1, 424) = 27.205, p < .001, η²p = .060, 90% CI [.029,.101]), indicating a reduction in negative symptom severity following treatment. A significant main effect of Sex was also observed (F(1, 424) = 7.28, p = .007, η² = .017, 90% CI [.003,.043]), with females exhibiting lower levels of negative symptoms overall. The Time × Sex interaction was not significant (F(1, 424) = 0.166, p = .684, η²p = .000, 90% CI [.000,.009]), suggesting that the magnitude of symptom change did not differ between males and females.

For general psychopathology (PANSS-G), a significant main effect of Time was found (F(1, 397) = 139.43, p < .001, η²p = .260, 90% CI [.202,.317]), reflecting improvement in general symptom severity following treatment. Neither the main effect of Sex (F(1, 397) = 1.21, p = .273, η²p = .003, 90% CI [.000,.019]) nor the Time × Sex interaction (F(1, 397) = 1.02, p = .313, η²p = .003, 90% CI [.000,.017]) reached statistical significance, indicating similar treatment responses across sexes.

Results of the pre-specified FEP sensitivity analysis are presented in Supplementary Table S3 in S1 File. Significant Time effects were observed for positive symptoms, negative symptoms, and total score, with general symptoms approaching significance. No significant main effect of Sex and no significant Time × Sex interaction were detected for any PANSS subscale, consistent with the primary analyses. Given the small subsample size, these results should be interpreted as exploratory.

In summary, the results demonstrated that significant reductions in psychotic symptomatology were observed across all measured domains following MCT. Females tended to exhibit lower levels of negative and overall symptoms. Both male and female participants experienced comparable reductions in positive, negative, general, and total psychotic symptoms following MCT, with sex showing no moderating effect on treatment outcomes.

Discussion

This study aimed to investigate whether biological sex moderates symptom trajectories following Metacognitive Training (MCT) in individuals with SSD, using harmonized retrospective data from the PERMEPSY project. Overall, the findings are consistent with prior evidence of MCT-associated reductions in psychotic symptoms across positive, negative, general, and total psychotic symptoms domains, with significant symptom changes observed following treatment in both male and female participants. Biological sex did not significantly moderate treatment outcomes, suggesting that the therapeutic benefits of MCT are broadly comparable across sexes.

The significant main effect of Time across all PANSS subscales is consistent with previous literature demonstrating the efficacy of MCT in reducing psychotic symptoms severity without directly confronting delusional content [15,24]. While it cannot be ruled out that other factors, such as concurrent treatments or regression to the mean, accounted for some of the observed symptom changes, findings are in line with prior evidence supporting the role of MCT in symptom improvement. Importantly, the absence of significant Time × Sex interactions suggests that both males and females experienced comparable symptom changes from MCT across symptom domains, reinforcing its applicability as a broadly effective psychological intervention.

Interestingly, despite the absence of Time × Sex interaction for any PANSS subscale, a significant main effect of Sex was observed for negative symptoms, with females exhibiting lower levels of negative symptoms overall. The main effect of Sex did not reach statistical significance for positive symptoms, general psychopathology or total score. Baseline comparisons confirmed that males presented with significantly higher positive and negative symptom severity and a distinct diagnostic distribution compared to females (see Table 1), consistent with well-documented sex differences in clinical presentation [2,5]. Importantly, when expressed as standardized mean differences (Cohen’s d_z) comparable to those reported in the MCT literature, the magnitude of symptom change for positive symptoms (d_z = 0.60) was broadly consistent with, albeit modestly higher than, meta-analytic estimates for positive symptoms (g = 0.50, [24]; g = 0.47, [15]), and the magnitude for negative symptoms (d_z = 0.25) closely matched meta-analytic estimates for negative symptoms (g = 0.23, [24]; g = 0.23, [15]. The effect for PANSS total score (d_z = 0.59) was somewhat larger than the pooled estimate reported for total psychotic symptoms by Meinhart et al. [15] (g = 0.39, 95% CI [0.25, 0.54]), although the respective confidence intervals showed marginal overlap. No directly comparable meta-analytic estimate for general psychopathology was identified in the MCT literature. Nonetheless, these group-level effect sizes should be interpreted with caution, particularly given the heterogeneous, retrospective nature of the sample. Future studies should complement group-level analyses with responder rates and clinically meaningful change indices to better capture individual-level clinical relevance. Furthermore, although statistically significant symptom reductions were observed, the absence of responder analyses and clinically meaningful change indicators prevents conclusions regarding the clinical importance of these changes at the individual patient level. Nonetheless, the consistency of symptom changes observed across domains and sexes is in line with the growing body of evidence supporting MCT as a transdiagnostic intervention for psychosis.

The lack of significant sex and time interactions contrasts with previous findings from Salas-Sender et al. [12], who reported greater sex-specific improvements for females in general symptoms within a FEP sample. This discrepancy may be explained by differences in sample composition, as the present study included participants at various stages of illness, with substantial variability in illness duration, age, and clinical presentation, as well as exposure to multiple MCT formats. Such heterogeneity may have obscured sex-specific effects that are more readily detectable in more homogeneous samples. Notably, the pre-specified sensitivity analysis restricted to FEP participants also yielded no significant Time × Sex interactions across any PANSS subscale. However, given the limited statistical power of this subgroup, these null findings should be interpreted with caution and do not preclude the existence of stage-specific sex effects. SSD onset and course differ by sex, with males more frequently experiencing onset in early adulthood, whereas females tend to show a later peak in onset during their late twenties, as well as a second peak around menopause, typically after the age of 40 [2,25,26]. Females with later-onset schizophrenia often exhibit fewer premorbid impairments compared to those with early-onset or compared to males [26,27], and generally demonstrate more favorable outcomes during early phases of psychosis [9,10]. However, longitudinal studies indicate that sex-related differences tend to diminish or reverse as illness progresses, with symptom profiles and functional outcomes converging in late-life schizophrenia [3,28]. Taken together, these findings suggest that sex may moderate the manifestation of psychosis over time, in a stage-dependant manner [29]. This dynamic may have limited our ability to detect sex differences in treatment response, as illness stage was not explicitly controlled for in the present analyses. Future research should therefore examine sex effects within more clearly defined clinical subgroups, such as individuals at early stages of illness or those receiving specific MCT formats.

Several limitations of the present study should be acknowledged. First, the retrospective nature of the dataset limits causal inference and introduces variability in assessment procedures. Concurrent treatments such as medication adjustments or other psychosocial interventions could not be controlled for and may have contributed to observed symptom changes. Second, missing data in sociodemographic variables and PANSS subscales may have affected statistical power and generalizability. Third, while the harmonization process ensured consistency across studies, residual heterogeneity in intervention delivery and clinical context may have influenced outcomes. Notably, the absence of study-of-origin identifiers and MCT format variables in the PERMEPSY harmonized database precluded the use of linear mixed-effects models with study as a random effect and sensitivity analyses stratified by MCT variant, meaning that potential clustering effects across the contributing studies could not be formally modelled [18,30]. This limitation reflects not only constraints of the final analytic dataset but also a limitation of the harmonization strategy itself. Future harmonization initiatives should preserve study-level identifiers and intervention-format information to enable more rigorous secondary analyses and better account for sources of between-study heterogeneity. Fourth, our analyses relied on the conventional PANSS subscales, which may not fully represent the five-factor structure originally proposed by Van der Gaag and colleagues [31] and later supported by Shafer and Dazzi [32]. These authors identified Positive, Negative, Disorganization, Affect, and Resistance dimensions that differ from the traditional subscale organization. As a result, the use of conventional subscales may have limited the sensitivity of the analyses to detect changes in specific symptom dimensions, potentially underrepresenting shifts in certain symptom dimensions within the five-factor model. Fifth, the observed baseline imbalances in age, diagnostic distribution, and symptom severity between males and females may have complicated the interpretation of longitudinal symptom trajectories. Differences in initial clinical characteristics could have increased variability in treatment-related change, potentially masking or diluting subtle Time × Sex interactions. Consequently, the absence of significant sex moderation effects should not be interpreted as definitive evidence that sex plays no role in treatment response, but rather as an indication that any such effects were not detectable within the context of the heterogeneous baseline profiles represented in the present sample. Despite these limitations, the study offers valuable insights into the generalizability of MCT across sexes. The findings suggest that symptom reductions following MCT are comparable in males and females, with no evidence of sex-specific symptom trajectories. Our results are consistent with another study from our research team [33], who found no moderating effect of sex on MCT efficacy on cognitive insight and the jumping to conclusions bias in heterogeneous samples, suggesting that sex differences may be more detectable in specific subgroups (e.g., FEP) or outcome domains rather than in overall symptom trajectories. On the other hand, the results contrast previous evidence suggesting sex differences may emerge more clearly in cognitive domains rather than in global symptom severity [34], since MCT was found to enhance specific neurocognitive functions in FEP, showing promising effects for women in verbal memory processes. Overall, these results support the integration of MCT into routine care for psychosis, while highlighting the need for further research into personalized approaches that consider sex and other individual differences.

It is also important to note that although this study examined biological sex (male/female) as recorded in clinical datasets, observed differences may partially reflect gendered experiences and socially constructed roles. Gender, as a social construct, influences help-seeking behavior, stigma, and therapeutic engagement, which may shape symptom presentation and treatment outcomes [35,36]. Social and cultural factors such as gender role expectations and help-seeking behaviors may influence therapeutic engagement and outcomes. For example, some studies suggest women with psychosis are more likely to seek treatment earlier and more frequently than men [37], whereas men may experience feelings of weakness or shame during early help-seeking attempts [38].

Moreover, stigma associated with psychosis appears to be both pervasive and gendered. Men may internalize stigma through societal norms emphasizing emotional suppression and self-reliance, potentially contributing to more pronounced negative symptoms, reduced emotional expressivity, and lower engagement in therapeutic processes [39,40]. Men are also often perceived as more dangerous or emotionally suppressed, which may exacerbate negative symptoms, social rejection and internalized stigma [41,42]. In contrast, women may face stigma related to perceived emotional instability or dependency, which can amplify affective symptoms and contribute to higher general psychopathology [41,42]. Public stigma tends to be particularly severe for individuals exhibiting positive symptoms, especially when presented by men, who are often perceived as more intentional or threatening [41]. These gendered stigma experiences may influence both baseline symptom profiles and perceived treatment gains. In terms of clinical importance, although previous studies have not found statistically significant differences in stigma improvement between males and females [43], different moderators of stigma improvement emerge in each sex, suggesting potential sex differences in stigma trajectories. Finally, of important note, however, the present analyses are limited to biological sex as recorded in clinical datasets, and no direct measures of gender-related constructs were available. Therefore, any discussion of gender-related mechanisms should be considered speculative and not directly supported by the data.

Although MCT’s normalizing and non-confrontational approach may help mitigate stigma-related barriers to engagement, future studies should explicitly incorporate measures of stigma, gender role expectations, and subjective treatment experiences to better understand their interaction with symptom trajectories and therapeutic outcomes. In this context, integrating qualitative methodologies, such as semi-structured interviews or focus groups, would allow for a deeper exploration of patients’ perspectives, expectations, and perceived treatment needs across genders, thereby complementing quantitative findings and advancing more personalized and sex- and gender-sensitive approaches to psychosis care.

Conclusions

This retrospective analysis of harmonized data from the PERMEPSY project demonstrates significant symptom reductions across positive, negative, general, and total domains following MCT, consistent with prior evidence supporting its broad applicability for psychosis regardless of biological sex.

These findings suggest that while sex-specific clinical profiles may influence baseline symptomatology, they do not appear to significantly alter the trajectory of symptom changes observed following MCT, supporting its use as a transdiagnostic and inclusive psychological intervention. Future research should explore whether more nuanced gender-related factors, such as social roles, stigma, and help-seeking behaviors, might interact with treatment engagement and outcomes. Prospective studies with stratified samples and standardized MCT formats could help clarify these dynamics and inform personalized approaches to care.

Supporting information

Acknowledgments

PERMEPSY project

The authors would like to inform you of all the members of the PERMEPSY project: Susana Ochoa, Maria Lamarca, Belen Ramos, Judith Usall, Regina Vila-Badia, Raquel Lopez-Carrilero, Trini Pelaez, Irene Birulés, Caroline König, Pedro Copado, Àngela Nebot, Alfredo Vellido, Cecilio Angulo (Spain); Steffen Moritz, Rabea Fischer, Merle Schlechte, Annika Schmueser, Antonia Meinhart, Jakob Scheunemann (Germany); Fabrice Berna, Adrien Goncalves (France); Łukasz Gawęda, Marytna Krezolek, Hanna Gelner, Adrianna Aleksandrowicz, Justyna Piwinska (Poland); Vanessa Acuña and Alvaro Cavieres (Chile).

The PERMEPSY team would like to thank and acknowledge the contribution to the retrospective database by researchers from the Spanish Metacognition Group and the following researchers: Ryan Balzan, Javier-David Lopez-Morinigo, Masayoshi Kobayashi, Ritta Kuokkanen, Linda Swanson, Mustafa Yildiz, Zita Fekete, Miguel Simón-Expósito, Christina Andreou, Jérôme Favrod, Lara Guedes de Pinho, Rita Roncone, Suzanne So, Hiroki Tanoue, Brooke Viertel, Erna Erawati, Ryotaro Ishikawa.