Introduction
The diagnostic boundary between psychotic disorders and borderline personality disorders (BPD) remains a persistent clinical challenge. Psychotic symptoms can be intertwined with the unstable emotional and behavioral traits characteristic of BPD, blurring the line between transient, stress-related psychotic experiences and more persistent psychotic disorders. This complexity underscores the importance of understanding psychosis not as a unitary condition but as a syndrome that can emerge across multiple diagnostic contexts, including, but not limited to, primary psychotic disorders.
Historically, borderline personality disorder was described as being ‘on the borderline’ between neurosis and psychosis (Kernberg, Reference Kernberg1967; Kernberg & Langs, Reference Kernberg and Langs1975). Kernberg’s model emphasized identity diffusion, primitive defenses such as splitting and sensitivity to stress, but also a largely intact assessment of reality. Subsequently, Gunderson’s empirical work consolidated BPD as a diagnostic entity in DSM-III, while maintaining its proximity to psychotic phenomena (American Psychiatric Association, 1980; Gunderson, Reference Gunderson1984). These initial perspectives highlighted both similarities and differences. Difficulties in differentiating between self and others and ipseity (‘sense of identity’) were often described as common vulnerabilities, while chronic psychotic processes and negative symptoms remained characteristic of schizophrenia spectrum disorders.
Psychotic disorders are defined by clinical syndromes and distinguished primarily by (i) duration (e.g. ≥ 6 months for schizophrenia symptoms and < 1 month for a brief psychotic disorders); (ii) pattern of symptoms (e.g. several psychotic symptoms in schizophrenia and only delusions in delusional disorder); (iii) relationship between psychotic symptoms and episodes of disturbed mood (i.e. whether psychotic symptoms occur during or after a mood disturbance); and (iv) underlying cause (i.e. whether the psychotic symptoms are due to substance use or medical conditions affecting the brain, such as epilepsy, autoimmune diseases, tumors, or dementia) (American Psychiatric Association, 2013). The term ‘psychotic symptom’ refers to a manifestation of cognitive and/or perceptual dysfunction (i.e. delusions or hallucinations), while ‘psychotic disorder’ refers to a condition in which psychotic symptoms meet the specific diagnostic criteria of a disease.
On average, the general population incidence of psychotic disorders is 26.6 per 100,000 persons/year (95% CI 22.0–31.7) (Jongsma, Turner, Kirkbride, & Jones, Reference Jongsma, Turner, Kirkbride and Jones2019). In the United States, it is estimated at 0.25–0.64%, with approximately 3 in 100 people experiencing a psychotic episode in their lifetime (Desai et al., Reference Desai, Lawson, Barner and Rascati2013; Kessler et al., Reference Kessler, Birnbaum, Demler, Falloon, Gagnon, Guyer, Howes, Kendler, Shi, Walters and Wu2005; Wu et al., Reference Wu, Shi, Birnbaum, Hudson and Kessler2006).
Psychotic disorders are complex mental disorders that are disabling both for the patient and their families. Psychotic disorders are associated with severe morbidity and high mortality (Kuo et al., Reference Kuo, Chen, Tsai, Chen, Ko and Chen2019; Schoenbaum et al., Reference Schoenbaum, Sutherland, Chappel, Azrin, Goldstein, Rupp and Heinssen2017), in particular when co-occurring with other disorders (Moran et al., Reference Moran, Walsh, Tyrer, Burns, Creed and Fahy2003). Personality disorders are a common comorbidity in people with psychotic disorders (Newton-Howes, Tyrer, North, & Yang, Reference Newton-Howes, Tyrer, North and Yang2008) and can contribute negatively to the course and outcome of the illness (Lenzenweger, Lane, Loranger, & Kessler, Reference Lenzenweger, Lane, Loranger and Kessler2007). The overall prevalence of personality disorders in schizophrenia is around 40%, although with wide confidence intervals (95% CI 25.2–55.8%) (Newton-Howes et al., Reference Newton-Howes, Tyrer, North and Yang2008). BPD is the most frequent personality disorder in clinical practice, present between 0.5% and 6% in general population and in more than 20% of users of psychiatric consultations (National Collaborating Centre for Mental Health [UK], 2009).
BPD causes severe functional impairment and is associated with a high suicidal risk (Kelleher, Ramsay, & DeVylder, Reference Kelleher, Ramsay and DeVylder2017) and a wide range of psychiatric and addictive comorbidities (around 85% of comorbid diagnoses) (Niemantsverdriet et al., Reference Niemantsverdriet, Slotema, Blom, Franken, Hoek, Sommer and van der Gaag2017). Shared vulnerability factors, such as childhood trauma, attachment disturbances, emotional dysregulation, social deprivation, and perhaps genetic susceptibilities also reinforce the plausibility of co-occurrence and challenge rigid diagnostic divisions (Gunderson & Sabo, Reference Gunderson and Sabo1993; Lysaker et al., Reference Lysaker, Meyer, Evans, Clements and Marks2001; Morgan, Reference Morgan2010). The likelihood of a co-occurrence of BPD with psychotic disorders is therefore high; however, this concept used to be controversial.
BPD has long been considered one of the most complex personality disorders due to its wide clinical heterogeneity (more than 200 polymorphisms), lack of consensus around its ontological entity, and poor response to pharmacological treatments (Mulder, Newton-Howes, Crawford, & Tyrer, Reference Mulder, Newton-Howes, Crawford and Tyrer2011). Nevertheless, emerging evidence suggests a substantial overlap between psychotic disorders and BPD. For instance, the prevalence rates of psychotic disorders among people with BPD reach 38% (Slotema et al., Reference Slotema, Blom, Niemantsverdriet, Deen and Sommer2018), while 17% to 25% of patients with schizophrenia or first-episode psychosis (FEP) meet BPD criteria (Bahorik & Eack, Reference Bahorik and Eack2010; Francey et al., Reference Francey, Jovev, Phassouliotis, Cotton and Chanen2018). The characteristics of BPD and psychotic disorders are generally first identified in adolescence and adulthood, and people who later develop BPD may have a prodrome of personality disorders during adolescence similar to the prodrome of schizophrenia (McClellan, Werry, & Ham, Reference McClellan, Werry and Ham1993). It is also likely that personality traits and personality disorders may influence the development of other diagnoses (Hulbert, Jackson, & McGorry, Reference Hulbert, Jackson and McGorry1996). Despite these overlaps, it remains commonly and wrongly believed in clinical settings that the two cannot coexist, leading many clinicians to feel compelled to choose between one diagnosis or the other, often at the cost of ignoring the complexity and comorbidity actually present.
Clarification of this co-occurrence has major clinical implications. It highlights the existence of a subgroup with specific functional profiles, raises the question of quasi-psychoses and pseudo-psychoses in the context of BPD, and calls for more appropriate early intervention and dimensional approaches to care. Confirming this co-occurrence would shed light on the phenomenology of these diseases and help answer clinical and therapeutic questions. These patients could correspond to a specific subgroup of patients with combined clinical and functional profiles. The co-occurrence of BPD or borderline personality traits in psychotic disorders could worsen prognosis and complicate therapeutic management. This systematic review and meta-analysis aims to (i) assess the prevalence of co-occurring BPD with psychotic disorders, and in FEP populations and (ii) describe the sociodemographic and clinical characteristics of this subgroup.
Methods
Search strategy
The systematic review and meta-analysis protocol was designed according to the PRISMA guidelines (Page et al., Reference Page, McKenzie, Bossuyt, Boutron, Hoffmann, Mulrow, Shamseer, Tetzlaff, Akl, Brennan, Chou, Glanville, Grimshaw, Hróbjartsson, Lalu, Li, Loder, Mayo-Wilson, McDonald and Moher2021) and registered at PROSPERO (CRD42024577525). The PECOS approach (Population, Exposure, Comparator, Outcome, and Setting) was used to define the research question, inclusion criteria, and search terms (see Supplementary Appendix A). Databases were searched from their inception until June 13, 2025, using PubMed, PsycInfo, PsycArticles, and Psychological and Behavioral Sciences Collection interfaces.
The search terms used with the PubMed database were ‘borderline personality disorder’ AND (‘schizophrenia’ OR ‘psychosis’ OR ‘psychotic disorder’). For the other databases, the search was performed using the terms ‘borderline personality disorder’ AND (schizophrenia or psychosis or psychoses or psychotic disorder or schizophrenic disorder). The differences between MeSH terms and keywords from one database to another were intentional, in order to adapt to the indexing systems specific to each database and optimize search sensitivity. Reference lists were manually searched to identify potential studies missed by the database search.
Inclusion and exclusion criteria
Studies had to be published in English or French in peer-reviewed scientific journals. There were no restrictions regarding year or country of publication, length of follow-up, or study design. Literature reviews, editorials, case reports, or case series were excluded. Studies had to examine the co-occurrence of psychotic disorders and BPD.
In accordance with the pre-registered protocol, only studies conducted in hospital or clinical service contexts were eligible. This criterion was applied to ensure that diagnoses of BPD and psychotic disorders were established using standardized clinical assessments, rather than on the basis of unstructured self-assessments conducted with community samples.
We distinguished two groups of patients with psychotic disorders: FEP and persistent psychotic disorders. For consistency, FEP has been defined as the first presentation of a psychotic disorder to clinical services, usually within the first 6 months of a symptom onset, regardless of the diagnosis posed (Birchwood, Todd, & Jackson, Reference Birchwood, Todd and Jackson1998; Francey et al., Reference Francey, Jovev, Phassouliotis, Cotton and Chanen2018). In addition, some operational definitions used in the literature also specify the presence of positive psychotic symptoms (delusions, hallucinations, or disorganization) for at least 1 week, occurring daily or at least three times a week for more than 1 hour each time, in people who have never received antipsychotic treatment prior to this initial episode (see review in Breitborde et al., Reference Breitborde, Srihari and Woods2009). Accordingly, diagnoses such as brief psychotic disorders and schizophreniform disorders, when reported, were included in the FEP group, as they are recognized in the DSM and ICD classifications as psychotic disorders that typically occur early in the course of the illness. The persistent psychotic disorders group included participants with established schizophrenia spectrum or other persistent psychotic disorders, including schizophrenia, schizoaffective disorder, delusional disorder, and unspecified psychotic disorders, typically characterized by a duration of illness longer than 6 months, with or without antipsychotic treatment. This distinction was based on diagnostic labels and the duration of illness as reported in each study.
High-risk patients in the prodromal phase (e.g. ultra-high risk or with attenuated psychosis syndrome) were excluded because, by definition, they do not meet the diagnostic criteria for a psychotic disorder.
Results of the systematic search
The studies obtained from the databases and reference lists of the included studies were manually selected by one author (JJ) and independently verified by two different authors (AS and CE). Disagreements were resolved by discussion until a consensus was reached between the three authors. Titles and abstracts of the screened studies were evaluated for eligibility, and full-text articles meeting the study criteria were retrieved for final review. A final manual search through the reference lists of screened records was performed. Then, the full-text articles were assessed for final inclusion.
Data extraction and quality assessment
Data extraction was performed by one author (JJ); standardized form was used to extract relevant data (Harris et al., Reference Harris, Quatman, Manring, Siston and Flanigan2014), including author names, publication date, study design, settings, study population (sample size, age, and sex), definition of psychotic disorders or FEP and subgroups, definition of BPD, scales or interviews used, results, and conclusions. The risk of bias of each study was initially evaluated by one author (JJ) using the Effective Public Health Practice Project (EPHPP) tool (Thomas et al., Reference Thomas, Ciliska, Dobbins and Micucci2004). The included studies were ordered alphabetically and split into two groups for independent assessment by a second evaluator (AS or CE), whereby AS assessed the first half to avoid assessing her own article (Schandrin et al., Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023). Differences were discussed until a consensus was reached. Biases were classified as ‘weak’, ‘moderate’, or ‘strong’ for the overall score and the individual categories.
Statistical analysis
Analyses were performed using R Statistical Software (R Core Team, 2021). Three separate meta-analyses were carried out, assessing (1) the prevalence of BPD in people with psychotic disorders, (2) the prevalence of psychotic disorders in people diagnosed with BPD, and (3) the prevalence of BPD in FEP. This approach allowed a more comprehensive and bidirectional examination of the co-occurrence between BPD and psychotic disorders, as well as an early stage clinical characterization of people experiencing FEP.
For each meta-analysis, results were calculated using common-effects and random-effects models. When both models gave similar results, confidence in the conclusions was reinforced. In cases of significant heterogeneity (high I 2), the random-effects model was preferred. All analyses were based on the inverse variance method, using a logit transformation of the raw proportions to account for distributional properties and stabilize variances between studies. The DerSimonian–Laird estimator was used to calculate the between-study variance (τ 2) for the three random-effects models. Heterogeneity was assessed using Cochran’s Q statistic and I 2, and 95% confidence intervals were calculated for τ 2 and τ using Jackson’s method. Prediction intervals were calculated to reflect the expected range of effect sizes in future studies. Confidence intervals for individual studies were calculated using the Clopper–Pearson method. Sensitivity analyses were carried out using two complementary approaches: a meta-analysis with exclusion of crude proportions (via ‘meta::metainf()’, with back-transformed logit estimates) and an analysis with exclusion based on logit using the ‘metafor::leave1out()’ function. These methods were used to assess the robustness of the estimates and identify potentially influential studies. Egger’s regression test for funnel skewness was used to assess potential publication bias and the effects of small studies, complemented by funnels with enhanced contours illustrating confidence zones (90%, 95%, and 99%).
We excluded Cailhol et al. (Reference Cailhol, Pelletier, Rochette, Renaud, Koch, David and Lesage2021) from all meta-analyses due to its extreme sample size (n = 7,000,005; 40,150 events), which would have disproportionately influenced the pooled estimates and artificially reduced the variance between studies (Borenstein, Hedges, Higgins, & Rothstein, Reference Borenstein, Hedges, Higgins and Rothstein2013; Higgins & Green, Reference Higgins and Green2008). Archer et al. (Reference Archer, Shnyien, Mansfield and Draycott2023) was also excluded because it did not report a clearly defined diagnostic prevalence, preventing the extraction of comparable effect sizes. Both studies were retained in the qualitative synthesis and discussed accordingly.
The study by Gleeson et al (Reference Gleeson, Chanen, Cotton, Pearce, Newman and McCutcheon2012) was included in the meta-analysis on FEP (Analysis 3) due to the limited number of studies available but was interpreted with caution due to its selective clinical population, which included people already enrolled in a specialized integrative intervention. Both studies were retained in the qualitative synthesis and discussed accordingly.
Finally, in the third analysis, Pelizza et al. (Reference Pelizza, Leuci, Quattrone, Azzali, Paulillo, Pupo, Pellegrini, Gammino, Biancalani and Menchetti2025) was included in the meta-analysis but interpreted with caution due to its small, highly selected sample and extreme prevalence (100%). Its impact was examined by means of sensitivity analyses and discussed explicitly in the results.
Results
Study selection
The initial database search generated 4,939 results. After the removal of 510 duplicates, 4,429 studies were screened by title and abstract, of which 4,396 were excluded. The full text of the remaining 33 studies was assessed, and 18 studies were definitively included in the qualitative synthesis. Among these, 16 were included in the quantitative meta-analyses, while 2 studies (Cailhol et al., Reference Cailhol, Pelletier, Rochette, Renaud, Koch, David and Lesage2021 and Archer et al., Reference Archer, Shnyien, Mansfield and Draycott2023) were retained for narrative review only due to, respectively, significant clinical/methodological differences and the absence of a clear prevalence estimate (see Methods) (Figure 1 and Supplementary Appendix B).
PRISMA 2020 flow diagram of included studies. Note: Flowchart detailing the identification, screening, eligibility, and inclusion of studies. A total of 4,939 records were identified through database searches. After duplicate removal and screening, 33 full-text articles were assessed for eligibility. Eighteen studies were included in the qualitative synthesis, among which 16 were included in the quantitative meta-analyses. Two studies were excluded from meta-analyses: Cailhol et al. (Reference Cailhol, Pelletier, Rochette, Renaud, Koch, David and Lesage2021), due to methodological and clinical discrepancies, and Archer et al. (Reference Archer, Shnyien, Mansfield and Draycott2023), due to the absence of a clearly defined prevalence estimate. Both were retained for narrative discussion.
Study design and characteristics
Study design, mean duration of follow-up if applicable, countries of inclusion, and sample characteristics (sample size, mean age, and initial diagnosis) are summarized in the supplementary material.
The studies analyzed were published in English between 2004 and 2024, with sample sizes ranging from 37 to 7,000,005 participants (Median = 80).
Eighteen studies were included in the systematic review, 16 of which were included in the meta-analyses. Cailhol et al (Reference Cailhol, Pelletier, Rochette, Renaud, Koch, David and Lesage2021) was excluded from the quantitative synthesis due to methodological and clinical considerations and Archer et al (Reference Archer, Shnyien, Mansfield and Draycott2023) was retained in the systematic review but excluded from the meta-analyses because it did not report a clear diagnostic prevalence of BPD (see Methods), but are the subject of a narrative discussion.
Seven studies examined the co-occurrence of BPD in people diagnosed with a psychotic disorder (Altunsoy et al., Reference Altunsoy, Sahiner, Cingi Kuluk, Okay, Ulusoy Kaymak, Aydemir and Goka2015; Bahorik & Eack, Reference Bahorik and Eack2010; Barral et al., Reference Barral, Rodríguez-Cintas, Grau-López, Daigre, Ros-Cucurull, Calvo, Ferrer and Roncero2018; Lysaker, Reference Lysaker, Wickett, Lancaster and Davis2004; Sar et al., Reference Sar, Taycan, Bolat, Özmen, Duran, Öztürk and Ertem-Vehid2010; Tschoeke, Steinert, Flammer, & Uhlmann, Reference Tschoeke, Steinert, Flammer and Uhlmann2014; Wickett et al., Reference Wickett, Essman, Beck-Jennings, Davis, McIlvried and Lysaker2006), four studies reported the prevalence of psychotic disorders in BPD samples (Barral et al., Reference Barral, Rodríguez-Cintas, Grau-López, Daigre, Ros-Cucurull, Calvo, Ferrer and Roncero2018; Kingdon et al., Reference Kingdon, Ashcroft, Bhandari, Gleeson, Warikoo, Symons, Taylor, Lucas, Mahendra, Ghosh, Mason, Badrakalimuthu, Hepworth, Read and Mehta2010; Pearse et al., Reference Pearse, Dibben, Ziauddeen, Denman and McKenna2014; Slotema et al., Reference Slotema, Blom, Niemantsverdriet, Deen and Sommer2018), and five studies assessed the prevalence of BPD in people with FEP (Chanen et al., Reference Chanen, Kerslake, Berubé, Nicol, Jovev, Yuen, Betts, McDougall, Nguyen, Cavelti and Kaess2024; Francey et al., Reference Francey, Jovev, Phassouliotis, Cotton and Chanen2018; Gleeson et al., Reference Gleeson, Chanen, Cotton, Pearce, Newman and McCutcheon2012; Pelizza et al., Reference Pelizza, Leuci, Quattrone, Azzali, Paulillo, Pupo, Pellegrini, Gammino, Biancalani and Menchetti2025; Schandrin et al., Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023).
Two studies involved participants with a primary diagnosis of psychotic disorder or BPD (Cailhol et al., Reference Cailhol, Pelletier, Rochette, Renaud, Koch, David and Lesage2021; Kingdon et al., Reference Kingdon, Ashcroft, Bhandari, Gleeson, Warikoo, Symons, Taylor, Lucas, Mahendra, Ghosh, Mason, Badrakalimuthu, Hepworth, Read and Mehta2010), and one study explored the impact of joint therapeutic intervention in participants with co-occurring BPD and psychotic disorders (Gleeson et al., Reference Gleeson, Chanen, Cotton, Pearce, Newman and McCutcheon2012). Barral et al. (Reference Barral, Rodríguez-Cintas, Grau-López, Daigre, Ros-Cucurull, Calvo, Ferrer and Roncero2018) was included in both the first and second meta-analyses, as the study reported prevalence estimates from distinct clinical subsamples (people with psychotic disorders and people with BPD) which could be independently extracted and analyzed.
Psychotic disorders were assessed using standardized clinical interviews based on the diagnostic criteria of the DSM or ICD. In most cases, the Structured Clinical Interview for DSM-Axis I Disorders was administered (SCID-I; Altunsoy et al., Reference Altunsoy, Sahiner, Cingi Kuluk, Okay, Ulusoy Kaymak, Aydemir and Goka2015; Barral et al., Reference Barral, Rodríguez-Cintas, Grau-López, Daigre, Ros-Cucurull, Calvo, Ferrer and Roncero2018; Francey et al., Reference Francey, Jovev, Phassouliotis, Cotton and Chanen2018; Pelizza et al., Reference Pelizza, Leuci, Quattrone, Azzali, Paulillo, Pupo, Pellegrini, Gammino, Biancalani and Menchetti2025; Sar et al., Reference Sar, Taycan, Bolat, Özmen, Duran, Öztürk and Ertem-Vehid2010; Schandrin et al., Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023; Slotema et al., Reference Slotema, Blom, Niemantsverdriet, Deen and Sommer2018; Tschoeke et al., Reference Tschoeke, Steinert, Flammer and Uhlmann2014). One study used the Mini International Neuropsychiatric Interview (Cailhol et al., Reference Cailhol, Pelletier, Rochette, Renaud, Koch, David and Lesage2021), while in another, the ICD-10 diagnosis was confirmed by psychiatrists based on medical files (Kingdon et al., Reference Kingdon, Ashcroft, Bhandari, Gleeson, Warikoo, Symons, Taylor, Lucas, Mahendra, Ghosh, Mason, Badrakalimuthu, Hepworth, Read and Mehta2010). Borderline personality disorder was assessed using validated structured or semi-structured interviews based on the diagnostic criteria of DSM-III-R or DSM-IV, such as the Structured Clinical Interview for DSM-IV Axis II Personality Disorders (SCID-II; Barral et al., Reference Barral, Rodríguez-Cintas, Grau-López, Daigre, Ros-Cucurull, Calvo, Ferrer and Roncero2018; Chanen et al., Reference Chanen, Kerslake, Berubé, Nicol, Jovev, Yuen, Betts, McDougall, Nguyen, Cavelti and Kaess2024; Gleeson et al., Reference Gleeson, Chanen, Cotton, Pearce, Newman and McCutcheon2012; Schandrin et al., Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023), the Structured Interview for DSM-III-R Personality Disorders (SIDP-R; Bahorik & Eack, Reference Bahorik and Eack2010), or the Millon Multiaxial Clinical Inventory – Third Edition (MCMI-III; Lysaker, Reference Lysaker, Wickett, Lancaster and Davis2004; Wickett et al., Reference Wickett, Essman, Beck-Jennings, Davis, McIlvried and Lysaker2006). One study used a self-report measure (PDQ-4+) for screening for BPD (Archer, Shnyien, Mansfield, & Draycott, Reference Archer, Shnyien, Mansfield and Draycott2023). Most studies were based on the DSM-IV or DSM-5 nosology, while a few used the ICD-10 classifications (Kingdon et al., Reference Kingdon, Ashcroft, Bhandari, Gleeson, Warikoo, Symons, Taylor, Lucas, Mahendra, Ghosh, Mason, Badrakalimuthu, Hepworth, Read and Mehta2010). Diagnoses were made by qualified clinicians or researchers, thus ensuring acceptable validity.
To ensure consistency across analyses, diagnostic definitions were carefully aligned with the methodological criteria described in the Methods section. In particular, the operational distinction between FEP and persistent psychotic disorders was verified in each study based on duration, treatment history, and diagnostic label. Indeed, the group of people who have experienced an FEP has been defined as comprising people who experience psychotic symptoms (hallucinations, delusions, and/or disorganization) for the first time for at least 1 week, significantly impairing their functioning in the absence of prior antipsychotic treatment (Francey et al., Reference Francey, Jovev, Phassouliotis, Cotton and Chanen2018; Pelizza et al., Reference Pelizza, Leuci, Quattrone, Azzali, Paulillo, Pupo, Pellegrini, Gammino, Biancalani and Menchetti2025; Schandrin et al., Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023). In contrast, persistent psychotic disorders referred to established psychotic disorders within the schizophrenia spectrum (schizophrenia, schizoaffective disorder, and delusional disorder), which are generally longer lasting and often treated with medication (Altunsoy et al., Reference Altunsoy, Sahiner, Cingi Kuluk, Okay, Ulusoy Kaymak, Aydemir and Goka2015; Barral et al., Reference Barral, Rodríguez-Cintas, Grau-López, Daigre, Ros-Cucurull, Calvo, Ferrer and Roncero2018; Kingdon et al., Reference Kingdon, Ashcroft, Bhandari, Gleeson, Warikoo, Symons, Taylor, Lucas, Mahendra, Ghosh, Mason, Badrakalimuthu, Hepworth, Read and Mehta2010; Tschoeke et al., Reference Tschoeke, Steinert, Flammer and Uhlmann2014; Wickett et al., Reference Wickett, Essman, Beck-Jennings, Davis, McIlvried and Lysaker2006).
A cross-sectional design was most common (Altunsoy et al., Reference Altunsoy, Sahiner, Cingi Kuluk, Okay, Ulusoy Kaymak, Aydemir and Goka2015; Archer et al., Reference Archer, Shnyien, Mansfield and Draycott2023; Chanen et al., Reference Chanen, Kerslake, Berubé, Nicol, Jovev, Yuen, Betts, McDougall, Nguyen, Cavelti and Kaess2024; Kingdon et al., Reference Kingdon, Ashcroft, Bhandari, Gleeson, Warikoo, Symons, Taylor, Lucas, Mahendra, Ghosh, Mason, Badrakalimuthu, Hepworth, Read and Mehta2010; Lysaker, Reference Lysaker, Wickett, Lancaster and Davis2004; Pearse et al., Reference Pearse, Dibben, Ziauddeen, Denman and McKenna2014; Sar et al., Reference Sar, Taycan, Bolat, Özmen, Duran, Öztürk and Ertem-Vehid2010; Slotema et al., Reference Slotema, Blom, Niemantsverdriet, Deen and Sommer2018; Tschoeke et al., Reference Tschoeke, Steinert, Flammer and Uhlmann2014). In these studies, participants were assessed in terms of sociodemographic and clinical characteristics.
In longitudinal studies (Bahorik & Eack, Reference Bahorik and Eack2010; Barral et al., Reference Barral, Rodríguez-Cintas, Grau-López, Daigre, Ros-Cucurull, Calvo, Ferrer and Roncero2018; Cailhol et al., Reference Cailhol, Pelletier, Rochette, Renaud, Koch, David and Lesage2021; Francey et al., Reference Francey, Jovev, Phassouliotis, Cotton and Chanen2018; Gleeson et al., Reference Gleeson, Chanen, Cotton, Pearce, Newman and McCutcheon2012; Pelizza et al., Reference Pelizza, Leuci, Quattrone, Azzali, Paulillo, Pupo, Pellegrini, Gammino, Biancalani and Menchetti2025; Schandrin et al., Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023; Wickett et al., Reference Wickett, Essman, Beck-Jennings, Davis, McIlvried and Lysaker2006), follow-up ranged from 3 months to 24 months.
Quality assessment
The main biases were related to (i) lack of information on blinding, (ii) absence of adjustment for clinical confounding factors, (iii) lack of randomization, and (iv) study designs with a low level of evidence (mainly transversal studies). Biases are summarized in Table 1.
Quality assessment according to the effective public health practice project tool
Note: The global rating was derived from the Effective Public Health Practice Project (EPHPP) quality assessment tool. A score of 1 indicates strong methodological quality (low risk of bias), 2 indicates moderate quality, and 3 indicates weak methodological quality (high risk of bias). The ‘Data collection methods’ domain was evaluated based on diagnostic reliability: structured clinical interviews (SCI-I/II; SIDP-R; MCMI-III) were rated as strong (1), diagnoses confirmed by a clinician or semi-structured were rated as moderate (2), and self-reported screening tools were rated as weak (3). As a result, Archer et al. (Reference Archer, Shnyien, Mansfield and Draycott2023) received a lower rating, while studies using standardized interviews (e.g. Slotema et al., Reference Slotema, Blom, Niemantsverdriet, Deen and Sommer2018; Schandrin et al., Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023) were rated as strong.
According to the EPHPP (Thomas, Ciliska, Dobbins, & Micucci, Reference Thomas, Ciliska, Dobbins and Micucci2004), 15 studies were considered as having a low methodological quality (score = 3), 3 studies as having a moderate methodological quality (score = 2), and no studies as having a strong methodological quality (score = 1). Scores were based on six domains (selection bias, study design, confounders, blinding, data collection, and withdrawals), and global ratings followed the EPHPP conventions where lower scores indicate higher methodological quality.
The variability of diagnostic procedures (structured interviews versus self-assessment) was explicitly taken into account in the ‘data collection methods’ domain. Studies based on self-assessment received a lower score for data validity, while those using structured clinical interviews (e.g. SCID-I/II; SIDP-R; MCMI-III) received a ‘strong’ rating in this area.
Prevalence of co-occurring BPD and psychotic disorders
Prevalence of BPD in people with psychotic disorders
A total of seven studies assessed the prevalence of BPD in people with psychotic disorders encompassing 515 participants, of whom 115 were diagnosed with BPD (Altunsoy et al., Reference Altunsoy, Sahiner, Cingi Kuluk, Okay, Ulusoy Kaymak, Aydemir and Goka2015; Bahorik & Eack, Reference Bahorik and Eack2010; Barral et al., Reference Barral, Rodríguez-Cintas, Grau-López, Daigre, Ros-Cucurull, Calvo, Ferrer and Roncero2018; Lysaker, Reference Lysaker, Wickett, Lancaster and Davis2004; Sar et al., Reference Sar, Taycan, Bolat, Özmen, Duran, Öztürk and Ertem-Vehid2010; Tschoeke et al., Reference Tschoeke, Steinert, Flammer and Uhlmann2014; Wickett et al., Reference Wickett, Essman, Beck-Jennings, Davis, McIlvried and Lysaker2006). The pooled prevalence estimate under a random-effects model was 22.7% (95% CI: 14.2%–34.3%), with substantial heterogeneity (I 2 = 85.3%, 95% CI: 71.7%–92.4%; τ 2 = 0.50; Q(6) = 40.91, p < .0001). The prediction interval ranged from 5.8% to 57.0%, reflecting considerable between-study variability (Figure 2). Sensitivity analyses showed that no single study substantially influenced the overall effect size. Leave-one-out estimates ranged from 20.5% to 26.5%, with consistently high heterogeneity (Supplementary Appendix C, Table C1). Secondary sensitivity analyses on logit-transformed estimates yielded convergent results (Supplementary Appendix C, Table C2). Visual inspection of the funnel plot revealed signs of asymmetry. Egger’s regression test was statistically significant (z = −2.12, p = .034), suggesting the presence of small-study effects or publication bias (Supplementary Appendix C, Figure C1).
Forest plot of the prevalence of BPD among people with psychotic disorders. Note: Each square represents the prevalence estimate for an individual study, with the size proportional to its weight in the meta-analysis. Horizontal lines indicate 95% confidence intervals. The diamond represents the pooled prevalence under the random-effects model. The prediction interval is shown below the diamond. Substantial between-study heterogeneity was observed (I 2 = 85.3%; τ 2 = 0.50).
Prevalence of psychotic disorders in people with BPD
Four studies reported the prevalence of psychotic disorders in samples of people diagnosed with BPD, encompassing a total of 233 participants, of whom 36 met the criteria for a psychotic disorder (Barral et al., Reference Barral, Rodríguez-Cintas, Grau-López, Daigre, Ros-Cucurull, Calvo, Ferrer and Roncero2018; Kingdon et al., Reference Kingdon, Ashcroft, Bhandari, Gleeson, Warikoo, Symons, Taylor, Lucas, Mahendra, Ghosh, Mason, Badrakalimuthu, Hepworth, Read and Mehta2010; Pearse et al., Reference Pearse, Dibben, Ziauddeen, Denman and McKenna2014; Slotema et al., Reference Slotema, Blom, Niemantsverdriet, Deen and Sommer2018). The pooled prevalence under a random-effects model was estimated at 14.3% (95% CI: 5.5% to 32.1%; k = 4), reflecting substantial between-study variability (Figure 3). Heterogeneity was high (I 2 = 85.4%, 95% CI: 64.1% to 94.1%; τ 2 = 0.95), and the heterogeneity test was statistically significant (Q(3) = 20.62, p = .0001). The prediction interval was wide (0.5% to 85.0%), indicating considerable dispersion in the expected true effect sizes across different populations. Sensitivity analyses revealed that no single study substantially altered the overall prevalence estimate. Leave-one-out analyses yielded prevalence values ranging from 7.7% (when excluding Kingdon et al., Reference Kingdon, Ashcroft, Bhandari, Gleeson, Warikoo, Symons, Taylor, Lucas, Mahendra, Ghosh, Mason, Badrakalimuthu, Hepworth, Read and Mehta2010) to 16.1% (when excluding Slotema et al., Reference Slotema, Blom, Niemantsverdriet, Deen and Sommer2018), with I 2 values remaining high except in the latter case (I 2 = 43.4%) (Supplementary Appendix C, Table C3). Results from a secondary leave-one-out analysis using logit-transformed estimates were consistent (Supplementary Appendix C, Table C4). Visual inspection of the funnel plot suggested mild asymmetry (Supplementary Appendix C, Figure C2). However, Egger’s regression test was not statistically significant (t(2) = −1.85, p = .2048), providing no clear evidence of small-study effects or publication bias.
Forest plot of the prevalence of psychotic disorders among people with BPD. Note: Each square represents the prevalence estimate for a people study, with the size proportional to its weight in the meta-analysis. Horizontal lines indicate 95% confidence intervals. The diamond at the bottom represents the pooled prevalence under the random-effects model. The prediction interval is represented as a horizontal line beneath the pooled estimate. Substantial heterogeneity was observed across studies (I 2 = 85.4%; τ 2 = 0.95).
Prevalence of BPD in people with FEP
Five studies reported the prevalence of BPD among people with FEP, totaling 669 participants, of whom 180 met the criteria for BPD (Chanen et al., Reference Chanen, Kerslake, Berubé, Nicol, Jovev, Yuen, Betts, McDougall, Nguyen, Cavelti and Kaess2024; Francey et al., Reference Francey, Jovev, Phassouliotis, Cotton and Chanen2018; Gleeson et al., Reference Gleeson, Chanen, Cotton, Pearce, Newman and McCutcheon2012; Pelizza et al., Reference Pelizza, Leuci, Quattrone, Azzali, Paulillo, Pupo, Pellegrini, Gammino, Biancalani and Menchetti2025; Schandrin et al., Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023). A wide range of point estimates was observed, from 18.4% (Schandrin et al., Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023) to 100% (Pelizza et al., Reference Pelizza, Leuci, Quattrone, Azzali, Paulillo, Pupo, Pellegrini, Gammino, Biancalani and Menchetti2025). The random-effects model yielded a pooled prevalence of 40.0% (95% CI: 21.9% to 61.3%; k = 5), with substantial heterogeneity (I2 = 89.5%, 95% CI: 78.3% to 94.9%; τ 2 = 0.72; Q(4) = 38.05, p < .0001), and a wide prediction interval ranging from 4.4% to 90.5% (Figure 4). The common-effect model yielded a more conservative estimate of 21.6% (95% CI: 18.4% to 25.1%). Visual inspection of the funnel plot suggested asymmetry, and Egger’s regression test was statistically significant (t = 5.43, p = .0123), indicating potential small-study effects or publication bias (Supplementary Appendix C, Figure C3). Leave-one-out sensitivity analyses confirmed that no single study dramatically altered the direction of the pooled estimate. However, removing Pelizza et al. (Reference Pelizza, Leuci, Quattrone, Azzali, Paulillo, Pupo, Pellegrini, Gammino, Biancalani and Menchetti2025) notably reduced heterogeneity (from I 2 = 89.5% to 85.6%) and shifted the estimate downward to 29.8% (95% CI: 17.6% to 45.7%) (Supplementary Appendix C, Table C5). This study reported a 100% co-occurrence rate, as all participants had both FEP and BPD by design, potentially inflating the overall estimate due to selection bias. Additional sensitivity analyses using logit-transformed data yielded consistent results (Supplementary Appendix C, Table C6).
Forest plot of the prevalence of BPD in people with FEP. Note: Pooled prevalence estimates are presented using a random-effects model with logit transformation. Diamonds represent the pooled estimates with 95% confidence intervals (CI); the horizontal line indicates the 95% prediction interval. The plot includes five studies ($N = 669$), with individual proportions ranging from 18% to 100%. The overall pooled prevalence was 40.0% (95% CI: 22.0%–61.3%), with high heterogeneity (I 2 = 89.5%).
Narrative synthesis of qualitative findings
Sociodemographic and clinical profiles
Across studies, the most frequently reported profile of patients with co-occurring BPD and psychotic disorders was that of young adult women, particularly those experiencing FEP or diagnosed with schizophrenia, who were more likely than men to meet the criteria for BPD (Bahorik & Eack, Reference Bahorik and Eack2010; Francey et al., Reference Francey, Jovev, Phassouliotis, Cotton and Chanen2018; Kingdon et al., Reference Kingdon, Ashcroft, Bhandari, Gleeson, Warikoo, Symons, Taylor, Lucas, Mahendra, Ghosh, Mason, Badrakalimuthu, Hepworth, Read and Mehta2010; Schandrin et al., Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023). Educational attainment was generally lower in this subgroup, with several studies reporting a reduced rate of post-secondary completion (Francey et al., Reference Francey, Jovev, Phassouliotis, Cotton and Chanen2018; Schandrin et al., Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023). In one study, the average duration of illness in patients with both schizophrenia and BPD reached approximately 10 years (Kingdon et al., Reference Kingdon, Ashcroft, Bhandari, Gleeson, Warikoo, Symons, Taylor, Lucas, Mahendra, Ghosh, Mason, Badrakalimuthu, Hepworth, Read and Mehta2010). However, among FEP patients, those with comorbid BPD had a shorter duration of untreated psychosis, suggesting earlier clinical presentation or referral (Pelizza et al., Reference Pelizza, Leuci, Quattrone, Azzali, Paulillo, Pupo, Pellegrini, Gammino, Biancalani and Menchetti2025).
Regarding diagnostic patterns, psychotic disorder not otherwise specified (PNOS) was the most frequently observed psychotic diagnosis in patients with BPD, with an estimated prevalence of 20% (Schandrin et al., Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023; Slotema et al., Reference Slotema, Blom, Niemantsverdriet, Deen and Sommer2018). Other diagnoses included schizophreniform disorder (6%), schizophrenia (2%), and schizoaffective disorder (4%). These figures stem from a limited number of studies and should be interpreted with caution, as most samples were small and came from specialized hospitals. In addition, PNOS labels may reflect different diagnostic concepts from one study to another, depending on whether the DSM or ICD criteria were applied, further contributing to the observed variability.
Clinical symptomatology and risk behaviors
Beyond diagnostic patterns, symptom profiles in patients with comorbid BPD and psychotic disorders appeared more severe and complex than in those with either disorder alone. One longitudinal study noted that comorbid patients initially presented with heightened levels of emotional distress, anxiety, and depression, although differences in affective symptoms diminished over time (Bahorik & Eack, Reference Bahorik and Eack2010). These observations were supported by other cross-sectional investigations (Archer et al., Reference Archer, Shnyien, Mansfield and Draycott2023; Wickett et al., Reference Wickett, Essman, Beck-Jennings, Davis, McIlvried and Lysaker2006). Across studies, the clinical profile of this subgroup was characterized by higher emotional reactivity, increased affective instability, and more intense interpersonal sensitivity, often leading to greater subjective distress and functional impairment. These characteristics have been repeatedly highlighted as dimensions common to BPD and psychotic disorders, reflecting a partially shared vulnerability to stress-related dysregulation.
Suicidal behavior was consistently elevated in patients with both diagnoses. In a prospective study, each additional BPD criterion was associated with a 44% increase in the likelihood of past suicide attempts in psychotic patients (Lysaker et al., Reference Lysaker, Wickett, Lancaster and Davis2004). Further evidence from recent FEP cohorts confirmed a higher baseline rate of suicide attempts in the comorbid group, although longitudinal differences in new attempts were not significant (Pelizza et al., Reference Pelizza, Leuci, Quattrone, Azzali, Paulillo, Pupo, Pellegrini, Gammino, Biancalani and Menchetti2025; Francey et al., Reference Francey, Jovev, Phassouliotis, Cotton and Chanen2018; Schandrin et al., Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023). However, the heterogeneity of suicide assessment tools across studies (structured interviews, medical file reviews, or self-report questionnaires) likely contributes to the discrepancies in reported rates.
Auditory verbal hallucinations (AVHs) were notably frequent in patients with co-occurrence. While AVHs affected approximately 27.6% of psychotic people with borderline traits in one study (Barral et al., Reference Barral, Rodríguez-Cintas, Grau-López, Daigre, Ros-Cucurull, Calvo, Ferrer and Roncero2018), other studies reported rates as high as 90% (Kingdon et al., Reference Kingdon, Ashcroft, Bhandari, Gleeson, Warikoo, Symons, Taylor, Lucas, Mahendra, Ghosh, Mason, Badrakalimuthu, Hepworth, Read and Mehta2010; Schandrin et al., Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023). Comparative figures with schizophrenia-only and BPD-only groups suggest an additive effect, although direct causality cannot be inferred. It should be noted that definitions of AVHs varied considerably, ranging from brief perceptual distorsions to prolonged hallucinatory experiences, depending on the diagnostic instrument used. AVHs in people with comorbidity have also been perceived as more distressing, leading to stronger negative emotional responses and resistance behaviors (Hepworth et al., Reference Hepworth, Ashcroft and Kingdon2013).
Despite methodological variability, the convergence of results between different study protocols and assessment tools confirms the robustness of these findings and highlights the clinical relevance of the overlap between affective and psychotic symptoms in comorbid presentations.
Mediating factors: trauma, dissociation, and substance use
Several studies have highlighted the role of adverse childhood experiences (ACEs), dissociation, and substance use disorder (SUD) as potential mediators in the development or exacerbation of comorbidities. In one study, borderline traits in people with psychotic disorders were linked to childhood sexual abuse (Lysaker et al., Reference Lysaker, Wickett, Lancaster and Davis2004). Another study confirmed that emotional, physical, and sexual abuse, as well as neglect, were significantly more frequent in people with BPD or BPD and schizophrenia than in people with schizophrenia alone (Kingdon et al., Reference Kingdon, Ashcroft, Bhandari, Gleeson, Warikoo, Symons, Taylor, Lucas, Mahendra, Ghosh, Mason, Badrakalimuthu, Hepworth, Read and Mehta2010). These results support the hypothesis that early trauma could be a common vulnerability factor contributing to both emotional dysregulation and the expression of psychotic symptoms. However, the implementation of ACEs varied considerably from one study to another (clinical interviews, self-assessment questionnaires, or file reviews), which limited the direct comparability of prevalence estimates.
Dissociative symptoms, including those meeting criteria for dissociative identity disorder, have been associated with ACEs and appear to mediate the link between trauma and psychotic symptomatology (Sar et al., Reference Sar, Taycan, Bolat, Özmen, Duran, Öztürk and Ertem-Vehid2010). Dissociation may contribute to the emergence or intensification of psychotic symptoms through impaired reality control and intrusive traumatic images, rather than through the production of purely transient or stress-induced phenomena. However, the number of studies specifically assessing dissociation in this population remains limited and longitudinal data are lacking.
Substance use is another recurring factor. Three studies have reported a higher prevalence of substance abuse in FEP and BPD samples than in FEP samples alone (Francey et al., Reference Francey, Jovev, Phassouliotis, Cotton and Chanen2018; Gleeson et al., Reference Gleeson, Chanen, Cotton, Pearce, Newman and McCutcheon2012; Pelizza et al., Reference Pelizza, Leuci, Quattrone, Azzali, Paulillo, Pupo, Pellegrini, Gammino, Biancalani and Menchetti2025). In a retrospective cohort, 62.8% of people with both BPD and SUD developed substance-induced psychosis, most often associated with cocaine or cannabis dependence (Barral et al., Reference Barral, Rodríguez-Cintas, Grau-López, Daigre, Ros-Cucurull, Calvo, Ferrer and Roncero2018). Substance use may act both as a trigger and as a maladaptive coping strategy for underlying affective instability and trauma-related distress. Longitudinal data also indicate higher persistence of SUD in BPD and schizophrenia samples (Cailhol et al., Reference Cailhol, Pelletier, Rochette, Renaud, Koch, David and Lesage2021), although findings remain scarce. Taken together, these findings suggest an interactive model in which trauma, dissociation, and substance use dynamically reinforce vulnerability to psychotic symptoms in people with BPD, supporting a dimensional rather than categorical view of comorbidity.
Social and professional functioning
The impact of co-occurrence on social and occupational functioning has yielded mixed results. One early study found no difference in functioning at baseline between groups but noted a decline in the comorbid group after 1 year (Bahorik & Eack, Reference Bahorik and Eack2010). Other studies have reported poorer interpersonal functioning at entry in people experiencing FEP with comorbid BPD (Schandrin et al., Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023) but paradoxically higher employment rates in the same group (Pelizza et al., Reference Pelizza, Leuci, Quattrone, Azzali, Paulillo, Pupo, Pellegrini, Gammino, Biancalani and Menchetti2025). These apparent contradictions may partly reflect differences between the measures of functional outcomes used in different studies, ranging from self-assessments of quality of life to objective indicators such as employment status or size of social network.
Some discrepancies emerged between studies concerning education and occupation. While some authors reported decreased educational and occupational functioning in comorbid patients (Bahorik & Eack, Reference Bahorik and Eack2010; Gleeson et al., Reference Gleeson, Chanen, Cotton, Pearce, Newman and McCutcheon2012), others observed better short-term outcomes, perhaps reflecting the heterogeneity of symptom profiles, particularly among people diagnosed with PNOS (Schandrin et al., Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023). In these cases, short-term improvements may be related to increased engagement in care or residual protective factors, rather than a true functional benefit.
Longitudinal data remain scarce, but the few follow-up studies available suggest that comorbidity may be associated with more unstable functional trajectories over time, alternating between phases of social recovery and relational withdrawal (Bahorik & Eack, Reference Bahorik and Eack2010; Gleeson et al., Reference Gleeson, Chanen, Cotton, Pearce, Newman and McCutcheon2012). Overall, the level of evidence remains limited and often comes from small, non-randomized cohorts.
Access to care and treatment
Evidence regarding access to care was heterogeneous, but revealed several important trends. In FEP cohorts, people with BPD comorbidity had lower access to specialized early intervention services than those with FEP alone (Francey et al., Reference Francey, Jovev, Phassouliotis, Cotton and Chanen2018; Schandrin et al., Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023). However, these people made greater use of general psychiatric services, particularly outpatient and psychotherapy services (Pelizza et al., Reference Pelizza, Leuci, Quattrone, Azzali, Paulillo, Pupo, Pellegrini, Gammino, Biancalani and Menchetti2025; Cailhol et al., Reference Cailhol, Pelletier, Rochette, Renaud, Koch, David and Lesage2021). This tendency may reflect both service-level exclusion criteria (e.g. diagnostic limitations restricting access to early psychosis intervention programs) and patient-related factors such as relationship instability, emotional dysregulation, or previous negative treatment experiences.
Hospitalization rates vary. In one study, women with comorbid BPD were hospitalized faster than their counterparts without BPD, a trend not observed in men (Bahorik & Eack, Reference Bahorik and Eack2010). Another study reported a hospitalization rate almost twice as high in the comorbid group (Wickett et al., Reference Wickett, Essman, Beck-Jennings, Davis, McIlvried and Lysaker2006). However, some results, particularly with regard to involuntary admissions and emergency room visits, are still inconclusive.
Only one randomized trial has evaluated a specific co-treatment intervention: Gleeson et al (Reference Gleeson, Chanen, Cotton, Pearce, Newman and McCutcheon2012) found that a combined program (Helping Young People Early [HYPE] for BPD + Specialist First-Episode Treatment for FEP) reduced psychotic symptoms and improved functioning, with full adherence from participants. This study confirms the feasibility and clinical value of integrated care models that simultaneously treat affective instability and psychotic symptomatology.
Pharmacological results are more contradictory. Francey et al. (Reference Francey, Jovev, Phassouliotis, Cotton and Chanen2018) reported higher antipsychotic doses and side effect rates in the comorbid FEP and BPD group, while Pelizza et al. (Reference Pelizza, Leuci, Quattrone, Azzali, Paulillo, Pupo, Pellegrini, Gammino, Biancalani and Menchetti2025) observed lower initial prescription rates but greater discontinuity over time. These differences likely stem from heterogeneous prescribing practices and clinicians’ uncertainty about the appropriate therapeutic goal: affect regulation or psychosis control.
Given the scarcity of high-quality interventional studies, these results should be interpreted as preliminary. Overall, current data highlight a gap between clinical needs and service organization, suggesting that integrative psychotherapeutic approaches such as mentalization-based treatment or other transdiagnostic models may offer more suitable frameworks for this complex population.
Discussion
This systematic review and meta-analysis reveals a significant co-occurrence between BPD and psychotic disorders across three complementary meta-analyses. The pooled prevalence of BPD in psychotic disorders was 22.7%, the prevalence of psychotic disorders in BPD was 14.3%, and the prevalence of BPD in FEP reached 40.0%. These results highlight a strong, bidirectional association between the two diagnostic entities, particularly in the early stages of psychotic disorders.
People with co-occurring disorders, often young women, tend to present with more severe clinical symptoms, including increased emotional distress, suicidality, and psychotic features such as AVHs. These findings are consistent with previous clinical description, perceptual abnormalities, and unstable self-experience (Gunderson, Reference Gunderson1984; Kernberg, Reference Kernberg1967).
Factors such as childhood trauma, dissociation, and substance use have been frequently reported and may play a role in the emergence and persistence of this comorbidity. Beyond psychosocial vulnerability, new data point to neurological mechanisms between BPD and psychotic disorders, particularly in the areas of social cognition, mentalization and differentiation between self and others people (Lysaker et al., Reference Lysaker, Kukla, Leonhardt, Hamm, Schnakenberg Martin, Zalzala, Gagen and Hasson-Ohayon2020; Weijers et al., Reference Weijers, Ten Kate, Viechtbauer, Rampaart, Eurelings and Selten2021).
Despite the promising results of combined treatment programs (Chanen et al., Reference Chanen, Kerslake, Berubé, Nicol, Jovev, Yuen, Betts, McDougall, Nguyen, Cavelti and Kaess2024; Gleeson et al., Reference Gleeson, Chanen, Cotton, Pearce, Newman and McCutcheon2012), co-occurrence remains associated with poorer functional outcomes and increased clinical complexity, including higher drug dosages and more frequent hospitalizations (Francey et al., Reference Francey, Jovev, Phassouliotis, Cotton and Chanen2018; Wickett et al., Reference Wickett, Essman, Beck-Jennings, Davis, McIlvried and Lysaker2006). Sensitivity analyses confirmed the robustness of our quantitative results, although publication bias may have influenced estimates in studies focusing on FEP. Quality assessment revealed substantial methodological heterogeneity between studies, notably due to non-randomized samples, cross-sectional designs, and limited adjustments for confounders.
Overall, these findings underline the importance of moving beyond a categorical view of psychiatric comorbidity. The frequent co-occurrence of BPD and psychotic disorders requires a dimensional and transdiagnostic perspective, integrating developmental, clinical, neurobiological, and contextual factors. Historically, Kernberg’s structural model of personality organization already emphasized the overlap between emotional instability, self-disruption, and reality testing. More recent frameworks, such as the Research Domain Criteria (RDoC; Insel et al., Reference Insel, Cuthbert, Garvey, Heinssen, Pine, Quinn, Sanislow and Wang2010) and the Hierarchical Taxonomy of Psychopathology (HiTOP; Kotov et al., Reference Kotov, Krueger, Watson, Achenbach, Althoff, Bagby, Brown, Carpenter, Caspi, Clark, Eaton, Forbes, Forbush, Goldberg, Hasin, Hyman, Ivanova, Lynam, Markon and Zimmerman2017), provide complementary tools for studying these overlaps in a dimensional manner. The RDoC approach organizes mental functions into domains (e.g. negative valence, cognitive control, and social processes) that transcend traditional diagnoses, while HiTOP proposes hierarchical dimensions of psychopathology (from specific symptoms to broad spectra) rather than distinct categories. Together, these models offer conceptual frameworks for understanding how affective dysregulation and psychotic symptoms may coexist within partially shared systems of vulnerability.
In the following sections, we synthesize the main results obtained in the different thematic areas in order to elucidate the mechanisms and implications of this comorbidity.
Exacerbated symptoms in psychotic disorders and BPD comorbidity
Across studies, people with co-occurring BPD and psychotic disorders consistently presented more severe, complex, and heterogeneous symptom profiles than those with psychosis alone. This exacerbation was evident across both affective and psychotic dimensions, with heightened emotional dysregulation, impulsivity, self-harm, and suicidal ideation frequently reported (Bahorik & Eack, Reference Bahorik and Eack2010; Lysaker, Reference Lysaker, Wickett, Lancaster and Davis2004; Sar et al., Reference Sar, Taycan, Bolat, Özmen, Duran, Öztürk and Ertem-Vehid2010). Comorbid patients also exhibited greater affective instability and higher prevalence of AVHs, particularly in early psychosis contexts (Francey et al., Reference Francey, Jovev, Phassouliotis, Cotton and Chanen2018; Gleeson et al., Reference Gleeson, Chanen, Cotton, Pearce, Newman and McCutcheon2012; Pelizza et al., Reference Pelizza, Leuci, Quattrone, Azzali, Paulillo, Pupo, Pellegrini, Gammino, Biancalani and Menchetti2025).
Indeed, AVHs in these patients were not only more prevalent but also perceived as more distressing, personalized, and emotionally salient (Slotema et al., Reference Slotema, Blom, Niemantsverdriet, Deen and Sommer2018; Kingdon et al., Reference Kingdon, Ashcroft, Bhandari, Gleeson, Warikoo, Symons, Taylor, Lucas, Mahendra, Ghosh, Mason, Badrakalimuthu, Hepworth, Read and Mehta2010). This is consistent with previous findings suggesting that trauma-related hallucinations often originate in interpersonal and emotional contexts, reflecting increased reactivity to attachment-related stressors and traumatic memories rather than a primary thought disorder process (Hepworth et al., Reference Hepworth, Ashcroft and Kingdon2013; Pearse et al., Reference Pearse, Dibben, Ziauddeen, Denman and McKenna2014).
From a dimensional perspective, the co-occurrence of BPD and psychotic disorders could reflect an interaction between stress sensitivity and socio-cognitive dysregulation mechanisms (Crowell, Beauchaine, & Linehan, Reference Crowell, Beauchaine and Linehan2009; Fonagy & Luyten, Reference Fonagy and Luyten2009; Lysaker et al., Reference Lysaker, Kukla, Leonhardt, Hamm, Schnakenberg Martin, Zalzala, Gagen and Hasson-Ohayon2020). Interpersonal instability, sensitivity to rejection, and mentalization deficits may lower the threshold for psychotic experiences in vulnerable people, leading to emotionally charged and clinically significant psychotic phenomena, rather than transient or secondary ones. Importantly, these mechanisms suggest that symptom exacerbation is not simply additive but rather synergistic, resulting in a distinct high-risk clinical phenotype characterized by emotional volatility, affective lability, and diminished coping resources.
Clinically, this pattern complicates early detection and diagnosis. The affective instability associated with BPD can mask emerging psychotic symptoms, leading to delayed recognition or misdiagnosis, particularly in young adults entering early intervention services (Cristea et al., Reference Cristea, Gentili, Cotet, Palomba, Barbui and Cuijpers2017; Paris, Reference Paris2018). In addition, the high risk of suicidal behavior and impulsive crises underscores the need for careful risk monitoring and integrated care pathways (Paris, Reference Paris2004).
In summary, the co-occurrence of BPD and psychotic disorders increases clinical severity and functional burden. Rather than representing two separate conditions, this presentation may define a subgroup with common vulnerability mechanisms that warrant specific assessment, early identification, and tailored interventions. These mechanisms are examined in more detail in the following section on trauma, dissociation, and substance use.
Trauma, dissociation, and substance use as transdiagnostic vulnerability factors
A considerable number of studies have identified childhood trauma exposure as a shared etiological factor in people with BPD and co-occurring psychotic disorders (Altunsoy et al., Reference Altunsoy, Sahiner, Cingi Kuluk, Okay, Ulusoy Kaymak, Aydemir and Goka2015; Sar et al., Reference Sar, Taycan, Bolat, Özmen, Duran, Öztürk and Ertem-Vehid2010; Zanarini et al., Reference Zanarini, Williams, Lewis, Reich, Vera, Marino, Levin, Yong and Frankenburg1997). This finding is consistent with the wider literature suggesting that early adverse experiences, particularly emotional neglect, physical abuse, and sexual violence, are major risk factors for both BPD and psychosis (Sitko et al., Reference Sitko, Bentall, Shevlin, O’Sullivan and Sellwood2014; Varese et al., Reference Varese, Smeets, Drukker, Lieverse, Lataster, Viechtbauer, Read, van Os and Bentall2012). Importantly, trauma does not imply a single nosological continuum between the two disorders but rather a set of common vulnerability mechanisms that may contribute independently to each diagnosis while facilitating their co-occurrence. Developmental disruptions in stress regulation, attachment, and self-integration may increase the risk of developing BPD and psychotic features, increasing the likelihood of their joint manifestation in vulnerable people.
In several studies, dissociation has emerged as a central mediator between trauma and symptom expression in this population. People with both BPD and psychotic disorders frequently report dissociative experiences that blur the boundaries between trauma-related intrusive and perceptual abnormalities such as AVHs (Pearse et al., Reference Pearse, Dibben, Ziauddeen, Denman and McKenna2014; Sar et al., Reference Sar, Taycan, Bolat, Özmen, Duran, Öztürk and Ertem-Vehid2010). These phenomena do not suggest diagnostic overlap or misclassification of BPD as psychosis but rather illustrate phenomenological intersections through which trauma can influence both affective and perceptual domains. From this perspective, dissociation may potentiate psychotic symptoms, such as AVHs, by amplifying emotional arousal, memory fragmentation, and impaired source control, without negating the distinct diagnostic validity of psychotic disorders.
Substance use has also been frequently reported, particularly among people with early psychosis associated with BPD (Chanen et al., Reference Chanen, Kerslake, Berubé, Nicol, Jovev, Yuen, Betts, McDougall, Nguyen, Cavelti and Kaess2024; Francey et al., Reference Francey, Jovev, Phassouliotis, Cotton and Chanen2018). Cannabis and alcohol use may serve as maladaptive strategies to regulate emotional dysregulation or to alleviate trauma-related distress but their use increases the risk of relapse, worsens clinical outcomes, and impairs compliance (Bahorik & Eack, Reference Bahorik and Eack2010; Kingdon et al., Reference Kingdon, Ashcroft, Bhandari, Gleeson, Warikoo, Symons, Taylor, Lucas, Mahendra, Ghosh, Mason, Badrakalimuthu, Hepworth, Read and Mehta2010). Beyond self-medication, substance use can interact with trauma and dissociation by destabilizing cognitive control and emotion regulation systems, which can exacerbate borderline and psychotic symptoms.
Taken together, trauma, dissociation, and substance use constitute transdiagnostic vulnerability factors that increase the risk, persistence, and complexity of the co-occurrence of BPD and psychotic disorders. They do not erase diagnostic boundaries but reveal interactive mechanisms that shape their simultaneous expression. These findings underscore the need for integrative care models that take trauma into account and are capable of treating both disorders simultaneously, each within its own clinical coherence but within a common developmental and neurobiological framework.
Diagnostic and nosological challenges
The significant co-occurrence of BPD and psychotic disorders raises complex diagnostic and nosological issues. Current classification systems, such as DSM-5 (American Psychiatric Association, 2013) and ICD-11 (World Health Organization, 2019), still maintain a categorical separation between personality disorders and psychotic disorders, reinforcing the idea of mutually exclusive entities. However, current findings challenge this separation: several studies have shown that BPD and psychotic disorders can coexist as distinct but interactive conditions, particularly when affective instability, dissociation, and psychotic features converge in the same clinical presentation (Kingdon et al., Reference Kingdon, Ashcroft, Bhandari, Gleeson, Warikoo, Symons, Taylor, Lucas, Mahendra, Ghosh, Mason, Badrakalimuthu, Hepworth, Read and Mehta2010; Paris, Reference Paris2018). Rather than indicating diagnostic confusion, this overlap highlights the need for a more integrative framework capable of capturing the simultaneous expressions of both disorders.
Several studies included in this review revealed considerable variability and ambiguity in diagnostic practices. For example, psychotic symptoms in people with BPD are sometimes misclassified as dissociative or ‘pseudo-hallucinatory’, which can delay the recognition and treatment of an underlying psychotic disorder (Pearse et al., Reference Pearse, Dibben, Ziauddeen, Denman and McKenna2014; Zanarini, Gunderson, Frankenburg, & Chauncey, Reference Zanarini, Gunderson, Frankenburg and Chauncey1990). Conversely, affective dysregulation, impulsivity, or identity disturbances during a FEP may be misinterpreted as a personality disorder, particularly in the absence of longitudinal follow-up (Cristea et al., Reference Cristea, Gentili, Cotet, Palomba, Barbui and Cuijpers2017). These bidirectional diagnostic errors reflect the limitations of current categorical systems, which struggle to take into account complex and overlapping clinical presentations without reducing their specificity.
The nosological tension is further compounded by the dimensional nature of both disorders. Empirical studies suggest that borderline traits and psychotic experiences are distributed continuously across the general population and share vulnerability factors such as childhood trauma, attachment disturbances, and cognitive-affective dysregulation (Fonagy & Luyten, Reference Fonagy and Luyten2009; Reininghaus et al., Reference Reininghaus, Kempton, Valmaggia, Craig, Garety, Onyejiaka, Gayer-Anderson, So, Hubbard, Beards, Dazzan, Pariante, Mondelli, Fisher, Mills, Viechtbauer, McGuire, van Os, Murray and Morgan2016). However, recognition of these common dimensions does not imply that BPD and psychotic disorders belong to a single continuum. Rather, it supports a pluralistic model in which dimensional constructs (e.g. affective instability, reality testing, and self-coherence) may overlap across distinct categorical diagnoses. In this sense, dimensional frameworks such as the HiTOP (Kotov et al., Reference Kotov, Krueger, Watson, Achenbach, Althoff, Bagby, Brown, Carpenter, Caspi, Clark, Eaton, Forbes, Forbush, Goldberg, Hasin, Hyman, Ivanova, Lynam, Markon and Zimmerman2017) and RDoC (Insel et al., Reference Insel, Cuthbert, Garvey, Heinssen, Pine, Quinn, Sanislow and Wang2010) offer conceptual tools for mapping these intersections without erasing nosological boundaries. HiTOP emphasizes empirically derived spectra (e.g. thought disorders and emotional dysregulation), while RDoC focuses on neurobiological and behavioral domains (e.g. negative valence systems and cognitive control). In this way, both promote a dimensional understanding that may be better suited to co-occurrence.
From a clinical perspective, the lack of a clear diagnostic consensus carries tangible risks. An ambiguous label can lead to fragmented care pathways, inadequate pharmacological management, or exclusion from specialized services. Furthermore, the persistent stigma surrounding personality disorders can lead clinicians to underestimate or ignore psychotic symptoms, particularly in emergency situations or in hospital settings (Aviram, Brodsky, & Stanley, Reference Aviram, Brodsky and Stanley2006; Lewis & Appleby, Reference Lewis and Appleby1988). Conversely, focusing solely on psychosis can obscure the enduring personality traits that determine the course, compliance, and recovery of these patients.
Overall, our findings highlight the need to refine diagnostic instruments and clinician training. Integrating longitudinal assessment, dimensional formulations, and trauma-informed assessment into practice could improve diagnostic accuracy while maintaining categorical integrity. This approach recognizes the coexistence of two distinct but often closely related conditions and promotes more consistent care strategies for this complex and underdiagnosed population.
Developmental and neurobiological underpinnings
Recent research points to interacting neurodevelopmental and neurobiological mechanisms that may underlie the frequent co-occurrence between BPD and psychotic disorders. Both disorders have been associated with early adversity and attachment disturbances, which can lead to persistent dysregulations in stress response systems such as the hypothalamic–pituitary–adrenal (HPA) axis. These early vulnerabilities appear to interact with abnormalities in fronto-limbic connectivity, notably reduced top-down regulation between the amygdala and the prefrontal cortex, and mechanisms consistently implicated in affective hyperactivity, impulsivity, and distortions in self-other representations (Goodkind et al., Reference Goodkind, Eickhoff, Oathes, Jiang, Chang, Jones-Hagata, Ortega, Zaiko, Roach, Korgaonkar, Grieve, Galatzer-Levy, Fox and Etkin2015; Herpertz et al., Reference Herpertz, Dietrich, Wenning, Krings, Erberich, Willmes, Thron and Sass2001). These alterations may constitute a neurobiological substrate promoting the simultaneous expression of borderline and psychotic symptomatic dimensions rather than a unitary pathway.
Sleep disturbances also appear central in this shared vulnerability. Insomnia, circadian misalignment, and trauma-related nightmares are widespread in both BPD and psychotic disorders and are linked to heightened emotional reactivity, dissociation, and perceptual abnormalities (Waite et al., Reference Waite, Sheaves, Isham, Reeve and Freeman2020; Reeve et al., Reference Reeve, Sheaves and Freeman2015). These findings suggest that sleep dysfunction acts as both a marker and mediator of psychiatric risk, influencing neural systems involved in affective regulation, cognitive control, and salience attribution. Chronic sleep disorders can therefore amplify existing vulnerabilities, thereby increasing the risk of simultaneous activation of symptoms in both diagnostic areas.
From a developmental standpoint, the early onset of BPD and FEP has been interpreted by some authors as reflecting a partially convergent pathway of altered social cognition and neurocognitive integration (Sharp et al., Reference Sharp, Pane, Ha, Venta, Patel, Sturek and Fonagy2011). Indeed, deficits in theory of mind, agency, and autobiographical reasoning have been observed in both conditions and may contribute to the emergence of interpersonal paranoia, mistrust, and unstable self-concept, features shared across categories (Dimaggio et al., Reference Dimaggio, Salvatore, Popolo and Lysaker2012; Lysaker et al., Reference Lysaker, Kukla, Dubreucq, Gumley, McLeod, Vohs, Buck, Minor, Luther, Leonhardt, Belanger, Popolo and Dimaggio2015; Sharp et al., Reference Sharp, Pane, Ha, Venta, Patel, Sturek and Fonagy2011). However, in cases of comorbidity, these dysfunctions appear to add up and interact rather than being identical, suggesting that each disorder may potentiate the cognitive and affective disturbances of the other through overlapping but distinct mechanisms.
Although no study to date has directly assessed neurocognitive performance in people diagnosed with both BPD and psychotic disorders, independent meta-analyses indicate that both groups exhibit comparable deficits in attention, working memory, and executive function (D’Iorio et al., Reference D’Iorio, Di Benedetto and Santangelo2024; Fenske et al., Reference Fenske, Lis, Liebke, Niedtfeld, Kirsch and Mier2015; Lysaker et al., Reference Lysaker, Erickson, Ringer, Buck, Semerari, Carcione and Dimaggio2011; Ventura et al., Reference Ventura, Hellemann, Thames, Koellner and Nuechterlein2009). These cognitive impairments may act synergistically in people with both disorders, contributing to increased clinical severity, disorganized thinking, and impaired adaptive functioning. Furthermore, neurocognitive dysfunction may weaken metacognitive abilities, thereby limiting the capacity to form integrated representations of self and others, a process essential for emotional regulation and identity coherence (Lysaker et al., Reference Lysaker, Erickson, Ringer, Buck, Semerari, Carcione and Dimaggio2011, Reference Lysaker, Kukla, Dubreucq, Gumley, McLeod, Vohs, Buck, Minor, Luther, Leonhardt, Belanger, Popolo and Dimaggio2015).
Overall, these findings highlight the importance of integrative, multilevel models that link neurobiological, developmental, and experiential processes. These models do not postulate the existence of a common disorder but rather a convergence of risk mechanisms that can combine to form two distinct clinical trajectories. Future studies should aim to identify common and disorder-specific neural correlates using longitudinal neuroimaging, sleep electroencephalogram, and digital phenotyping approaches to clarify co-occurrence mechanisms and guide targeted interventions. Figure 5 illustrates a proposed integrative model synthesizing these multilevel interactions across developmental, neurobiological, and experiential domains.
Integrative model of the co-occurrence of psychotic disorders and BPD. Note: This conceptual model synthesizes the multilevel interactions involved in the frequent co-occurrence between BPD and psychotic disorders. Early vulnerabilities, such as childhood trauma and attachment insecurity (Level 1), contribute to long-lasting alterations in stress response systems and emotional regulation, including dysregulation of the HPA axis, abnormalities in fronto-limbic connectivity, and deficits in social cognition and executive function (Level 2). These neurobiological and cognitive vulnerabilities converge to produce a common clinical phenotype (Level 3), characterized by emotional dysregulation, impulsivity, identity instability, transient psychotic experiences, and social dysfunction. Aggravating factors (Level 4) such as chronic stress, stigmatization, sleep disorders, and discontinuity of care contribute to the persistence and intensification of symptoms, reinforcing a trajectory of functional decline and psychiatric chronicization (Level 5). Arrows indicate assumed causal or bidirectional relationships between levels. Although not represented graphically for reasons of legibility, chronic psychiatric conditions can, in turn, progressively alter neurocognitive and emotional functioning via prolonged exposure to stress and social exclusion, thus reinforcing the cycle of dysregulation.
Finally, chronic stress, persistent sleep dysfunction, and cumulative social exclusion may reinforce each other over time, creating a self-perpetuating feedback loop that increases symptom persistence and the risk of long-term functional impairment and psychiatric chronicization.
Functioning and access to care
The co-occurrence of BPD and psychotic disorders is consistently associated with substantial impairments in real-life functioning, including reduced social and occupational functioning, diminished autonomy, and higher rates of unemployment. Across studies, these functional difficulties appear greater than those observed in single-diagnosis groups, particularly in early intervention cohorts (e.g. Francey et al., Reference Francey, Jovev, Phassouliotis, Cotton and Chanen2018; Gleeson et al., Reference Gleeson, Chanen, Cotton, Pearce, Newman and McCutcheon2012), although the magnitude of this difference varies. For instance, Schandrin et al. (Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023) found no significant difference in baseline functioning between FEP with or without BPD comorbidity, while Pelizza et al. (Reference Pelizza, Leuci, Quattrone, Azzali, Paulillo, Pupo, Pellegrini, Gammino, Biancalani and Menchetti2025) included only participants with both diagnoses, precluding group comparison. These discrepancies underscore the need for longitudinal designs to assess the trajectories of functioning and adaptation over time, ideally controlling for symptom severity and treatment exposure. Despite these inconsistencies, most data converge toward a model of aggravated disability, where difficulties in maintaining interpersonal relationships, daily routines, and employment reflect the additional burden of co-occurring disorders (Gunderson et al., Reference Gunderson, Stout, McGlashan, Shea, Morey, Grilo, Zanarini, Yen, Markowitz, Sanislow, Ansell, Pinto and Skodol2011; Lysaker et al., Reference Lysaker, Carcione, Dimaggio, Johannesen, Nicolò, Procacci and Semerari2005; Zanarini et al., Reference Zanarini, Frankenburg, Reich and Fitzmaurice2010). Further research is needed to determine whether this dual-diagnosis group represents a qualitatively distinct functional trajectory or an exacerbation of preexisting vulnerabilities.
Access to care is another major challenge for this population. People with both BPD and psychotic disorders often oscillate between several types of psychiatric care (e.g. general psychiatry, early psychosis intervention programs, and units specializing in personality disorder) without benefiting from a stable or integrated therapeutic framework. Diagnostic uncertainty often leads to fragmented care, inconsistent formulations, and fragile therapeutic alliances. In some cases, the comorbidity of BPD leads to exclusion from early psychosis intervention services, even in the presence of obvious psychotic symptoms, and the reverse is also true. This ‘diagnostic ping-pong’ effect delays appropriate intervention, promotes disengagement from care, and contributes to the chronicity of symptoms (Schandrin et al., Reference Schandrin, Francey, Nguyen, Whitty, McGorry, Chanen and O’Donoghue2023).
Barriers to care are also reinforced by clinician attitudes and structural limitations. Negative stereotypes surrounding BPD, including assumptions of manipulation or treatment resistance (Aviram et al., Reference Aviram, Brodsky and Stanley2006; Lawn & McMahon, Reference Lawn and McMahon2015), reduce clinicians’ willingness to engage and may perpetuate stigma. Furthermore, organizational silos and the lack of common protocols for managing the dual diagnoses, BPD and psychotic disorders, hinder the coordination of interventions. New models incorporating transdiagnostic, trauma-informed, and stage-based approaches such as HYPE and EPPIC, offer promising prospects for more consistent and developmentally appropriate care. Their expansion and systematic evaluation in different healthcare settings is a crucial next step in improving access and outcomes for this underserved group.
Clinical and research implications
These findings emphasize the importance of moving beyond rigid diagnostic boundaries in both clinical and research contexts. Rather than adhering strictly to categorical nosologies, clinicians should be trained to recognize and treat co-occurring affective and psychotic symptoms as interacting yet distinct processes within the same people. Emotional dysregulation, dissociation, and psychotic features can co-occur in clinically meaningful constellations that warrant integrated, multimodal interventions (Lysaker et al., Reference Lysaker, Erickson, Ringer, Buck, Semerari, Carcione and Dimaggio2011; Zanarini et al., Reference Zanarini, Frankenburg, Reich, Conkey and Fitzmaurice2015).
Transdiagnostic and modular therapeutic frameworks appear particularly suited to this comorbid population. Evidence-based models, such as dialectical behavior therapy and cognitive behavioral therapy for psychosis, have demonstrated their effectiveness in treating affective instability and psychotic experiences, respectively (Lynch, Trost, Salsman, & Linehan, Reference Lynch, Trost, Salsman and Linehan2007; van der Gaag, van den Berg, & Ising, Reference van der Gaag, van den Berg and Ising2019). When adapted to early intervention and youth mental health contexts, these therapies can also address identity diffusion, metacognitive disorders, and interpersonal dysfunction, which are characteristics common to both diagnostic categories (Cottraux et al., Reference Cottraux, Note, Boutitie, Milliery, Genouihlac, Yao, Note, Mollard, Bonasse, Gaillard, Djamoussian, Guillard Cde, Culem and Gueyffier2009; Dimaggio & Norcross, Reference Dimaggio and Norcross2008; Gleeson et al., Reference Gleeson, Chanen, Cotton, Pearce, Newman and McCutcheon2012). Combining these models in progressive or modular frameworks may improve flexibility and facilitate engagement in diverse clinical trajectories.
From a research perspective, there is an urgent need for longitudinal, multimethod studies to clarify the developmental mechanisms leading to co-occurrence. Future studies should explore how exposure to trauma, attachment disorders, neurocognitive vulnerabilities, and sleep disorders interact dynamically over time to shape risk trajectories (Insel, Reference Insel2017; Reeve, Sheaves, & Freeman, Reference Reeve, Sheaves and Freeman2015). Innovative digital methodologies, including ecological momentary assessment, mobile sleep tracking, and digital phenotyping can capture these micro-level processes and refine our understanding of symptom fluctuations and environmental sensitivity.
It is also essential to address gender and diagnostic biases. The over-representation of women in BPD diagnoses and the possible under-recognition of psychotic symptoms in this group highlight how gender biases can skew prevalence estimates and delay appropriate treatment (Bayes & Parker, Reference Bayes and Parker2020; Paris, Reference Paris2003). Future research should disentangle sociocultural and biological influences on these disparities and promote diagnostic equity across populations.
At the healthcare system level, tiered and stepped care models that incorporate trauma-informed and developmentally appropriate interventions may help reduce fragmentation and improve long-term outcomes (Francey et al., Reference Francey, Jovev, Phassouliotis, Cotton and Chanen2018; Gleeson et al., Reference Gleeson, Chanen, Cotton, Pearce, Newman and McCutcheon2012; McGorry et al., Reference McGorry, Purcell, Hickie, Yung, Pantelis and Jackson2007). Programs such as HYPE and EPPIC illustrate how flexible, person-centered services can be integrated into existing psychiatric infrastructures. However, scaling up these models requires sustained institutional commitment, interprofessional collaboration, and specialized staff training.
Ultimately, this co-occurrence of BPD and psychotic disorders requires a dimensional, developmentally based, person-centered paradigm. Such a paradigm should address not only symptoms but also functioning, identity development, and subjective experience, thereby promoting diagnostic clarity, therapeutic engagement, and recovery-oriented outcomes.
Limitations
This systematic review and meta-analysis presents several limitations that should be acknowledged. First, the number of studies explicitly examining the co-occurrence of BPD and psychotic disorders remains limited, restricting the generalization of conclusions. Existing studies are characterized by heterogeneous samples, diagnostic criteria, and methodological approaches which may contribute to variability across findings. Moreover, most included studies had small sample sizes and cross-sectional designs, precluding the robust examination of developmental or longitudinal trajectories in comorbid populations.
Second, our quantitative synthesis was based on three separate meta-analyses assessing prevalence from different diagnostic perspectives (BPD in psychosis, psychosis in BPD, and BPD in FEP). Although sensitivity analyses supported the robustness of the results, publication bias cannot be excluded, as studies reporting null or moderate associations are less likely to be published, potentially inflating prevalence estimates.
Third, the overall methodological quality of the included studies was rated as ‘weak’ using the EPHPP tool. Common sources of bias included non-randomized sampling, lack of blinding, and incomplete control for confounding variables, such as trauma history or substance use disorder, factors known to influence both disorders’ expression and interaction. In several cases, diagnostic procedures relied on self-report measures rather than structured clinical interviews, further limiting diagnostic reliability.
Finally, few studies provided detailed clinical or pharmacological data, preventing an in-depth exploration of treatment effects or medication–symptom interactions in the co-occurring group. Similarly, information on psychosocial interventions and longitudinal outcomes was often missing, restricting insight into functional recovery and care trajectories.
Taken together, these methodological shortcomings highlight the urgent need for well-powered, longitudinal, and multisite studies that use standardized diagnostic tools, dimensional measures, and trauma-informed frameworks to clarify the mechanisms and clinical implications of this complex comorbidity.
Conclusion
Overall, this systematic review and meta-analysis provides strong evidence for the clinical and nosological relevance of the co-occurrence between BPD and psychotic disorders, particularly during the early stages of the illness. Across three complementary meta-analyses, high prevalence rates were observed in both psychotic and FEP samples, confirming that the association between these disorders is not incidental but reflects a meaningful clinical phenomenon. Shared vulnerability factors, such as childhood trauma, dissociation, emotional dysregulation, and neurocognitive impairment, support the view that these conditions may interact dynamically within a common developmental framework.
Rather than representing diagnostic overlap or artifact, this co-occurrence may delineate a distinct clinical phenotype characterized by heightened emotional distress, identity instability, and recurrent psychotic experiences. Such profiles challenge traditional categorical classifications and call for dimensional, transdiagnostic, and developmentally grounded approaches capable of capturing the fluidity and interaction of affective and psychotic processes over time.
Clinically, these findings underscore the need for integrated and flexible care models that can accommodate psychiatric complexity without reinforcing diagnostic silos. People experiencing both BPD and psychotic disorders are particularly vulnerable to exclusion, stigmatization, and fragmentation care pathways as they move across diagnostic and service boundaries. Building clinical systems capable of sustaining continuity of care, while tailoring interventions to people’s needs, represents a critical step toward more equitable and recovery-oriented mental health services.
Future research should extend beyond categorical comparisons and symptom aggregation to investigate longitudinal mechanisms of interactions between trauma, neurocognition, self-structure, and affective regulation. This endeavor will require combining fine-grained clinical assessment with ecological and digital tools to capture the dynamic and context-dependent processes shaping these trajectories.
Ultimately, addressing the complexity of dual diagnoses will not only enhance clinical outcomes but also contribute to redefining psychiatry toward a person-centered, developmentally sensitive, and integrative paradigm, one that recognizes co-occurrence as an opportunity to deepen, rather than fragment, our understanding of psychopathology.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0033291726103432.
Competing interests
The authors declare that they have no conflicts of interest relevant to the content of this article.
Open access
