A meta-analysis on the neuropsychological correlates of Borderline Personality Disorder: An update

Previous research on Borderline Personality Disorder (BPD) demonstrated dysfunction across a broad range of cognitive domains. However, the limited number of neuropsychological studies on BPD and their occasionally conflicting results have precluded a clear characterization of the neuropsychological features associated with this personality disorder. Therefore, the main aim of the present study is to provide an updated overview of neuropsychological functions related to BPD. A meta-analysis of 36 studies was performed, comparing the performance of BPD patients and healthy controls (HCs) across several cognitive domains. Significant differences between BPD patients and HCs in multiple cognitive domains were found. The smallest effect size was observed for general executive function, while the largest effect sizes were found in the long-term spatial memory and inhibition domains. In conclusion, the neuropsychological profile of BPD, characterized by deficits in inhibition as well as attention, memory, and executive functions, can result in difficulties in performing everyday activities. Accordingly, assessing neuropsychological functions could assist clinicians in developing more targeted non-pharmacological treatments.


Introduction
Borderline Personality Disorder (BPD) is characterized by a pervasive pattern of instability in interpersonal relationships, self-image, and affects, as well as marked impulsivity (APA, 2013) (American Psychiatric Association, 2013).Clinical manifestations of BPD include emotional dysregulation, impulsive aggression, self-injury, and an increased risk of suicide (Lieb et al., 2004;Fonagy and Luyten, 2009).BPD is widely recognized as a debilitating condition that results in functional impairment (Skodol et al., 2002;Soeteman et al., 2008a;Gunderson et al., 2011) and psychological suffering (Zanarini et al., 1998).A recent review (Winsper et al., 2020) reported that the global prevalence of personality disorders is 7.8 %, with cluster B disorders alone accounting for 2.8 %.The global prevalence of BPD is estimated to be 1.8 %.Although BPD is predominantly diagnosed in women, with a prevalence ratio of three to one (approximately 75 %) (APA, 2013), some studies suggest that it may be equally prevalent in men (Torgersen et al., 2001;Lenzenweger et al., 2007).It should be noted that the gender differences observed in some clinical studies may be due to selection bias (Ten Have et al., 2016).Despite its prevalence being no higher than other personality disorders in the general population, BPD has a high prevalence in treatment settings (Kulacaoglu and Kose, 2018) and a significant impact on mental health services (Soeteman et al., 2008b;Hörz et al., 2010;Zanarini et al., 2004Zanarini et al., , 2012)).Specifically, BPD is observed in 6 % of primary care patients (Gross et al., 2002) and in 15-20 % of patients in psychiatric hospitals and outpatient clinics (Gunderson and Links, 2008).However, some follow-up studies have shown a favorable course for the disorder, with remission rates ranging from 85 % to 88 % within ten years (Zanarini et al., 2007;Gunderson et al., 2011).Zanarini et al. (2007) observed that while self-mutilation, suicide risk and regression to treatment tended to decrease over time, affective symptoms such as anger and feelings of emptiness and loneliness remained stable.Another prospective study (Links et al., 1999) found that impulsivity remained stable over time.It is important to note that clinical remission alone is not a sufficient indicator of recovery (Winsper, 2021), as it does not indicate the absence of symptoms but rather that diagnostic criteria are no longer met.A large proportion of individuals with BPD may continue to experience persistent functional impairments over time ( Álvarez-Tomás et al., 2019).While many acute symptoms of BPD may remit in earlier years, underlying personality traits may persist for longer periods and may not be fully addressed by current treatments (Biskin, 2015).
In addition to pharmacological treatments, psychotherapy is a primary non-pharmacological treatment for BPD (Paris, 2009).A recent review and meta-analysis (Cristea et al., 2017) of 33 clinical trials involving 2256 participants found that psychotherapies are effective for treating borderline symptoms and related problems.Another review (Storebø et al., 2020) found that BPD-tailored psychotherapy had beneficial effects on symptom severity, self-harm, suicide-related outcomes and psychosocial functioning compared to treatment-as-usual (TAU).While specialized psychotherapy can be helpful for BPD, achieving long-term functional recovery can be challenging (Winsper, 2021).Moreover, impulsive behaviors and chaotic relationships, which are characteristic of BPD, may impede progress in treatment.
BPD patients have also been found to have cognitive deficits (e.g., in memory, attention, language, and executive functions) that may contribute to difficulties in daily life and psychological interventions.A review by Monarch et al. (2004) found that 71 % of primary studies reported dysfunction in a wide range of cognitive domains among BPD patients.Many studies highlighted the occurrence of executive dysfunction in BPD, including deficits in inhibition, which may be associated with abnormalities in the prefrontal cortex (Swirsky-Sacchetti et al., 1993;Bazanis et al., 2002;Berlin and Rolls, 2004).Some studies suggested that executive dysfunction in BPD patients, including a tendency towards risky decision-making and planning difficulties (Bazanis et al., 2002;Lenzenweger et al., 2004;Haaland and Landrø, 2007), may reflect dysfunction of the orbitofrontal cortex (Swirsky-Sacchetti et al., 1993;Bazanis et al., 2002).Berlin et al. (2005) found that patients with BPD and those with lesions affecting the orbitofrontal cortex had similar neuropsychological deficits.This finding suggested that some BPD symptoms, particularly impulsivity, may be epiphenomena of damage to the orbitofrontal cortex.To date, few neuropsychological studies on borderline personality disorder were performed and the results are sometimes conflicting, making difficult to clearly characterize the neuropsychological profile of this disorder.Several studies have found that BPD patients exhibit deficits in many cognitive functions (e.g., O′Leary et al., 1991;Judd and Ruff, 1993;Swirsky-Sacchetti et al., 1993), while others (e.g., Kunert et al., 2003) suggested that BPD patients perform similarly to healthy subjects.
The identification of the neuropsychological correlates of BPD could have an impact on psychological interventions.For instance, Monarch et al. (2004) noted that difficulties with attention and vigilance may have direct and indirect effects on other cognitive domains, limiting individuals' ability to solve problems and encode and retrieve new information.Nonverbal memory deficits have also been found to be consistent with BPD patients' difficulties with visual recall in daily life and treatment (Swirsky-Sacchetti et al., 1993).A study by Fertuck et al. (2012) found that worse executive function and visual memory were associated with a higher risk of treatment dropout.Interestingly, a review by LeGris and van Reekum (2006) found that impaired executive function, particularly disinhibition, may represent a dominant executive pathway to suicide attempts.
These neuropsychological features could contribute to unfavorable treatment outcomes.A study by Unoka and Richman (2016) suggested that neuropsychological functioning in BPD patients may play an important role in therapy planning.Ruocco (2005) highlighted that deficits in attention, learning, and memory may affect these patients' ability to communicate effectively and engage successfully in treatment.Impairment in several cognitive domains among BPD patients could result not only in difficulties in daily activities but also affect the efficacy of psychotherapeutic treatment.Thus, the identification of cognitive deficits related to BPD could have significant clinical implications.
The aim of the current study is to provide an update on neuropsychological functioning related to BPD and to better elucidate which cognitive functions are impaired in these patients.To that end, we conducted a meta-analysis of primary studies comparing performances on cognitive tests between BPD patients and healthy controls (HCs), focusing on a wide range of cognitive domains including global cognitive functioning, memory and its subdomains (short-term verbal memory; short-term spatial memory; long-term verbal memory; long-term spatial memory), executive functions and their subdomains (general, shifting, abstraction ability/concept formation, inhibition, generativity), visuospatial abilities, language and processing speed/attention/working memory.

Methods
This review was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Page et al., 2021) and was registered with The International Prospective Register of Systematic Reviews (PROSPERO) at number CRD42023308851.

Search strategy, study eligibility criteria, outcome
A systematic literature search was conducted, including studies published up to May 2023, using PsycInfo (PROQUEST), PubMed and Scopus entering the search terms: ("borderline personality disorder" OR borderline) AND (cognit* OR executive OR frontal OR memory OR language OR visuospatial OR attention).Hand searches of reference lists cited in the original and review articles were adopted.We included only primary studies in the meta-analysis if: (1) they were published in peerreviewed journals in English; (2) they provided results comparing neuropsychological test performance between BPD patients and healthy subjects; (3) they described statistical results (i.e., mean, standard deviation); and (4) they had a Newcastle-Ottawa Scale (NOS) score of 6 or higher.Studies were excluded if they: (1) did not report performance data for each individual task; (2) were reviews, meta-analyses, or overviews; or (3) were conference proceedings, letters to the editor, commentaries, theses, studies performed on animals, single case studies, or studies conducted on subjects in developmental age (Fig. 1).When two or more studies were performed by the same center, we included in the meta-analysis the article with the largest sample after contacting the authors to clarify whether there was overlap in the respective samples.
As for the outcomes of the meta-analysis, we identified global cognitive functioning, memory with its subdomains (i.e., short-term verbal memory, short-term spatial memory, long-term verbal memory and long-term spatial memory), executive functions with their subdomains (i.e., general, shifting, abstraction ability/concept formation, inhibition, generativity), visuospatial abilities, language and processing speed/attention/working memory.Table 1 reported the specific cognitive tests included in each cognitive domain.

Data extraction, coding and assessment of study quality
Data extracted and coded from the primary articles included: (1) publication characteristics (e.g., authors, publication status, year of publication, journal); (2) sample characteristics (i.e., total sample size, sex coded as the frequency of men in the sample, age, education, disease duration, pharmacological treatment, comorbidities); and (3) the neuropsychological test used to explore each cognitive domain.
After identifying primary studies, we used the Newcastle-Ottawa Scale (NOS) to evaluate their quality.The components of the NOS include selection (representativeness of the sample, sample size, nonrespondents, ascertainment of exposure), comparability and outcome assessment (assessment of the outcome and statistical test).The quality rating ranges from 0 to 10 stars, with a score of 6 or higher indicating high-quality articles.Two reviewers (AD, GLDB) independently performed quality assessments and any disagreements were resolved through discussion with a third reviewer (GS).

Statistical analysis
Effect sizes (ES) from data reported in the primary studies were computed using Hedges' g unbiased approach.The conventions used to interpret Hedges' g are similar to those of Cohen's d (Cohen, 1988).Therefore, values less than 0.20 are considered small effects, values around 0.50 are considered moderate effects, and values around 0.80 are considered large effects.Negative values of Hedges' g indicate that BPD patients had lower scores than HCs on each cognitive domain.Additionally, we calculated the 95 % confidence interval, variance, standard error and statistical significance for each ES.Heterogeneity among studies was assessed using the Q and I 2 statistics index (Huedo-Medina et al., 2006).A significant Q value indicates a lack of homogeneity; the proportion of observed variance that reflects real differences in effect sizes was estimated by I 2 , with values of 25 %, 50 %, and 75 % considered as low, moderate, and high, respectively (Higgins et al., 2003).
Several meta-regressions were performed to explain heterogeneity among study results and to examine the influence of demographic (age, sex, education) and clinical (disease duration and pharmacological treatment) variables.
Publication bias was assessed through funnel plots and tests for their asymmetry, i.e., Begg and Mazumdar's rank correlation and Egger's linear regression method (Egger et al., 1997;Higgins and Green, 2011), which statistically test the asymmetry of the funnel plot.Nonsignificant results indicate the absence of publication bias.The trim-and-fill procedure, a nonparametric iterative statistical technique (Duval, 2005), was applied to assess the effect of censoring potential data on the outcome of meta-analyses: the absence of publication bias is indicated by zero-to-zero studies or by the difference between observed and estimated effect sizes.In instances where publication bias was detected by both Egger's test and Begg and Mazumdar's test, or in the case of trim-and-fill studies, sensitivity analyses were conducted.
In each meta-analysis, mean scores and standard deviations of each group were recorded for all cognitive test across all primary studies.Statistical analyses were conducted using the meta-analytic software ProMeta 3.0.
No significant differences were found between the two groups on long-term verbal memory (Table 3; Fig. 8a) and on abstraction ability/ concept formation (Table 3; Fig. 8b).

Heterogeneity and publication bias
While no heterogeneity was observed in the domains of verbal and spatial short-term memory across the studies, heterogeneity was identified in a medium to large range (range 49.15-76.01) in the domains of long-term spatial memory, global executive functioning, and specific executive functions such as shifting, inhibition, and generativity.Similarly, heterogeneity was found across studies in the domains of processing speed/attention/working memory and visuospatial and constructional ability.The heterogeneity values (I 2 ) for each cognitive domain are showed in Table 3.
Publication bias was detected in the domains of short-term verbal memory, long-term spatial memory, global executive functioning, specific executive functions such as inhibition and generativity, and processing speed/attention/working memory.Due to the presence of publication bias, sensitivity analyses were conducted.After conducting a sensitivity analysis in the domain of long-term spatial memory and excluding one study (Judd and Ruff, 1993), the effect size was slightly reduced but remained large at − 0.88 (Table 3; Fig. 3a).Similarly, after conducting a sensitivity analysis in the domain of inhibition and excluding one study (Koudys and Ruocco, 2021), the effect size slightly increased but remained large at − 0.88 (Table 3; Fig. 4b).In the case of processing speed/attention/working memory, three studies (Dinn et al., 2004;Folesani et al., 2022;Palomares et al., 2019) were excluded, resulting in a slight decrease in effect size, but it remained moderate at − 0.28 (Table 3; Fig. 6).Lastly, for generativity, two studies were removed from the trim-and-fill analysis resulting in a slight increase in effect size to − 0.79 (Table 3; Fig. 5), and a sensitivity analysis was also conducted by excluding one study (Hurtado et al., 2016), thus maintaining a moderate effect size at − 0.47.

Association with demographic and clinical variables
In agreement with Borenstein et al. (2009), we did not conduct meta-regressions to assess the influence of demographic and clinical variables in cognitive domains where fewer than ten studies were included.Meta-regression analyses revealed that sex significantly moderated BPD patients' performances on Processing Speed/Attention/Working Memory and Shifting, while age did not influence these variables.

Discussion
This meta-analytic study found significant differences between BPD patients and HCs in multiple cognitive domains.Specifically, BPD patients performed worse than HCs in memory, executive functions, processing speed/attention/working memory, visuospatial/constructional ability, and language.Effect sizes were primarily medium to large, with the smallest effect size observed for the processing speed/attention/ working memory domain and the largest effect sizes in the long-term spatial memory and inhibition domains.No significant difference was observed in long-term verbal memory and abstraction ability/concept formation.
Focusing on processing speed/attention/working memory and executive functions, particularly inhibition, the meta-analytic results showed that BPD patients performed worse on tasks assessing these cognitive domains than HC.These findings indicated that a dysfunction in prefrontal areas in BPD (Kelley et al., 1998;Johnson et al., 2001).Specifically, reduced performance on tasks assessing working memory may indicate dysfunction in the dorsolateral prefrontal cortex (Cabeza and Nyberg, 2000).Reduced inhibitory abilities and interference control may indicate dysfunction in the anterior cingulate cortex and orbitofrontal cortex.This is consistent with other studies that have found a decrease in volume of the prefrontal cortex (Lyoo, Han, and Cho, 1998), particularly in the left orbitofrontal cortex and right anterior cingulate cortex, among BPD patients (Hazlett et al., 2005).Additionally, reduced inhibitory abilities may be related to the impulsivity that characterizes this personality disorder (Silbersweig et al., 2007).
Our results also revealed that BPD patients had worse performance on tasks assessing language and generativity compared to HCs.However, it should be noted that most of the studies included in the metaanalysis (e.g., Beblo et al., 2006;Mensebach et al., 2009b;Kobeleva et al., 2014;Hurtado et al., 2016;Palomares et al., 2019;Zegarra--Valdivia and Chino Vilca, 2019) explored the language domain using verbal fluency tasks that involve several cognitive processes, including selective attention, inhibition, set-shifting, and self-monitoring (Patterson, 2011).Reduced performance on verbal fluency tasks may therefore reflect dysfunction in one of these cognitive processes, which are mainly mediated by the frontal lobes (Patterson, 2011).
Regarding memory, BPD patients performed worse than HCs in short-term verbal memory, short-term and long-term spatial memory, but not for long-term verbal memory.Since short-term memory is the holding information part of working memory, the results of impairment of short-term memory in BPD might indicate that a difficulty in shortterm memory contributes to difficulties in working memory and other executive functions (e.g., fluency) that are found to be impaired in BPD as aforementioned.Moreover, the results of a deficit of long-term spatial memory in BPD might be interpreted considering that in many primary studies long-term memory has been assessed with ROCF and that this cognitive task is a measurement of visual-constructional ability, memory and also a measure of executive function (EF), specifically requiring planning and organizing skills (Fertuck et al., 2006).However, neuroimaging studies on BPD patients have reported decreased amygdala and hippocampal volume (Lyoo, Han, and Cho, 1998;Tebartz van Elst et al., 2003;Rüsch et al., 2003;Schmahl et al., 2003;Brambilla et al., 2004;Minzenberg et al., 2008) and temporal cortex.In particular, a study by Driessen et al. (2000) showed that the hippocampus and amygdala of BPD patients were 16 % smaller than controls; moreover, a meta-analysis on neuroimaging studies on grey matter volumes abnormalities in BPD patients showed reduction in bilateral temporal gyri and hippocampi and other cerebral areas (i.e.right inferior and superior frontal gyrus and the left precentral gyrus) (Schulze et al., 2016).Therefore, our result of impaired long-term memory might be an epiphenomenon of abnormalities in amygdala, hippocampus and temporal cortex.This issue deserves to be better investigated by studies of correlation between neuropsychological scores and neuroimaging parameters.
We found that BPD patients had a worse performance than HCs on cognitive tasks assessing visuospatial and constructional abilities revealing dysfunction of visuo-constructional abilities in BPD.Given that visuospatial abilities are mainly mediated by the parietal lobes (Husain and Nachev, 2007), and some study revealed abnormalities of the parietal cortex in BPD (Swinton, 2003), our result might suggest that dysfunction of visuo-constructional abilities are a consequence of a damage of parietal cortex in BPD patients.However, since many studies explored visuo-constructional abilities by copying task of ROCF that measures executive functions, the poor performance of BPD patients on visuo-constructional abilities might be interpreted as a consequence of executive dysfunction as deficit of spatial planning and organization and therefore, of a damage of prefrontal cortex reported in neuroimaging studies on BPD.
A meta-regression analyses revealed that BPD patients' performances on processing speed/attention/working memory and shifting were significantly moderated by sex, while age did not influence the     variables of interest.A possible sex-related explanation could be drawn from the observation that across adulthood there is a female advantage in processing speed (Siedlecki et al., 2019).
Overall, the above mentioned results seem to suggest that BPD patients showed dysfunctions in nearly every area of cognitive functioning.These findings may be in line with the idea that BPD patients showed diffuse cognitive impairment caused by a variety of source.Accordingly, despite the heterogeneity among neuropsychological studies on BPD, a review by Monarch et al. (2004) found some consistent findings, including the fact that BPD patients consistently performed below comparison groups on the Digit Symbol subtest, which is thought to be a sensitive indicator of brain damage (Lezak, 1995;Wechsler, 1981).In the present meta-analysis, the only cognitive domains in which BPD patients did not perform worse than controls were the long-term verbal memory and abstraction ability/concept formation.A possible explanation is the wide heterogeneity among studies with respect to comparison groups, inpatient/outpatient status and test administered.Focusing on neural underpinnings, a meta-analytic study (Visintin et al., 2016) that aimed to quantitatively identify intrinsic neural functional alterations associated with BPD, revealed an altered activity of the regions spanning across the default mode network (DMN) and in the midline core, as well as in the dorsal subsystems.In more details, hyperactivity of the medial prefrontal cortex and anterior cingulate at rest in BPD was observed, probably reflecting a dysfunction of fronto-limbic circuits, which in turn resulted in a reduced ability to activate prefrontal regions while performing emotion-processing tasks and an inability to effectively modulate limbic hyperactivity (Ruocco et al., 2013;Schulze et al., 2016).Visintin et al. (2016) also observed altered activations at the level of the precuneus and posterior cingulate, in the right lateral temporal complex, as well as decreased activity in the resting orbitofrontal cortex.Moreover, reduced activity in the right lateral temporal complex -involved in semantic memory, language, visual perception (the "what" pathway) and integration of information from different senses -including the inferior and middle temporal gyri, within the dorsal DMN, was observed in patients with BPD.
Taking together, all these evidences seem to suggest that the impaired functioning of several cognitive abilities in BPD could be associated with the altered functioning of several cerebral areas included in the DMN.
Assuming that neurocognitive functioning could moderate the clinical manifestation of BPD (Judd, 2005), as suggested by Poletti (2011), an early intervention focused on the rehabilitation of neurocognitive functions, especially attention and executive functions, could be a relevant protective factor against the severity of clinical manifestation of BPD.Thus, an intervention on neurocognitive functioning could also lead to better outcomes of psychotherapeutic interventions.From a clinical perspective, psychotherapy is considered the first line of treatment for BPD (Paris, 2009).Interestingly, a study by Schmitt et al. (2016) revealed an attenuation of limbic hyperarousal along with elevated coupling between limbic regions and prefrontal and inferior parietal control regions in patients with BPD after successful therapy.In a recent review (Marceau et al., 2018), decreased activity of the anterior cingulate, amygdala, insula, and prefrontal cortex (PFC), especially in ventrolateral and right hemisphere regions was observed after psychotherapy treatment.Increased activation of the PFC, particularly the dorsolateral prefrontal cortex (dlPFC), and dorsal anterior cingulate cortex (dACC) was also observed after psychotherapy.
The present meta-analysis is characterized by some limits: since up to 96 % of patients with BPD receive at least one psychotropic medication (Gartlehner et al., 2021) that can affect neuropsychological performance especially in tasks exploring executive functions, we did not perform a meta-regression analysis to evaluate the influence of drug treatments on neuropsychological performance.Moreover, it is known that in BPD poor performance on executive frontal tasks has been correlated with comorbidity with other psychiatric disorders such as co-occurrence of severe personality disorders or eating disorders (Unoka and Richman, 2016).However, we acknowledge that our review did not include a meta-regression analysis to specifically evaluate the influence of these comorbidities on neuropsychological performance.Another limit may be a considerable heterogeneity among the primary studies regarding the use of different tests to assess the same cognitive function, the co-occurrence of other disorders in patients with BPD (e.g., major depression).Moreover, it should be mentioned that we could not perform meta-regression analysis to assess the influence of the type of assessment instruments, demographic, and clinical variables in all those cognitive domains in which fewer than ten studies were included.Another aspect that may have limited the generalizability of the results is the fact that the samples of the included studies were predominantly composed of females.Therefore, the results should be generalized to male patients with BPD with caution.
Future research should aim to disentangle the unique contribution of BPD to neuropsychological deficits, independent of comorbid psychiatric disorders and explore potential sex differences in the manifestation and impact of BPD on cognitive functions in order to gain a clearer understanding of the distinctive cognitive profile associated with BPD and inform the development of targeted interventions.
In conclusion, the neuropsychological profile of borderline personality disorder is mainly characterized by deficits in inhibition, as well as in attention, memory, and executive functions.Therefore, assessing neuropsychological functions in individuals with a diagnosis of BPD could help clinicians to identify cognitive dysfunctions and, consequently, perform more targeted non-pharmacological treatments such as non-invasive stimulation which, in turn, could prevent the conversion from cognitive deficits to a frank dementia.

Declaration of Interest
None.

Fig. 2 .
Fig. 2. Forest plot for Short-Term Verbal Memory (A) and Short-Term Spatial Memory (B), displaying effect size (Hedges' g) calculated using a random effects model.EF, Effect size; CI, Confidence Intervals; V, Variance; N, total number of participants; BPD, Borderline Personality Disorder; HCs, Healthy Controls.

Fig. 3 .
Fig. 3. Forest plot for Long-Term Spatial Memory related to sensitivity analysis (A) and Executive Functions -General (B), displaying effect size (Hedges' g) calculated using a random effects model.EF, Effect size; CI, Confidence Intervals; V, Variance; N, total number of participants; BPD, Borderline Personality Disorder; HCs, Healthy Controls.

Fig. 4 .
Fig. 4. Forest plot for Executive Functions -Shifting (A) and Executive Functions -Inhibition related to sensitivity analysis (B), displaying effect size (Hedges' g) calculated using a random effects model.EF, Effect size; CI, Confidence Intervals; V, Variance; N, total number of participants; BPD, Borderline Personality Disorder; HCs, Healthy Controls.

Fig. 5 .
Fig. 5. Forest plot for Executive Functions -Generativity related to sensitivity analysis, displaying effect size (Hedges' g) calculated using a random effects model.EF, Effect size; CI, Confidence Intervals; V, Variance; N, total number of participants; BPD, Borderline Personality Disorder; HCs, Healthy Controls.

Fig. 6 .
Fig. 6.Forest plot for Processing Speed/ Attention/ Working Memory related to sensitivity analysis, displaying effect size (Hedges' g) calculated using a random effects model.EF, Effect size; CI, Confidence Intervals; V, Variance; N, total number of participants; BPD, Borderline Personality Disorder; HCs, Healthy Controls.

Fig. 7 .
Fig. 7. Forest plot for Visuospatial and Constructional Ability (A) and Language (B), displaying effect size (Hedges' g) calculated using a random effects model.EF, Effect size; CI, Confidence Intervals; V, Variance; N, total number of participants; BPD, Borderline Personality Disorder; HCs, Healthy Controls.

Fig. 8 .
Fig. 8. Forest plot for Long-Term Verbal Memory (A) and Executive Functions -Abstraction Ability/ Concept Formation (B), displaying effect size (Hedges' g) calculated using a random effects model.EF, Effect size; CI, Confidence Intervals; V, Variance; N, total number of participants; BPD, Borderline Personality Disorder; HCs, Healthy Controls.

Table 1
Classification of cognitive tests for cognitive functions.

Table 2
Characteristics of primary studies included in the meta-analysis.

Table 2
(continued ) MMSE Mini-Mental State Examination; MoCA Montreal Cognitive Assessment; OCD Obsessive-compulsive disorder; PTSD Post-traumatic stress disorder; ROCF Rey Osterrieth Complex Figure; SCWT Stroop Color-Word Interference Test; SNRIs Serotonin and norepinephrine reuptake inhibitors; SPM Raven's Progressive Matrices; SRT Buschke Selective Reminding Test; SSRIs Selective serotonin reuptake inhibitors; SUD Substance use disorder; TMT-A/B Trail Making Test, part A and part B; ToL Tower of London; VLMT Verbaler Lern-und Merkfa¨higkeitstest (German version of the Rey Auditory Verbal Learning Test); WCST Wisconsin Card Sorting Test; WRAT Wide Range Achievement Test.
N number; M males; NR Not reported; AVLT Auditory Verbal Learning Test; BCST Berg card sorting test; BDAE Boston Diagnostic Aphasia Examination test; BVRT Benton Visual Retention Test -Revised; COWA Controlled Oral Word Association test; D-KEFS Delis-Kaplan Execution Function System; GAD Generalized anxiety disorder; HSCT Hayling Sentence Completion Test; HVLT-R Hopkins Verbal Learning Test -Revised; IQR Interquartile range; MDD Major depressive disorder; a BPD with PTSD group.b BPD with Auditory verbal hallucinations group.A. D'Iorio et al.Neuroscience and Biobehavioral Reviews 165 (2024) 105860

Table 3
Summary of meta-analytic results of the following cognitive dimensions.
K Number of studies; N Total number of participants; HCs Healthy Controls; LL Lower Limit; UP Upper Limit; Q and I 2 Heterogeneity statistics; df degrees of freedom.Statistically significant values are reported in bold.