هوشیاری بیش از حد در بیماران مبتلا به اختلال شخصیت مرزی: ویژگی، خودکاری و پیش بینی

Abstract According to cognitive theory, an important factor in borderline personality disorder (BPD) is hypervigilance. The aim of the present study was to test whether BPD patients show schema-related biases, and to explore relations with childhood trauma, schemas, and BPD symptoms. Sixteen BPD patients were compared with 18 patients with a cluster C personality disorder, 16 patients with an axis I disorder, and 16 normal controls. An emotional Stroop task was applied with schema-related and unrelated, negative and positive, supra- and subliminal person-related stimuli. BPD patients showed hypervigilance for both negative and positive cues, but were specifically biased towards schema-related negative cues. Predictors were BPD schemas, childhood sexual traumas, and BPD anxiety symptoms. Both BPD and axis I disorder patients showed a trend for a bias for negative schema-related subliminal stimuli. More attention to hypervigilance in BPD is recommended for clinical practice.

Introduction According to the Beckian cognitive model of borderline personality disorder (BPD), an important factor in the development and maintenance of this disorder is cognitive bias (Arntz, 2004 and Pretzer, 1990). The model hypothesizes that BPD patients process information through a specific set of three core beliefs or schemas of themselves and others, i.e., ‘I am powerless and vulnerable’, ‘I am inherently unacceptable’, and ‘Others are dangerous and malevolent’. Needing support in a dangerous world but not trusting others brings BPD patients in a state of hypervigilance. Schema-specific information is highly prioritized or difficult to inhibit in this state, resulting in biases in early information processing phases such as selective attention. Selective attention has been studied extensively in various disorders, and has been shown to play a crucial role in the etiology and maintenance of pathological anxiety in particular (Kindt and Van den Hout, 2001 and Williams et al., 1997). In BPD, however, anxiety in general and early biases in particular, have not been paid much attention to by researchers. Physiological affective hyperarousal in BPD has been investigated more often (e.g., Herpertz et al., 2000, Herpertz et al., 2001a and Herpertz et al., 2001b; Schmahl et al., 2003), however, with contradictory findings, i.e., from hypo-arousal to hyperarousal. The scarcity of selective attention studies of BPD is in contrast with the acknowledgement of anxiety as a significant aspect of BPD already in the earliest papers on ‘borderline patients’ ( Hoch and Cattell, 1959 and Stern, 1932), the relation of BPD with childhood trauma ( Herman et al., 1989, Sabo, 1997 and Zanarini, 1997), and the relatively high comorbidity of BPD both with anxiety disorders ( Zanarini et al., 1998b and Zimmerman and Mattia, 1999) and anxious cluster personality disorders (PDs) ( Zanarini et al., 1998a). Diverging hypotheses exist about the specificity of early biases in BPD. BPD is often conceptualized as a post trauma disorder (Gunderson and Sabo, 1993 and Herman et al., 1989), a view that is supported by data on high prevalences of interpersonal childhood traumas in BPD (Herman et al., 1989, Sabo, 1997 and Zanarini, 1997). Cognitive-behavioral theories conceptualize these traumas as learning experiences resulting in specific trauma-related cognitive schemas. These schemas facilitate but also bias information processing (Arntz, 2004 and Pretzer, 1990), or result in relatively isolated memory structures generating pathological fear behaviors and cognitions (Foa, Steketee, & Rothbaum, 1989). Both views hypothesize specific trauma-related biases in BPD, like these have also been found in patients with post-traumatic stress disorder (PTSD) (McNally, 1998). Other theories on BPD hypothesize structural deficits causing general handicaps in emotional functioning instead of specific biases. Amongst them are Linehan's dialectical model (Linehan, 1993) assuming a general emotional dysregulation, and Kernberg's psychoanalytic theory (e.g., Kernberg, 1967) assuming immature cognitive-emotional functioning in these patients. Studies focusing on biological deficits such as serotonin dysregulation (Coccaro, 1989), limbic system irritability (Andrulonis et al., 1981), or attentional network deficits (Posner et al., 2002) also seem to take this stand. Previous studies have found evidence for selective attention in BPD but did not resolve the stimulus specificity issue. Waller and Button (2002) investigated emotional Stroop interferences of BPD patients, patients with an anxiety or depressive disorder, and nonpatient controls. They found specific biases in BPD for stimuli related to self-criticism (e.g., failure), but not for stimuli related to criticism by others (e.g., ridiculed). A PD control group was not assessed, so it could not be tested whether the effect was specific for BPD or for PDs in general. Arntz, Appels, and Sieswerda (2000) who also applied an emotional Stroop task did compare BPD patients to other PD patients. They found a bias for emotionally negative stimuli related and unrelated to the BPD schemas in both BPD and cluster C PD (CPD) patients. Bias for emotionally positive stimuli was not investigated, leaving the question unanswered whether BPD patients show a really general emotional bias. The present study, that included control groups for both axis I and II, as well as emotionally positive stimuli, might clarify the specificity issue further. We hypothesized that patients with BPD would show deviating strong attentional biases, in particular for schema-related stimuli. A second unresolved issue is selective attention in BPD for not consciously perceived, i.e., subliminal stimuli. Schemas characteristically operate unconsciously and automatically (Beck, 1976), therefore, they can be expected to be triggered by subliminal as well as supraliminal stimuli (Williams et al., 1997). Some studies even suggest that early bias for subliminal threats is a better predictor for emotional vulnerability than early bias for supraliminal threats (MacLeod and Hagan, 1992, van den Hout et al., 1995 and Verhaak et al., 2004). Biases of subliminal stimuli in BPD were not found previously (Arntz et al., 2000) but this might be explained by methodological artifacts, like too short presentation times, no priming by supraliminal stimuli, and too small sample size. The present study applied less stringent calibration resulting in longer presentation times, presented the sub- and supraliminal stimuli mixed instead of blocked, and included more participants. Our hypothesis was that the interference of schema-related subliminal stimuli of patients with BPD would be larger than we would find for the whole group. A final objective of the present study was to explore potential predictors and consequences of hypervigilance in BPD. We investigated the relation between cognitive bias on the one hand and childhood trauma, BPD schemas, and BPD symptoms on the other hand. The symptom clusters we focused on were disinhibition and negative affectivity (anxiety in particular), which are personality traits characteristic for patients with BPD (Trull, 2001), and identity disturbance, which is another generally acknowledged problem in BPD (e.g., Pretzer, 1990 and Kernberg, 1967).

Results Questionnaires Mean scores on the questionnaires are presented in Table 2. The BPD group showed higher scores than the control groups. They showed significantly higher scores on the BPD Checklist and the three VBG subscales than all other groups. Their scores on the STAI subscales were significantly higher than those of the non-PD groups. Table 2. Mean scores on the questionnaires per groupa Variable Group BPD (n=16)(n=16) CPD (n=18)(n=18) AID (n=16)(n=16) NP (n=16)(n=16) BPD Checklist*** 116 (24)aa 85 (23)bb 74 (24)bb 52 (7.7)cc PDBQ-120 borderlineb,*** 38 (28)aa 37 (27)aa 18 (18)bb 4.4 (4.0)bb VBGc Sexual** 8.6 (8.5)aa 2.2 (5.1)bb 2.0 (6.1)bb 1.5 (3.8)bb Physical*** 18 (9.2)aa 8.7 (11)bb 2.5 (5.3)bb 7.4 (9.3)bb Emotional*** 32 (18)aa 19 (17)bb 16 (16)b,cb,c 6.4 (11)cc STAI Trait*** 59 (9.4)aa 54 (14)a,ba,b 50 (10)bb 30 (7.6)cc State 1*** 49 (14)aa 47 (14)a,ba,b 40 (12)bb 27(4.9)cc State 2***,d 48 (15)aa 46 (15)aa 37 (12)bb 28 (5.3)bb a,b,ca,b,c Indicators of significant simple contrasts. a Standard deviations or frequencies are between parentheses. b Scores of one BPD and one CPD subject missing. c Score of one BPD subject missing. d Score of one AID subject missing. ** Groups differ significantly, p<0.01p<0.01. *** Groups differ significantly, p<0.005p<0.005. Table options Stroop interference Supraliminal stimuli Emotional interference scores for supraliminal negative and supraliminal positive stimulus types are shown in Figs. 2 and 3, respectively. Mean emotional interference scores (ms) (+SE) of the BPD (n=16), CPD (n=18), AID ... Fig. 2. Mean emotional interference scores (ms) (++SE) of the BPD (n=16)(n=16), CPD (n=18)(n=18), AID (n=16)(n=16), and NP (n=16)(n=16) group for the supraliminal negative schema-related and schema-unrelated stimulus types. Figure options Mean emotional interference scores (ms) (+SE) of the BPD (n=16), CPD (n=18), AID ... Fig. 3. Mean emotional interference scores (ms) (++SE) of the BPD (n=16)(n=16), CPD (n=18)(n=18), AID (n=16)(n=16), and NP (n=16)(n=16) group for the supraliminal positive schema-related and schema-unrelated stimulus types. Figure options We performed four separate ANOVAs of the EMO and SCHEMA scores for supraliminal negative and positive stimuli. Groups were compared with deviation contrasts (BPD, CPD, AID, and NP vs whole group). Results are presented in Table 3. Table 3. Deviation contrasts for EMO scores and SCHEMA scores for supraliminal negative and positive stimuli (ms) Contrast MGroupMGroup (SD) MContrastGroupMContrastGroup (SD) tt pp Negative stimuli EMO score BPD vs all 32.4 (40.7) 16.6 (29.6) 2.66 <0.01<0.01 CPD vs all 19.6 (21.7) 16.6 (29.6) 0.53 ns AID vs all 15.4 (25.6) 16.6 (29.6) −0.19 ns NP vs all −1.35 (18.2) 16.6 (29.6) −2.99 <0.005<0.005 SCHEMA score BPD vs all 8.48 (50.7) −9.35 (35.9) 2.41 <0.05<0.05 CPD vs all −8.46 (28.6) −9.35 (35.9) 0.13 ns AID vs all −9.25 (29.7) −9.35 (35.9) 0.02 ns NP vs all −28.3 (22.2) −9.35 (35.9) −2.55 <0.05<0.05 Positive stimuli EMO score BPD vs all 21.0 (32.0) 7.14 (24.5) 2.27 <0.01<0.01 CPD vs all 4.66 (17.1) 7.14 (24.5) −0.52 ns AID vs all 5.45 (18.0) 7.14 (24.5) −0.35 ns NP vs all −2.25 (24.7) 7.14 (24.5) −1.78 =0.07=0.07 SCHEMA score BPD vs all −1.49 (46.3) −4.61 (36.3) 0.41 ns CPD vs all −2.30 (27.6) −4.61 (36.3) 0.32 ns AID vs all −18.3 (35.4) −4.61 (36.3) −1.75 =0.09=0.09 NP vs all 3.39 (34.4) −4.61 (36.3) 1.03 ns Table options The ANOVA of the EMO scores for the supraliminal negative stimuli yielded significant contrasts for the BPD group showing higher scores, and the NP group showing lower scores. BPD patients also showed significantly higher SCHEMA scores for the supraliminal negative stimuli than the overall group, whereas the contrasts for the control groups for the SCHEMA scores were not significant or in the opposite direction. Direct tests also showed that BPD patients had higher scores than the whole group (whereas the others had not) for emotional interference scores of schema-related stimuli, t(65)=2.80t(65)=2.80, p<0.005p<0.005, one-tailed, but not for emotional interference scores for schema-unrelated stimuli, t(65)=0.01t(65)=0.01, ns. The high EMO scores for the supraliminal negative stimuli of the BPD group, could thus be attributed to the interference of schema-related stimuli. In order to examine whether the BPD group showed differential emotional interference scores across the three schema-related stimulus types, we performed a repeated measures analysis with Schema-Related Stimulus Type (powerless, unacceptable, and malevolent words) as within-subjects factor, BPD Diagnosis (BPD, not BPD) as between-subjects factor, and Emotional Interference score as dependent factor. This analysis showed no effect of Schema-Related Stimulus Type, F(1.75,112)=0.55F(1.75,112)=0.55, ns, nor an interaction effect of Schema-Related Stimulus Type ×× BPD Diagnosis, F(1.75,112)=0.58F(1.75,112)=0.58, ns. Apparently, the BPD group showed similar emotional interference scores for all three schema-related stimulus types. Analyses for the supraliminal positive stimuli revealed that the four groups had dissimilar EMO scores again, but SCHEMA scores were now similar across the groups. The BPD group stood out with significantly higher EMO scores. A repeated measures analysis of the emotional interference scores with Stimulus Type (powerful, worthy, reliable, and joyful words) and BPD Diagnosis (BPD, not BPD) as factors, showed no effect of Stimulus Type, F(3,192)=1.64F(3,192)=1.64, ns, nor an interaction effect of Stimulus Type ×× BPD Diagnosis, F(3,192)=1.16F(3,192)=1.16, ns. BPD patients thus showed similar emotional interference scores across all supraliminal positive stimulus types. Analyses of EMO and SCHEMA scores for the restricted supraliminal positive schema-unrelated stimulus set yielded similar effects. We controlled for effects of general state and trait anxiety by running four follow-up ANCOVAs of the EMO scores and SCHEMA scores for supraliminal negative and supraliminal positive stimuli with STAI state 1, state 2, and trait scores as covariates. This changed the NP group's contrast for EMO scores for supraliminal negative stimuli into a trend, t(65)=1.91t(65)=1.91, p=0.06p=0.06, and this group's contrast for supraliminal positive stimuli into a nonsignificant effect, t(65)=1,50t(65)=1,50, ns. Contrasts for the BPD group remained essentially the same. Subliminal stimuli Presentation times selected by the calibration tasks were similar across the groups, Kruskall–Wallis View the MathML sourceχ2s(3,N=66)<1.93, ns. Mean hit rate in awareness task 1 (M=0.52M=0.52, SD=0.068SD=0.068, range (0.36–0.73)) was higher than 0.50, t(65)=2.12t(65)=2.12, p<0.05p<0.05, one-tailed. Mean hit rate in awareness task 2 (M=0.50M=0.50, SD=0.11SD=0.11, range (0.29–0.81)) was not higher than 0.50, t(65)=-0.13t(65)=-0.13, ns, one-tailed. Six Ss in task 1 and one S in task 2 showed hit rates close to mean hit rate plus 3 SD. After having removed their responses in these tasks, mean hit rate in task 1 (M=0.51M=0.51, SD=0.063SD=0.063, range (0.36–0.60)) and task 2 (M=0.49M=0.49, SD=0.10SD=0.10, range (0.29–0.75)) were both not higher than 0.50, |t|s<0.72|t|s<0.72, ns, one-tailed. RTs on subliminal stimuli of Stroop task 1 of the six Ss whose hit rates were too high in awareness task 1, and RTs on subliminal stimuli Stroop task 2 of the S whose hit rate was too high in awareness task 2, were removed from the analyses. Interference and effect scores for the subliminal stimuli of these Ss were calculated with their remaining RTs. EMO and SCHEMA scores for the subliminal negative and positive stimuli were also analyzed with four ANOVAs. Table 4 summarizes the results. Analyses only revealed a significant contrast for the NP group showing lower SCHEMA scores for subliminal negative stimuli. Contrasts for the BPD and AID patients’ SCHEMA scores for the subliminal negative stimuli were in the other direction, but did both not reach significance. Analyses of EMO and SCHEMA scores for the restricted set of subliminal positive schema-unrelated stimuli yielded similar effects. Adding STAI state 1, state 2, and trait scores weakened the AID group's contrast, t(65)=1.69t(65)=1.69, p=0.10p=0.10, and removed the effects for the NP and BPD group. Table 4. Deviation contrasts for EMO scores and SCHEMA scores for subliminal stimuli (ms). Contrast MGroupMGroup (SD) MContrastGroupMContrastGroup (SD) tt pp Negative stimuli EMO score BPD vs all 0.08 (37.2) 4.28 (23.6) −0.67 ns CPD vs all 7.64 (16.9) 4.28 (23.6) 0.72 ns AID vs all −4.17 (14.4) 4.28 (23.6) −1.67 ns NP vs all 12.4 (17.8) 4.28 (23.6) 1.64 ns SCHEMA score BPD vs all 16.9 (27.9) 5.08 (30.1) 1.87 =0.06=0.06 CPD vs all −3.87 (25.1) 5.08 (30.1) −1.57 ns AID vs all 17.1 (27.5) 5.08 (30.1) 1.92 =0.06=0.06 NP vs all −8.72 (33.0) 5.08 (30.1) −2.30 <0.05<0.05 Positive stimuli EMO score BPD vs all −5.29 (28.1) 1.31 (21.8) −1.39 ns CPD vs all 6.14 (16.1) 1.31 (21.8) 1.12 ns AID vs all −3.29 (18.1) 1.31 (21.8) −0.96 ns NP vs all 7.07 (22.6) 1.31 (21.8) 1.27 ns SCHEMA score BPD vs all −6.43 (37.3) 1.89 (31.6) −1.20 ns CPD vs all 4.09 (34.4) 1.89 (31.6) 0.34 ns AID vs all 3.43 (27.8) 1.89 (31.6) 0.23 ns NP vs all 6.18 (26.8) 1.89 (31.6) 0.63 ns Table options Predictors and comorbidity In order to detect potential predictors of cognitive bias, as well as to disentangle relations between comorbid diagnoses and bias, we performed nine multiple regression analyses on the whole sample (N=66)(N=66). Independent factors in these analyses were diagnosis (fulfilling DSM-IV criteria for BPD, a CPD, an anxiety disorder, or a mood disorder), BPD symptom cluster (anxiety, impulsiveness, identity disturbance), or childhood trauma (sexual, physical, emotional). The dependent variables were those scores on which the BPD patients stood out in the previous analyses, i.e., the SCHEMA score for the supraliminal negative stimuli, the EMO score for the supraliminal positive stimuli, and the SCHEMA score for the subliminal negative stimuli. Results of these analyses are summarized in Table 5. In order to control for an artificial enhancement by including a nonclinical sample, we controlled whether results were similar when only patients were included. This appeared to be the case, except for a clearly weaker relation (β=0.15,p=0.32)(β=0.15,p=0.32) between the diagnosis anxiety disorder and the SCHEMA score for the subliminal negative stimuli. Table 5. Summary of nine regression analyses for variables predicting SCHEMA scores for supraliminal negative stimuli, EMO scores for supraliminal positive stimuli, or SCHEMA scores for subliminal negative stimuli (N=66)(N=66) Dependent variable SCHEMA score EMO score SCHEMA score Supraliminal negative Supraliminal positive Subliminal negative Independent variable ββ pp ββ pp ββ pp Diagnosis BPD 0.28 <0.05<0.05 0.32 <0.05<0.05 0.24 =0.05=0.05 CPD 0.25 =0.08=0.08 0.066 ns −0.11 ns Anxiety disorder 0.061 ns −0.049 ns 0.28 <0.05<0.05 Mood disorder −0.10 ns −0.003 ns −0.14 ns BPD symptom clustersa Anxiety 0.44 <0.05<0.05 −0.38 =.08=.08 0.02 ns Impulsiveness −0.16 ns 0.086 ns 0.04 ns Identity disturbance 0.016 ns 0.59 <0.01<0.01 0.10 ns Childhood traumab Sexual 0.38 <0.005<0.005 0.19 ns −0.09 ns Physical 0.25 =0.08=0.08 −0.02 ns −0.04 ns Emotional −0.14 ns 0.23 ns 0.21 ns Note. SCHEMA score supraliminal negative stimuli: R2=0.14R2=0.14, ns for Diagnosis; R2=0.19R2=0.19, p<0.001p<0.001 for Childhood trauma; R2=0.15R2=0.15; p<0.05p<0.05 for BPD symptom clusters. EMO score supraliminal positive stimuli: R2=0.11R2=0.11, ns for Diagnosis; R2=0.10R2=0.10, ns for Childhood trauma; R2=0.15R2=0.15, p<0.05p<0.05 for BPD symptom clusters. SPE score subliminal negative stimuli: R2=0.14R2=0.14, p=0.06p=0.06 for Diagnosis; R2=0.04R2=0.04, ns for Childhood trauma; R2=0.02R2=0.02; ns for BPD symptom clusters. a BPD Checklist subscale scores. b VBG subscale scores. Table options BPD diagnosis was the only diagnosis that showed significant relations with the effect scores for supraliminal schema-related negative stimuli and supraliminal schema-related and unrelated positive stimuli. Effect scores for subliminal schema-related negative stimuli showed similar borderline significant or significant ββs for the diagnoses BPD and anxiety disorder, respectively. The latter however was no longer significant in an analysis without the nonclinical group and might thus be artificial. BPD severity, especially the severity of BPD anxiety symptoms (such as separation anxiety), showed a significant relation with the SCHEMA score for supraliminal negative stimuli. Severity of BPD impulsiveness and identity disturbance symptoms were not significantly related to the SCHEMA score for supraliminal negative stimuli. The EMO score for the supraliminal positive stimuli on the other hand was significantly and positively related to severity of BPD identity disturbance and tended to be negatively instead of positively related to severity of BPD anxiety symptoms. Having experienced (more severe) childhood sexual or physical traumas showed significant or nearly significant ββs with the SCHEMA score for negative supraliminal stimuli. No significant relations were found between having experienced childhood traumas and the other dependent variables.