The current study is a comparative cross-sectional study, whereas a convenient sample was taken.

The study was conducted in two psychiatric hospitals located in Cairo, Egypt, which are the Okasha Institute of Psychiatry (Ain Shams University) and Al Mashfa Private Hospital. They represent both the low-middle and high socio-economic class, respectively.

Patients were eligible for the study if they fulfilled DSMIV diagnostic criteria for BPD, and their age was between 18 and 60 years. Patients were ineligible if they had any SCID I or SCID II diagnosis disorder, received recent electroconvulsive therapy, or have associated chronic medical or neurological disorder.

Controls were chosen from healthy employees and their relatives in both hospitals, and they did not have any axis I or II psychiatric comorbidities.

Sample size calculation: Assuming a high proportion of BPD patients with history of suicidal attempts (50%) and only a small proportion of healthy people having history of suicidal attempts (5% or less), it is estimated that 60 patients and 30 controls will achieve more than 90% power to detect difference between two groups setting alpha error at 0.05.

The programme used for sample size calculation is the Epi Info 7 programme.

Rigid family: Authoritarian leadership, which controls behaviour with strict rules and roles.

Structured family: Rules and roles are where everyone is aware and on board. This empowers children to help create rules and roles when appropriate. This creates individual autonomy.

Flexible family: Rules, roles and decision-making are susceptible to change when situations warrant. The power dynamics are shared when appropriate.

Diffused family: Chaotic, and erratic rules and roles, ineffective and confusing leadership are dominant.

All subjects (cases and controls) who accepted to participate in the study were interviewed using the Arabic Version of SCID II (Structured Clinical Interview for DSM-IV Axis II Disorders)^11,12 and the Arabic version of SCID I (Structured Clinical Interview for DSM-IV Axis I Disorders).^13,14

Sixty-seven subjects diagnosed as BPD by SCID II were selected. Out of these 67 patients of borderline, six of them were not included in the study, as they withdrew their consent. The remaining 60 patients were further subdivided into: 30 patients without Axis I comorbidities (group B) with a mean age of 28.9 years (SD ± 5.5) and 31 patients with Axis I comorbidities (group C) with a mean age of 26 years (SD ± 6.8).

Thirty healthy subjects matched for age and sex were included as a control group (group A); they were chosen from the employees and their relatives of both hospitals.

All participants (cases and controls) were asked to complete the following tools:

Item 1 taps into lifetime suicide ideation and/or suicide attempt; item 2 assesses the frequency of suicidal ideation over the past 12 months; item 3 assesses the threat of suicide attempt; item 4 evaluates self-reported likelihood of suicidal behaviour in the future.

The test has a sensitivity of 93% and specificity of 95%, while cutoff scores in adult patients are equal or more than 8. Reliability values of the Arabic version were found to be high as well (Cronbach’s alpha = 0.88).

Its scoring is as follows: 60–70 is mild, 70–80 is moderate and if more than or equal to 80 then impulsivity is severe. Total scoring of Barratt’s scoring: 30–120, (AI): 5–20, (AII): 3–12, (MI): 7–28, (MII): 4–16, (PI): 6–24, (PII): 5–20.

Low scores on both (AI) and (AII) show good attention span, cognitive stability and the qualities of non-impulsivity. Low scores on both (MI) and (MII) show good control of motor actions and persevere in holding off on impulsive actions. Low scores on both (PI) and (PII) show good self-control in planning for future and possessing good cognitive ability for complexity, which is the reverse of an impulsive attitude.

All analyses were carried out using SPSS, version 20 (SPSS Inc., Chicago, Illinois, USA). The results were tabulated and statistically analysed using the following tests: Student’s t test and the c² test. ANOVA test and post hoc analysis were used to compare values between three groups. Furthermore, Spearman correlation was used to test the association between suicidality, impulsivity, Global Assessment of Functioning (GAF) and socio-economic state of patients. Significance level was set at p < 0.05.

The FMASU REC is organised and operated according to guidelines of International Council on Harmonization (ICH) and the Islamic Organization for Medical Services (IOMS), the United States Office for Human Research Protections and the United States Code of Federal Regulations and operates under Federal Wide Assurance number: FWA 000017585. Patients were informed about the nature of the research and the confidentiality of the obtained information.

There was significant difference between the three groups with regard to age (p = 0.002) and the occupation (p < 0.001).

Family role differed significantly between the three groups; structured family was observed more in group A (43.3%), while diffused family role was more evident in group B (40%) and group C (45.2%), respectively, with p-value 0.007.

Furthermore, the present history of abuse differed significantly between the three groups (p = 0.015), which nearly quadrupled in group B 53.3% (n = 16) and group C 51.6% (n = 16) in comparison to group A 13.3% (n = 4). Emotional abuse was the most evident type of abuse in the three groups.

Table 1 clarifies that borderline personality subjects were significantly exposed to perpetration from their first degree family member (p = 0.005).

There was a statistically significant difference between the two groups of borderline personality in total mean score, item 2 which denotes the frequency of suicidal ideation over the past 12 months and item 4 showing the likelihood of suicidal behaviour in the future, which means that suicidality was found to be significantly more prevalent in borderline personality with comorbidity (group C) than borderline personality without comorbidity (group B) (54.8% in group C vs 36.7% in group B).

There was significant difference between the three groups regarding total score and sub-items of Baratt’s impulsiveness scale. There was high significant difference between the three groups regarding the total score of Barratt’s impulsiveness scale of Barratt (AII), Barratt (MI) and Barratt (PI) with p < 0.001. The difference between the three groups regarding Barratt (MII) and Barratt (PII) was significant with p = 0.049 and p = 0.046, respectively.

On post hoc test, there was significant difference between group A and B regarding Baratt’s (AII), Baratt’s (MI) and Baratt’s (PI). All items of Barrat’s impulsiveness scale increased in group C versus group A. On the other hand, when comparing the two groups of bipolar with each other (group B and C), Baratt’s (PII) was increased in group C.

Significant difference was found in GAF mean score when comparing the three groups. Highest scores of GAF were in controls (group A), then in BPD group (Group B), while the lowest scores were in BPD cases with comorbidities (group C).

Moreover, we compared, in our study, the GAF scores in association with suicidality. It showed highly significant results where the mean score of GAF in BPD cases with low suicidality was higher than those with high suicidality (Table 2).

Among the BPD without axis I comorbidities (group B), suicidality (as measured by SBQ-R) was significantly correlated to most of the impulsivity scores that was measured by BIS-R (Table 2). Suicidality was positively correlated with AI item: lack of span Attention in Attentional Facet (r = 0.489, p = 0.006), PI item: lack of self-control in planning facet (r = 0.401, p = 0.028), as well as MII item: lack perseverance in holding off impulsive acts in motor facet (r = 0.471, p = 0.009) of Barratt’s impulsiveness score (BIS-R), while a strong negative correlation was found between SBQ-R items and GAF (r = -0.671, p < 0.001). When we studied the association between suicidality measured by SBQ-R scale and level of impulsivity shown by BIS-11 scale, all the results related to the group of BPD without Axis I comorbidities showed no statistical significance using the t test. Total score of Barratt’s impulsiveness score was mildly lower (mean = 77.58) with (11.07 ± SD) in those who had low suicidality by SBQ than those who were high in suicidality (mean = 82.64) with (13.32 ± SD).

Among the BPD with axis I comorbidities (group C), results show total score of Barratt’s impulsiveness score was mildly higher (mean = 80.86) with (10.86 ± SD) in those who had low suicidality by SBQ than those who were high in suicidality (mean = 79.76) with (14.03 ± SD) which is the opposite of what was found in the group of BPDs without Axis I comorbidities.

Suicidality showed negative correlation between total score of SBQ-R with (PII item: lack of cognitive complexity in nonplanning facet) of BIS-R (r = –0.394, p = 0.028), and negative correlation between item 2 in SBQ-R and (PII item) in BIS-R (r = –0.483, p = 0.006), while it showed no correlation with GAF (Table 3).