The British Journal of Psychiatry
Attention-deficit hyperactivity disorder as a potentially aggravating factor in borderline personality disorder
Alexandra Philipsen, Matthias F. Limberger, Klaus Lieb, Bernd Feige, Nikolaus Kleindienst, Ulrich Ebner-Priemer, Johanna Barth, Christian Schmahl, Martin Bohus

Abstract

Background

Clinical experience suggests that people with borderline personality disorder often meet criteria for attention-deficit hyperactivity disorder (ADHD). However, empirical data are sparse.

Aims

To establish the prevalence of childhood and adult ADHD in a group of women with borderline personality disorder and to investigate the psychopathology and childhood experiences of those with and without ADHD.

Method

We assessed women seeking treatment for borderline personality disorder (n=118) for childhood and adult ADHD, co-occurring Axis I and Axis II disorders, severity of borderline symptomatology and traumatic childhood experiences.

Results

Childhood (41.5%) and adult (16.1%) ADHD prevalence was high. Childhood ADHD was associated with emotional abuse in childhood and greater severity of adult borderline symptoms. Adult ADHD was associated with greater risk for co-occurring Axis I and II disorders.

Conclusions

Adults with severe borderline personality disorder frequently show a history of childhood ADHD symptomatology. Persisting ADHD correlates with frequency of co-occurring Axis I and II disorders. Severity of borderline symptomatology in adulthood is associated with emotional abuse in childhood. Further studies are needed to differentiate any potential causal relationship between ADHD and borderline personality disorder.

Borderline personality disorder consists of pervasive affective instability, impulsivity, unstable relationships and self-image disturbances.1 It affects 1–2% of the general population and is characterised by severe psychosocial impairment25 and a high suicide rate.6 According to Fossati et al (2002),7 60% of adults with borderline personality disorder meet criteria for childhood attention-deficit hyperactivity disorder (ADHD).8 The two disorders share some similar clinical features, e.g. emotional dysregulation and impulsivity. This suggests that childhood ADHD may be a risk factor for borderline personality disorder in adulthood. The prevalence rate of ADHD in children ranges from 3–12% (depending on sample selection and diagnostic criteria),9,10 and 40–60% thereof have persisting symptoms.1114 The prevalence of adult ADHD is estimated to range from 1 to 4%.14,15 No study has investigated the prevalence of adult ADHD in people with borderline personality disorder or the impact of ADHD symptoms on the severity of the latter and co-occurring psychopathologies. This study investigated the prevalence of both childhood and adult ADHD symptomatology in people with borderline personality disorder, the influences of both on adult borderline personality disorder and the association between histories of childhood ADHD and traumatic childhood experiences.

Method

Participants

We recruited 118 women with borderline personality disorder (mean age 29.2 years, s.d.=7.6) from the out-patient clinics of our departments (Freiburg, Mannheim) that offer diagnostic and therapeutic services (dialectical behaviour therapy4,16,17) for people with the disorder seeking treatment. The ethical committees of the University of Freiburg and Mannheim approved the study protocol prior to data collection. Written informed consent was obtained from patients prior to study participation.

General diagnostic assessments

Co-occurring Axis I disorders were assessed by the Structured Clinical Interview for DSM–IV Axis I Disorders (SCID–I).18 Personality disorders were examined by the German version of the International Personality Disorder Examination (IPDE).19

The IPDE and the SCID–I were administered by experienced clinical psychologists. To determine interrater reliability, a random sample of 10% of the interviews were independently rated by a second observer. Interrater reliability values were within the acceptable range for both the IPDE (κ=0.77) and the SCID–I (κ=0.70).

Assessment of ADHD symptomatology

Participants rated their ADHD symptoms in childhood retrospectively, using the short version of the Wender Utah Rating Scale (WURS-k), German version,20 which includes 25 items on a five-point Likert-scale (`not at all' to `severe'). Following Fossati et al (2002),7 we used a very conservative cut-off score of ⩾46 to indicate the presence of a diagnosis of ADHD in childhood. Participants rated adult ADHD symptoms with the ADHD–Checklist (ADHD–CL),21 which includes 18 items on a three-point Likert-scale corresponding to the diagnostic criteria of DSM–IV (0–2, `not at all' to `severe'). To minimise the likelihood of overestimating the prevalence of adult ADHD (especially the inattentive subtype), we used a cut-off of ⩾25 to indicate that participants met criteria for the combined subtype of ADHD. Only patients who fulfilled both the WURS-k criteria and the ADHD–CL criteria were diagnosed as having adult ADHD.

Assessment of borderline symptomatology

Severity of borderline symptomatology was assessed by the total score of the Borderline Symptom List (BSL).22 The BSL is a self-report questionnaire of symptoms of borderline personality disorder that includes 95 items on seven sub-scales (self-image, affect regulation, self-destruction, dysphoria, loneliness, intrusion and hostility) and is based on the DSM–IV criteria for the disorder (as assessed by the Diagnostic Interview for Borderline Personality Disorder – Revised Version). Participants rate the severity of symptoms on a five-point Likert scale (`not at all' to `very strong'). The BSL has demonstrated high internal consistency and test–retest reliability, strong construct validity and low correlations with gender, age and level of education.

Assessment of childhood trauma experiences

Childhood history of abuse and neglect was assessed by the Childhood Trauma Questionnaire (CTQ). The CTQ is a 28-item self-report inventory that assesses five types of maltreatment – emotional, physical and sexual abuse, and emotional and physical neglect. Also included in this questionnaire is a three-item minimisation/denial scale for detecting false-negative trauma reports.23

Statistical analyses

Statistically significant differences were assessed at a two-tailed alpha level of <0.05. Statistical analyses were carried out using SPSS for Windows, version 12, and `R', version 2.4.1 (R Foundation for Statistical Computing, http://www.R-project.org/). To reduce the number of independent tests performed, multivariate logistic regression was used. Eight separate regression models were computed: for each of the two target variables (childhood ADHD as defined by WURS-k ⩾46 and adulthood ADHD symptomatology defined by ADHD symptom sum score ⩾25), we tested the influence of each of the four following sets of variables:

  1. Axis

    I disorders as assessed by the SCID–I: bipolar and unipolar affective disorders, acute psychotic disorders, substance abuse/dependence, alcohol abuse/dependence, panic disorder, agoraphobia, social phobia, specific phobia, post-traumatic stress disorder, anorexia, bulimia nervosa, other eating disorders, obsessive–compulsive disorder, somatoform disorders;

  2. Axis

    II disorders assessed by the IPDE: paranoid, schizoid, antisocial, histrionic, avoidant, dependent and schizotypal personality disorder (criteria for narcissistic personality disorder were not met by any participants);

  3. borderline

    symptom severity measured by the BSL and number of DSM–IV criteria for borderline personality disorder;

  4. adverse

    events in childhood assessed by the CTQ: emotional, physical and sexual abuse, and emotional and physical neglect.

In order to assess the multivariate significance of each model, likelihood ratio statistics are reported; for each variable within the model, Wald Z-values are given.

Results

Patient characteristics

Of the 118 participants, 62.7% were single, 17.8% had a partner or were married, 11% were divorced and 8.5% were widowed or separated. The majority (60.2%) were childless. Just over half (50.6%) were working full time or studying (27.1% and 23.5% respectively) and 21.2% were unemployed or had retired early.

Exploratory Mann–Whitney U-tests revealed no differences between participants who scored above v. below the threshold for childhood ADHD on the WURS-k in terms of age or education. Similarly, there were no significant differences in age or education between participants who scored above v. below the thresholds on the ADHD–CL for adult ADHD.

Axis I disorders

Almost all participants (99.2%) had at least one co-occurring Axis I disorder (Table 1) (mean=5.04, s.d.=2.45). For Axis I disorders including probable diagnoses this was 5.69 (2.46).

View this table:
Table 1

Co-occurring Axis I disorders (lifetime and current) in 118 women with borderline personality disorder (SCID–I)

Axis II disorders

Of the 118 participants, 36 (30.5%) fulfilled only the diagnostic criteria for borderline personality disorder, assessed by the IPDE. The remaining 69.5% also fulfilled criteria for at least one other personality disorder: 43.2% met criteria for one co-occurring personality disorder; 16.9% met criteria for two, 6.8% for three, and 2.5% for four or five personality disorders. An overview of co-occurring Axis II disorders is given in Table 2.

View this table:
Table 2

Co-occurring Axis II disorders in 118 women with borderline personality disorder measured by the International Personality Disorder Examination

Retrospective diagnosis of childhood ADHD

Of the 118 individuals studied, 49 (41.5%) fulfilled criteria for childhood ADHD as retrospectively diagnosed by the WURS-k. The mean (s.d.) WURS-k score among these 49 participants was 55.7 (6.8).

Adult ADHD symptomatology

Using the predefined cut-off of 25 on the ADHD–CL, 16.1% of the 118 participants were diagnosed with severe adult ADHD (combined subtype, mean (s.d.) ADHD–CL score 28.79 (2.99)).

Influence of a diagnosis of ADHD on co-occurring disorders

Axis I disorders

The logistic regression analysis significantly discriminated between participants with and without adult ADHD (model likelihood ratio=29.3, P=0.022). Specific phobias and somatisation disorders were positively associated with adult ADHD (Wald Z=2.24, P=0.025, and Wald Z=2.25, P=0.024 respectively). Furthermore, the association between adult ADHD and panic disorder as well as anorexia nervosa just missed the level of significance (Wald Z=1.95, P=0.052, and Wald Z=1.79, P=0.073 respectively). Logistic regression analyses indicated that Axis I disorders did not significantly discriminate between participants who met criteria for childhood ADHD and those who did not (P=0.117).

Axis II disorders

Logistic regression indicated that personality disorders significantly discriminated participants who were above v. below the thresholds for adult ADHD (model likelihood ratio=16.92, P=0.031). Co-occurring paranoid personality disorder was positively associated with adult ADHD (Wald Z=2.19, P=0.029). For schizoid and dependent personality disorders, and there was a trend (Wald Z=1.80, P=0.072, and Wald Z=1.66, P=0.098 respectively).

Logistic regression analysis revealed a trend for a positive association between childhood ADHD and co-occurring personality disorders (model likelihood ratio=13.81, P=0.087) with a trend towards significance for a higher prevalence of paranoid personality disorder (Wald Z=1.93, P=0.054) among participants who reported childhood ADHD.

Influence of a diagnosis of ADHD on severity of borderline symptomatology

Logistic regression analyses revealed a trend towards a positive association between symptoms of borderline personality disorder and adult ADHD (Wald Z=1.72, P=0.085) and a significant positive association between severity of current symptoms and childhood ADHD (Wald Z=2.33, P=0.020, see Table 3). Regarding the DSM–IV criteria for borderline personality disorder, criterion 8 (difficulty controlling anger) and criterion 9 (stress-related dissociative symptoms/paranoid ideations) were significantly more pronounced in participants who reported childhood ADHD (WURS-k ⩾46) compared with those with WURS-k scores below the threshold for the childhood disorder.

View this table:
Table 3

Severity of borderline symptomatology and childhood trauma experiences among participants (118 women with borderline personality disorder) with retrospectively assessed childhood ADHD compared with participants without

Association of a diagnosis of ADHD with negative childhood experiencesy

Logistic regression analysis demonstrated a strong association between the retrospective diagnosis of childhood ADHD and reported emotional abuse in childhood measured by the CTQ (Wald Z=2.62, P=0.009). There were no differences in terms of reported sexual or physical abuse and physical or emotional neglect in childhood (Table 3).

In terms of adult ADHD and reported adverse childhood experiences, statistical analyses yielded no significant associations.

Discussion

ADHD in people with borderline personality disorder

Our findings of a high prevalence of reported childhood ADHD among women with borderline personality disorder support the earlier findings of Fossati et al (2002)7 and suggest that childhood ADHD may be a risk factor for the development of borderline personality disorder in adulthood. Although our prevalence rate of childhood ADHD (41.8%) was lower than that of Fossati et al (59.5%), this difference may be because Fossati et al also included men (18 men, 24 women) in their analysis; ADHD is more common among boys and men than among girls and women.24

In terms of adult ADHD symptomatology we found a prevalence rate of 16.1% when only including participants who also fulfilled predefined criteria for childhood ADHD. To avoid the risk of overestimating the prevalence of adult ADHD of the inattentive type, which is especially high in people with borderline personality disorder with co-occurring disorders (such as substance misuse and affective disorders), we only included participants with ADHD of the combined subtype. Therefore, we cannot exclude a possible effect of other subtypes (inattentive, hyperactive/impulsive) or of participants in partial remission. A recently published meta-analysis on follow-up studies in childhood ADHD found a persistence rate of only 15% meeting full criteria for ADHD at age 25 years.14 Whereas the prevalence rate of adult ADHD in the present study is rather low (16.1%), the persistence rate is slightly higher (38.8%), but in line with recently published findings which revealed a persistence rate of 36.3% in respondents to a comorbidity survey study.25 Adverse family environment variables such as low social class and family conflicts are considered important risk factors for increased ADHD symptomatology26 and severity, and lack of treatment for the disorder in childhood predicts persistence into adulthood.25 Thus, it is possible that all three factors – negative environmental factors, ADHD severity and lack of treatment in childhood – contributed to the higher rate of persistence of ADHD that we observed in our sample. One could expect ADHD in childhood, especially when untreated, to have a negative impact on educational achievement, but we did not find differences in achievement among those with and those without ADHD. One intrepretation is that strong predictor variables such as sexual abuse, emotional neglect and physical neglect, which were equally distributed between the two patient groups, might have a greater impact on educational level than co-occurring ADHD.

Adult ADHD and Axis I and II disorders in borderline personality disorder

Interestingly, we found an influence of adult ADHD symptomatology on co-occurring disorders among our participants. Both Axis I and Axis II disorders (specific phobias and somatisation, a tendency towards panic disorder and anorexia nervosa, paranoid personality disorder, elements of schizoid and dependent personality disorder) were associated with a greater likelihood of adult ADHD. In contrast, Axis I and II disorders were less consistently associated with the presence of childhood ADHD.

History of childhood ADHD and severity of borderline personality disorder

Our findings also revealed that participants with more severe current symptoms of borderline personality disorder were more likely to report childhood ADHD. The cause of the association between childhood ADHD and more severe borderline symptomatology in adulthood is not clear. In ADHD, various genetic and neuroimaging studies support a genetic and neurobiological origin largely associated with the central dopaminergic and noradrenergic systems.27 Furthermore, other organic causes such as pregnancy and delivery complications, maternal smoking and alcohol misuse during pregnancy have been identified as risk factors for ADHD in offspring.28 Adverse family environment variables such as low social class and family conflicts are thought to be important risk factors for the exacerbation of ADHD symptomatology rather than for the development of ADHD. In contrast, adverse events such as sexual or physical abuse in childhood are well documented as serious risk factors for the development and severity of borderline personality disorder.2932 Thus, one could speculate that childhood ADHD associated with severe negative childhood experiences predispose to the development of borderline personality disorder in adulthood in a subgroup of individuals. In our sample, women with borderline personality disorder retrospectively diagnosed with childhood ADHD (WURS-k ⩾46) reported a higher rate of emotional abuse in childhood than those with WURS-k scores <46. These groups, however, did not differ in terms of other negative childhood experiences (e.g. physical or sexual abuse or neglect). Our findings regarding emotional abuse are consistent with those of a recent study33 that found that emotional abuse and neglect were more common among adults with ADHD compared with a control group. Therefore, the higher rate of emotional abuse of our participants with childhood ADHD symptomatology may have led to even more severe borderline symptomatology in adulthood.

As reported by Watson et al (2006)34 and by Simeon et al (2003)35 dissociative features are primarily correlated with emotional abuse and neglect as well as physical abuse in borderline personality disorder. In these studies, however, participants with the disorder were not screened for co-occurring ADHD. Thus, our findings of a more severe borderline symptomatology as well as the enhanced stress-related dissociative symptoms among women with borderline personality disorder retrospectively diagnosed with childhood ADHD could be explained by the increased risk of having been emotionally abused in childhood.

The precise mechanism for the high association between ADHD and borderline personality disorder found in our study is not clear. The high co-occurrence of these disorders may be due to overlapping clinical features and diagnostic criteria of the two disorders. In particular, intense anger and difficulty controlling anger (borderline personality disorder criterion 8) may overlap considerably with ADHD features. However, transient stress-related dissociative symptoms or paranoid ideation are not part of ADHD criteria. Moreover, the scales used for the assessment of ADHD (WURS-k, ADHD–CL) mainly focus on core symptoms of ADHD such as persistent inattention, distractability and hyperactivity, which are also not part of the current diagnostic criteria of borderline personality disorder.

Thus, further studies investigating the influence of the relationship between ADHD and adverse events in childhood on borderline symptom severity in adulthood are warranted.

Limitations

Some study limitations must be considered. First, ADHD symptoms, severity of borderline personality disorder and negative childhood experiences were assessed using self-report questionnaires, and we did not investigate data reliability. Thus, the influence of current mood state or severity of symptoms on our data cannot be conclusively excluded. Second, this was not a prospective study; childhood ADHD was assessed retrospectively by the WURS-k. Thus, it is unclear whether childhood ADHD would be similarly associated with symptoms of borderline personality disorder or co-occurring disorders if childhood ADHD symptoms had been assessed in childhood. Nevertheless, to avoid an overestimation of ADHD diagnosis in childhood, we used a very conservative cut-off score (⩾46), as described previously by Fossati et al (2002),7 and the prevalence of childhood ADHD remained high. Third, we used a self-report measure (rather than a structured interview) of adult ADHD symptomatology based on DSM–IV criteria for ADHD (the ADHD–CL). There are, however, no specific DSM–IV criteria for adult ADHD, and the SCID–I does not include questions for diagnosing adult ADHD. To avoid an overestimation of co-occurring adult ADHD diagnoses, we used a very conservative cut-off score. Using these criteria, we found persistence rates comparable to those published by Kessler et al (2005),25 which led us to conclude that these participants most likely had persisting adult ADHD. The conservative cut-off scores and the applied scales for the assessment of ADHD, which mainly detect the core features of inattention and hyperactivity, speak against a merely methodologically based association of ADHD and borderline personality disorder.

Finally, the findings of our study are based on observations from only female and treatment-seeking participants recruited from the out-patient departments of our two clinics. Therefore, the study results cannot be generalised to all people with borderline personality disorder.

Implications

Our findings indicate that childhood ADHD is associated with greater emotional abuse in childhood as well as more severe borderline psychopathology in adult women with borderline personality disorder. Thus, ADHD in childhood may be considered a risk factor that predisposes to borderline personality disorder in adulthood in a subgroup of patients. Clinicians should be aware of childhood ADHD and co-occurring adult ADHD among patients with borderline personality disorder. Future treatment development might focus on whether different interventions are needed for subgroups of patients with borderline personality disorder who either have current ADHD or reported childhood ADHD. In addition, the effect of methylphenidate and noradrenergic psychopharmacological agents should be systematically investigated in patients with borderline personality disorder and co-occurring ADHD.

Acknowledgments

The results of this paper partially stem from Johanna Barth's and Matthias Limberger's dissertations which are in preparation at the Medical Faculty of the Ruprecht-Karls-Universität Heidelberg. The study has been funded by the Borderline Personality Disorder Research Foundation, New York.

  • Received January 27, 2007.
  • Revision received August 2, 2007.
  • Accepted August 23, 2007.

References

View Abstract