Sage Journals: Discover world-class research

Abstract

Objective:

Despite considerable morbidity and functional losses associated with adolescent borderline personality disorder, little is known about psychopathological outcomes. This study examined associations between adolescent borderline personality disorder symptoms and subsequent depressive, psychotic and hypomanic symptoms.

Methods:

We used data from the Avon Longitudinal Study of Parents and Children. Participants were adolescents living in the community who had data for all longitudinal outcomes (N = 1758). We used logistic regression and path analysis to investigate associations between borderline personality disorder (five or more probable/definite symptoms) reported at age 11–12 years and depressive and psychotic symptoms reported at age 12 and 18, and lifetime hypomanic symptoms reported at age 22–23 years.

Results:

Adolescent borderline personality disorder symptoms were associated with psychotic symptoms (odds ratio: 2.36, confidence interval: [1.82, 3.06]), diagnosis of depression at age 18 years (odds ratio: 1.30, confidence interval: [1.03, 1.64]) and hypomanic symptoms (odds ratio: 2.89, confidence interval: [2.40, 3.48]) at 22–23 years. Path analysis controlling for associations between all outcomes indicated that borderline personality disorder symptoms were independently associated with depressive symptoms (β = 0.97, p < 0.001) at 12 years and hypomanic (β = 0.58, p < 0.01) symptoms at 22–23 years. Borderline personality disorder symptoms were also associated with psychotic symptoms at age 12 years (β = 0.58, p < 0.01), which were linked (β = 0.34, p < 0.01) to psychotic symptoms at age 18 years.

Conclusion:

Adolescents with borderline personality disorder symptoms are at future risk of psychotic and hypomanic symptoms, and a diagnosis of depression. Future risk is independent of associations between psychopathological outcomes, indicating that adolescent borderline personality disorder symptoms have multifinal outcomes. Increasing awareness of borderline personality disorder in early adolescence could facilitate timely secondary prevention of these symptoms subsequently, helping to prevent future psychopathology.

Keywords

Borderline personality disorder outcome depression hypomania psychotic path analysis adolescents ALSPAC

Introduction

Borderline personality disorder (BPD) is a common and serious mental disorder with community prevalence ranging from 0.7% to 1.8% (Lieb et al., 2004; Winsper et al., 2019). The validity and clinical importance of adolescent BPD is increasingly recognized (Sharp and Fonagy, 2015; Winsper et al., 2016a). Indeed, BPD may be better conceptualized as a youth disorder, arising in early adolescence and then slowly declining from young adulthood onwards (Chanen et al., 2014). Adolescents with BPD are common among help-seeking populations. Adolescent BPD is reported in up to 11% of psychiatric outpatients (Chanen et al., 2004) and 33–49% of psychiatric inpatients (Chanen et al., 2014; Grilo et al., 1998). These individuals often experience long-term social, educational, work and financial impairment and have a substantial need for healthcare (Winsper et al., 2015). Nevertheless, our understanding of the outcomes of adolescent BPD remains rudimentary.

Our previous systematic review indicated a limited number of prospective studies investigating links between adolescent BPD and subsequent psychopathology (Winsper et al., 2015). Existing studies are limited by methodological problems. Clinical studies are confounded by treatment effects and the sampling of individuals with severe psychopathology. Community studies are restricted by the lack of validated, structured assessment tools specifically designed to screen for adolescent BPD, and moderate sample sizes limiting power for detecting specific diagnostic or subsyndromal-level associations (Cohen et al., 2007). Some community studies tend to focus on high-risk populations, e.g. oversampling for conduct disorder (Stepp et al., 2012), or are focused on twin populations (Belsky et al., 2012). Furthermore, few studies have controlled for earlier life experiences (e.g. abuse), which have been found to link to later psychopathology (Jaffee, 2017). Another major issue is that the extant literature primarily focuses on associations between adolescent BPD and future personality disorders (Chanen et al., 2004; Lofgren et al., 1991), anxiety and depression symptoms (Johnson et al., 1999; Wenning, 1990), substance misuse (Cohen et al., 2007) and self-harming behaviours (Johnson et al., 1999). There is very limited evidence examining associations with subthreshold or threshold psychotic (Thompson et al., 2012) or bipolar spectrum disorders. These omissions need to be addressed as psychotic symptoms are relatively common in adult BPD (Kaess et al., 2014) and concurrent comorbidity between BPD and bipolar disorder is approximately 25–30% (George et al., 2003). Finally, previous studies have not simultaneously modelled pathways between BPD and multiple psychopathological outcomes, despite the known high levels of comorbidity between psychopathologies (Winsper et al., 2016a). Such an approach is now needed to determine whether associations represent independent links between adolescent BPD symptoms and subsequent psychopathologies, or whether such associations are an artefact of symptom overlap.

Aims

We bridge these gaps in the literature by addressing the following questions:

Are BPD (five or more probable/definite) symptoms in early adolescence associated with subsequent depressive (reported at 12 and 18 years), psychotic (reported at 12 and 18 years) and lifetime hypomanic (reported at 22–23 years) psychopathology following adjustment for early environmental risks?

Are the associations between BPD symptoms and psychopathology reported in late adolescence/early adulthood (i.e. 18 and 22–23 years) significantly mediated by psychopathological symptoms in early adolescence (i.e. 12 years)?

Methods

Sample

The Avon Longitudinal Study of Parents and Children (ALSPAC) is a UK birth cohort study examining the determinants of development, health and disease during childhood and beyond. The study is described in detail elsewhere (Boyd et al., 2013). ALSPAC recruited pregnant women in Avon with expected dates of delivery between the 1 April 1991 and 31 December 1992. A total of 14,541 pregnant women were initially enrolled in the study and had returned at least one questionnaire or attended a ‘Children in Focus’ clinic by 19 July 1999. Out of the initial 14,541 pregnancies, all but 69 had known birth outcome. These 14,472 pregnancies yielded 14,676 foetuses in the initial sample. Of these, 14,477 were singleton, 195 were twin, 3 were triplet and 1 was quadruplet. There were 14,062 live births, of which 13,988 children were alive at 1 year of age (see Figure 1 for the flowchart of sample numbers at each stage of the study). When the oldest children were approximately 7 years old, the sample was bolstered with eligible cases who had failed to join the study originally. Consequently, when considering variables collected from the age of 7 onwards, there are data available on 14,701 children (an additional 713 children). The study website contains details of the data available through a fully searchable data dictionary (www.bris.ac.uk/alspac/researchers/data-access/data-dictionary/). From the first trimester of pregnancy, parents completed postal questionnaires about the study child’s health and development. The child attended annual assessment clinics, including face-to face interviews and psychological and physical tests. Ethical approval for the study was obtained from the ALSPAC Law and Ethics Committee and the local research ethics committees.

Figure 1.

Flow diagram detailing numbers of participants who completed key follow-up assessments during the Avon Longitudinal Study of Parents and Children (ALSPAC).

Assessments

BPD symptoms at 11–12 years

Adolescents were interviewed to assess BPD symptoms over the past 2 years. Trained psychologists used a face-to-face semi-structured interview: the UK Childhood Interview for DSM-IV Borderline Personality Disorder [UK-CI-BPD] (Zanarini et al., 2004). The UK-CI-BPD was adapted from the US version of the CI-BPD and both are based on the module for BPD in the Diagnostic Interview for DSM-IV Personality Disorders. The inter-rater and test–retest reliability of the Diagnostic and Statistical Manual of Mental Disorders (3rd ed.; DSM-III), Diagnostic and Statistical Manual of Mental Disorders (3rd ed.; rev., DSM-III-R) and Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV) versions of this measure are good to excellent (Zanarini and Frankenberg, 2001). In a test of convergent validity, the CI-BPD diagnosis was significantly associated with clinician diagnosis (kappa = 0.34, p < 0.001), and other measures of BPD (i.e. Borderline Personality Features Scale, Personality Assessment Inventory – Borderline subscale) reported by patients and parents (Sharp et al., 2012). The inter-rater reliability of the UK-CI-BPD within the current sample ranged from 0.36 to 1.0 (median: 0.88); 86% of the kappa values were within the excellent range of >0.75 (Zanarini et al., 2011).

Consistent with the DSM-IV and Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5) Section II conceptualizations of BPD, the UK-CI-BPD covers nine symptoms: intense inappropriate anger; affective instability; emptiness; identity disturbance; paranoid ideation; abandonment; suicidal or self-mutilating behaviours; impulsivity and intense unstable relationships. Each symptom was rated as absent (score = 0), probably present (score = 1) or definitely present (score = 2). Definitely present refers to very frequent occurrence (daily or at least 25% of the time), and probably present refers to repeated but not very frequent occurrence. As in previous studies (Winsper et al., 2015), we constructed a dichotomous BPD symptom variable. Participants were assigned a positive status if they scored 1 or 2 on five or more of the DSM-IV diagnostic criteria. This symptom threshold (i.e. five or more symptoms) is consistent with the current DSM-5 diagnosis of BPD (American Psychiatric Association, 2013). However, we lowered the frequency threshold to include the repeated (score of 1) and very frequent (score of 2) occurrence of each symptom to ensure that sufficient positive cases were identified in our community sample. Therefore, our BPD assessment is more sensitive than current clinical diagnosis.

Depressive symptoms at 12 years

Child-reported depressive symptoms over the past 2 weeks were assessed using the Short Mood and Feelings Questionnaire (SMFQ). The SMFQ is a 13-item scale (items on a three-point scale) with high validity and reliability (Zwierzynska et al., 2013). Responses to items (no = 0; sometimes = 1; true = 2) were summed yielding a maximum score of 26 points. The total score for each participant was collapsed into a dichotomous variable (threshold score ⩾ 11), with higher scores consistent with clinically relevant levels of depressive symptoms (Zwierzynska et al., 2013).

Psychotic symptoms at 12 years

Psychotic symptoms were assessed with the semi-structured Psychosis-Like Symptom Interview, which contains 12 core questions, which are hallucinations (visual and auditory), delusions (spied on, persecution, thoughts read, reference, control and grandiosity) and experiences of thought interference (broadcasting, insertion and withdrawal). Unspecified delusions were also rated. Items are derived from the Diagnostic Interview Schedule for Children, Version IV (DISC-IV) and the Schedules for Clinical Assessment in Neuropsychiatry, Version 2.0 (SCAN). Trained interviewers rated each item as absent, suspected or definitely present. The average kappa was 0.72 indicating good inter-rater reliability. Consistent with previous research, we used at least one definite psychotic symptom (Singh et al., 2014), to demarcate the presence of relevant experiences.

International Classification of Diseases, 10th Revision, depression at 18 years

Depression was assessed with the Revised Clinical Interview Schedule (CIS-R; Lewis et al., 1992). The CIS-R has been validated in community populations (Lewis et al., 1992) and is widely used in the Adult Psychiatric Morbidity Surveys of Great Britain. The CIS-R was self-administered via computerised interview during a research clinic. This scale establishes the severity of symptoms of depressive disorders in the past month and uses diagnostic cut-off points based on International Classification of Diseases, 10th Revision (ICD-10) criteria. We used a categorical depression variable representing an ICD-10 diagnosis of mild, moderate or severe depression at the age of 18 years.

Psychotic symptoms at 18 years

Psychotic symptoms since age 12 were assessed with the semi-structured Psychosis-Like Symptom Interview when participants were age 18 (Winsper et al., 2016b) (described above). The average kappa value was 0.83. Test–retest reliability was assessed with 162 adolescents re-interviewed after approximately 47 days (kappa = 0.76, standard error [SE] = 0.078), 46 of whom were re-interviewed by the same interviewer (kappa = 0.86, SE = 0.136) (Zammit et al., 2013). Consistent with previous research, the psychotic symptom outcome represented at least one definite psychotic symptom (Winsper et al., 2016b).

Hypomanic symptoms at 22 years

Participants completed the ‘Your Life Now (at age 21+)’ postal questionnaire at age 22–23 years, including the Hypomania Checklist Questionnaire (HCL-32). Hypomanic symptoms were not assessed before this point. The HCL-32 is a self-report questionnaire pertaining to the lifetime experience of manic symptoms (Angst et al., 2005). Respondents are asked to consider a time when they were in a ‘high or hyper’ condition and endorse various statements about their emotions, thoughts and behaviours during this time. There are 32 symptom statements, e.g. ‘I think faster’ and ‘I need less sleep’. The HCL-32 was developed to screen for possible subthreshold hypomanic syndromes in people with depressive disorders and has been used previously in non-clinical populations of youth.

Consistent with this previous work (Marwaha et al., 2017), we constructed a lifetime history of hypomania variable, with a threshold of 14+ out of 32 hypomanic features. In addition, participants had to report the following: at least one occurrence of either ‘negative consequences’ or ‘negative plus positive consequences’, due to hypomanic symptoms; and that mood changes caused a response (neutral, negative or negative and positive) in close others; and that hypomanic symptoms lasted for a duration of at least ‘2–3 days’. A duration of 2–3 days or more was originally chosen because it was deemed important that the screening tool was sensitive to experiences of brief hypomania (i.e. syndromes lasting 2–3 days rather than the usual diagnostic threshold of 4 days). This duration increases the detection of individuals at risk of bipolar disorder that meets diagnostic criteria and previous research had demonstrated that individuals who experience brief hypomania have similar clinical and family histories to individuals with bipolar I disorder (BD-I) and bipolar II disorder (BD-II). We choose this conservative threshold (i.e. severity, duration and impairment) to identify clinically relevant symptoms.

Psychosocial confounders

We controlled for gender, family adversity and physical and sexual abuse, given that they are associated with depression, psychotic symptoms and hypomania (Marwaha and Bebbington, 2015; Marwaha et al., 2016). Multiple family risk factors were assessed using the Family Adversity Index (FAI) during pregnancy (‘long index’), 2 years (‘long index’) and 4 years (‘short index’). The FAI ‘long index’ comprises 18 items (Wolke et al., 2012). The short index has 15 of the same items (e.g. financial status, housing). Each item was given one point (if yes). Points were summed at each time-point for a total FAI score.

Physical and sexual abuse were mother reported (postal questionnaire) when children were 1.5, 3.5, 4.8, 5.8 and 6.8 years old. We constructed two variables representing physical and sexual abuse, respectively, at any time-point.

Statistical analysis

Missing data

While 6091 completed the BPD interview at age 11 years, inclusion in the final sample depended on the child also having completed the depression computerised interview at 18 years, the psychotic symptoms interview at 18 years and the hypomania questionnaire at 22–23 years. This meant that the sample for the final path model was 1758 (see Figure 1). We conducted logistic regressions to determine the predictors of missing data. These were male sex, low birthweight and family adversity (Supplemental Table S1). We therefore constructed weights based on these predictors, to account for missing data within our analysis.

Logistic regression analysis

We chose to take an approach that used dichotomous variables so that measures of effect were odds ratios (ORs), which are generally more easily understood among clinicians than coefficients. Where possible, the dichotomous variables are also linked with relevant clinical diagnoses, aiding the extent to which results are clinically interpretable. We conducted logistic regressions with SPSS version 25. BPD was the independent variable and depressive, psychotic and hypomania symptoms were the dependent variables.

We developed our model in four stages, iteratively adjusting for confounders: (1) unadjusted associations; (2) adjusted for gender, family adversity and physical/sexual abuse; (3) plus depressive and psychotic symptoms at 12 years and (4) plus depressive and psychotic symptoms at age 18 for the hypomanic outcome, depression at 18 years for the psychotic outcome and psychotic symptoms at age 18 years for the depression outcome.

Path analysis

We conducted path analysis modelling using Mplus version 7.11. We controlled for multiple pathways (concurrent and prospective) between psychopathological measures (e.g. depressive symptoms at 12 and 18 years; depressive and psychotic symptoms at 18 years) to determine the independent associations between BPD and subsequent psychopathologies. We included our weighting variable in the path model.

Results

Descriptive and unadjusted statistics

At age 11–12 years, 6.7% of adolescents reported five or more probable/definite symptoms over the past 2 years. At 18 years, 7.9% had any ICD-10 diagnosis of depression (experienced during the past month), while 4.1% reported experiencing one or more psychotic symptoms since age 12 years. At 22–23 years, 8% reported lifetime clinically relevant hypomanic symptoms. Psychotic and depressive symptoms were strongly associated (OR: 5.50; 95% confidence Interval [CI]: [3.20, 9.45]). Associations between psychotic experiences and hypomania symptoms (OR: 1.69; 95% CI: [0.82, 3.47]) and depression and hypomania symptoms were not significant (OR: 1.44; 95% CI: [0.82, 2.54]).

Logistic regression models

Results of the weighted analysis are reported in Table 1. BPD symptoms significantly predicted each of the psychopathological outcomes in both unadjusted and adjusted analyses. Associations remained largely unchanged following control for psychosocial confounders (Model 2). Overall, the biggest attenuation effect followed the incorporation of depressive and psychotic symptoms at 12 years (Model 3).

Table 1.

Weighted unadjusted and adjusted associations between BPD in early adolescence and psychopathological outcomes in late adolescence/early adulthood.

Depression at 18 years		Psychotic symptoms at 18 years		Hypomanic symptoms at 22–23 years
Odds ratio (95% CI)		Odds ratio (95% CI)		Odds ratio (95% CI)
Unadjusted	Adjusted	Unadjusted	Adjusted	Unadjusted	Adjusted
Model 12.31 [1.90, 2.80]	Model 22.09 [1.72, 2.55]	Model 13.05 [2.42, 3.86]	Model 22.84 [2.24, 3.60]	Model 12.90 [2.43, 3.47]	Model 22.94 [2.46, 3.51]
	Model 31.46 [1.17, 1.82]		Model 32.39 [1.85, 3.08]		Model 32.96 [2.46, 3.56]
	Model 41.30 [1.03, 1.64]		Model 42.36 [1.82, 3.06]		Model 42.89 [2.40, 3.48]

BPD: borderline personality disorder.

Significant associations are in boldface.

Analysis was weighted for significant predictors of attrition (gender, birthweight and family adversity); results are reported as odds ratios with 95% confidence intervals (CIs).

Model 2: adjusted for gender, family adversity and any abuse in early childhood; Model 3: all psychosocial confounders and additionally controlled for psychotic symptoms and depression at 12 years; Model 4: All confounders in Model 3 and additionally controlled for depression for the psychotic outcome, psychotic symptoms for the depression outcome and depression and psychotic symptoms for the hypomania outcome.

Path analysis

The path model is shown in Figure 2. The model had adequate (confirmatory fit index [CFI] = 0.94) to good (χ² = 23.84, p = 0.07; root mean square error of approximation [RMSEA] = 0.02) fit. Four independent trajectories from BPD were evident in the path analysis. First, BPD symptoms were independently associated with clinically relevant depressive symptoms, one or more psychotic symptoms at 12 years and clinically relevant hypomanic symptoms at 22–23 years. Second, BPD symptoms were indirectly associated with ICD-10 depression (mild, moderate or severe) at 18 years via depressive symptoms at 12 years. BPD symptoms were not significantly indirectly associated with psychotic symptoms at 18 years via psychotic symptoms at 12 years. However, BPD symptoms were significantly linked to psychotic symptoms at age 12 years, which in turn were significantly associated with psychotic symptoms at age 18 years (Table 2 and Figure 2).

Figure 2.

The final path model including associations between adolescent BPD and subsequent psychopathologies.

Table 2.

Direct and indirect associations between BPD, depression, psychotic and hypomanic symptoms within the final path model.

Predictor variables	Psychopathological outcomes
Predictor variables	Depressive symptoms at 12 years	Psychotic symptoms at 12 years	Depression at 18 years	Psychotic symptoms at 18 years	Hypomanic at 22 years
Direct associations
BPD at 11–12 years	0.97 (p < 0.001)	0.58 (p < 0.001)	0.22 (p = 0.28)	0.22 (p = 0.34)	0.58 (p < 0.01)
Depression at 12 years	NA	0.32 (p < 0.001) ^a	0.25 (p < 0.01)	0.20 (p = 0.07)	NA^b
Psychotic symptoms at 12 years	0.32 (p < 0.001) ^a	NA	0.02 (p = 0.90)	0.34 (p < 0.01)	NA^b
Depression at 18 years	NA	NA	NA	0.37 (p < 0.001) ^a	0.03 (p = 0.79)
Psychotic symptoms at 18 years	NA	NA	0.37 (p < 0.001) ^a	NA	0.11 (p = 0.30)
Indirect associations
BPD via depression at 12 years	NA	NA	0.24 (p < 0.05)	0.20 (p = 0.08)	NA
BPD via psychotic symptoms at 12 years	NA	NA	0.01 (p = 0.90)	0.20 (p = 0.057)	NA
BPD via depression at 18 years	NA	NA	NA	NA	0.01 (p = 0.80)
BPD via psychotic symptoms at 18 years	NA	NA	NA	NA	0.02 (p = 0.48)

BPD: borderline personality disorder; NA: not applicable.

Boldface indicates significant associations.

Results are reported as probit coefficients; probit coefficients indicate the strength of the relationship between the predictor variable and the probability of group membership, representing the change in the probability of ‘caseness’ associated with a unit change in the independent variable. Analysis was weighted for predictors of attrition.

These are correlations.

We did not include these pathways in the final model as it led to saturation and thus no model fit information. However, when included in the model, reported associations did not change.

Discussion

Our study represents a first step towards explicating the psychopathological sequelae of BPD symptoms (five or more repeated or very frequent) in a community population. Adolescent BPD symptoms increased the risk of subsequent psychotic symptoms, a diagnosis of depression reported at 18 years and clinically relevant lifetime hypomanic symptoms reported at 22–23 years. The prospective direct association with depression diagnosis at 18 was not significant following control for all confounders (including inter-relationships between all psychopathological symptoms) in the path analysis. However, BPD was significantly indirectly associated with depression diagnosis via depressive symptoms at 12 years.

The associations between adolescent BPD symptoms and subsequent psychotic and hypomanic symptoms are notable given that they have received limited prospective investigation. It is suggested that many psychiatric disorders are sequentially comorbid, recurrent or chronic (Caspi et al., 2014). Our findings suggest that young adolescents with substantial BPD symptomatology are at increased risk of experiencing a range of clinically relevant psychiatric symptoms across several developmental stages (early/late adolescence, early adulthood). Independent developmental pathways linked to hypomania and psychotic symptoms were not likely attributable to symptom overlap between outcomes, as we simultaneously controlled for all associations between each of the psychopathological outcome assessments. These findings are consistent with notions of multifinality (i.e. BPD has divergent outcomes).

Our results are in line with the hypothesis that early BPD symptoms can be seen as an adolescent manifestation of the confluence of internalizing and externalizing (IE) psychopathology on the severity pathway that leads to subsequent major mental disorders (Sharp and Wall, 2018). These outcomes may be determined by the severity level of IE, mediating and moderating contexts and transactional feedback loops with the manifest syndrome (Beauchaine and Cicchetti, 2016). Indeed, Hayes et al. (2017) hypothesize that the genetic architecture of maladaptive personality traits and serious mental illness may be shared, with emotional instability (a key feature of BPD), in particular predicting serious mental illness. Maladaptive interpersonal functioning (e.g. self–other relatedness, including maladaptive self-perception and social function), which is another key feature of BPD, may be particularly important as a precursor to other personality pathology (Sharp et al., 2018; Sharp and Wall, 2018). Thus, personality pathology can be considered an important severity indicator in adolescence that should not be ignored.

Previous research suggests three trajectories for individuals with adolescent BPD symptoms: (1) continuation of BPD symptoms over time, (2) remission of psychiatric symptoms and (3) the development of other disorders (Winsper et al., 2015). Our analyses highlight that there is a risk that individuals develop other types of clinically relevant psychopathology and supports the idea of heterotypic continuity as a developmental pathway. It also supports the notion that adolescents with BPD symptoms are a clinically vulnerable group by virtue of their current symptomatology, but also due to risks of future psychopathology. This helps explain the long-term morbidity associated with the condition and why the outcomes can be for so devastating into adulthood.

Our study has strengths and limitations. Strengths include the use of data from a large, well-defined birth cohort, with repeated assessments of important psychopathologies covering a developmental period largely neglected in BPD research. Using a validated interview for BPD, we could identify adolescents with five or more BPD symptoms, thus creating a variable comparable in composition to the DSM-5 diagnosis for BPD (American Psychiatric Association, 2013). In our community sample, 6.7% of adolescents met our criteria for substantial BPD symptoms. This rate is somewhat higher than the 1.8% (95% CI: [1.2%, 2.5%]) reported in adult community populations (Winsper et al., 2019) and is likely attributable to our lower symptom frequency threshold (i.e. inclusion of repeated and very frequent symptom counts). Other small-scale community studies with younger populations have suggested a higher prevalence of BPD (e.g. 11% using algorithms for DSM-III-R diagnosis), with a peak in prevalence at 12–13 years of age (Bernstein et al., 1993). This suggests that our data handling strategy is likely to have some clinical utility (i.e. in identifying subclinical youth at heightened risk of BPD). While our BPD symptom variable does not equate to a clinician-derived BPD diagnosis, it does represent a good balance between clinical and scientific needs. We were able to study a relatively large group of youths demonstrating five or more BPD symptoms, while avoiding the inevitable resource issues associated with providing clinical assessments in epidemiological research. At age 18 years, 7.9% of the sample had any ICD-10 diagnosis of depression consistent with previous epidemiological data (Costello et al., 2005), again supporting the validity of our approach to data handling and analysis. Another strength was our use of advanced analytical methods, e.g. path analysis, which allowed us to understand the prospective connections between psychopathological clusters. Future analyses should consider the use of network analysis to further understand the interconnections between multiple psychopathological and other parameters in studies of adolescent BPD.

There were also limitations. First, while the prospective design enabled us to model predictive pathways based on temporal ordering, some assessments lacked temporal precision, e.g. psychotic symptoms at 18 referred to events occurring since the previous assessment (12 years), and the hypomanic scale referred to lifetime occurrence. However, it should be highlighted that pre-pubertal hypomania is very rare (Kroon et al., 2013), and it is unlikely that respondents would remember symptoms occurring several years ago. Second, there was considerable selective attrition in the cohort reducing statistical power and potentially biasing our results. We weighted both our logistic and path analyses to account for this attrition (Kinner et al., 2007). Third, while we included clinically relevant levels of psychiatric symptoms where possible (e.g. ICD depression diagnosis, hypomanic symptoms according to severity, duration and impairment), we could not include psychotic disorder as the frequency was too low. Nevertheless, psychotic symptoms are a marker for, and significantly increase risk of, psychotic disorder in adults (Fisher et al., 2013). Fourth, childhood risk factors, including family adversity and physical and sexual abuse, were reported by the child’s mother. This could have led to issues such as under-reporting as mothers may have been unaware of abuse experiences or were reluctant to report these factors as they could imply neglected care. Thus, there is a small chance that we could have underestimated the impact of confounders on associations between psychopathologies. Furthermore, some of our psychopathological outcomes were assessed via interview, either face to face or computerized (i.e. psychotic symptoms reported at 12 and 18 years and depression symptoms reported at 18 years), and others via self-report postal questionnaire (i.e. depression symptoms reported at 12 years and hypomania symptoms reported at 22–23 years). Thus, there were variations in mode of administration, which could have potentially influenced responses in reference to the different psychopathological outcomes (Bowling, 2005). These differences could have led to an under-estimation of the strength of associations between psychopathologies. Ideally, we would use interview data for all psychopathologies as the gold standard approach to assessing mental disorders. However, this was not possible due to the considerable resource demands, e.g. young adults assessed at 22–23 years of age likely moved to locations across the country. Fifth, BPD symptoms were only assessed once in this cohort. This is unfortunate and likely reflects the substantial costs involved in administering clinical interviews to very large samples, and the current underfunding of BPD in comparison to other psychiatric disorders (Zimmerman and Gazarian, 2014). Subsequently, we could not examine whether adolescents with BPD symptoms continued to manifest chronic BPD symptoms in early adulthood or whether associations with other psychopathological outcomes (e.g. psychotic and hypomania symptoms) remained significant following control for concurrent BPD symptoms. Therefore, it is possible that associations between adolescent psychopathology and subsequent psychotic, depression and hypomania symptomatology were explained by a continuation of BPD symptoms into late adolescence and early adulthood. Clinical studies indicate that most episodes of psychotic and dissociative symptoms in patients with BPD are transient, and of a different structural, functional and physiologic nature to those observed in psychotic disorder (e.g. amygdala over-reactivity vs decreased amygdala activity). This suggests that associations were not solely attributable to concurrent BPD in late adolescence. However, studies do indicate that a subset of BPD patients may experience permanent and more severe psychotic symptoms over time (Barnow et al., 2010). Similarly, clinical studies suggest that BPD and bipolar disorders are independent, though there may be a subgroup of individuals with characterological bipolar disorder, i.e. they demonstrate both BPD and BD-II symptoms that are less tied to BD-I (Gunderson et al., 2014). Relatedly, given the overlap between diagnostic clusters of mental disorders and the fact that those who did not meet criteria for BPD symptoms in early adolescence may have done so subsequently (e.g. at 18 or 22–23 years), it is possible that we measured the same transdiagnostic symptom pool over several time-points. We cannot completely discount this possibility, but it is likely that the reliable and valid measures such as the UK-CI-BPD, CIS-R and HCL are at least to some extent able to accurately differentiate different symptom clusters. Finally, we did not examine the influence of treatment within our analyses. However, in the UK context, services and specialist programmes for adolescents with BPD are very limited and it is unlikely that specific treatment for BPD in adolescence will have had a major influence on our findings, especially given that few UK psychiatrists will diagnose BPD before age 18 years.

In conclusion, in a large birth cohort study, we found that adolescent syndromal BPD was associated with a range of psychopathological sequelae up to 10 years later. Our findings add to the growing literature on the adverse outcomes of adolescent BPD (Winsper et al., 2015) and should raise clinical awareness that if BPD symptoms are measured in early adolescence, a significant subpopulation already appears to have a BPD-like syndrome and these individuals are at risk of a range of future psychopathologies. This study raises the possibility that early secondary intervention (Chanen et al., 2014) may help reduce the risk of poor psychopathological outcomes potentially offering an avenue for primary prevention of other major mental disorders. The next step in our investigation is to identify the determinants (e.g. environmental risks) of different developmental trajectories to increase our understanding of the varied outcomes of adolescent BPD (Forbes et al., 2016).

Supplemental Material

Supplementary_Tables – Supplemental material for Psychopathological outcomes of adolescent borderline personality disorder symptoms

Supplemental material, Supplementary_Tables for Psychopathological outcomes of adolescent borderline personality disorder symptoms by Catherine Winsper, Dieter Wolke, Jan Scott, Carla Sharp, Andrew Thompson and Steven Marwaha in Australian & New Zealand Journal of Psychiatry

Footnotes

Acknowledgements

The authors are extremely grateful to all the families who took part in this study, the midwives for their help in recruiting them and the whole ALSPAC team, which included interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists and nurses. The authors give special thanks to Dr Andrea Waylen, PhD, and Dr Jeremy Horwood, PhD, who helped in the conduct of the study. The UK Medical Research Council and Wellcome (Grant ref. 102215/2/13/2) and the University of Bristol provide core support for ALSPAC. This publication is the work of the authors and C.W. and S.M. will serve as guarantors for the contents of this paper.

Declaration of Conflicting Interests

S.M. attended a World Psychiatric Association consensus meeting on early intervention sponsored by an unrestricted educational grant from Sonuvion. All the other authors declare no conflicts of interest.

Funding

The author(s) received no financial support for the research, authorship and/or publication of this article.

ORCID iDs

Jan Scott

Steven Marwaha

Supplemental Material

Supplemental material for this article is available online.

References

American Psychiatric Association (2013) Diagnostic and Statistical Manual of Mental Disorders (DSM-5®). Washington, DC: American Psychiatric Association.

Angst

Adolfsson

Benazzi

, et al. (2005) The HCL-32: Towards a self-assessment tool for hypomanic symptoms in outpatients. Journal of Affective Disorders 88: 217–233.

Barnow

Arens

Sieswerda

, et al. (2010) Borderline personality disorder and psychosis: A review. Current Psychiatry Reports 12: 186–195.

Beauchaine

Cicchetti

(2016) A new generation of comorbidity research in the era of neuroscience and research domain criteria. Development and Psychopathology 28: 891–894.

Belsky

Caspi

Arseneault

, et al. (2012) Etiological features of borderline personality related characteristics in a birth cohort of 12-year-old children. Development and Psychopathology 24: 251–265.

Bernstein

Cohen

Velez

, et al. (1993) Prevalence and stability of the DSM-III-R personality disorders in a community-based survey of adolescents. American Journal of Psychiatry 150: 1237–1243.

Bowling

(2005) Mode of questionnaire administration can have serious effects on data quality. Journal of Public Health 27: 281–291.

Boyd

Golding

Macleod

, et al. (2013) Cohort profile: The ‘children of the 90s’ – The index offspring of the Avon Longitudinal Study of Parents and Children. International Journal of Epidemiology 42: 111–127.

Caspi

Houts

Belsky

, et al. (2014) The p factor one general psychopathology factor in the structure of psychiatric disorders? Clinical Psychological Science 2: 119–137.

10.

Chanen

Jackson

McGorry

, et al. (2004) Two-year stability of personality disorder in older adolescent outpatients. Journal of Personality Disorders 18: 526–541.

11.

Chanen

McCutcheon

Kerr

(2014) HYPE: A cognitive analytic therapy-based prevention and early intervention programme for borderline personality disorder. In: Sharp

Tackett

(eds) Handbook of Borderline Personality Disorder in Children and Adolescents. New York: Springer, pp. 361–383.

12.

Cohen

Chen

Crawford

, et al. (2007) Personality disorders in early adolescence and the development of later substance use disorders in the general population. Drug and Alcohol Dependence 88: S71–S84.

13.

Costello

Egger

Angold

(2005) 10-year research update review: The epidemiology of child and adolescent psychiatric disorders: I – Methods and public health burden. Journal of the American Academy of Child & Adolescent Psychiatry 44: 972–986.

14.

Fisher

Caspi

Poulton

, et al. (2013) Specificity of childhood psychotic symptoms for predicting schizophrenia by 38 years of age: A birth cohort study. Psychological Medicine 43: 2077–2086.

15.

Forbes

Tackett

Markon

, et al. (2016) Beyond comorbidity: Toward a dimensional and hierarchical approach to understanding psychopathology across the life span. Development and Psychopathology 28: 971–986.

16.

George

Miklowitz

Richards

, et al. (2003) The comorbidity of bipolar disorder and axis II personality disorders: Prevalence and clinical correlates. Bipolar Disorders 5: 115–122.

17.

Grilo

McGlashan

Oldham

(1998) Course and stability of personality disorders. Journal of Psychiatric Practice® 4: 61–75.

18.

Gunderson

Stout

Shea

, et al. (2014) Interactions of borderline personality disorder and mood disorders over 10 years. Journal of Clinical Psychiatry 75: 829–834.

19.

Hayes

Osborn

Lewis

, et al. (2017) Association of late adolescent personality with risk for subsequent serious mental illness among men in a Swedish Nationwide Cohort Study. JAMA Psychiatry 74: 703–711.

20.

Jaffee

(2017) Child maltreatment and risk for psychopathology in childhood and adulthood. Annual Review of Clinical Psychology 13: 525–551.

21.

Johnson

Cohen

Skodol

, et al. (1999) Personality disorders in adolescence and risk of major mental disorders and suicidality during adulthood. Archives of General Psychiatry 56: 805–811.

22.

Kaess

Brunner

Chanen

(2014) Borderline personality disorder in adolescence. Pediatrics 134: 782–793.

23.

Kinner

Alati

Najman

, et al. (2007) Do paternal arrest and imprisonment lead to child behaviour problems and substance use? A longitudinal analysis. Journal of Child Psychology and Psychiatry 48: 1148–1156.

24.

Kroon

Wohlfarth

Dieleman

, et al. (2013) Incidence rates and risk factors of bipolar disorder in the general population: A population-based cohort study. Bipolar Disorders 15: 306–313.

25.

Lewis

Pelosi

Araya

, et al. (1992) Measuring psychiatric disorder in the community: A standardized assessment for use by lay interviewers. Psychological Medicine 22: 465–486.

26.

Lieb

Zanarini

Schmahl

, et al. (2004) Borderline personality disorder. The Lancet 364: 453–461.

27.

Lofgren

Bemporad

King

, et al. (1991) A prospective follow-up study of so-called borderline children. American Journal of Psychiatry 148: 1541–1547.

28.

Marwaha

Bebbington

(2015) Mood as a mediator of the link between child sexual abuse and psychosis. Social Psychiatry and Psychiatric Epidemiology 50: 661–663.

29.

Marwaha

Gordon-Smith

Broome

, et al. (2016) Affective instability, childhood trauma and major affective disorders. Journal of Affective Disorders 190: 764–771.

30.

Marwaha

Winsper

Bebbington

, et al. (2017) Cannabis use and hypomania in young people: A prospective analysis. Schizophrenia Bulletin 44: 1267–1274.

31.

Sharp

Fonagy

(2015) Practitioner review: Borderline personality disorder in adolescence – Recent conceptualization, intervention, and implications for clinical practice. Journal of Child Psychology and Psychiatry 56: 1266–1288.

32.

Sharp

Wall

(2018) Personality pathology grows up: Adolescence as a sensitive period. Current Opinion in Psychology 21: 111–116.

33.

Sharp

Michonski

, et al. (2012) Borderline personality disorder in adolescents: Evidence in support of the childhood interview for DSM-IV borderline personality disorder in a sample of adolescent inpatients. Comprehensive Psychiatry 53: 765–774.

34.

Sharp

Vanwoerden

Wall

(2018) Adolescence as a sensitive period for the development of personality disorder. Psychiatric Clinics of North America 41: 669–683.

35.

Singh

Winsper

Wolke

, et al. (2014) School mobility and prospective pathways to psychotic-like symptoms in early adolescence: A prospective birth cohort study. Journal of the American Academy of Child & Adolescent Psychiatry 53: 518–527.e511.

36.

Stepp

Smith

Morse

, et al. (2012) Prospective associations among borderline personality disorder symptoms, interpersonal problems, and aggressive behaviors. Journal of Interpersonal Violence 27: 103–124.

37.

Thompson

Nelson

Bechdolf

, et al. (2012) Borderline personality features and development of psychosis in an ‘Ultra High Risk’ (UHR) population: A case control study. Early Intervention in Psychiatry 6: 247–255.

38.

Wenning

(1990) Borderline children: A closer look at diagnosis and treatment. American Journal of Orthopsychiatry 60: 225–232.

39.

Winsper

Bilgin

Thompson

, et al. (2019) The prevalence of personality disorders in the community: A global systematic review and meta-analysis. The British Journal of Psychiatry. Epub ahead of print 12 July 2019. DOI: 10.1192/bjp.2019.166.

40.

Winsper

Lereya

Marwaha

, et al. (2016a) The aetiological and psychopathological validity of borderline personality disorder in youth: A systematic review and meta-analysis. Clinical Psychology Review 44: 13–24.

41.

Winsper

Marwaha

Lereya

, et al. (2015) Clinical and psychosocial outcomes of borderline personality disorder in childhood and adolescence: A systematic review. Psychological Medicine 45: 2237–2251.

42.

Winsper

Wolke

Bryson

, et al. (2016b) School mobility during childhood predicts psychotic symptoms in late adolescence. Journal of Child Psychology and Psychiatry 57: 957–966.

43.

Wolke

Schreier

Zanarini

, et al. (2012) Bullied by peers in childhood and borderline personality symptoms at 11 years of age: A prospective study. Journal of Child Psychology and Psychiatry 53: 846–855.

44.

Zammit

Kounali

Cannon

, et al. (2013) Psychotic experiences and psychotic disorders at age 18 in relation to psychotic experiences at age 12 in a longitudinal population-based cohort study. American Journal of Psychiatry 170: 742–750.

45.

Zanarini

Frankenberg

(2001) Attainment and maintenance of reliability of Axis I and II disorders over the course of a longitudinal study. Comprehensive Psychiatry 42: 369–374.

46.

Zanarini

Horwood

Waylen

, et al. (2004) The UK Version of the Childhood Interview for DSM-IV Borderline Personality Disorder (UK-CI-BPD). Bristol: University of Bristol.

47.

Zanarini

Horwood

Wolke

, et al. (2011) Prevalence of DSM-IV borderline personality disorder in two community samples: 6330 English 11-year olds and 34,653 American adults. Journal of Personality Disorders 25: 607–619.

48.

Zimmerman

Gazarian

(2014) Is research on borderline personality disorder underfunded by the National Institute of Health? Psychiatry Research 220: 941–944.

49.

Zwierzynska

Wolke

Lereya

(2013) Peer victimization in childhood and internalizing problems in adolescence: A prospective longitudinal study. Journal of Abnormal Child Psychology 41: 309–323.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB