Borderline personality disorder (BPD) is a severe mental disorder that is characterised by a pervasive pattern of impulsivity, emotional instability, interpersonal dysfunction and disturbed self-image [1]. BPD affects 0.7–2.7% of the general adult population [2, 3], 9.3–22.5% of psychiatric outpatients, and in some settings, over 40% of inpatients [4].The outcome for BPD in adulthood is now reliably characterised by attenuation of diagnostic criteria over time but severe and continuing functional disability across a broad range of domains that is comparable to or greater than that associated with many mental state disorders [5]. Patients with BPD also have continuing high rates of health service utilisation [6] and a suicide rate of around 8% [7]. Although effective interventions exist for adults with BPD [1], the overall outcomes from such interventions are modest and their availability is limited.

Despite longstanding general agreement that personality disorders (PDs) have their roots in childhood and adolescence [8], diagnosing PDs prior to age 18 years has been more controversial than diagnosing PDs in adults [9], but this is no longer justified [10]. BPD is increasingly seen as a lifespan developmental disorder [11] that is just as reliable and valid in adolescence as it is in adulthood [12], is not reducible to Axis I diagnoses [13], and can be identified in day-to-day clinical practice [14].

When the diagnosis is applied, BPD occurs in approximately 3% of community-dwelling adolescents and young adults [15, 16]. Indeed, BPD might be better considered as a disorder of younger people, with a rise in prevalence from puberty and a steady decline with each decade from young adulthood [17, 18]. Over 3% of the population will meet BPD diagnostic criteria between ages 14 and 22 years [18]. Limited data suggest that BPD occurs in up to 22% of outpatient adolescents and young adults [14, 19].

BPD (or dimensional representations of BPD) in young people demarcates a group with high morbidity and a particularly poor outcome. BPD uniquely and independently predicts current psychopathology, general functioning, peer relationships, self-care and family and relationship functioning [13]. It also uniquely predicts poor outcomes up to 2 decades into the future, such as a future BPD diagnosis, increased risk for axis I disorders (especially substance use and mood disorders), interpersonal problems, distress and reduced quality of life [20–22].

The above data suggest that BPD is a leading candidate for developing empirically based prevention and early intervention programs because it is common in clinical practice, it is among the most functionally disabling of all mental disorders, it is often associated with help-seeking (cf. schizotypal or antisocial PDs) and it has been shown to respond to intervention, even in those with established disorder. Moreover, BPD can be reliably diagnosed in its early stages and it demarcates a group with high levels of current and future morbidity and mortality. Data also suggest considerable flexibility and malleability of BPD traits in youth [23], making this a key developmental period during which to intervene, and adolescent BPD features have been shown to respond to intervention [24, 25].

Prevention and early intervention for BPD should primarily aim to alter the life-course trajectory of young people with borderline personality pathology by attenuating or averting associated adverse outcomes and promoting more adaptive developmental pathways. It should not be narrowly focussed upon the diagnostic features of BPD, as these naturally attenuate over time. Such objectives might be realised through identifying appropriate risk factors and antecedents for intervention.

Evidence supports both gene–environment interaction and correlation in the development of BPD [26]. This means that individuals with a ‘sensitive’ genotype are at greater risk of BPD in the presence of a predisposing environment. Furthermore, the genes that influence BPD features also increase the likelihood of being exposed to certain adverse life events. Overall, the findings regarding neurobiological risk factors for BPD (summarised in [27]) are preliminary and do not provide clear and consistent targets for preventive interventions. For some findings, such as frontolimbic network abnormalities in adults with BPD [1], it is unclear whether these findings are a cause, an effect, or an epiphenomenon of BPD [28] and their specificity for BPD appears to be limited [29].

Prospective, longitudinal data are more consistent in demonstrating that a range of childhood and parental demographic characteristics, adverse childhood experiences, early relational difficulties, parental problems and forms of maladaptive parenting are risk factors for adolescent and adult BPD.

Although there is a strong association between BPD and adverse childhood experiences, the precise role of childhood adversity in the aetiology of BPD remains unclear because putative risk factors, such as childhood abuse, adverse familial environment and a family history of psychopathology are highly intercorrelated [30]. One study has demonstrated that shared genetic influences, causal effects and an interaction between genes and environment, can explain the association between life events and BPD features, depending on the type of life event [26].

Specific data on prospectively assessed risk factors for BPD are scarce, with the Children in the Community (CIC) study [20] being the only study to have published prospective risk factors over multiple waves from childhood through to adulthood. A series of CIC publications (summarised in [20, 31]) report childhood abuse or neglect, childhood and parental demographic characteristics, maladaptive parenting and maladaptive school experiences as risk factors for adolescent and adult PD.

Prospective longitudinal data have found that childhood physical abuse, sexual abuse and neglect [32], along with low family of origin socio-economic status [33] are independently associated with elevated features of BPD up to two decades later. Also, maternal inconsistency in child rearing in the presence of high maternal over-involvement [34];maladaptive parenting behaviour present during the child-rearing years [35]; early separations of offspring from their mothers before age 5 [36] and early relational experiences including attachment disorganization and maltreatment, maternal hostility and boundary dissolution, family disruption related to father’s presence and family life stress [37] all predict elevations in BPD features from 2 to 30 years later.

Prospective longitudinal data indicate that certain temperamental characteristics and early onset mental state or behavioural problems that are analogous to characteristics of BPD are precursors to the emergence of the BPD phenotype but do not predict its onset with certainty. These include attention deficit hyperactivity disorder (ADHD), oppositional defiant disorder (ODD), conduct disorder (CD), substance use, depression and deliberate self-harm (DSH), along with the actual features of BPD. However, it is technically imprecise to refer to many of these phenomena as ‘risk factors’ [38], as these same phenomena are mostly trait-like and are later used to define PD. Eaton et al. [39] refer to the signs and symptoms from a diagnostic cluster that precede a disorder but do not predict its onset with certainty as precursor signs and symptoms.

Maternal reports of childhood temperament are related to BPD in adolescence or adulthood, up to 30 years later [36, 37]. Substance use disorders during adolescence, particularly alcohol use disorders, also specifically predict young adult BPD [40, 41] and there are strong prospective data that disturbances in attention, emotional regulation and behaviour, especially the disruptive behaviour disorders (CD, ODD and ADHD) in childhood or adolescence, are independent predictors of young adult BPD [37, 42, 43]. Moreover, in one study, the rate of growth in ADHD scores from age 10 to 13 and the rate of growth in ODD scores from 8 to 10 uniquely predicted higher BPD symptoms at age 14, suggesting that for adolescent BPD symptoms, difficulties with emotion regulation and relationships might precede problems with impulse control [42].

DSH is a core feature of BPD [1] and retrospective reports from adults with BPD indicate childhood-onset of DSH in more than 30% and adolescent-onset in another 30% [44]. However, DSH is surprisingly under-researched as a potential precursor to BPD. Although DSH is relatively common among adolescents and young adults [45] and is associated with a range of clinical syndromes, there is evidence that repetitive DSH, which is less frequent, might differ from occasional DSH [46]. BPD can be diagnosed in the majority of female adolescent inpatients with DSH [47] and the likelihood of meeting the diagnosis of BPD is greater in adolescents endorsing both DSH and suicide attempts, compared with individuals reporting DSH or suicide attempts alone [48]. Also, the number of BPD criteria met is predictive of whether or not an adolescent has engaged in DSH or attempted suicide [49].

There is now clear evidence that dimensional representations of BPD features have similar stability in adolescence and adulthood [12]. Evidence is emerging that the underlying dimensions of BPD features (conceptualized as impulsivity, negative affectivity and interpersonal aggression) are also stable in children [50, 51]. Only the CIC has specifically measured childhood or adolescent PD features as a predictor of later PD over multiple assessments from childhood to adulthood [20]. PD symptoms in childhood or adolescence were the strongest long-term predictors, over and above disruptive behaviour disorders and depressive symptoms [20, 52–54] of later DSM-IV cluster A, B or C PD. Overall, the CIC data support a normative increase in BPD traits after puberty, perhaps bringing the problems associated with BPD to clinical attention. As this wanes in early adulthood, partly due to maturational or socialization processes [20], a group is revealed that is increasingly deviant compared with their peers [55] and perhaps conforms more to the ‘adult’ BPD phenotype. This suggests that young people displaying BPD features are a major group from which the adult BPD phenotype arises.

In short, signs and symptoms might appear from childhood through to adolescence that resemble aspects of the BPD phenotype and presage its later appearance in adolescence or emerging adulthood. Certain early temperamental and personality features, internalizing and externalizing psychopathology, and specific BPD criteria are all candidate precursor signs and symptoms. However, more work needs to be done to gain a better understanding of the role these factors play in the developmental pathways to BPD and to increase their specificity for BPD.

Stand-alone universal (whole population) prevention of BPD is not currently feasible because BPD is not sufficiently prevalent to justify whole population approaches and it is unclear what form or ‘dose’ of intervention would be appropriate. Similarly, selective prevention (targeting those with risk factors for BPD) is currently impractical because many of the risk factors for BPD (particularly environmental factors) are non-specific and more commonly lead to, or are associated with, outcomes other than BPD. This should not diminish the importance of intervention for some risk factors (e.g., child abuse and neglect) as primary objectives because they are undesirable, immoral or unlawful. However, many factors (e.g. poverty) require major social and political change and are unlikely to have a major impact on BPD prevention in the near future. Also, it is difficult to design studies with adequate statistical power to demonstrate the efficacy or effectiveness of universal and selective prevention [56]. Some of these problems would be overcome if current universal and selective programs (e.g., parent training programs) were to measure multiple syndromes as outcomes, and the above data constitute a strong case for including BPD as one of these syndromes.

The data reviewed above suggest that ‘indicated prevention’ [12] is currently the ‘best bet’ for prevention of BPD. This targets individuals displaying precursor (i.e., early) signs and symptoms of BPD. Although the BPD phenotype is not clearly identifiable in children, its underlying dimensions can be measured, appear to be relatively stable and could be directly targeted. Moreover, typical child and adolescent psychopathology (e.g., disruptive behaviour disorders, DSH, substance use and depressive disorders) might additionally be regarded as targets for indicated prevention of BPD, rather than separate domains of psychopathology that might then be renamed in adulthood. Two programs, described below, have been developed that directly target sub-syndromal borderline pathology in adolescents [24, 25, 57], while concurrently targeting syndromal BPD.

Early detection and intervention for BPD is now justified and practical in adolescence and emerging adulthood [10, 14] and consequently, novel early intervention programs have been developed and researched in Australia [24, 57] and the Netherlands [25]. Such programs should be differentiated from conventional BPD treatment programs that are applied to individuals who have established, complex and severe BPD but happen to be less than 18-years old. Intervention for this latter group should now be considered part of routine clinical practice in adolescent mental health [10].

Indicated prevention and early intervention also offer a unique platform for investigating BPD earlier in its developmental course, where duration of illness factors that complicate the psychopathology and neurobiology of BPD can be minimised [28].

The Australian Helping Young People Early (HYPE) and Dutch Emotion Regulation Training (ERT) treatment programs have several features in common. They have broad inclusion criteria, with limited exclusions for co-occurring psychopathology (which is common in BPD). They view BPD dimensionally, combining sub-syndromal (indicated prevention) and syndromal (early intervention) BPD. BPD and other personality pathology are carefully diagnosed, often supported by semi-structured interview. Both HYPE and ERT are time-limited, being 16–24 and 17 sessions, respectively. Both have adapted interventions designed for adults with BPD to make them developmentally suitable. HYPE uses Cognitive Analytic Therapy (CAT) [58], whereas ERT uses Systems Training for Emotional Predictability and Problem Solving (STEPPS) [59].

The major difference between these programs is that ERT is delivered in a group format as an adjunct to treatment as usual (TAU), whereas HYPE employs a comprehensive, team-based integrated intervention. In a randomised controlled trial (RCT), ERT + TAU was not substantially different to TAU alone [25]. In contrast, a quasi-experimental comparison of the HYPE intervention and TAU [60] found that HYPE achieved faster rates of improvement in internalising and externalising psychopathology and lower levels of psychopathology at 2-year follow-up. This suggests that some or all of the elements of a team-based, integrated intervention (outlined in Box 24.1) might be important for early intervention and matches clinical experience working with this population.