Childhood abuse and borderline personality disorder features in Chinese undergraduates: the role of self-esteem and resilience

A multifactorial model suggested that the development of BPD, to a large extent, is an end product of childhood traumatic experiences such as emotional abuse, physical abuse and sexual abuse [6]. Additionally, researchers have found childhood abuse to be an important predictor of BPD in adolescence and adulthood [7‐9]. Individuals who experienced childhood abuse tended to report higher scores for BPD features than those without that [9].

However, examining the effects of a specific subtype of childhood abuse on BPD features without controlling for the overlap between childhood abuse subtypes does not yield accurate results because most abused children have been exposed to multiple forms of abuse, which exhibit a high rate of co-occurrence [10, 11]. Considering that previous studies have confirmed that among the different forms of childhood abuse, only emotional abuse was a unique predictor of BPD features [12‐14], we hypothesized that only emotional abuse would be uniquely associated with BPD features when other forms of abuse were controlled (Hypothesis 1).

Furthermore, the potential mechanisms of how childhood abuse influences BPD features are unclear. We attempted to explore the potential mediators between childhood abuse and BPD features, which may provide some meaningful guidance for school staff and mental health professionals to prevent the development of BPD features. A growing number of studies have suggested that both resilience and self-esteem were significantly associated with childhood abuse and BPD features [15‐31]. In light of the mediating role of self-esteem and resilience between childhood abuse and mental health problems [32‐34], it was predicted that self-esteem and resilience might play a significant mediating role in the relationship between childhood abuse and BPD features.

As suggested by the ecological systems theory, microsystem (e.g. family environment) was the most proximal factor of individual development [15]. Childhood abuse as a hostile environment for child that may threaten the ability to cope with risk or adversity and positive self-evaluations [16]. Extant research has demonstrated a strong negative relationship between childhood abuse and self-esteem [17‐19], childhood abuse and resilience [20, 21]; that is, the more childhood abuse an individual experienced, the lower his or her self-esteem and resilience. As an overall view of or feeling towards oneself as worthy or unworthy [22], self-esteem is beneficial for the development of human behavior, motivation, cognition, and emotion [23]. Sandler emphasizes that positive self-evaluation has a positive effect, whereas negative self-evaluation leads to a negative effect [16]. Furthermore, there were evidence suggesting that self-esteem may play a protective role against the development of BPD features [24‐26].

Resilience, which refers to a dynamic system for the maintenance of positive adaptation in the face of trauma or adversity [27], may be another promising mediator of this relationship. High resilience has been proven to be linked to better health outcomes in the face of trauma or adversity, while low resilience has been found to lead to adverse consequences on an individual’s mental health [28, 29]. The emotional flexibility theory of resilience also suggests that resilient people can flexibly change their affective and physiological responses to match the demands of frequently changing environmental circumstances [30]; thus, high resilient people could successfully cope with adversity or risk and are less likely to have BPD features [31].

Despite resilience and self-esteem are found to be closely related [35], there are still controversies about their relationship in the literature. Some studies found that resilience could exert a positive impact on the development of self-esteem, potentially through positive affect [36, 37]; Other studies, however, found that the levels of self-esteem could predict resilience [38, 39]. Based on the preceding rationale and previous findings, we hypothesized that resilience and self-esteem mediated the relationship between childhood abuse and BPD features (Hypothesis 2). Considering the uncertainty of the relationship between self-esteem and resilience, we establish three hypothesis models (Model A, Model B, Model C). Model A: two simple mediators (resilience and self-esteem) and one three-path mediator (resilience to self-esteem). Model B: two simple mediators (resilience and self-esteem) and one three-path mediator (self-esteem to resilience). Model C: two simple mediators (resilience and self-esteem).

The objectives of this study were to develop a better understanding of the mediating role of resilience and self-esteem in the development of BPD in the context of different types of childhood abuse in Chinese undergraduates. Understanding the role of self-esteem and resilience in the association between childhood abuse and BPD features may provide school staff and mental health professionals with a number of meaningful directions for preventing the development of BPD features.

A multistage stratification method was used to select participants. First, we randomly selected four universities from Anhui Province of China. Then, we randomly selected 2–5 classes for each grade in four schools, and excluded undergraduates under the age of 18. In total, 4287 undergraduates were invited to complete the questionnaires anonymously; of these undergraduates, 173 undergraduates refused to participate in this survey, and 54 undergraduates were absent. After removing 26 undergraduates who accidentally missed one or more tables of the questionnaire, 4034 undergraduates completed the questionnaire effectively (response rate: 94.1%). The ethics committee of Anhui Medical University approved the study (No. 20180083). We obtained written informed consent from all participants after providing them with a complete and extensive description of the study.

The data analyses were performed using SPSS 21.0 and AMOS 24.0 statistical software. Spearman’s correlation analysis was used to evaluate the correlations between variables. Then, structural equation modeling (SEM) was carried out to analyze the mediation effects. Based on the recommendations of Hooper et al. [47], a model was considered to have an acceptable fit if the comparative fit index (CFI) and Tucker-Lewis index (TLI) were 0.95 or above, the root mean square error of approximation (RMSEA) was less than 0.08, and the χ²/degrees of freedom (χ²/df) was less than 5.00.

A total of 4034 students participated in the study, including 1070 (26.5%) freshmen, 1048 (26.0%) sophomores, 936 (23.2%) juniors, 980 (24.3%) seniors. The participants were aged between 18 and 23 years [mean = 20.38, standard deviation (SD) = 1.35)], and 1692 (41.9%) were females (see Table 1).

As shown in Table 2, BPD features were positively correlated with the three types of childhood abuse and negatively associated with self-esteem and resilience, while the three types of childhood abuse were negatively correlated with resilience and self-esteem (P < 0.001).

Each type of childhood abuse was entered into Model 1, Model 2, and Model 3 separately, and then all types of childhood abuse were simultaneously entered into one comprehensive model (Model 4). All variables were latent variables in our models. Using factorial algorithm to create separate item parcels for variables of emotional abuse, physical abuse, sexual abuse, and self-esteem [48]. Affect dysregulation, impulsivity, self and interpersonal disturbances were indicators of the latent variable of BPD features. Tolerance for stress and goal orientation, adaptability and acceptance of change, trust in one’s instinct, optimism and sense of security were indicators of the latent variable of resilience (see Fig. 1). There was evidence that the data departed significantly from multivariate normality (Model 1 to Model 4, Mardia’s coefficients were 210.163, 319.993, 623.714, 944.968) [49]; thus, these models were estimated with the maximum likelihood estimation, and fit statistics were corrected by using the Bollen-Stine bootstrapping procedure. Indirect effects and direct effects were tested by using bias-corrected bootstrapping procedures (5000 bootstrap samples and 95% confidence interval [CI]) [50, 51]. The indirect effect or direct effect was considered to be statistically significant if the bias-corrected 95% CI did not include zero [52]. The SEM results showed that all of the models had good fits (see Fig. 1, Additional files 1, 2 and 3).

When childhood emotional abuse, physical abuse or sexual abuse was considered individually (see Additional files 1, 2 and 3 for an illustration of Model 1, Model 2 and Model 3), the indirect effects and direct effects and their associated 95% CIs are shown in Additional files 4, 5 and 6. All the indirect effects and direct effects were significant, indicating that resilience and self-esteem were mediators between these three types of childhood abuse and BPD features.

When these three types of abuse were considered simultaneously (see Fig. 1 for an illustration of Model 4), the indirect effects and direct effects and their associated 95% CIs are shown in Table 3. Model 4A, Model 4B and Model 4C all showed that only the direct and indirect effects of emotional abuse on BPD features were significant (see Fig. 1 and Table 3). Consequently, when other forms of childhood abuse were controlled for, the indirect effects and direct effects of both physical abuse and sexual abuse were not found to be significant, and the association between emotional abuse and BPD features was partly mediated via resilience and self-esteem.

Although our study has a number of strengths, including the large sample size, the adoption of a random sampling approach and the separate and simultaneous testing of three types of childhood abuse, our study has several limitations. First, given that the cross-sectional design of our study can’t provide time-ordering of resilience, self-esteem, and BPD features, there may be other models [59]. Future prospective researches are required to explore the relationship between childhood abuse, resilience, self-esteem, and BPD features. Second, childhood abuse, resilience, self-esteem, and BPD features were assessed by self-report; thus, recall bias was inevitable. Third, the sociodemographic variables such as age, gender, area of family residence, grade and academic disciplines were not included in these models, in other words, we did not control for the sociodemographic variables. Fourth, only four universities in one city were included in this study, so the conclusions cannot be extended to all college students in China.