Lifetime and Current Self-Harm Thoughts and Behaviors and Their Relationship to Parent and Peer Attachment

Abstract. Background: Previous research suggests attachment is a vulnerability factor for self-harm thoughts and behaviors in adults. Yet, few studies have investigated this relationship during adolescence, although adolescence is a critical period for changes in attachment relationships and self-harm onset. Whether and how attachment relates to self-harm thoughts and behaviors as measured in daily life is also unknown. Aims: To investigate whether and how paternal, maternal, and peer attachment are associated with lifetime and current adolescent self-harm thoughts and behaviors. Additionally, to examine how different attachment bonds interact in relation to lifetime and current adolescent self-harm thoughts and behaviors. Method: Pre-existing data from N = 1,913 adolescents of the SIGMA study were used. Attachment and lifetime history of self-harm thoughts and behaviors were measured via retrospective questionnaires. Current self-harm thoughts and behaviors were assessed 10 times per day for 6 days using the experience sampling method (ESM). Results: Paternal and maternal attachments were associated with lifetime self-harm thoughts and behaviors and current self-harm thoughts. No significant associations were found between peer attachment and self-harm outcomes. Limitations: Some analyses were underpowered. Conclusion: Our results highlight the importance of parent–child attachment relationships, which may be intervention targets for prevention and treatment of adolescent self-harm.

Self-harm refers to any act of self-poisoning or self-injury carried out by an individual, irrespective of motivation (National Institute for Health and Care Excellence, 2011).It is among the leading causes of death and injury worldwide (World Health Organization, 2021), and engaging in self-harm significantly increases suicide risk (Hawton et al., 2020).Approximately 90% of people who self-harm began during adolescence (Nock & Prinstein, 2004), with the typical age of onset being around 14 years (Gandhi et al., 2018).To enable early intervention and prevention of self-harm thoughts and behaviors, it is crucial to understand the psychological factors that underpin them.
Previous studies have investigated the role of family functioning in self-harm.The results demonstrated that adolescents reporting reduced family support and excessive behavioral parental control were more likely to report self-harm (Baetens et al., 2015;Palmer et al., 2016).Despite the indication that family functioning may play an important role in self-harm, few studies have substantively examined self-harm in relation to what is potentially the most important aspect of family functioning: attachment.
Bowlby's attachment theory (1969) contends that secure attachment bonds develop when children generally experience their parents as available, responsive, and attuned.These secure attachment bonds promote the development of stress buffering intrapersonal and interpersonal skills (Bowlby, 1973).Conversely, children with predominantly negative past attachment experiences will develop an insecure parental attachment bond (i.e., lack of trust, poor communication quality, and high interpersonal alienation).Consequently, these children may adopt other, possibly maladaptive, strategies to cope with intense negative emotions, such as self-harm.
Existing studies on attachment and self-harm report conflicting results, and one study found only indirect effects between parental attachment and self-harm via coping strategies (Glazebrook et al., 2016).While some support the importance of maternal and peer bonds (Gandhi et al., 2016), others emphasize the paternal bond (Santens et al., 2018).One potential explanation for these conflicting results is that secure attachment in one domain could protect against insecure attachment in another (Buyse et al., 2011;Mota et al., 2016).Similarly, double-insecure children (i.e., insecurely attached to both parents) may do worse than those with a secure relationship to at least one parent (Kochanska & Kim, 2013).To our knowledge, no published research has investigated the relative and buffering effects of different types of attachment bonds on self-harm within a large, general population sample.Furthermore, most research has focused on individuals from clinical or adult populations (Claes et al., 2016;Glazebrook et al., 2015).As most young people do not present to health services for self-harm (Geulayov et al., 2018;McMahon et al., 2014), research with a general population adolescent sample could increase generalizability and open up new directions for interventions to prevent selfharm in this underserved population.
A further limitation of previous research on attachment and self-harm is that studies have rarely directly investigated whether attachment is differentially associated with thinking (ideating) about self-harm versus engaging in selfharm behaviors.However, contemporary ideation-to-action theoretical models of suicidal behavior distinguish between the psychosocial processes associated with self-harm thoughts and those associated with self-harm behaviors (Klonsky & May, 2015;O'Connor & Kirtley, 2018).To this end, the current study builds upon recent research (Zortea et al., 2019) highlighting the utility of the Integrated Motivational-Volitional model (IMV; O'Connor & Kirtley, 2018) for understanding how attachment is related to self-harm thoughts and behaviors.Within the IMV, attachment is posited as a premotivational factor, associated with both thoughts and behaviors (Zortea et al., 2019).To date, however, the relationship between attachment and self-harm in the context of the IMV model has been investigated only in adult samples.Determining whether attachment is differentially related to self-harm thoughts and behaviors in adolescence is of practical importance for early intervention efforts, as most self-harm thoughts and behaviors begin during adolescence.
In addition to investigating whether attachment may differentially relate to self-harm thoughts or behaviors, it is also crucial to measure these thoughts and behaviors where they naturally occur in daily life.Previous research on selfharm in adolescence has relied on retrospective, self-report questionnaires that do not assess the dynamic nature of selfharm.Yet, research assessing self-harm thoughts and behaviors repeatedly within short periods (i.e., hours) is critical to meaningfully capture fluctuations as they occur (Czyz et al., 2019;Glenn et al., 2020).ESM is an intensive longitudinal technique assessing self-reports of behaviors and experiences during daily life (Csikszentmihalyi & Larson, 1987).Participants complete questionnaires on a smartphone, several times a day for multiple days, during their normal everyday life.This technique offers a powerful solution to issues of recall bias and ecological validity by providing us with reliable information about individuals' self-harm thoughts and behaviors within their real-world context (Myin-Germeys et al., 2018).
In the current study, we aimed to investigate the association between different attachment relationships and selfharm thoughts and behaviors in adolescents.Using preexisting data from a large, adolescent cohort study (Kirtley et al., 2021), we investigated attachment in relation to both lifetime history and current daily life experiences of selfharm thoughts and behaviors.Specifically, we investigated how mother, father, and peer attachment are associated with lifetime and current self-harm thoughts and behaviors.Additionally, we examined how mother, father, and peer attachment interact with each other in relation to lifetime and current self-harm thoughts and behaviors.We hypothesized that adolescents with higher levels of paternal, maternal, and peer attachment insecurity would be more likely to report lifetime and current self-harm thoughts and behaviors.Furthermore, we predicted that the negative effect of an insecure attachment bond on lifetime and current self-harm thoughts and behaviors would be reduced by security in other attachment bonds.

Method Participants and Recruitment
The data used in this study were drawn from Wave I of the SIGMA study (Kirtley et al., 2021), a large-scale longitudinal study investigating mental health in Flemish adolescents.The original data set includes cross-sectional data from 1,913 adolescents, recruited from the general population via 22 schools across Flanders (Belgium).Schools distributed information letters to potential participants and their caregivers inviting them to participate.Participants were included if they were in their first, third, or fifth year of mainstream secondary education, had an adequate command of the Dutch language, and provided informed consent (from themselves and a caregiver).Most participants within the sample were female (n = 1,207; 63%).The age range was 11-20 years (M = 13.76 years, SD = 1.86 years), with n = 1,048 (55%) in their first year, n = 424 (22%) in their third year, and n = 441 (23%) in their fifth year of mainstream secondary education.
From the full sample (N = 1,913), n = 1,507 have valid lifetime self-harm data and n = 1,788 have valid daily life selfharm data.We used a 10% subsample from both data sets to calculate power and conduct a confirmatory factor analysis (CFA).For the main analyses, samples comprised n = 1,272 with valid lifetime self-harm data and at least one observation in the attachment variable and n = 1,450 with valid daily life self-harm data and at least one observation in the attachment variable.For more details on this selection procedure, see the supplementary material at https://osf.io/8eu9y/.

Procedure
The full procedure for the SIGMA project is detailed elsewhere (Kirtley et al., 2021).Data were collected via selfreport questionnaires completed during a 100-min in-class testing session, and in daily life using the experience sampling method (ESM), between January 2018 and June 2019.Participants completed a battery of self-report questionnaires via tablets during the testing session.Afterward, participants received a smartphone to complete the ESM questionnaires via the MobileQ application (Meers et al., 2020).Participants were asked to complete 10 ESM questionnaires during a 6day period.Each day, notifications were randomly distributed in 90-min blocks.The questionnaire consisted of 39-46 items with an estimated completion time of 2-4 min.Participants had 90 s to complete each item.As reimbursement, all participants received a 10-Euro shopping voucher.Participants received no feedback on their compliance or other data during or after the ESM week.Various steps were taken to ensure participant safety and well-being, including provision of support information.For details, see the supplementary material at https://osf.io/8eu9y/.This study received ethical approval from the UZ/KU Leuven Medical Ethics Committee (S61395).

Measures
The full list of items and details on scoring and construction are included in the supplementary material at https://osf.io/8eu9y/.

Attachment
The Dutch version of the Inventory of Parent and Peer Attachment (IPPA; Armsden & Greenberg, 1987; Dutch translation by Noom et al., 1999) is a self-report questionnaire developed to assess the quality of parental and peer attachment.Three dimensions of attachment security are measured: trust, communication, and alienation.A CFA on two subsamples stratified according to the prevalence of lifetime (n = 149) and current (n = 178) self-harm thoughts and behaviors confirmed a 3-factor structure.
Lifetime Self-Harm Thoughts and Behaviors Four items were used to assess lifetime nonsuicidal and suicidal self-harm thoughts and behaviors, adapted from the Child and Adolescent Self-Harm in Europe study (Madge et al., 2008) questionnaire.All items were translated and back-translated from English to Dutch.Based on these items, three groups were created: 0 = no self-harm thoughts/ behaviors, 1 = thoughts only, and 2 = (thoughts and) behaviors.

Depression
Depressive symptoms were assessed using the 6-item Depression subscale of the Brief Symptom Inventory (Derogatis & Spencer, 1993), and a sum-score was calculated.

Self-Harm Thoughts and Behaviors During Daily Life
In the ESM questionnaire, three self-constructed items were used to assess current self-harm thoughts and behaviors.The three items were branched and presented in a fixed order.The first item was presented at every notification to assess self-harm thoughts.If self-harm thoughts were indicated, one item on self-harm behavior and one item on suicidal intent were followed.

Missing Data
Due to completely missing data on the self-harm and attachment variable, our final sample sizes were n = 1,272 to investigate hypotheses regarding lifetime self-harm and n = 1,450 to investigate hypotheses regarding daily life self-harm.Missing attachment data were imputed using multiple imputation using chained equations.For details on missing data analyses, see the supplementary material at https://osf.io/8eu9y/.

Data Analyses
A model selection procedure was conducted for all models.For further details of the analyses, including power calculations, see https://osf.io/8eu9y/.

Attachment and Lifetime Self-Harm Thoughts and Behaviors
To investigate the relationship between attachment (paternal, maternal, and peer) and lifetime self-harm thoughts and behaviors, we first conducted an ordinal logistic regression with self-harm history (0 = no thoughtsno behaviors group, 2 = only thoughts group, and 3 = [thoughts and] behaviors group) as the outcome to test main effects.Paternal, maternal, and peer attachment insecurities were all set simultaneously as the predictor variables in the model.We then investigated each combination of the following interaction terms: Paternal × Maternal × Peer attachment.Given convergence issues during analysis of the stratified subsample (10%; n = 149), interaction analyses were exploratory.

Attachment and Current Self-Harm Thoughts
The relationships between attachment (paternal, maternal, and peer) and self-harm thoughts in daily life were analyzed using a 2-part mixed-effects model for semicontinuous data (Tom et al., 2016).The model specified a logistic regression for the dichotomous indicator that the outcome is zero or not (0 = having no thoughts or 1 = having thoughts).Then, a standard linear mixed model was estimated for the logarithmic transformation of the nonzero responses (having thoughts with a strength from 1 to 6), allowing for varying intercepts.This model was estimated by specifying a joint distribution that links separate mixed-effects models (i.e., logistic and a linear mixed-effects model for the log-transformed nonzero responses).The two models were linked by covariance between their respective random effects with the results that the two parts of the model were estimated simultaneously (Tooze et al., 2002).Given the general population sample, we expected an excess of zero observations (0 = no self-harm thoughts), and therefore, the 2-part mixed-effects model was especially suitable for analyzing these intensive longitudinal data (Blozis et al., 2020;Farewell et al., 2017).

Attachment and Current Self-Harm Behaviors
To investigate the relationship between attachment and self-harm behaviors in daily life (measured using ESM), we conducted a binary logistic mixed-effects regression with the presence of self-harm behavior (0 = no self-harm behavior, 1 = self-harm behavior) as the outcome.We included random intercepts for persons.
In all models, we simultaneously entered paternal, maternal, and peer attachment as predictor variables and controlled for age, gender, and depression assessed at baseline, given their associations with self-harm thoughts and behaviors.To investigate the interaction effect of attachment on current self-harm thoughts and behaviors, we included each combination of the following interaction terms: Paternal × Maternal × Peer Attachment.

Power Analysis
A 2-step Monte Carlo simulation approach for power calculation was used, as distributions for the variables of interest could not be found within the existing literature.
For the hypotheses regarding lifetime self-harm, the results showed that the model performed with sufficient power (.78-1) when investigating the probability of reporting no thoughts versus thoughts.Analysis on the probability of reporting thoughts versus thoughts and behaviors was underpowered (.22-.69).
For analyses on current self-harm, the model was sufficiently powered (1), except for analysis of the relationship between paternal attachment insecurity and self-harm thoughts (the zero part of the model), which was underpowered (.10).

Deviations From Stage 1
During data analysis, it became clear that we had to deviate slightly from the Stage 1 manuscript.See the supplementary materials at https://osf.io/8eu9y/for a transparent changes document.

Exploratory Analyses
Due to space constraints, all exploratory analyses are reported in the supplementary materials at https://osf.io/8eu9y/.

Confirmatory Analyses
Consistent with previous literature (Gandhi et al., 2016;Santens et al., 2018) and attachment theory (Bowlby, 1969,  Note. a For full sample on a scale from 1 to 7. b For thoughts only, and thoughts and behaviors groups on a scale from 1 to 7. 1973), we found that adolescents with higher levels of paternal and maternal attachment insecurity were more likely to report lifetime and current self-harm thoughts and lifetime self-harm behaviors.Given the association between attachment insecurity and both lifetime self-harm thoughts and behaviors and current self-harm thoughts, our findings also offer partial support for attachment insecurity as a premotivational variable within the IMV model (O'Connor & Kirtley, 2018), where parent-child attachment relationships may increase vulnerability for self-harm thoughts and behaviors (Zortea et al., 2019).Conversely, self-harm may influence attachment relationships (Ferrey et al., 2016), and this should be investigated in future studies.The findings support prevention and treatment initiatives for adolescent self-harm that restore trust and communication in parent-child relationships, e.g., Attachment-Based Family Therapy (Diamond et al., 2010).However, we found no significant associations between paternal or maternal attachment insecurity and the intensity of current self-harm thoughts or presence of current self-harm behaviors.Since our models of the association between attachment insecurity and lifetime self-harm behaviors, and paternal attachment insecurity and the presence of current self-harm thoughts were underpowered, the results should be interpreted with caution and require replication in larger samples.
Despite the equivocal empirical evidence to date, the lack of an association between peer attachment and selfharm was unexpected (Gandhi et al., 2016).A possible explanation for these results could be the younger average age of the current sample (13.8 years) relative to those of previous studies (≥15.6 years) that found an association between peer attachment insecurity and self-harm (Claes et al., 2010;Jarvi et al., 2013).The effect of peer relationships on psychological outcomes increases with age in the transition to adulthood, where peers become increasingly important (Moretti & Peled, 2004;Wilkinson, 2004); therefore, this relationship may only emerge in older adolescents.Moreover, research has often found only an indirect effect of peer attachment on self-harm via parental attachment or self-esteem (Gandhi et al., 2016;Wilkinson & Walford, 2001).

Strengths and Limitations
To our knowledge, this is the first study investigating whether adolescents' attachment insecurity with both parents and peers is associated with self-harm.Given that the current study provides evidence from both retrospective questionnaires and ESM for the relationship between attachment and self-harm, we believe this study contributes to research and practice.Finally, we consider the use of open science practices (the Registered Report article format, open materials, and open code)still the exception rather than the rule in clinical psychology research (Tackett et al., 2019)a major strength.
However, several limitations should be discussed.First, researchers have argued that the IPPA assesses parent and peer relationship quality rather than attachment (Gandhi et al., 2019); therefore, our study may be more appropriately characterized as investigating relationship quality.Second, some analyses were underpowered.However, the high degree of transparency in our methodology and analysis facilitates future direct replication in higherpowered studies.

Conclusions
The current study provides evidence for the association of both paternal attachment and maternal attachment with lifetime and current self-harm thoughts and lifetime selfharm behaviors, during adolescence, and provides support for attachment as a premotivational variable within the IMV model.Parent-child attachment relationships may be promising intervention targets for the prevention and treatment of adolescent self-harm.Future research should further investigate the buffering effects of different types of attachment bonds on self-harm.
Achterhof uses experience sampling to find out which aspects of daily life social interactions are most relevant for adaptive adolescent development.
No ëmi Hagemann, PhD, graduated from the Center for Contextual Psychiatry, KU Leuven, Belgium.Her research focuses on the lifestylepsychopathology link in adolescence.In particular, she investigates physical activity and sleep using accelerometry.To understand how to make interventions meaningful, Dr. Hagemenn investigates correlates of lifestyle while considering methodological aspects of accelerometry research.Karlijn S. F. M. Hermans, PhD, graduated from the Center for Contextual Psychiatry at KU Leuven, Belgium.Dr. Hermans research focuses on understanding social interaction and its role in psychosis using the experience sampling method and the perceptual crossing experiment to capture real-time social interaction and capacity in adolescents.
Anu P. Hiekkaranta, MA, is a doctoral researcher at the Center for Contextual Psychiatry, KU Leuven, Belgium.Her research uses experience sampling methods to study the role of psychopathology and contextual factors in emotion regulation in daily life in young people and adolescents.Aleksandra Lecei, MA, is a doctoral researcher at the Center for Clinical Psychiatry, KU Leuven, Belgium.Her research focuses on prospectively assessing neurobiological and psychosocial alterations in adolescents following childhood adversity.Specifically, she investigates alterations in the fear-and memorycircuitry using fMRI and psychosocial experiences in a virtual reality setting that may underlie risk and resilience in mental health.
Olivia Kirtley, PhD, is a Research Foundation Flanders senior postdoctoral research fellow at the Center for Contextual Psychiatry, KU Leuven, Belgium.Dr. Kirtley's research uses experience sampling methods to investigate dynamic processes involved in suicidal ideation and behavior.She cochairs the IASP Early Career Group and is an editorial board member of Crisis.