Introduction
Non-suicidal self-injury (NSSI) is defined as the deliberate, direct destruction or alteration of body tissue without conscious suicidal intent [
1]. In the adolescent psychiatric sample, the prevalence of one-time-only NSSI was as high as 60% and the incidence of recurrent NSSI was approximately 50% [
2]. Suicide attempts (SAs) refer to direct efforts to intentionally end one’s own life [
3]. A meta-analysis showed that the overall incidence of SAs among adolescents within 12 months was 6%, internationally [
4]. In China, the prevalence of SAs among adolescents is 1.5–3.6% [
5].
Prospective studies among adolescents suggest that a history of NSSI is a stronger predictor of future suicidal ideation (SI) [
6] and SAs [
7] than previous SAs. Overall, 70% of adolescents engaging in recent NSSI reported a lifetime history of at least one SA [
3]. Having experienced NSSI strongly predicted concurrent or later suicidal thoughts and behaviors (STB), according to the findings of a follow-up study comprising 2,320 college students (aOR = 2.8, 95%Cl = 1.9–4.1). More than 20 lifetime NSSI occurrences among those with prior or present NSSI indicate the risk of STB (aOR = 3.8, 95% Cl = 1.4–10.3) [
8]. This shows that NSSI is a highly important risk factor for death and subsequent SAs.
According to neuropsychological studies, SAs display executive function (EF) abnormalities that may be linked to suicidality [
9]. Patients with recent SAs and current SI have been observed to have impaired executive functioning [
10‐
14]. Recent SAs, in contrast, have revealed significant EF impairment [
13]. This suggests that EF might be particularly impaired around the time that SAs are made. Pu et al. [
15] found that impairments in EF, motor speed function, and overall neuropsychological functioning were associated with SI in patients with major depression. Studies comparing depressed patients with and without SI suggest that SI may be caused by dysfunctional executive decision-making [
16]. Keilp and colleagues [
17] found that depressed, high-lethality suicide attempters performed significantly worse than low-lethality suicide attempters on tests of executive functioning. Additionally, SA risk may be associated with better problem-solving skills but worse inhibitory control [
9].
Research shows that suicidal individuals are characterized by “cognitive rigidity” [
18]. Neurocognitive functioning, such as decision-making and EF, has been identified as a main candidate endophenotype of suicidal behaviors [
19]. According to the integrative model, EF includes mental set shifting (‘‘Shifting’’) and information updating and monitoring (‘‘Updating’’) [
20]. EF deficits can lead to a wide range of difficulties in an individual’s emotional regulation, thoughts, and actions, which may lead to suicidal thoughts or behaviors and increase the risk of suicide [
21]. Additionally, impaired attentional control has been found in suicide attempters and in individuals who are at high risk of suicide, particularly when words related to suicide are used [
22]. Furthermore, deficits in memory performance have been associated with SAs [
23]. A meta-analysis showed that long-term memory and working memory were both more impaired in suicide attempters than in patients and healthy controls [
24], which may have prevented these individuals from using past experiences to solve current problems and envision the future, as well as altered inhibitory processes [
24]. However, it is unclear whether these deficits underlie the executive dysfunction found in other studies [
25].
Several studies have addressed the neurocognitive functioning of adolescents with NSSI behavior. A recent study showed little evidence of neurocognitive (e.g., processing speed, attention, memory, executive functioning) differences, apart from intelligence quotient, between adolescents with NSSI and control subjects [
26]. Research has also found that adolescents with a current history of self-harm exhibit impaired decision-making skills compared to adolescents with a previous history of self-harm, adolescents with depression, and healthy controls [
27]. However, findings across studies are inconsistent. For example, researchers assessing EF in a high-severity NSSI group (n = 33), low-severity NSSI group (n = 29), and healthy control group (n = 35) found distinct significant EF deficits in the NSSI subgroups, with working memory deficits in the high-severity NSSI group and impaired inhibitory control in the low-severity NSSI group [
28]. Zhang et al. [
29] also found that depressed adolescents with NSSI may have executive dysfunction. Therefore, the neurocognitive functioning of depressed adolescent patients with NSSI behavior needs to be further investigated.
NSSI [
30], SI, SAs, and suicide completion [
31] are all substantially correlated with major depressive disorder (MDD) in children and adolescents. Attention, memory and learning, EF, and psychomotor processing are the domains that are most relevant to MDD [
32]. Impairments in these cognitive functions are strongly associated with SI, NSSI, SAs, and death. Self-harm behaviors (i.e., NSSI [
33] and SAs [
34]) are common in borderline personality disorder (BPD), and NSSI was discovered in earlier studies to be an easily accessible marker in the early detection of people at risk of developing BPD [
33]. Executive dysfunction in BPD is associated with suicidality and treatment adherence and may serve as an endophenotype [
35].
Some theories on suicide incorporate depression and hopelessness as necessary or sufficient causes of suicidal thoughts and behaviors (e.g., Interpersonal Psychological Theory of Suicide [
36,
37] and Hopelessness Theory of Suicide [
38]). According to a recent meta-analysis, MDD diagnosis, depression scale score, and hopelessness were the best indicators of SI. For SAs, MDD diagnosis yielded the strongest effect [
39]. Essentially, ruminative thinking is a cycle of unfavorable cognitive processes [
40]. Rumination-related cognitive impairment includes attentional impairment [
41], EF impairment [
42], and set-shifting deficits [
43]. It is evident that both hopelessness and rumination have important effects on SI and SAs, and this study will further investigate the factors associated with cognitive impairment in different groups.
This study aimed to explore differences in cognitive function between NSSI and SAs groups in a clinical sample of adolescents and the factors associated with cognitive impairment in the different groups. For this purpose, the neuropsychological performance of depressed adolescents with NSSI, SAs, and without NSSI or SAs (control group) were compared in the following cognitive domains: processing speed, attention, working memory, emotion recognition, and EF. This was followed by further exploration of the factors associated with cognitive impairment in the different groups, such as depression scores, hopelessness, rumination and borderline personality traits. It was assumed that: (a) NSSI and SAs may be associated with impairments in cognitive function, with worse EF performance in suicidal patients; (b) adolescent depression with NSSI and SAs have different influencing factors, and BPD traits may be indicative of NSSI.
Materials and methods
Participants
142 adolescent depressed patients (12–18 years old) were recruited from the outpatient department of the Department of Psychiatry, First Hospital of Wenzhou Medical University. The sample was categorized into the following groups: 52 adolescents with depression who had a history of SAs within one year, 65 patients who had NSSI within one year, and 25 patients with no history of SAs or NSSI. Inclusion criteria for NSSI and SAs were a history of any self-harm and SAs in the past 12 months, respectively.
The inclusion criteria for the study were as follows: (1) 12–18 years old. (2) Diagnosed with depression by a senior psychiatrist according to the Diagnostic Criteria and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), see Additional file 1. Of these, the subjects in the NSSI group also had to meet the criteria of NSSI (see Additional file 2), and the SAs group had attempted suicide within the past year. (3) Signed informed consent form.
The exclusion criteria were as follows. (1) Presence of psychotic symptoms and other comorbid psychiatric disorders. (2) NSSI group excluded adolescents who had SI and/or SAs within the past year. (3) Neurological illness, intellectual disability, dementia, organic diseases that compromise cognitive functioning, and cognitive syndromes.
Procedure
Following evaluation by a primary care psychiatrist or above, all participants were enrolled. Before the official start of the study, research assistants (testers) were trained to check the accuracy of the data. On weekdays, participants worked together to complete anonymous surveys. Each exam took about 20 min to complete and was conducted one-on-one in a quiet room using a tablet computer. The cognitive tests were completed by each participant on their own. Before the research was conducted, the Wenzhou Medical University Research Ethics Committee evaluated and approved the study protocol. The Declaration of Helsinki was followed in carrying out the study’s methods. The researchers introduced the project to the group of participants and legal guardians at the start of the study. The information provided included the study’s objectives, its methods of data and sample collection, the potential benefits and drawbacks of participation, anticipated outcomes of the research, privacy and confidentiality principles, a statement of voluntary participation, and the researcher’s contact information. Potential volunteers were advised that they could leave at any time. Informed consent was obtained by all the participants and/or their legal guardians.
Measures
A group of professionally trained research assistants aided the participants to complete the sociodemographic data and the clinical assessments of depressive symptoms, hopelessness, rumination, borderline personality traits, and neuropsychology.
Self-reported demographic survey
This section collected general information from the participants, such as age, gender, grade, residence, siblings, number of parents, and left-behind experience.
Depression
The patient health questionnaire (PHQ-9) [
44] is one of the most widely-used self-reporting measures in clinical practice. It consists of nine items. The response options for each item range from not at all (0 points) to almost every day (3 points), reflecting how often each symptom has affected respondents in the past two weeks. Higher scores indicate more severe depressive symptoms. The Chinese version of PHQ-9 is considered to have a good internal consistency [
45] and the scale had a Cronbach’s internal validity value of α = 0.86.
Suicide attempts
A single item was used to assess SAs. Participants were asked to respond to the questions “Have you thought about suicide in the past 12 months?” and “Have you attempted suicide in the past 12 months?”. These one-item measures of SI and SAs have been used in previous studies [
46,
47].
Hopelessness
Hopelessness was measured by the Beck hopelessness scale (BHS). The BHS is a 20-item self-reporting instrument that is used to assess a respondent’s negative attitudes towards future events. The Chinese version of the BHS has satisfactory reliability and validity in adolescents [
48]. This BHS consists of three subscales: expectations, loss of motivation, and feelings about the future [
49]. The Cronbach’s alpha value obtained in this study was 0.79.
Rumination
The Chinese version of the Nolen-Hoeksema ruminative response scale was used to assess a respondent’s tendency to focus passively on the reasons for their suffering [
50]. Participants responded to the 10 items on a Likert-type scale ranging from 1 (never) to 4 (always). This scale has been used in a sample of Chinese adolescents with good reliability and validity [
51]. The Cronbach’s alpha value obtained in this study was 0.92.
Borderline personality traits
The borderline personality features scale for children (BPFS-C) [
52] is a reliable and effective assessment tool for children and adolescents. The self-reporting scale consists of 24 items and is a Likert scale ranging from 1 (never) to 5 (always). The Chinese version of BPFS-C has been widely used with good validity and reliability [
53]. The Cronbach’s alpha obtained in this study was 0.89.
Neurocognitive functioning
The participants underwent several computerized neuropsychological assessments using a standardized test in the following cognitive domains: processing speed [
54], attention [
55], working memory [
54], emotion recognition [
56], and EF [
57]. The test and measures used in each domain were adapted from the existing literature on cognitive assessment in mental disorders [
12,
58].
Processing speed
This test measures the subject’s hand-eye coordination, cognitive processing speed, and attention. It is a component of the Wechsler Intelligence Scale for Children, Fourth Edition (WISC-IV). The participants must determine the matching decoding symbols in accordance with the system’s prescribed order after three consecutively correct learning attempts. The test lasts 90 s and the better the function, the higher the score.
Attention
Attention was measured with the Stroop color word test (SCWT), using a single-item presentation and a button press response. On a numeric keypad, the participants used their fingers to press “yes” for consistency and “no” for inconsistency. Stimuli were presented individually and cleared after the participant’s response, with a 30-millisecond delay between stimuli. All responses received auditory feedback, both correct (beep) and incorrect (buzz). The percentage interference (percentage change in the median RT to color/word vs. color responses) was used to summarize performance [
59].
Working memory
The WAIS-IV digit span (DS) was used to assess working memory. Participants entered a string of numbers they had just heard in reverse order on a computer (DS Backward). The length of the trial was increased by one unit for each trial of a given length that was completed correctly until two trials with the same number of digits were answered incorrectly. The total number of correctly repeated trials was summed to calculate the total score [
60].
Emotion recognition
This test measures the ability to recognize emotional faces. There were 49 images, including happiness, sadness, fear, anger, disgust, and surprise. Following the start of the test, all the images appeared at random and the participants had to select the emotion category that best matched their response. The final analysis index was the total score of these seven basic emotions that were correctly recognized.
Executive function
Utilizing the Wisconsin sorting card test (WCST), EF was evaluated. The four stimulus cards were a red triangle, two green stars, three yellow crosses, and four blue circles. Participants were given two sets of 64 response cards, which they could sort by color, form, and number, but they were not given instructions on how to do so. Each subject was instructed to deduce the proper sorting rules and each trial received feedback. As the rules changed, they had to summarize the rules and flexibly transform the classification principle [
61].
Statistical analyses
The sociodemographic and clinical rating data of the participants were summarized using frequencies and percentages for categorical variables. The Chi-square test was used to assess the differences between three independent qualitative datasets. ANOVA was used for normally distributed quantitative data and the Kruskal-Wallis H Test was used for non-normal distributions. Post-hoc comparisons were applied. Subsequently, a partial correlation analysis was performed after controlling for age and gender. For the analysis of cognitive performance, measures selected from each test were gathered into the corresponding cognitive domain to be evaluated. All statistical analyses in the present study were performed using SPSS version 22.0. A significance level of 5% was set.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.