Journal of Affective Disorders 298 (2022) 301–307

Contents lists available at ScienceDirect

Journal of Affective Disorders

journal homepage: www.elsevier.com/locate/jad

The psychological characteristics and risk factors of suicidal attempt among

mood disorders adolescents accompany with non-suicidal self-injury: A
multi-center study
Yanni Wang a, #, Xin Zhou b, c, #, Bo Cao d, Lijuan Chen e, Ruoxi Wang f, Ling Qi g, Linlin Meng h,
Lingyun Zeng i, Xia Liu i, Wenjia Wang j, Chuanxiao Li j, Jiezhi Yang k, Xueyan Gu l, Zezhi Li m, **,
Yongjie Zhou c, *
a
Department of Maternal, Child and Adolescent Health, School of Public Health, Lanzhou University, Lanzhou, China
b
Research Center for Psychological and Health Sciences, China University of Geosciences, Wuhan, China
c
Department of Psychiatric Rehabilitation, Shenzhen Kangning Hospital, Shenzhen, China
d
Department of Psychiatry, Faculty of Medicine & Dentistry, University of Alberta, Edmonton, Canada
e
School of Literature, Journalism & Communication, South-Central University for Nationalities, Wuhan, China
f
School of Medicine and Health Management, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China
g
School of Health Science and Nursing, Wuhan Polytechnic University, Wuhan, China
h
Linyi Mental Health Center, Linxi, China
i
Shenzhen Kangning Hospital, Shenzhen, China
j
CAS Key Laboratory of Mental Health, Institute of Psychology, Chinese Academy of Sciences, Beijing, China
k
Shenzhen Health Development Research Center, Shenzhen, China
l
Department of Toxicology, School of Public Health, Lanzhou University, Lanzhou, China.
m
Department of Psychiatry, The Affiliated Brain Hospital of Guangzhou Medical University, Guangzhou, China

A R T I C L E I N F O A B S T R A C T

Keywords: Background: Identifying the risk factors of suicide attempts(SA) in adolescents with mood disorders(MD) who engage in
Suicide attempts non-suicidal self-injury(NSSI) is of great significance for suicide prevention. The aim of the present study was to explore
Non-suicidal self-injury the psychological characteristics and risk factors of SA among MD adolescents engaged in NSSI.
Adolescents
Methods: We recruited MD outpatients accompany with NSSI aged 12–18 years. SA, NSSI methods and function,
Mood disorders
suicidal ideation(SI), psychological distress(PD), self-esteem, stress mindset and perceived social support were
assessed by valid scales. Classification and regression tree analysis (CART) was employed to explore the char­
acteristics and risk factors of SA among MD adolescent with NSSI.
Results: We included 658 participants in this study. Of 58.1% participants reported SA during the past 12 months.
Compared with the adolescents without SA, the attempters used more different NSSI methods and reported more
frequent NSSI. SA, SI, PD, self-esteem and amount of thinking time before engaging in self-injury were risk factors of
SA among MD adolescents. Interactions between the four risk factors resulted in varying degrees of risk of SA.
Compared to adolescents with the characteristics of low level of SI — little consideration before self-injury, adolescents
who having multiple characteristics of the high level of SI —high level of PD — low self-esteem were associated with a
15.1-fold increased risk of SA(P < 0.001), and those with the characteristics of high SI — deliberated before engaging
in self-injury were associated with a 28.1-fold increased risk of attempted suicide(P < 0.001).
Conclusions: Our findings identify multiple correlates for SA in MD adolescents accompany with NSSI, including
SI, PD, self-esteem and deliberate time they thought before self-injury, which may contribute to the development
of suicidal behaviors in an interactive manner.

* Corresponding authors: Department of Psychiatric Rehabilitation, Shenzhen Kangning Hospital, No.77 Zhenbi Road, Shenzhen, 518118, China.
** Corresponding authors: Department of Psychiatry, The Affiliated Brain Hospital of Guangzhou Medical University, No. 36 Mingxin Road, Fangcun, Guangzhou,
510370, China.
E-mail addresses: [email protected] (Z. Li), [email protected] (Y. Zhou).
#
Contributed equally.

https://doi.org/10.1016/j.jad.2021.10.130
Received 4 July 2021; Received in revised form 11 September 2021; Accepted 23 October 2021
Available online 31 October 2021
0165-0327/© 2021 Elsevier B.V. All rights reserved.
Y. Wang et al. Journal of Affective Disorders 298 (2022) 301–307

1. Introduction five days of NSSI behaviors in the past 12 were included. Participants
were excluded if they have a history of mental retardation. Before the
Suicide attempts (SA), a severe suicide behavior, are defined as non- investigation, all participants and their guardians were informed of the
fatal self-directed injuries with implicit or explicit intent to kill oneself, purposes and procedures of the study in detail. Self-administrated
is a severe suicide behavior. While non-suicidal self-injury (NSSI) is questionnaires were used to collect participants’ socio-demographic
suggested to be a disorder characterized by five or more days of self- characteristics, methods and functions of NSSI behaviors and psycho­
injury without suicidal intent occurring during the past 12-months logical characteristics. A trained investigator was present in case par­
where the purposes are not socially sanctioned (American Psychiatric ticipants have any confusions about the question.
Association, 2013). In the past decades, an increase in the incident rate
of SA and NSSI in adolescents was observed, which had become public 2.3. Demographic profile
health concerns affecting adolescents worldwide(McManus et al., 2019).
SA and NSSI are closely related, but they have difference in terms of We collected the participant’s socio-demographic data, including
frequencies, method, and purpose (Coppersmith et al., 2017). Adoles­ age, sex (male or female), grade (junior high school or senior high
cents who engaged in SA often reported more frequent self-injury be­ school), and residential area(urban or rural), household income
haviors and more serious methods of self-injury than adolescents who (<80000CNY, 80,000~200000CNY, ≧200,000CNY/year) (Gan et al.,
engaged in NSSI (Victor and Klonsky, 2014). While existing studies 2015).
suggested that NSSI have the function of emotional regulation, such as in
stopping worse emotions or in increased feelings of relaxation (Harris 2.4. Measurement
et al., 2018).Therefore, in DSM-5, NSSI does not belong to suicidal be­
haviors although it is one of the strongest predictors of suicide. Ado­ The participants’ MD were confirmed by a pediatric psychiatrist
lescents with NSSI have a higher risk of developing SA (Brausch and through a structured clinical interview according to the DSM-IV. In
Gutierrez, 2010; Victor and Klonsky, 2014). It is quite important to order to assess SA, participants were asked the following question:
understand how NSSI develops into SA, as SA is the closest to suicide “During the past 12 months, did you actually attempt suicide?”
death. Response options for the question were “yes” and “no”.
Previous studies documented the risks of SA in adolescents with NSSI The methods and function of participants’ NSSI over the past 12
in community samples, and most all of them noted that mood disorders months were assessed using the Chinese version of Functional Assess­
(MD) are associated with markedly elevated levels of suicidality (Car­ ment of Self-Mutilation (FASM) (Lloyd, 1997; Yuhui et al., 2018). A list
ballo et al., 2020; MacPherson et al., 2018). A meta-analysis of 52 of 12 NSSI methods were specified assessed, including (hitting, head
studies showed the strongest risk factors of SA among NSSI adolescents banging, stabbing, pinching, scratching, biting, burning, cutting, tat­
were having suicidal ideation, and using more NSSI frequency and tooing, and pulling hair). The severity of NSSI was assessed by the
number of NSSI method (Victor and Klonsky, 2014). A population-based question “Did you received medical treatment due to the injury?” Ad­
birth cohort study reported that substance use, sleep, psychological olescents reported their level of agreement with 15 statements of reasons
traits were the possible predictors of future SA among adolescents with for self-injury on a scale ranging from 0 (never) to 3 (Always) (e.g., “I
NSSI (Mars et al., 2019). However, no study has explored risk factors of engaged in self-injury to stop bad feelings”). The reasons for self-injury
SA among NSSI adolescents in MD sample to the best of our knowledge. were documented in a three-factor structure of functions: emotion
It is well established that NSSI and SA in adolescence are closely regulation, attention seeking and social avoidance according to our
associated with MD, such as depression and bipolar disorder (Carballo previous study. Additionally, the age of onset of this behavior, level of
et al., 2020; MacPherson et al., 2018). Importantly, adolescents with MD subjective pain during self-injury, amount of thinking time before
are at an increased risk for suicide (Patel et al., 2020). Therefore, it is engaging in NSSI, and the use of alcohol or drugs during self-injury was
important to determine the essential factors of SA in MD adolescents collected by the FASM (Taylor et al., 2018).
who have NSSI, as this affords great insight into those at risk for suicide, Suicidal ideation was assessed by the self-report Beck Suicide Idea­
which may be critical for improving suicide prevention strategies. An tion Scale (BSI)-part I(A.T. Beck et al., 1979), including 5 items: 1)wish
increasing evidence showed that suicidal ideation, impulsivity, psy­ to live, 2) wish to die, 3) reasons for living/dying, 4) desire to make a
chological distress, low social support and low self-esteem were strongly suicide attempt and 5) passive suicidal thoughts. The items are rated
associated with SA (Carballo et al., 2020; Patel et al., 2020). Therefore, it between 1and 5 and gage the intensity of a patient’s thoughts about
is important to determine the essential factors of SA in MD adolescents committing suicide. The items are rated on a 3-point Likert scale, and the
who have NSSI. In this study, we recruited MD adolescents accompany scores range from 0 to 10. A higher score indicates the adolescent has
with NSSI to explore the psychological characteristics and risk factors of higher active suicidal desire.
SA among MD adolescents who engaged in NSSI. The 10-item Kessler Psychological Distress Scale (K10 scale) (Kessler
et al., 2002) was used to measure psychological distress. The K10 scale
2. Methods consists of 10 items assessing the frequency of non-specific psychologi­
cal distress in the past month. These items are rated on a 5-point Likert
2.1. Ethical considerations scale with scores ranging from 10 to 50. The K10 score≧31 may indicate
severe psychological distress in adolescents (Huang et al., 2009).
The protocol of this study was approved by the ethics committee of The Chinese version of the Rosenberg self-esteem scale(RSE)
the Institutional Review Board (IRB) of the Shenzhen Kangning Hospital (Rosenberg, 1965.) was used to measure overall self-esteem of partici­
(IRB:2020-K021–01). Informed consents were obtained from all partic­ pants. RSE consists of 10 items with each item rated on a 4-point scale
ipants and their guardians. from 1: strongly disagree to 4: strongly agree. The total score ranges
from 10 to 40. According to Yang (Wu et al., 2017), item 8 is treated as a
2.2. Participants and data collection positively worded item. After recording the reverse-scored items, the
total scores of all items were calculated. Higher scores indicate higher
This was a case seriess study. All the paticipants was selected from self-esteem.
psychiatric departments in 20 psychiatric hospitals or general hospitals The Multidimensional Scale of Perceived Social Support (MSPSS)
in 9 provinces across China. Participants were consecutively recruited (Zimet et al., 1990) was used to assess the level of social support. The
from August 2020 to November 2020. Outpatients who aged 12–18 MSPSS is a brief 12-item, self-administered scale with each item rated on
years met DSM-V criteria for any mood disorders and reported at least a 7-point Likert scale ranging from 1 (very strongly disagree) to 7 (very

302
Y. Wang et al. Journal of Affective Disorders 298 (2022) 301–307

strongly agree). A higher score indicates the greater social support fitting after the tree was generated. The risk of these subgroups was
perceived from families, friends and significant others. In addition, the computed using logistic regression analysis using treating smallest per­
MSPSS includes three subscales: Family (4 items), Friends (4 items) and centage of participants with SA as the reference. All ORs and 95% CIs
Significant Others (4 items). were adjusted for gender and age.
The Chinese version of the Stress Mindset Measure (SMM) (Jiang All analyses were performed using SPSS version 20.0 software (SPSS
et al., 2019) was used to measure views on the effects of stress. SMM is Inc., Chicago, Illinois, USA). All tests were two-tailed, and statistical
an 8-item instrument that assesses the extent to which participants significance was defined at P < 0.05.
believe the impact of stress enhances or debilitates their life (e.g., the
effects of stress are positive and should be utilized, the effects of stress 3. Results
are negative and should be avoided). After recording the reverse-scored
items, the average scores for all items were calculated; higher scores 3.1. Demographic profile of participants
reflect a stress-is-enhancing mindset, while those with lower scores
reflect a stress-is-debilitating mindset (Crum et al., 2013). A total of 701 adolescents were recruited in the study. Forty-three
adolescents were excluded as they did not complete the questionnaire.
2.5. Statistical analysis Finally, 658 participants (102 boys and 556 girls) were included in the
current analysis. Of them, 433 adolescents had depressive disorder, 102
Kolmogorov–Smirnov one-sample test was applied to examine the adolescents had bipolar disorder and 123 adolescents had unspecified
normal distribution. Analysis of variance (ANOVA) for continuous var­ mood disorders).
iables, chi-square test for categorical variables, and Mann-Whitney U test Among 658 adolescents included, 382(58.1%) adolescents reported
for non-normally distributed variables were utilized to compare differ­ engaging in SA during the past 12 months. Table 1 showed the de­
ences in the characteristics of the two groups(participants with SA and mographic characteristics of the participants by the status of SA. There
participants without SA). Classification and regression tree (CART) were no significant differences in sex, age, residential area, annual
analysis (Breiman et al., 1984) was performed to identify associated household income, type of diagnosed mood disorders, or antipsychotic
factors of SA among adolescents who reported NSSI. Psychological medication between the two groups (all P > 0.05).
characteristics, methods and function of self-injury were included in the
analysis. The CART tree is grown by splitting the root node into two 3.2. Psychological characteristics of participants with and without SA
offspring nodes and repeating the process over the nodes via recursive
partitioning according to splitting criteria of the Gini index. The root Table 1 shows the psychological characteristics of participants with
node contained the entire sample. The minimum number for splitting a and without SA. The adolescents with SA had higher BSI and K10 scores,
node was 20, and the optimal tree was selected based on the lowest but had lower RSE and MSPSS scores, including family support and
misclassification error rate. Ten-fold cross-validation was used to eval­ friend support subscores, when compared to adolescents without SA (all
uate the model fit. Post-process pruning was performed to avoid over P < 0.05). There were no differences in SMM score or significant others

Table 1
Demographic and psychological characteristics of participants with and without SA.
Characteristics Total(N = 658) Participants without SA (N = 276) Participants with SA (N = 382) F/χ2 P-value

Diagnosis, N(%) 6.64 0.094

Depression 433(65.8) 180(65.2) 253(66.2)
Bipolar Disorder 102(15.5) 52(18.8) 50(13.1)
Unspecified mood disorder 123(18.7) 44(15.9) 79(20.7)
Age(yrs), Mean(SD) 15.05 ± 1.67 15.14 ± 1.73 14.99 ± 1.63 1.20 0.273
Sex, N(%) 3.50 0.082
boy 102(15.6) 51(18.7) 51(81.3)
girl 556(84.4) 225(13.4) 331(86.6)
Grade, N(%) 1.63 0.201
Junior high school 353(53.6) 140(39.7) 213(60.3)
Senior high school 305(46.4) 136(50.7) 169(55.8)
Residential area, N(%) 4.29 0.034
urban 453(68.6) 203(73.0) 250(65.4)
rural 205(31.4) 73(27.0) 132(34.6)
Annual household income(CNY), N(%) 2.96 0.227
<80,000 261(39.7) 100(36.3) 161(42.1)
80,000~ 283(43.0) 129(46.8) 154(40.3)
≧200,000 114(17.3) 47(16.9) 67(17.5)
Only-child, N(%) 199(30.2) 87(31.7) 112(29.3) 0.41 0.519
Antipsychotic medication, N(%) 482(74.7) 207(75.8) 275(73.9) 0.30 0.647
BSI score, Mean(SD) 6.08 ± 2.88 4.72 ± 2.76 7.07 ± 2.55 133.57 <0.001
K10 score, Mean(SD) 35.90 ± 7.89 33.14 ± 7.59 37.91 ± 7.48 64.48 <0.001
SMM score, Mean(SD) 2.64 ± 0.63 2.69 ± 0.62 2.61 ± 0.63 2.59 0.108
RSE score, Mean(SD) 20.18 ± 5.79 22.00 ± 6.79 18.86 ± 5.79 40.86 <0.001
MSPSS score, Mean(SD) 43.26 ± 12.95 45.50 ± 12.15 41.62 ± 13.29 14.91 <0.001
Family support subscore 14.86 ± 5.77 15.71 ± 5.40 14.25 ± 5.95 10.46 0.001
Friend support subscore 15.98 ± 6.31 16.81±5.88 15.38 ± 6.55 8.42 0.004
Significant others support subscore 12.41 ± 5.84 12.97 ± 5.71 11.99 ± 5.91 4.57 0.033

SA:suicide attempts.
BSI: Beck Suicide Ideation Scale.
K10: The 10-item Kessler Psychological Distress Scale.
SMM: Stress mindset measure.
RSE: Rosenberg self-esteem scale.
MSPSS: Multidimensional Scale of Perceived Social Support.

303
Y. Wang et al. Journal of Affective Disorders 298 (2022) 301–307

support subscore between the two groups (both P > 0.05). engaging in self-injury) were associated with a 28.1-fold increased risk
of attempted suicide(P < 0.001).
Table S2 shows the importance of independent variables in the CART
3.3. Self-injury methods and function reported by participants with and
analysis. The importance of BSI score, K10 score, amount of thinking
without SA
time before engaging in self-injury, RSE score and family support sub­
score were 100%, 50.8%, 28.6%, 28.4% and 20.8%, respectively.
Table S1 shows different types of self-injury methods of the two
groups (participants with SA vs. participants with SA). In both groups,
4. Discussion
cutting or carving on skin was the most common method of self-injury.
The adolescents with SA had a higher prevalence of the different types of
To the best of our knowledge, our study is the largest multi-center
methods of self-injury than participants without SA (all P < 0.05), except
study to identify behavioral and psychological risk correlates for SA in
the methods of cutting on skin, burning skin and total number of self-
MD adolescents accompany with NSSI. The main findings of this study
injury methods. The adolescents engaged in more different NSSI
were as follows: 1) About 58.1% of MD adolescents reported SA when
methods and higher NSSI frequency in the past 12 months (Table 2). As
they engaged in self-injury in the past 12 months. 2) Compared with
shown in Table 2, there were no significant differences between the two
NSSI adolescents without SA, the adolescents with SA reported using a
groups in physical pain during self-injury or age of onset. Compared to
greater variety of methods of self-injury and spending more time before
participants without SA, more participants with SA deliberated before
starting the self-injury. 3) Suicidal ideation, psychological distress, self-
engaging in self-injury (they thought for hours or days before they
esteem of participants, and deliberation before engaging in self-injury
started self-injury) (P = 0.001). A higher score of NSSI function was
were associated with SA. Compared to adolescents without high level
observed in the SA group than in the group of participants without SA.
of suicidal ideation - little or no consideration before self-injury, ado­
Most participants without SA reported engaging in NSSI impulsively
lescents who experienced high levels of suicidal ideation, high levels of
(They started thinking about self-harm less than an hour before
psychological distress and low self-esteem were associated with a 15.1-
engaging in self-injury).
fold increased risk of attempted suicide, and those with the character­
istics of high suicide ideation and deliberate before engaging in self-
3.4. Associated factors of the SA in participants by CART analysis injury were associated with a 28.1-fold increased risk of attempted
suicide.
CART analysis was performed to evaluate the associated factors of SA Accumulating studies have emphasized that adolescence is a
in the participants. As shown in Fig. 1, the tree was initially split from vulnerable phase for developing NSSI and committing suicide, as the
BSI scores (Node 0), which indicated that suicidal ideation exhibited prevalence of NSSI and SA rises in early adolescence and peaks in the
strong significant associations with SA. BSI score, K10 score, RSE score middle of adolescence (Grandclerc et al., 2016; Stewart et al., 2017).
and amount of thinking time before engaging in self-injury. Our present study indicated that NSSI and SA co-occur frequently in MD
As shown in the Table 3, individuals in the subgroup of Node 10 and adolescents, which is consistent with findings in the nonclinical samples
Node 11 showed the highest risks for SA than other terminal nodes. (Victor and Klonsky, 2014). According to a recent report, 12-month
Treating the individuals in the subgroup of Node3 (with the character­ prevalence of NSSI without SA was 23.7% and of NSSI with SA was
istics of BSI score≦5.0 - little or no consideration before self-injury) as 2.3% among the general Chinese adolescents (Liu et al., 2018). In our
the reference, adolescents in the subgroup of Node11, who having study, SA was reported approximately 60% of MD adolescents who
multiple characteristics of the high level of suicide ideation(BSI score > engaged in self-injury, which is significantly higher than that of in the
5.0), high level of psychological distress(K10 score > 39.5) and low self- nonclinical population.
esteem(RSE score≦19.5), were associated with a 15.1-fold increased risk A previous meta-analysis has mentioned that the frequency and
of attempted suicide(P < 0.001), and those in the subgroup of Node10 number of methods of self-injury have been considered as moderate risk
(with the characteristics of high suicide ideation and deliberated before

Table 2
Methods and function of self-injury by of participants with and without SA.
Variables Total(N = 658) Participants without SA (N = 276) Participants with SA (N = 382) F/χ2 P-value

Frequency of self-injury, Median (IQR) 13(8,20) 9(6,15) 13(8,21) − 6.11 0.001

Frequency of severe self-injury, N(%) 21.45 0.091
1 132(20.1) 52(19.2) 80(21.1)
2 60(9.1) 30(11.1) 30(7.9)
≧3 136(20.7) 48(17.4) 88(23.0)
Number of NSSI methods, Median (IQR) 5(3,7) 4(3,6) 6(4,8) − 6.46 0.001
Function of self-injury, Median (IQR)
Social avoidance 2(0,4) 2(0,4) 2(0,5) − 3.361 0.001
Emotional regulation 9(5,11) 8(4.5,10) 9(6,12) − 3.476 0.001
Attention seeking 2(0,6) 2(0,5) 3(0,7) − 3.158 0.002
Age of onset of the behavior, Mean(SD) 13.10±2.08 13.21±2.10 13.03±2.06 1.09 0.296
Use of alcohol/drugs during self-injury, N(%) 67(10.2) 15(5.4) 52(13.6) 17.71 0.001
#
Pain, N(%) 0.815 0.846
No pain 203(30.9) 81(29.3) 122(31.9)
Litter pain 337(51.2) 147(53.3) 190(49.7)
Moderate pain 96(14.6) 39(14.1) 57(14.9)
Severe pain 22(3.3) 9(3.3) 13(3.4)
*Time, N(%) 31.34 0.001
Zero or several minutes 333(50.7) 163(59.1) 170(44.5)
less than an hour 163(24.8) 72(26.1) 91(23.8)
Several hours or days 162(24.6) 41(14.8) 121
(31.6)
#
Physical pain during self-injury.
*
Amount of thinking time before engaging in self-injury.

304
Y. Wang et al. Journal of Affective Disorders 298 (2022) 301–307

Fig. 1. CART analysis of psychological characteristics, NSSI characteristics (methods and functions) for SA risks. Frequency and percentage of participants with and
without SA of subgroup is showed in each node. Predicted percent correct for SA was 88.2%. BSI: Beck Suicide Ideation Scale
K10: The 10-item Kessler Psychological Distress Scale
RSE: Rosenberg self-esteem scale.

indicated that suicide attempters with a mood disorder used more NSSI
Table 3
methods than non-attempters. Some researchers observed that,
Risk estimates of the CART terminal nodes.
compared with participants without SA, suicide attempters used more
Node Factors Participants Participants OR(95%CI) P NSSI methods but did not engage in self-injury more frequently (Stewart
without SA, with SA, N value
et al., 2017). On the other hand, other researchers believed the NSSI
N(%) (%)
frequency was the strongest predictor of suicide(Duarte et al., 2020).
3 BSI score≦5.0 154(55.8) 65(17.0) Ref. According to the theoretical model of suicide, more NSSI behaviors will
- little or no
consideration
increase one’s experience with different types of pain, which may reduce
before self- barriers to suicide, such as pain tolerance (Victor and Klonsky, 2014),
injury and the experience would enable adolescents to acquire the capability
9 BSI score 76(27.5) 103(27.0) 3.05 < for suicide (Hamza et al., 2012) more readily.
>5.0 – K10 (2.03–4.60) 0.001
The functional model of FASM provides insight into possible reasons
score≦39.5 -
little or no for engaging in NSSI. Our results showed that emotion regulation (using
consideration self-injury to alleviate acute negative emotion or aversion arousal) was
before self- the most likely cause of self-injury in adolescents with recent SA, which
injury suggested that MD adolescents with SA generally perform poorly in
10 BSI score>5.0 3(1.1) 35(9.2) 28.07
emotional management or regulation (Harris et al., 2018) and therefore
<
– K10 score (8.34–94.52) 0.001
≦39.5 - lack strategies to cope with emotional distress (Klonsky, 2007).
deliberate Suicidal behavior in patients with MD is an inherent phenotype.
before However, the majority of patients with MD never commit suicide or
engaging in
attempt suicide (Rihmer, 2012), thus, MD do not explain all suicidal
self-injury
11 BSI score>5.0 20(7.2) 137(35.9) 15.11 < behaviors (Carballo et al., 2020). What triggers the transition to suicide
- K10 score > (8.725–26.16) 0.001 among adolescents with MD is an ongoing question. According to the
39.5 - RSE etiological model of suicidal behavior, both NSSI and suicide seem to
score≦19.5 exist on a continuum of suicide behaviors, in which NSSI is one extreme
12 BSI score>5.0 10(3.6) 21(5.5) 4.78
and suicide is the other extreme (Duarte et al., 2020). A
<
- K10 score > (2.13–10.69) 0.001
39.5 - RSE population-based birth cohort study demonstrated that 21% of adoles­
score>19.5 cents who reported both suicidal thoughts and NSSI would escalate to SA
The ORs and 95%CIs of the terminal nodes were calculated by logistic regression
during adulthood (Mars et al., 2019). In this study, we identified that
analysis adjusted for sex and age. suicidal ideation exhibited the strongest association with SA among MD
BSI: Beck Suicide Ideation Scale. adolescents who engage in NSSI. In the SA group, 77.5% participants
K10: The 10-item Kessler Psychological Distress Scale. reported a high desire for active suicide when they engaged in NSSI,
RSE: Rosenberg self-esteem scale. which indicates that MD adolescents who reported both strong suicidal
ideation and NSSI might be an especially high-risk group for suicide
factors of SA (Victor and Klonsky, 2014). Burke et al. (2016) suggested (Perez et al., 2019). In addition to suicidal ideation, severe psychological
that the presence and number of NSSI scars were objective physical in­ distress and low self-esteem were also strongly associated with SA. The
dicators of the risk of suicidal ideation and SA. In adults with eating predictive effects of psychological distress and low self-esteem on sui­
disorders, NSSI methods, frequency of NSSI and cutting were risk factors cide have been well documented in previous studies (Martin et al.,
for differentiating suicide ideators from suicide attempters (Perez et al., 2005). Some studies have shown that impulsivity is an important risk
2019). Our results showed that adolescents with SA reported more factor for SA (Carballo et al., 2020; Dougherty et al., 2009). However, in
frequent and more types of self-injury methods than those without SA. It this study the association between impulsivity and SA was not identi­
fied. On the contrary, we found that MD adolescents who have an

305
Y. Wang et al. Journal of Affective Disorders 298 (2022) 301–307

attempted suicide history showed deliberation before starting to hurt Key Medical Discipline Construction Fund (No. SZXK072), and the
themselves, compared to those without SA history. About 1 in 3 of MDs Fundamental Research Funds for the Central Universities (No. lzujbky-
adolescents who attempted suicide thought about the self-injury act for a 2019–5). The funders had no role in the study design, data collection,
few hours or days before they started, while only 1 in 7 of adolescents analysis and interpretation, writing of the report, and the decision to
without SA did it. submit the article for publication.
The transition from NSSI to SA in MD adolescents is complex, and
there are likely interactions among the factors that lead to SA. Therefore,
Availability of data and material
we performed CART analysis to identify the risk factors associated with
SA in NSSI adolescents. We identified four factors in the classification
The data used are available and will be provided by the corre­
trees to be “optimal” risk factors of SA. Individuals with the combination
sponding author if necessary.
of high suicidal ideation, high level of psychological distress and low
self-esteem were associated with a 15.1-fold increased risk of SA. The
combination of high suicidal ideation and deliberation before engaging Supplementary materials
in self-injury was associated with a 28.1-fold increased risk of SA.
Therefore, assessments of suicidal ideation, psychological distress, self- Supplementary material associated with this article can be found, in
esteem and whether deliberating before engaging in self-injury could the online version, at doi:10.1016/j.jad.2021.10.130.
provide information for risk assessment and help clinicians identify the
risk of SA in adolescents with NSSI. References
Interestingly, the results of CART showed that low perceived family
American Psychiatric Association, 2013. Diagnostic and Statistical Manual of Mental
support was a crucial factor in MD adolescents without SA, which may Disorders, 5th edn. American Psychiatric Publishing, Arlington, VA.
explain the effectiveness of the family-based behavioral therapy for Beck, A.T., Kovacs, M., Weissman, A., 1979. Assessment of suicidal intention: the Scale
NSSI, such as dialectical behavioral therapy for adolescents (Mehlum for Suicide Ideation. J. Consult. Clin. Psychol. 47.
Brausch, A.M., Gutierrez, P.M., 2010. Differences in non-suicidal self-injury and suicide
et al., 2016), which includes weekly multi-family skills training and
attempts in adolescents. J. Youth Adolesc. 39, 233–242.
necessary family therapy sessions. Therefore, for NSSI with or without Breiman, L., Friedman, J., Stone, C.J., 1984. Classification and Regression Tree. CRC
SA, it could be necessary for clinicians to tailor interventions differently Press.
Burke, T.A., Hamilton, J.L., Cohen, J.N., Stange, J.P., Alloy, L.B., 2016. Identifying a
for various individuals.
physical indicator of suicide risk: non-suicidal self-injury scars predict suicidal
There were several limitations should be considered when inter­ ideation and suicide attempts. Compr. Psychiatry 65, 79–87.
preting these findings. First, selection bias may be existed as 84% of Carballo, J.J., Llorente, C., Kehrmann, L., Flamarique, I., Zuddas, A., Purper-Ouakil, D.,
participants were female. Previous evidence has suggested that female Hoekstra, P.J., Coghill, D., Schulze, U.M.E., Dittmann, R.W., Buitelaar, J.K., Castro-
Fornieles, J., Lievesley, K., Santosh, P., Arango, C., Consortium, S., 2020.
MD adolescents were more likely to have SA than male MD adolescents Psychosocial risk factors for suicidality in children and adolescents. Eur. Child
(Patel et al., 2020). Second, all the scales used in the current study were Adolesc. Psychiatry 29, 759–776.
self-administrated questionnaires. Some items may not be understood or Coppersmith, D.D.L., Nada-Raja, S., Beautrais, A.L., 2017. Non-suicidal self-injury and
suicide attempts in a New Zealand birth cohort. J. Affect. Disord. 221, 89–96.
misunderstood by the participants, although we have trained all the Crum, A.J., Salovey, P., S, A., 2013. Rethinking stress: the role of mindsets in
study investigators with a standardized procedure. Third, this was a determining the stress response. J. Pers. Soc. Psychol. 104, 716–733.
cross-sectional study, which may limit the interpretation of a causal Dougherty, D.M., Mathias, C.W., Marsh-Richard, D.M., Prevette, K.N., Dawes, M.A.,
Hatzis, E.S., Palmes, G., Nouvion, S.O., 2009. Impulsivity and clinical symptoms
relationship. among adolescents with non-suicidal self-injury with or without attempted suicide.
To identify potential risk factors that predict the development of Psychiatry Res. 169, 22–27.
NSSI into SA is important to improve suicide prediction and prevention. Duarte, T.A., Paulino, S., Almeida, C., Gomes, H.S., Santos, N., Gouveia-Pereira, M.,
2020. Self-harm as a predisposition for suicide attempts: a study of adolescents’
Our study identifies the associated risk correlates for attempted suicide
deliberate self-harm, suicidal ideation, and suicide attempts. Psychiatry Res. 287,
in NSSI adolescents with MD including suicidal ideation, psychological 112553.
distress, self-esteem and some characteristics of NSSI. Furthermore, Gan, L., Yin, Z., Tan, J., 2015. Investigation report of Chinese Family Finance 2015.
Chengdu.
these risk factors may contribute to the development of suicidal be­
Grandclerc, S., De Labrouhe, D., Spodenkiewicz, M., Lachal, J., Moro, M.R., 2016.
haviors in a cumulative and interactive manner in adolescents with Relations between nonsuicidal self-injury and suicidal behavior in adolescence: a
NSSI. The findings might have clinical implications on SA assessment systematic review. PLoS ONE 11, e0153760.
and treatment. Hamza, C.A., Fau, Stewart Sl, Willoughby, T., Willoughby, T., 2012. Examining the link
between nonsuicidal self-injury and suicidal behavior: a review of the literature and
an integrated model. Clin. Psychol. Rev. 32, 482–495.
CRediT authorship contribution statement Harris, L., Chelminski, I., Dalrymple, K., Morgan, T., Zimmerman, M., 2018. Suicide
attempts and emotion regulation in psychiatric outpatients. J. Affect. Disord. 232,
300–304.
Yanni Wang: Writing – original draft, Formal analysis. Xin Zhou: Huang, J.P., Xia, W., Sun, C.H., Zhang, H.Y., Wu, L.J., 2009. Psychological distress and
Investigation. Bo Cao: Writing – review & editing. Lijuan Chen: its correlates in Chinese adolescents. Aust. N. Z. J. Psychiatry 43, 674–680 quiz 681.
Investigation, Data curation. Ruoxi Wang: Methodology. Ling Qi: Jiang, Y., Zhang, J., Ming, H., Huang, S., Lin, D., 2019. Stressful life events and well-
being among rural-to-urban migrant adolescents: the moderating role of the stress
Formal analysis. Linlin Meng: Investigation. Lingyun Zeng: Investi­ mindset and differences between genders. J. Adolesc. 74, 24–32.
gation. Xia Liu: Data curation. Wenjia Wang: Data curation. Chuan­ Kessler, R.C., Fau, Andrews G, Colpe, L.J., Fau, Colpe Lj, Hiripi, E., Fau, Hiripi E,
xiao Li: Data curation. Jiezhi Yang: Data curation. Xueyan Gu: Formal Mroczek, D.K., Fau, Mroczek Dk, Normand, S.L.T., Fau, Normand Sl, Walters, E.E.,
Fau, Walters Ee, Zaslavsky, A.M., Zaslavsky, A.M., 2002. Short screening scales to
analysis. Zezhi Li: Writing – review & editing. Yongjie Zhou: Super­
monitor population prevalences and trends in non-specific psychological distress.
vision, Conceptualization, Project administration. Psychol. Med. 32, 959–976.
Klonsky, E.D., 2007. The functions of deliberate self-injury: a review of the evidence.
Clin. Psychol. Rev. 27, 226–239.
Declaration of Competing Interest
Liu, Z.Z., Chen, H., Bo, Q.G., Chen, R.H., Li, F.W., Lv, L., Jia, C.X., Liu, X., 2018.
Psychological and behavioral characteristics of suicide attempts and non-suicidal
The authors declare that there is no conflict of interest. self-injury in Chinese adolescents. J. Affect. Disord. 226, 287–293.
Lloyd, E., 1997. Self-mutilation in a community sample of adolescents.
MacPherson, H.A., Weinstein, S.M., West, A.E., 2018. Non-Suicidal Self-Injury in
Acknowledgments Pediatric Bipolar Disorder: clinical Correlates and Impact on Psychosocial Treatment
Outcomes. J. Abnorm. Child Psychol. 46, 857–870.
This work was supported by the Sanming Project of Medicine in Mars, B., Heron, J., Klonsky, E.D., Moran, P., O’Connor, R.C., Tilling, K., Wilkinson, P.,
Gunnell, D., 2019. Predictors of future suicide attempt among adolescents with
Shenzhen (No. SZSM202011014), Shenzhen Fund for Guangdong Pro­ suicidal thoughts or non-suicidal self-harm: a population-based birth cohort study.
vincial High-level Clinical Key Specialties (No. SZGSP013), Shenzhen Lancet Psychiatry 6, 327–337.

306
Y. Wang et al. Journal of Affective Disorders 298 (2022) 301–307

Martin, G., Richardson, A.S., Bergen, H.A., Roeger, L., Allison, S., 2005. Perceived Rosenberg, M., 1965. Society and the Adolescent Self-Image. Princeton University Press,
academic performance, self-esteem and locus of control as indicators of need for Princeton, NJ, USA.
assessment of adolescent suicide risk: implications for teachers. J. Adolesc. 28, Stewart, J.G., Esposito, E.C., Glenn, C.R., Gilman, S.E., Pridgen, B., Gold, J., Auerbach, R.
75–87. P., 2017. Adolescent self-injurers: comparing non-ideators, suicide ideators, and
McManus, S., Gunnell, D., Cooper, C., Bebbington, P.E., Howard, L.M., Brugha, T., suicide attempters. J. Psychiatr. Res. 84, 105–112.
Jenkins, R., Hassiotis, A., Weich, S., Appleby, L., 2019. Prevalence of non-suicidal Taylor, P.J., Jomar, K., Dhingra, K., Forrester, R., Shahmalak, U., Dickson, J.M., 2018.
self-harm and service contact in England, 2000-14: repeated cross-sectional surveys A meta-analysis of the prevalence of different functions of non-suicidal self-injury.
of the general population. Lancet Psychiatry 6, 573–581. J. Affect. Disord. 227, 759–769.
Mehlum, L., Ramberg, M., Tørmoen, A.J., Haga, E., Diep, L.M., Stanley, B.H., Miller, A.L., Victor, S.E., Klonsky, E.D., 2014. Correlates of suicide attempts among self-injurers: a
Sund, A.M., Grøholt, B., 2016. Dialectical behavior therapy compared with enhanced meta-analysis. Clin. Psychol. Rev. 34, 282–297.
usual care for adolescents with repeated suicidal and self-harming behavior: Wu, Y., Zuo, B., Wen, F., Yan, L., 2017. Rosenberg self-esteem scale: method effects,
outcomes over a one-year follow-up. J. Am. Acad. Child Adol. Psychiatry 55, factorial structure and scale invariance across migrant child and urban child
295–300. populations in China. J. Pers. Assess. 99, 83–93.
Patel, R.S., Onyeaka, H., Youssef, N.A., 2020. Suicidal ideation and attempts in unipolar Yuhui, W., Wan, L., Jiahu, H., Fangbiao, T., 2018. Development and evaluation on
versus bipolar depression: analysis of 131,740 adolescent inpatients nationwide. reliability and validity of adolescent non-suicidal self-injury assessment
Psychiatry Res. 291, 113231. questionnaire. Chin. J. School Health 39, 170–173.
Perez, S., Ros, M.C., Folgado, J.E.L., Marco, J.H., 2019. Non-suicidal self-injury Zimet, G.D., Powell Ss Fau - Farley, G.K., Farley Gk Fau - Werkman, S., Werkman S Fau -
differentiates suicide ideators and attempters and predicts future suicide attempts in Berkoff, K.A., Berkoff, K.A., 1990. Psychometric characteristics of the
patients with eating disorders. Suicide Life Threat Behav. 49, 1220–1231. Multidimensional Scale of Perceived Social Support. J. Pers. Assess. 55, 610–617.
Rihmer, Z., 2012. Depression and suicide: an update. Minerva Psichiatr. 53, 277–288.

307