Introduction

The outbreak of coronavirus disease 2019 (COVID-19) has a considerable effect on human life including psychological impacts (Asmundson & Taylor, 2020) and mental health problems such as post-traumatic stress, anxiety, depressive symptoms, insomnia, denial, anger, fear as well as grief-related symptoms (Torales et al., 2020; Guessoum et al., 2020; Rajkumar, 2020; Duan et al., 2020). The pandemic has also worsened existing mental health problems (Golberstein et al., 2020), and brought many challenges to the health care system (Bouali et al., 2020; Ilesanmi & Afolabi, 2020). After the world economic crisis in 2008, the impact of recession had a great effect on self-harming behaviors (Hawton et al., 2016). Consequently, it can be assumed that there may also be an upsurge of self-injurious behavior in youth as a result of the COVID-19 pandemic (Fegert et al., 2020). Many researches target the prevention of the disease from the transmission end (Hanscom et al., 2020; Tipaldi et al., 2020). However, to our knowledge, there is no research investigating the effect on nonsuicidal self-injury (NSSI) during COVID-19 outbreak. NSSI has become a major public health concern among adolescents. Defined as the deliberate destruction of body tissue without intent to die, NSSI is often viewed with social disproval and distinguished from behavior whose harmful consequences are unintended (Nock & Favazza, 2009). NSSI has become a growing health problem worldwide. The lifetime prevalence of NSSI was as high as 17.2% among adolescents and 13.4% among young adults (Swannell et al., 2014). Similar prevalence were found across countries ((Giletta et al., 2012; Plener et al., 2009). Moreover, NSSI is a common problem among adolescents and a prominent risk factor for later suicidality (Asarnow et al., 2011; Guan et al., 2012). The DSM-5 emphasized the behavior by including NSSI disorder as a “condition requiring further study” (American Psychiatric Association, 2013). Therefore, it is important to examine the prevalence and psychosocial risk factors associated with the development and maintenance of NSSI, in order to take necessary preventive actions during the COVID-19 pandemic.

Numerous empirical studies have investigated the associations between NSSI and psychosocial risk factors in adolescents (Baetens et al., 2014; Fox et al., 2015; Hankin & Abela, 2011; Martin et al., 2016; Tuisku et al., 2014). A number of models pertaining to NSSI have been proposed, including Nock’s (2009) integrated theoretical model. According to the integrated theoretical model, NSSI is maintained by intrapersonal or interpersonal vulnerability factors through reinforcement processes to regulate affective experiences and social situations (Nock, 2009). This study incorporated the integrated theoretical model as a conceptual framework for understanding psychosocial risk factors for NSSI in adolescents during the COVID-19 outbreak. Intrapersonal and interpersonal vulnerability factors were included to examine NSSI risk factors. In this study, intrapersonal factors included neuroticism, self-esteem, alexithymia, impulsivity, depression, and subjective well-being. Interpersonal factors included family function, actual social support, virtual social support, and dissatisfaction with academic performance. Previous studies have also pointed out that higher levels of neuroticism (Allroggen et al., 2014; Brown, 2009; Kiekens et al., 2015; Mullins-Sweatt et al., 2013), greater impulsivity (Claes & Muehlenkamp, 2013; Hamza et al., 2015), higher levels of depression (Hankin & Abela, 2011; Klonsky et al., 2003; Muehlenkamp & Gutierrez, 2007; Ross & Heath, 2002), greater alexithymia (Cerutti et al., 2018; Garisch & Wilson, 2015; Gatta et al., 2016; Howe-Martin et al., 2012; Lambert and De Man, 2007; Laukkanen et al., 2013; Lüdtke et al., 2016), lower levels of self-esteem (Cawood & Huprich, 2011; Garisch & Wilson, 2015), and lower levels of subjective well-being (Vansteenkiste & Ryan, 2013) were associated with NSSI. In addition, past research also revealed intrapersonal factors, including lower levels of actual social support (Andrews et al., 2014), lower levels of family function (Baetens et al., 2014), and dissatisfaction with academic performance (Mortier et al., 2015) were all correlated with NSSI. Furthermore, no research has investigated the association of virtual social support and NSSI. Concerning the quarantine periods during the COVID-19 outbreak, social-distancing may have an important effect on the mental health of adolescents. Adolescents may be more likely to connect and gain support through the internet. Therefore, we want to examine the association between virtual social support and NSSI during the COVID-19 outbreak.

Past studies found that various psychosocial risk factors are related to NSSI. Nevertheless, few studies have encompassed an inclusive examination that integrated psychological and sociological risk factors aimed at a large sample of junior high school students. Moreover, most research were relatively finite for the sake of investigating a small number of psychosocial risk factors, which limits the comprehensiveness in understanding the association between NSSI and related variables. Owing to limited resources and time during the COVID-19 pandemic, educational agencies, mental health organizations, and educational practices may only focus on a maximum of three to four psychosocial risk factors per time in preventive education. This limits the understanding of NSSI. It is essential to incorporate relevant and multiple critical psychosocial risk factors relating to NSSI. Most importantly, as far as we know, this was the first study to examine the prevalence of NSSI and its psychosocial risk factors in a large sample of adolescents during the COVID-19 outbreak. This study aimed to probe at the prevalence of NSSI and its associations with psychological and sociological risk factors in a large sample of adolescents in Taiwan, and the results will be discussed. Through this investigation, we can provide information to educational agencies and mental health organizations geared towards NSSI prevention policies during the COVID-19 pandemic.

Methods

Participants and Procedure

Cross-sectional design was used to construct this study. The data in junior high school students were recruited during March 2 through March 27, 2020. COVID-19 new cases rapidly increased in February, 2020, forcing schools in Taiwan to extend an additional two-week winter break. Junior high schools opened on February 25th. This study recruited 1244 junior high school students in three schools located in northern Taiwan by stratified and cluster sampling methods. 1060 valid surveys were completed (Mage = 14.66, SD = 0.86 years) in total. Invalid questionnaires and those who did not took part in the administration of the surveys were excluded from the data analysis. The response rate is 85.21%.

The present study protocol was reviewed and approved by the institutional review board at the National Taiwan Normal University. Before assessment implementation, completed informed consent forms were obtained from the teachers of the administered classes, the guidance counselor, and the school principals/guidance director. The research team also provided a pre-training session regarding the procedure of the administration to the guidance counselors. Survey administration was conducted in groups during class time. In attempt to fain truthful responses, students were fully informed of the goal, procedure, research ethics and confidentiality. Written informed consent was also obtained from the participants and their guardians. Data analysis was conducted on the participants who provided two sets of informed consent. At the completion of the surveys, participants were informed that individual results and response feedbacks will be given at the end of the semester. The research methodology is presented in Fig. 1.

Fig. 1
figure 1

Flow diagram of current study

Full size image

Measures

The Socio-Demographic Measures

Gender (female = 1 and male = 2), age, and dissatisfaction with academic performance were assessed.

NSSI (You et al., 2012)

Twelve NSSI behaviors were examined in this study. Participants were asked, “In the past year, have you ever engaged in the following behaviors to deliberately injure yourself but without suicidal intent?” All of the twelve NSSI behavior items were rated on a 6-point scale, ranging from 0 “never” to 5 “five times or more”. A dichotomous variable of NSSI status was computed based on the twelve items. A code of “0” was entered when participants endorsed “never” on all items, and “1” when participants reported having engaged in one or more NSSI acts. The internal consistency coefficient in the current study was .86.

Shortened Chinese Version of Five-Factor Inventory_ Neuroticism Subscale (Chen & Piedmont, 1999; Chien et al., 2007)

We assessed neuroticism personality traits by the neuroticism subscale of Shortened Chinese version of Five-Factor Inventory (Chen & Piedmont, 1999; Chien et al., 2007). This inventory has 31 items, selected from the NEO Five-Factor Inventory (Chen & Piedmont, 1999), and showed good reliability and validity among Chinese adolescents (Chien et al., 2007). The neuroticism subscale contains 6 items and the Cronbach’s alpha coefficient was .81 in this study.

The Short-Form of the Impulsivity Scale (Fu et al., 2007; Li et al., 2002)

The impulsivity scale was an adaptation from the Barratt Impulsivity Scale, which included 15-items. The confirmatory factor analysis displayed two subscales: motor impulsiveness and non-planning (Li et al., 2002). Fu et al. (2007) assessed the construct validity of this scale using confirmatory factor analysis and showed a satisfactory construct validity among adolescents. The scale was rated on a 4-point Likert scale, and the Cronbach’s alpha coefficient was .79 in this study.

Depression Anxiety Stress Scale _ Depression Subscale

The current study incorporated the depression subscale of the Chinese version of the Depression Anxiety Stress Scale (DASS; Taouk et al., 2001). The subscale has seven items, and responses were made on a 4-point Likert scale. The DASS displayed satisfactory factor structure and internal consistency (Taouk et al., 2001). In the present study, the depression subscale showed good internal consistency (α = .85).

The Toronto Alexithymia Scale-20 (TAS-20; Bagby et al., 1994; Parker et al., 2003)

We used the TAS-20 to assess the alexithymia (Bagby et al., 1994). TAS-20 is a self-report scale comprising 20 items on a 5-point scale. Good internal consistency, test-retest reliability (Bagby et al., 1994), and factorial validity (Parker et al., 2003) was found in this scale. In this study, the TAS-20 had a Cronbach’s α value of .80.

The Rosenberg Self-Esteem Scale (RSES; Rosenberg, 1965)

The RSES was used in the current study to evaluate global self-esteem. This scale contains 10 items, which yields a single overall score for self-esteem and has good reliability and constructs validity (Rosenberg, 1965). The RSES has also been successfully used among adolescents in Taiwan (Chang et al., 2013), and the internal consistency coefficient for this scale was .89.

Chinese Happiness Inventory (CHI; Lu, 2006)

This study used CHI to assess subjective well-being (Lu, 2006), which is a ten item self-report rated on a 4-point scale. Higher scores represented higher levels of happiness. CHI has displayed good reliability and validity among Chinese adolescents (Lu, 2006). The Cronbach’s alpha coefficient was .90 in the current study.

Social Support Scale (Yeh et al., 2008)

This scale contains 16 items that evaluates perceived social support from friends and parents (Yeh et al., 2008). Higher scores represented a greater degree of perceived actual social support. Lin, Wu, You, Chang, Hu, & Xu (2018) revealed good reliability and validity of this scale among adolescents in Taiwan, and the internal consistency coefficient was .90 in the present study.

Virtual Social Support Scale (VSSS; Yeh et al., 2008)

This scale (Yeh et al., 2008) was revised from the Social Support Scale, and assesses perceived social support from those who are acquainted only through the Internet as distinguished from those known in real life. Moreover, Yeh et al. (2008) added two questions to this scale in accordance with the online environment, and the final VSSS has 10 items. Higher scores represented a greater degree of perceived virtual social support. The scale was rated on a 4-point Likert scale, and the Cronbach’s alpha coefficient was .95 in this study.

Brief Family Function Questionnaire (Chang et al., 2017; Ren et al., 2018)

We used the Brief Family Function Questionnaire to assess family function (Ren et al., 2018), which is a 22 item self-report questionnaire rated on a 5-point Likert scale. This questionnaire includes eight aspects of family functioning, labeled as family cohesion, problem solving, role responsibility, independence, communication, affective responsiveness, family conflict, and emotional involvement. This questionnaire has showed good reliability and validity among adolescents in Taiwan (Chang et al., 2017). The family conflict and emotional involvement subscales were reversed scored, and higher scores indicated healthier family functioning. The internal consistency coefficient for the total scale was .92 in the current study.

Statistical Analysis

SPSS version 18.0 for Windows was used to analyze the data, and the significance level was set at .05. The prevalence for NSSI was calculated. The association between NSSI in the groups and psychosocial risk factors were examined by t-test and χ2 test. Significant factors were further selected and included in the forward logistic regression analysis to examine their associations with NSSI among adolescents.

Results

The outbreak of COVID-19 has worsened existing mental health problems, and NSSI is a common problem among adolescents. This study examined the prevalence of NSSI during the COVID-19 outbreak and discovered that the prevalence of NSSI among sample of 1060 adolescents was 40.9% (95% confidence interval, 37.9%–43.9%), and girls reported significantly higher rates of NSSI than did boys. Among self-injurers, 34.2% (n = 145) reported using only one method, and 65.8% (n = 279) reported using multiple methods. Additionally, 18.1% (n = 78) of self-injurers engaged in NSSI only once, and the remaining 81.9% (n = 354) performed NSSI twice or more. Among participants who engage in NSSI, self-cutting was the most prevalent method (21.6%, n = 229), followed by skin scratching (18.0%, n = 190), hair pulling (16.4%, n = 173), biting (14.0%, n = 148), banging the head or other parts of the body against the wall (12.4%, n = 131), carving words or symbols using sharp objects on the skin that leads to bleeding (11.7%, n = 124), punching (9.9%, n = 105), inserting objects to the nail or skin (8.1%, n = 86), erasing skin (7.2%, n = 76), burning (4.1%, n = 43), and dripping acid onto skin (1.2%, n = 13). Scrubbing skin using bleach or cleaner (1.0%, n = 11) was the least prevalent method of NSSI during the COVID-19 pandemic.

In order to take necessary preventive actions to NSSI during the COVID-19 pandemic, this study also examined the association between NSSI and psychosocial risk factors by t-test and χ2 test. Significant factors were further selected and included in the forward logistic regression analysis to explore multiple critical psychosocial risk factors among adolescents. The results in Table 1 showed that all demographic, psychological, and sociological characteristics were significantly different between the self-injurers and non-injurers groups. Participants in the self-injurers group reported higher levels of neuroticism, impulsivity, depression, alexithymia, virtual social support, and dissatisfaction with academic performance, and lower levels of subjective well-being, family function, actual social support, and self-esteem.

Table 1 Descriptive statistics of all study variables between the self-injurers and non-injurers groups
Full size table

Table 2 showed that high neuroticism, high impulsivity, high alexithymia, low self-esteem, and high virtual social support were associated with NSSI in the forward logistic regression analysis. However, gender, depression, subjective well-being, actual social support, family function, and dissatisfaction with academic performance were not related to NSSI in the forward logistic regression analysis. We put each factor into the logistic regression analysis separately, and the result showed that depression (Wald. coeff. = 109.208, p < .001), subjective well-being (Wald. coeff. = 86.722, p < .001), family function (Wald. coeff. = 44.246, p < .001), dissatisfaction with academic performance (Wald. coeff. = 12.794, p < .001), actual social support (Wald. coeff. = 11.477, p < .01), and being female (Wald. coeff. = 10.963, p < .01) were capable of predicting NSSI independently and significantly, but was not able to significantly predict NSSI when other psychosocial risk factors were put into the forward logistic regression analyses. In brief, neuroticism was the first variable to enter the regression model, followed by self-esteem and virtual social support.

Table 2 Forward logistic regression analyses in predicting NSSI from psychosocial risk factors
Full size table

Discussion

This study aims to uncover and examine the prevalence of NSSI and related psychosocial risk factors among junior high school students in Taiwan during the COVID-19 outbreak. Overall, the results indicated a 40.9% NSSI prevalence among the participants. The prevalence in our study was increasingly higher than those found in previous researches regarding adolescent samples. For example, Barrocas et al. (2012) indicated 12.7% of ninth-graders engaged in NSSI. You et al. (2011) examined 6374 secondary school students and found a prevalence of 15%; Muehlenkamp et al. (2012) conducted a systematic review of 52 empirical studies and found an average lifetime estimate of 18% for NSSI. Giletta et al. (2012) found 24% of 1862 cross-country adolescents; Claes et al. (2014) assessed 532 high school students and indicated 26.5% of adolescents engaged in NSSI. Zetterqvist et al. (2013) found 35.6% among 3060 Swedish adolescents aged 15–17 years old conducted NSSI. In Taiwan, COVID-19 cases rapidly exploded from February 2020, which resulted in a two-week postponement of the second school semester. Schools started on February 25th and data was gathered during March 2 through March 27, 2020. The extended winter vacation may bring about several psychological stress due to the COVID-19 outbreak. Anxiety, lack of peer contact, and reduced opportunities for stress regulation may be the main concern/affect during COVID-19 among adolescents. Additional challenges may arise to adolescents with trauma experiences and for those who have existing mental health problems (Fegert et al., 2020). COVID-19 brings various aspects of change and stress to adolescents and their families. In order to cope with stress associated with COVID-19, junior high school students may engage in NSSI to regulate negative emotions. The COVID-19 pandemic is still ongoing and affects multiple aspects of daily life. With regard to a high prevalence of NSSI, it is necessary for educational agencies and mental health organization to develop intervention strategies.

Among intrapersonal risk factors, this study found that neuroticism was positively associated with NSSI. This finding corresponds with the findings of previous researches (Allroggen et al., 2014; Brown, 2009; Kiekens et al., 2015; Mullins-Sweatt et al., 2013). Furthermore, this study also found that neuroticism had the greatest influence on NSSI among other intrapersonal and interpersonal categories in the logistic regression model. Neuroticism individuals have the tendency to experience negative emotions, such as anxiety, anger, and depressed mood (Brown, 2009). They may perceive more stress, interpret situations as more threatening and view minor frustrations as more hopelessly difficult (Hettema et al., 2006; Penley & Tomaka, 2002). NSSI is likely to be engaged as a way to cope or tolerate negative feelings. Individuals with high neuroticism tend to be more sensitive to COVID-19-related stress and interpret the situation as more difficult to deal with, increasing the chance of NSSI to regulate negative feelings, which increases the risk of NSSI.

This study showed NSSI was negatively associated with self-esteem, which is consistent with past studies (Cawood & Huprich, 2011; Lundh et al., 2007; Tatnell et al., 2014). NSSI may also serve the function of self- punishment (Klonsky, 2007; Nock, 2009). Self-criticism, conceptually similar to low self-esteem, is consistently associated to NSSI (Glassman et al., 2007; Swannell et al. 2012). Compared to indirect forms of self-injury, NSSI individuals showed more self-criticism (Germain & Hooley, 2012). On the contrary, people who are high on self-esteem may be less likely to conduct NSSI. Self-esteem is defined as one’s general sense of his or her value or worth (Rosenberg, 1979), and may be hypothesized to act as a mediator in the relationship between several factors and NSSI (Cawood & Huprich, 2011). Past empirical studies have shown that social support significantly predicted one’s level of self-esteem (Hoffman et al., 1988; Kong & You, 2013). However, the decrease of interaction and social support during the COVID-19 quarantine may affect adolescents’ self-esteem.

NSSI is positively related to impulsivity and alexithymia, which corresponds with the results of past studies (Claes & Muehlenkamp, 2013; Crowell et al., 2012; Cawood & Huprich, 2011; Norman & Borrill, 2015). Individuals with high impulsive may act rashly when experiencing negative emotions in order to gain immediate relief from distress without considering negative consequences of their behavior (lack of premeditation) and tend to quit when faced with frustration or boredom (lack of perseverance) (Glenn & Klonsky, 2010; Lynam et al., 2011; Mullins-Sweatt et al., 2013). NSSI serves as an effective way to regulate aversive emotions (Klonsky & Glenn, 2009, Nock & Prinstein, 2004), therefore, impulsive individuals may be at risk for NSSI (Hamza et al., 2015). On the other hand, adolescents who are high on alexithymia tend to have difficulties regulating and expressing their emotion. Adolescents with high impulsivity may be more likely to engage in NSSI during the COVID-19 outbreak, especially when exposed to negative feelings, experience stress and lack of opportunities for stress regulation, in order to gain immediate relief without considering the consequences. Likewise, adolescents with high alexithymia may engage in NSSI in order to regulate their feelings.

Among interpersonal risk factors, virtual social support is the only risk factor that was able to predict NSSI in the logistic regression model, which contradicts previous research (Lin et al., 2017). With the global use of the Internet, there is an increasing use of online services to get in touch with peers. Social relations were disrupted during the COVID-19 outbreak. Social distancing triggered isolation and aggravated mental problems. The Internet might ease social interaction (Guessoum et al., 2020; Zalite & Zvirbule, 2020), and online social connections is important for mental health outcomes during the COVID-19 quarantine (Pancani et al., 2020) However, excessive media consumption during times of crisis may be a concern for depression, elevated stress, anxiety and psychological distress (Garfin et al., 2020; Keles et al., 2020). Ellis et al. (2020) found that adolescents’ stress during COVID-19 was related to more loneliness and more depression, especially for adolescents who spend more time on social media. Therefore, virtual social support may be related to excessive use of the internet, which aggravates mental problems and trigger NSSI.

The present study found that depression and dissatisfaction with academic performance were significantly higher, actual social support, subjective well-being and family function were significantly lower in the NSSI group compared to the non-NSSI group, which corresponded with prior studies on NSSI (Barrocas et al., 2015; Kiekens et al., 2016; Vansteenkiste & Ryan, 2013; Andrews et al., 2014; Baetens et al., 2014). However, in the forward logistic regression analyses, these factors were unable to significantly predict NSSI when entered after other risk factors. Moreover, this study also added each factor separately into the logistic regression analysis, and found that depression (Wald. coeff. = 109.208, p < .001), subjective well-being (Wald. coeff. = 86.722, p < .001), family function (Wald. coeff. = 44.246, p < .001), dissatisfaction with academic performance (Wald. coeff. = 12.794, p < .001), actual social support (Wald. coeff. = 11.477, p < .01), and being female (Wald. coeff. = 10.963, p < .01) were all able to independently and significantly predict NSSI, but was unable to significantly predict NSSI when other psychosocial risk factors were added into the logistic regression analyses. It is probable that the relationship is mediated by other psychosocial risk factors. Possible mediating factors need to be examined in future studies. Comprehensively, the main feature of this study falls on the incorporation of multiple psychosocial risk factors associated with NSSI during the COVID-19 outbreak.

Despite the significance of our findings, the following limitations should be noted. First, recruiting only junior high school students may limit the generalizability of results, and findings may not be applicable to all NSSI students of similar age who did not attend school due to the pandemic. Second, no causal relationships could be determined due to the cross-sectional nature of the study design. Furthermore, all information was acquired from self-report questionnaires. The reliability of the results depended heavily on the honesty of participants and that they have a clear understanding of the meaning of questions. Response bias may exist. Future work using multiple-method assessments may provide a richer and more thorough understanding of NSSI.

Conclusion

The COVID-19 pandemic influenced people’s lives to a great extent, leading to psychological reactions and mental health problems. NSSI appears to be an important problem for adolescents. As far as we know, this was the first study to investigate the prevalence of NSSI and examined the associations between NSSI and several psychosocial risk factors among junior high school students during the COVID-19 outbreak. This study found that the prevalence of NSSI among junior high school students in Taiwan was extremely high during the COVID-19 outbreak. Furthermore, this study also indicated that a majority of students in the self-injurers group was female, and showed significantly higher scores in neuroticism, impulsivity, depression, alexithymia, virtual social support, dissatisfaction with academic performance, and lower scores in self-esteem, subjective well-being, actual social support, and family function than the non-injurers group. More importantly, high neuroticism, high impulsivity, high alexithymia, high virtual social support and low self-esteem were all independently predictive in the forward logistic regression analysis. Neuroticism was the first variable to be entered into the regression model, followed by self-esteem and virtual social support. The limitations of this study were the recruitment of solely junior high school students and having a cross-sectional research design. However, the results can serve as a guideline for mental health organizations and educational agencies to create policies and designs suitable to NSSI prevention programs geared toward the junior high school population during the COVID-19 pandemic.