Construction and verification of nomogram prediction model for non-suicidal self-injury in adolescents with depression

doi:10.21203/rs.3.rs-5232562/v1

Download PDF

Research Article

Construction and verification of nomogram prediction model for non-suicidal self-injury in adolescents with depression

https://doi.org/10.21203/rs.3.rs-5232562/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Background

Accurate identification of adolescents with depression at high risk of non-suicidal self-harm and implementation of appropriate interventions are critical to reducing the risk of self-harm. This study used the nomogram technique to develop a predictive model for predicting non-suicidal self-injury (NSSI) behavior in adolescents with depression.

Methods

Convenience sampling was used to examine 596 adolescents with depression. Data from one hospital were used as training and internal validation sets (n = 455) and data from two other hospitals as external validation sets (n = 144). In the former, nine optimal predictors were identified by combining the results of one-way analysis, LASSO regression, and binary logistic regression. These predictors were used to construct a graphical column-line model of the NSSI. The receiver operator characteristic (ROC) curve, the area under the curve (AUC), the Hosmer-Lemeshow test, and calibration curves were used to test the discrimination and calibration of the training and internal and external validation sets of the predictive model, and decision curve analysis (DCA) was used to measure clinical applicability.

Results

Predictors included in the delivery model, history of peer NSSI, parental psychiatric history, sleep duration, social life events, self-esteem, psychological resilience, social support, and degree of depression (P < 0.05). The area under the ROC curve of the model was 0.880 (P < 0.001), and the Hosmer-Lemeshow test (χ2 = 7.19, P = 0.516), showed a maximum Jordan index of 0.668, a specificity of 0.765 and a sensitivity of 0.933. For clinical applicability analyses, the nomogram prediction model had a significant net gain for almost all risk threshold probabilities. In particular, there was a significant net gain for threshold probabilities of 0.1-1.0. Both internal and external validation of the model showed good discrimination, calibration, and clinical applicability.

Conclusion

The constructed nomogram risk prediction model performs well and can effectively predict the occurrence of NSSI in adolescents with depression, which is scientifically instructive and worthy of further promotion.

adolescent depression

non-suicidal self-injury(NSSI)

nomogram

prediction

model

Non-suicidal self-injury (NSSI) refers to repeated, direct, and deliberate injury to a body part or tissue by an individual without suicidal intent. This behavior is usually not socially acceptable, and although it does not directly lead to death, it may reveal deep psychological or emotional distress[1].In the Chinese adolescent population, the detection rate of NSSI ranges from 5.4–23.2%, and among adolescents suffering from depression, this rate is significantly higher, as high as 44.0%-61.2%[2]. NSSI can transiently increase the levels of endogenous opioids, resulting in transient pleasurable sensations, making it an addictive and difficult-to-treat behavior[3]. In particular, studies from other countries have shown that only 50% of young people confide in others after self-harm[4]. There is research evidence the onset of NSSI is not fixed, and if patients choose not to share their distress, family members may find it difficult to recognize their abnormal behavior because they are unable to pay constant attention. However, these recurrent self-injurious behaviors are an outward manifestation of the adolescent's inner pain and struggle, and failure to recognize and treat them in time greatly increases their risk of suicide[5]. Some studies have shown that once NSSI behaviors occur in depressed patients, their risk of suicide increases to more than seven times the original risk[6]. In the context of the current continuous increase in suicide rates, NSSI is considered to be one of the most important factors in predicting future suicide risk, and its potential threat cannot be ignored[7]. Preventing NSSI in young people with depression is therefore crucial to preventing the further development of suicidal behavior. We need to identify the high-risk group for NSSI at an early stage and provide them with targeted care interventions to effectively prevent the occurrence of NSSI and thus prevent its development into suicidal behavior. Zuo Sweet's study indicated that age, education level, childhood trauma, depression level, and peer rejection were all predictors of NSSI in adolescents with depression[8].In contrast, Yang Yuping's study highlighted the role of factors such as poor family environment, impulsivity, anxiety, and poor interpersonal relationships[9]. However, relying on one or a few factors is not sufficient to accurately predict the likelihood of NSSI in adolescent depressed patients. For accurate prediction, we need to consider the predictive weights of each influencing factor and develop a valid, reliable, and accurate quantitative risk assessment tool to screen individuals at high risk for NSSI accurately. In this study, we used a multi-center cross-sectional survey to investigate the factors influencing self-injurious behavior in adolescents with depression and to construct a prediction model to screen out high-risk groups and provide them with interventions in advance to prevent the occurrence of self-injurious behavior and to promote the physical and mental health of adolescents.

2.1 Study participants

The study encompassed a population of adolescents with depression, and the sample comprised adolescents diagnosed with depression and treated at Nantong Fourth People's Hospital, Nanjing Brain Hospital, and Suzhou Guangji Hospital from December 2022 to July 2023. Inclusion criteria: ① age range of 11–18 years[10]; ② fulfillment of the diagnostic criteria for depression (experiencing a persistently depressed mood or reduced interest in activities for at least two weeks, accompanied by difficulty in concentrating, feelings of worthlessness, or excessive and inappropriate guilt and self-guilt, hopelessness, etc.) as outlined in the ICD-11 Classification of Mental and Behavioral Disorders: Clinical Description and Diagnostic Points[11]; ③ possession of adequate communication skills, including the ability to read and express oneself accurately. Exclusion criteria: ① individuals diagnosed with other mental disorders; ② those with a history of alcohol and drug abuse; ③ presence of severe physical illnesses; ④ abnormalities in hearing or vision; ⑤ suicidal behaviors (jumping from heights, hanging, ingesting lethal doses of drugs, and other obvious suicidal and self-injurious behaviors for death).

2.2 Sample size

The model construction was based on the minimum of observations for both types of outcomes in the dichotomous outcome. The literature indicates a prevalence of 51% for NSSI among adolescents with depression. Given the inclusion of 10 model variables, the final sample size requirement was assumed to be 205 cases(minimum of 100), adhering to the ten events per variable (EPV) principle. Considering a 10% missing rate, invalid questionnaires, and a substantial number of missing values, 228 cases were included for the training set and internal validation, with 70% allocated to the training set and 30% for internal validation. External verification consisted of the same sample size as the internal validation, requiring an additional 68 cases. Therefore, the total planned sample size for this study was 296 cases. Ultimately, 596 patients were included.

2.3 Methods

2. 3.1 General information survey

A self-administered questionnaire was utilized to collect data on age, sex, whether there is self-injury, being a single child, relationships and love experiences, sleep duration, online activity duration, parental education, cohabitation status, parental marital status, parental relationships, residence, and family income.

2.3.2 The Eysenck personality questionnaire (EPQ): It was initially developed by Donald E[12], and subsequently sinicized and revised by Gong Y.X[13]. It was employed to assess the personality dimensions of the patients. The questionnaire consists of 88 questions and includes three personality subscales: extraversion (E), neuroticism (N), and psychoticism (P), and a validity subscale (L). The cumulative scores of each subscale were transformed into standardized T-scores based on age and gender norms.The Cronbach's α-coefficient is 0.81.

2.3.3 The Family Intimacy and Adaptability Scale (Family Adaptation and Cohesion Evaluation Scales II, FACES II): It was developed by Olson in 1982[14], and subsequently sinicized and revised by Fei L.P[15]. It encompasses two subscales: 'Family Intimacy' (the emotional interconnectedness of individual family members) and 'Family Adaptability' (the capacity of individual members to make appropriate adjustments in response to varying family dynamics and stages). FACES II is a tool for evaluating family functioning by measuring intimacy and adaptability. The scale demonstrated satisfactory reliability and validity, with Cronbach's α-coefficient of 0.85. 2.3.4 Self-esteem scale (Rosenberg self—esteem scale, SES): It was initially developed by Rosenberg M[16], and later sinicized and revised by Ji C.F[17].This scale examines an individual's positive or negative self-attitude. Comprising ten questions graded on a four-level scale from "very consistent" to "very inconsistent," the ranking reflects the level of self-esteem, with higher scores indicating elevated self-esteem. The scale's robust internal consistency is characterized by Cronbach's α-coefficient of 0.90.
2.3.5 Conner–Davidson Resilience Scale (CD-RISC10): CD-RISC10 was derived from the work of Campbell-Sills and Stein[18], later sinicized and revised by Dai X.Y[19]. It serves as a concise iteration of the CD-RISC, gauging an individual's resilience and adaptive capacity when confronted with adversity. The scale, involving ten self-rated items, employs a 5-point Likert scale from 0 ("not true at all") to 4 ("true nearly all the time"). The resulting scores, ranging from 5 to 50, correlate with higher levels of resilience. The CD-RISC10 exhibits robust internal reliability (Cronbach's α = 0.89).
2.3.6 Childhood trauma questionnaire (CTQ): A simplified adaptation of the CTQ, initially revised by Bernstein et al[20]. and later by Zhao. X.F[21]. in 2003 assesses the abuse and neglect experienced during childhood. The CTQ comprises 28 items categorized into five subscales: emotional abuse, physical abuse, sexual abuse, emotional neglect, and physical neglect. The Cronbach's α-coefficient is 0.82.
2.3.7. Adolescent Social Support Scale(Social support rate scale, SSRS): According to Xiao.'s social support theory[22], Yei and Dai compiled a social support scale for adolescents in 2008[23].This scale comprises three dimensions: subjective social support, objective social support, and utilization of social support. These dimensions evaluate the social support available to adolescents and their interaction with this support system. The scale encompasses 17 items, each rated on a five-point scale, with "Meets" receiving 5 points, "Somewhat Meets" = 4 points, "Unsure" = 3 points, "Somewhat" = 2 points, and "Not sure" = 1 point. The scale's internal consistency, Cronbach's α-coefficient is 0.82.
2.3.8 Adolescent Society Life Events Scale (ASLEC): Compiled by Liu[23], this scale assesses the frequency and intensity of stressful life events among adolescents, particularly middle school and college students. The correlation coefficients between individual event scores and the total score range from 0.24 to 0.57, averaging 0.45. This scale demonstrates satisfactory reliability and validity, with Cronbach's α-coefficient of 0.85.
2.3.9 Barratt Impulsivity Scale (BIS-11) The 11th edition of the Impulsivity Scale was compiled by Barratt et al. and revised by Zhou L. et al[24]. was used to assess the impulsivity trait, which has a Cronbach's α-coefficient of 0.76, retest reliability of 0.85, and a good structural validity index. The questionnaire consists of 26 items covering three dimensions, namely attentional impulsivity, locomotor impulsivity and unplanned impulsivity, and is scored on a 4-point scale (from 1 'never' to 4 'all the time'), with higher scores indicating greater impulsivity.
2.3.10 Self-Rating Anxiety Scale (SAS) SAS was derived from the work of Zung W.W[25]. It is a unidimensional scale with 20 items, including five reverse-scored questions. The sum of the scores for each item was multiplied by 1.25 and rounded to the nearest whole number to obtain the standardized anxiety score. A 4-point Likert scale was used, with one indicating none or very little time, four indicating most or all of the time, and four indicating most or all of the time. The total standardized score ranges from 25 to 100, with higher scores indicating greater levels of anxiety. Higher scores indicate greater levels of anxiety. The scale has a cut-off for anxiety, with < 50 indicating no anxiety, 50–59 indicating mild anxiety, 60 − 59 indicating moderate anxiety, and ≥ 70 indicating severe anxiety. The Cronbach's α-coefficient is 0.86.
2.3.11 Self-Rating Depression Scale (SDS) The scale was developed by Zung to quantify the amount of money received. Measure the severity and treatment of depressive state Changes in therapy(zung 1971). There are 20 items on the scale. Using a 4-level scoring method, each of the 20 items. If you add up all the scores, you get a rough score; Multiply by the rough parts. Take the integer part after 1.25. Obtained standard Points. The higher the standard score, the more severe the depression. According to the results of the Chinese norm, the SDS standard is divided into 53 points, 53 to 62 points for mild depression, 63 to 72 was classified as moderate depression, and ≥ 72 was classified as severe depression. The Cronbach's α-coefficient is 0.86.
2.4. Statistical analysis
2.4.1 Analytical methods
Two individuals entered all survey data into Excel software. There were four missing data concerning parental relationships, one regarding parental mental health, six related to family income, and six others regarding delivery mode. Data with less than 5% missing values required no filling. Statistical analysis was conducted using SPSS 25.0 and R language 4.1.1. Normally distributed measurement data in two samples were represented by mean and standard deviation and compared using an unpaired two-tailed t-test. Count data variables were expressed as frequencies and percentages, with comparisons made using the χ2 or Fisher's exact tests.Significant.Statistical significance was set at P < 0.05.

2.4.2 Variable selection and prediction model establishment

Initial screening of variables was performed using single-factor analysis. Subsequently, LASSO regression was employed to identify predictive factors. Factors selected from the LASSO regression were then incorporated into binary logistic regression for further analysis. Only those factors demonstrating statistical significance (P < 0.05) were retained for inclusion in the risk prediction model.The dataset was randomly partitioned into training and validation sets, with 70% allocated for model development. The performance of the risk prediction model was evaluated using the following metrics. The Area Under the Curve (AUC) was computed to assess the model's discriminatory ability in distinguishing between different outcomes. Calibration curves of the ROC were also generated to assess model calibration.This test was employed to evaluate the goodness-of-fit of the model.Decision Curve Analysis (DCA) was utilized to assess the clinical utility of the risk prediction model across various threshold probabilities.

2.4.3 Internal and external validation of the model

30% of the information was used for internal validation of the risk prediction model, and data from other hospitals were conveniently extracted for external validation, and the area under the ROC curve, AUC, Hosmer-Lemeshow goodness-of-fit test, calibration curve, and DCA were used to evaluate the discriminability, calibration and clinical utility of the NSSI prediction model, respectively.

2.5 Data collection

①Setting and Participant Selection: Data collection was conducted in two distinct settings: outpatient and ward environments. For outpatient data collection, patients were selected based on their preference for a quiet, comfortable, and well-lit single-room clinic environment following consultation. In the ward setting, patients were chosen based on their stable mood and preference for a quiet, comfortable, and private meeting room.②Participant Recruitment and Consent༚Adolescent patients diagnosed with depression were recruited for this study. Prior to participation, detailed explanations regarding the study's purpose, content, and significance were provided to both the adolescent patients and their parents or legal guardians. Informed consent was obtained from all participants, and assent was secured from adolescent patients. A signed informed notification letter was provided, outlining the study's scope and providing guidance for completing the questionnaire.③Questionnaire Administration༚A standardized approach using uniform instructions and terminology was employed to guide adolescent patients with depression through the questionnaire assessment process. Additional explanations and clarifications were offered to participants to address any queries or uncertainties they may have had during the completion of the questionnaire.

2.6 Ethical considerations

This paper adheres to the ethical principles set forth in the Declaration of Helsinki by the World Medical Association. The ethical considerations guiding this study encompass:

2.6.1 Informed Consent: Prior to participation, all subjects provided informed consent. They were informed comprehensively about the study's objectives, procedures, potential risks, and their right to withdraw at any time without repercussion.

2.6.2 Confidentiality: Strict measures were implemented to safeguard participant confidentiality. All data collected were anonymized and securely stored to prevent unauthorized access.
2.6.3Beneficence and Non-maleficence: The study aimed to maximize benefits for participants while minimizing potential harm. Stringent measures were taken to mitigate risks during data collection and analysis.
2.6.4 Research Design and Approval: The study was approved by the Ethics Committee of the Fourth People’s Hospital of Nantong City (approval no.: 2022-Ko37)
2.6.5Conflict of Interest Disclosure: The researchers affirm no conflicts of interest that could have influenced the study's outcomes or interpretations.
2.6.6 Participant Treatment: Participants were treated with utmost respect and dignity throughout the study. Their voluntary participation was paramount, and efforts were made to ensure their comfort and safety.
2.6.7 Regulatory Compliance: This study adhered to all relevant laws and regulations governing research involving human subjects in China.

3.1. Single factor analysis of influencing factors

Our study comprised 455 participants, including 131 males and 324 females, ages 11–18 years (14.96 ± 1.88) .272 had engaged in NSSI, while 183 had not. Univariate analysis revealed significant factors, such as sex, delivery mode, love status, the experience of heartbreak, history of peer NSSI, parental psychiatric history, score, sleep duration, online time spent, parental education, cohabitants, parental relationship, parents' marital status, monthly household income, scores of FACES II, SSRS, SES, CD-RISC10, ASLEC, SDS, SAS, and Barratt scale. (P < 0.05) (Multimedia Appendix 1).

3.2. LASSO regression identifies potential predictors

Following the univariate analysis, 22 variables were included in this study to screen the characteristic variables in the LASSO regression, and the LASSO regression analysis was carried out according to the 'cv. glmnet'. package of R4.1.1. By changing the penalty function (λ), the estimates of the coefficients of the variables in the model were increasingly compressed until the coefficients of the variables that had a small effect on the results were compressed to 0. Each independent variable was represented by a curve corresponding to one independent variable. The coefficients of the variables that had a small effect on the results were compressed to zero, and the change in the coefficients of each independent variable during the punishment process was represented by a curve, with each curve corresponding to one independent variable. The optimal lambda values were determined using the 'cv. glmnet' function for fivefold cross-validation (Fig. 1. A). The independent variables were filtered according to the optimal lambda value (Fig. 1. B). The 'predict' function was used to identify 13 potential predictors of the optimal lambda value, including gender, mode of production, history of peer self-harm, history of parental psychiatric illness, academic performance, sleep duration, personality traits, level of family closeness and adjustment, social life events, level of psychological resilience, level of self-esteem, level of social support, and level of depression.

3.3.Binary logistic regression to determine variables for inclusion in the model

Based on the results of the LASSO regression analysis, we included 13 potential predictors to be analyzed in a binary logistic regression to determine the variables that were ultimately included in the model, and a multivariate stepwise logistic regression analysis narrowed the predictors of self-injury to nine (Table 1). These predictors included delivery model, history of peer NSSI, parental psychiatric history, sleep duration, social life events, self-esteem, psychological resilience, social support, and depression. (p < 0.05).

3.4.Construction of the Risk Prediction Model

The clinical sample data were randomly split into a training set (70%) and a validation set (30%) and divided into self-injurious and non-self-injurious groups based on the presence or absence of NSSI. A nomogram risk prediction model chart for the prediction of NSSIs was constructed using the nine predictors finally identified by univariate, LASSO regression, and dichotomous logistic regression analyses (Fig. 2). The nomogram risk prediction model diagram consists of a score corresponding to production mode, history of peer self-injury, history of parental psychiatric illness, sleep duration, social life events, level of self-esteem, psychological resilience, social support, and self-rated depression, total score, and a prediction probability. Each factor corresponds to a corresponding score, and the sum of each score is the total score, from which the corresponding probability of NSSI occurrence is derived. For this risk prediction model, the area under the ROC curve was 0.880 (P < 0.001), and the discriminative performance of this model was good (Fig. 3. A). The optimal cut-off value of the prediction model corresponds to the maximum value of the Jordan index. The results of the modeling group test showed that the Hosmer-Lemeshow goodness of fit test (χ2 = 7.19, P = 0.516) had a maximum Yoden index of 0.668, a specificity of 0.765, and a sensitivity of 0.933. A calibration plot was plotted to assess the calibration of the predictive model, and the high overlap between the data point connectors and the diagonal lines suggests that the predictive model is well calibrated (Fig. 4. A). The DCA of the training set shows that the column-line plot prediction model has a significant net gain for almost all risk threshold probabilities, especially when the threshold probabilities are in the range of 0.1-1.0 (Multimedia Appendix 2. A).

For example, in a case of adolescent depression where the mode of delivery was normal (0 points), peers had a history of self-harm (20 points), parents had no history of psychiatric disorders (0 points), sleep duration was 4–6 hours per day (35 points), social life events scored 90 points (47 points), self-esteem was 38 (0 points), psychological resilience was 50 (28 points), social support was 60 (17 points), and self-rated depression was 80 (36 points), for a total score of 183, giving a probability of 70%.

3.5. Internal validation of the prediction models

Thirty percent of the information was used for internal validation of the nomogram risk prediction model, with an area under the ROC curve of 0.863 (p < 0.001) in the internal validation set test (Fig. 3. B), a Hosmer-Lemeshow goodness-of-fit test (χ2 = 11.366,p = 0.182), a maximum Yoden's exponent of 0.986, a specificity of 0. 994 and sensitivity of 0.992. The internal validation set calibration curves were plotted after 1000 replicate samples of the internal data. The internal validation set calibration curves fit the diagonal line well, and the internal validation set calibration curves predicted that the NSSI showed strong performance (Fig. 4. B). The internal validation set DCA showed that the nomogram risk prediction model had a significant net gain for almost all risk threshold probabilities, especially for threshold probabilities in the range of 0.25–0.95 (Multimedia Appendix 2. B).

3.6. External validation of the prediction model

Adolescent patients with depression who were admitted to Nanjing Brain Hospital and Suzhou GuangJi Hospital from December 2022 to August 2023 were selected as study subjects, and the information collected was used as the external validation data set of the model. The area under the ROC curve of the nomogram risk prediction model was 0.820 (Fig. 3. C), and the maximum Yoden index of the model was 0.523, sensitivity was 0.645, and specificity was 0.878. Hosmer-Lemeshow test (χ2 = 9.466, P = 0.305). The nomogram calibration curve for predicting NSSI also performed well (Fig. 4. C). The DCA of the external validation set showed that the Nomogram risk prediction model had a significant net benefit for almost all risk threshold probabilities, particularly for threshold probabilities in the range of 0.45–0.85 (Multimedia Appendix 2. C).

In this study, we constructed a risk prediction model that included nine key variables: depression severity, sleep duration, social life events, parental history of mental illness, peer history of NSSI, mode of birth, level of mental resilience, level of self-esteem, and level of social support. All these variables showed a certain degree of correlation with NSSI behavior. This finding provides important clues for us to understand the multi-dimensional influencing factors of NSSI behavior.

Previous studies have deeply explored the close relationship between the severity of depression and NSSI tendency in adolescents, and it is generally believed that the severity of depression is one of the important factors affecting the incidence of NSSI in adolescents[9, 26].Further studies also found that the serotonin receptor binding index in the frontal lobes of depressed patients with NSSI was decreased compared with depressed patients without NSSI[27].This suggests that differences at the neurobiological level may be related to NSSI behavior. In addition, the levels of proinflammatory cytokines in depressed patients with NSSI were significantly increased, and these patients tended to exhibit more aggressive personality traits. Further analysis showed that the levels of proinflammatory cytokines were positively correlated with impulsivity[28] .

Numerous studies have revealed that a significant reduction in sleep duration is an important predictor of non-suicidal self-injury (NSSI) behavior[29, 30]. In addition, other studies have further pointed out that both nightmares and irregular sleep are significantly correlated with NSSI The results of this study further confirm that sleep duration is an independent influence on NSSI in adolescents with depression, but the study did not go into depth regarding sleep patterns, so future studies could focus more on the specific effects of sleep patterns on NSSI behaviors in adolescents with depression to provide more comprehensive intervention strategies.

This study significantly revealed that social life events have a significant impact on the development of NSSI behavior. Negative life events, such as academic stress, interpersonal distress, and family conflict, have been widely recognized as important triggers for NSSI. Young people have a relatively limited ability to deal with problems, adapt to the environment, and face difficulties, and they are prone to fall into pain and produce strong negative emotions when they encounter challenges. In such cases, NSSI often becomes the preferred way for them to try to get rid of negative emotions[31]. Families, schools, and all sectors of society should pay great attention to the mental health of adolescents, actively prevent and reduce the stimulation of adverse life events, and provide them with the necessary support to promote their healthy growth.

Mental disorders, such as schizophrenia, mood disorders, anxiety, and related disorders, will not only cause individuals to have abnormalities in thinking, will, attention, and behavior, such as thinking disorders, inattention, impulsive behavior, etc., but also may cause individuals to severely limit their functions in various fields such as personal, family, social and occupational[32]. Parents suffering from mental disorders will undoubtedly hurt the family environment, which has been widely verified in domestic and foreign studies[33–34].

Studies have shown that adolescents who have witnessed their relatives' suicide or self-injury tend to imitate their relatives' behavior to cope with distress and frustration[35]. Through qualitative interviews with adolescents with depression associated with NSSI, we explored their psychological experience in-depth and found that patients tend to have the impulse to imitate after witnessing their peers' NSSI[36]. However, there are relatively few studies on the influence of this factor on NSSI at home and abroad, and they lack sufficient theoretical support.

Foreign studies have shown that infants born by cesarean section are exposed to unique hormonal, physiological, bacterial, and medical exposures that may subtly alter the physiological state of the newborn[37].More importantly, multiple studies have revealed an association between cesarean section delivery and a range of mental and behavioral problems, such as attention deficit hyperactivity disorder, autism spectrum disorder in children, and behavioral problems[38]. General anesthesia used during cesarean section may produce neurotoxic substances and lead to oxytocin dysregulation, factors that have been closely linked to the development of autism in children[39]. To explore the effect of production mode on NSSI behavior in adolescents with depression, the results show that cesarean section is an important factor affecting NSSI in adolescents with depression. Given the lack of previous studies on this type, we call on future scholars to conduct more in-depth research on cesarean section and the mental and behavioral problems of the offspring to provide a more scientific basis for the rational use of cesarean section and protect the physical and mental health of children.

People with a higher level of psychological resilience can adopt more effective coping strategies when encountering adversity, integrate surrounding resources more flexibly and intelligently, and thus avoid self-harm[40]. In the family, a good parent-child relationship provides a solid backstop for children, not only provides social support and risk resistance when dealing with setbacks but also helps children establish a sense of self-worth. When communicating with children, pay attention to listening and feeling the child's inner world, help children grow healthily, and cultivate stronger psychological resilience. Therefore, by improving the psychological resilience of adolescents, we are expected to prevent and reduce the occurrence of NSSI behaviors more effectively.

People with high self-esteem tend to have a positive self-perception, have confidence in themselves, and be able to face challenges bravely. However, people with low self-esteem tend to hold a negative view of themselves and lack self-confidence and are more likely to choose to escape or self-harm when facing difficulties[41].To improve the self-esteem level of teenagers, we can start from many aspects, such as cultivating their interests and helping them to set and achieve goals so that they feel their value and achievement.

Social support is the key emotional and material support given to individuals. When individuals feel a lack of social support for a long time, they are prone to emotional disorders, interpersonal tension, and cognitive bias, which will significantly increase the incidence of NSSI[42]. Rebuilding and improving social support systems is essential to prevent NSSI in adolescents. At the societal level, we need to build an integrated system involving hospitals, communities, schools, and families, focusing on those at risk. By providing emotional support and solving practical problems, young people can feel loved and respected. At the same time, provide them with effective communication channels to help them get timely support and help when they are confused or encounter difficulties, to promote the physical and mental health of adolescents.

5.Advantages and limitations

This research has a certain degree of advantages. First, unique choice of research subject: In today's society, NSSI behaviours have attracted much attention, and adolescents, as a major group among them, are even a hotspot for research. However, this study uniquely focuses on the special group of adolescents with depression and explores in depth the factors influencing their NSSI behaviours. This unique research perspective not only enriches the research content in related fields, but also provides valuable clues for understanding and solving the NSSI problems of adolescents with depression. Second, Deepening of research content: Unlike previous studies that mainly focused on the analysis of NSSI influencing factors, this study further explored in depth the independent influencing factors on the occurrence of NSSI in adolescent depressed patients, and successfully constructed a risk prediction model. This model can accurately predict the probability of occurrence of NSSI behaviours in adolescent depression patients, providing a scientific basis for clinical intervention and risk assessment.Finally, Comprehensive innovation of research methodology: This study adopted a variety of advanced statistical methods, such as single-factor analysis, LASSO regression analysis and binary logistic regression equation analysis, which effectively avoided the problems of covariance and overfitting and ensured the scientificity and reliability of the model. However, this study does have some shortcomings. First of all, This is a cross-sectional study and data were collected mainly through questionnaires. However, patients may be at risk of avoiding certain questions or recall bias when completing the questionnaire, and it failed to provide an in-depth analysis of NSSI behavioral causality. To overcome this limitation, future studies should aim to conduct longitudinal follow-up studies to explore the dynamic evolution of NSSI behaviors in greater depth. Secondly, this study relied heavily on the analyses provided by the scale data to identify key variables and did not explore this in terms of imaging and laboratory information. To gain a more comprehensive understanding of the biological mechanisms behind NSSI behaviors, future studies should incorporate imaging indices and laboratory values to provide more comprehensive scientific evidence.

This study adopted a combination of single-factor analysis, LASSO regression, and binary Logistic regression to screen the nine optimal predictors for self-injury and construct a nomogram to predict the risk of self-injury. Among them, the mode of delivery, peer self-injury history, parental history of mental illness, sleep time, social life events, self-esteem level, psychological resilience, social support level, and depression degree are predictors of NSSI in adolescents with depression. The constructed Nomogram risk prediction model has good performance and can effectively predict the occurrence of NSSI in adolescents with depression, which is scientific and instructive. It provides theoretical support for accurately screening high-risk populations, providing effective intervention measures, and preventing the occurrence of NSSI. External validation found that the model has good discrimination and calibration, which is worthy of further promotion and use.

Funding

This work was supported by the Nantong Municipal Health and Wellness Committee [Approval Number: MS2022080]. Nantong City Social Livelihood Science and Technology project [Approval Number: MS2023023]

Availability of data and materials

The data of this study are presented in the attachment.

Declaration of Competing Interest

The authors declare no conflicts of interest related to this work.

Acknowledgments

We extend our heartfelt gratitude to all individuals who contributed to the realization and documentation of this project. We sincerely appreciate the guidance and assistance of our supervisors, Prof. Xu Xujuan and Associate Prof. Zhang Feng, provided throughout the design, execution, and writing phases. Furthermore, we are indebted to the project team members who dedicated their leisure time to diligently gathering and organizing the data. Lastly, we acknowledge the invaluable support of the Nantong Municipal Health and Wellness Commission leaders and the Nantong Fourth People's Hospital. Their selfless aid has profoundly underpinned the successful implementation of this endeavor.

Ethics approval and consent to participate

Ethical approval was granted for this study by the Fourth People’s Hospital of Nantong City (approval no.: 2022-Ko37). This study was performed in accordance with the Declaration of Helsinki.

Consent for publication

Not applicable.

Availability of data and material

The data that support the findings of this study are available upon reasonable request from the corresponding author Xujuan Xu. The data are not publicly available because it contains information that could compromise the privacy of the research participants.

Competing interests

The authors declare that they have no competing interests.

Acknowledgements

Thanks to Dr. Zhang Feng for his guidance and to Professor Xun Pengcheng for his statistical guidance

Author contributions

YHG designed the research scheme, prepared the charts, reviewed the drafts, and approved the final drafts. YCW collated the data, analyzed the data, and approved the final draft. XJX and HMH directed and supervised the entire research plan and approved the final draft. YC and JJS are responsible for data collection. JHW、XLM and JLH were responsible for data entry and sorting.

Ghinea DF, Parzer A, Koenig P, et al. Psychosocial functioning in adolescents with non-suicidal self-injury: the roles of childhood maltreatment, borderline personality disorder and depression. Borderline Personal Disord Emot Dysregul 2021;8(1):21.
Liu XL, Zhou XY, Huang, XP, et al. Study on related factors of depression patients with non-suicidal self-injury behavior. Chongqing Medical Science 2023;52(14):1-11.
Li ZY, Ban C, Song JY, et al. Suicidal ideation in depressed adolescents with non-suicidal self-injury. Chinese Journal of Mental Health 2021;25(12):1783-85.
Ammerman BA, Mccloskey MS. The Development of a Measure to Assess Social Reactions to Self-Injury Disclosure. Assessment 2021;28(1):225-37.
Yan SY. Research on detection methods of non-suicidal self-injury behavior based on indoor activities[M]. Guizhou University 2023.
Shen XL, Dong ZQ, Lou SX, et al. Study on alexithymia and family education in adolescents with depression accompanied by non-suicidal self-injury behavior. Neurological disorders and mental health 2020;20(2):101-03.
Mars B, Heron J, Klonsky ED, et al. Predictors of future suicide attempt among adolescents with suicidal thoughts or non-suicidal self-harm: a population-based birth cohort study. The Lancet Psychiatry 2019;6(4):327-37.
Zuo TT, Wang KC, Hu YF. The predictive effect of childhood trauma and peer environment on non-suicidal self-injury behavior in adolescents with depression. Chinese Journal of Psychiatry 2022;55(4):272-280.
Yan YP, Cheng BB, Wu SY, et al. To explore the relationship between non-suicidal self-injury and negative emotion and impulsivity in depressed adolescents. Chinese and foreign medical treatment 2021;36:77-79.
Teicher MH, Samson JA, Anderson CM, et al. The effects of childhood maltreatment on brain structure, function and connectivity. Nat Rev Neurosci 2016;17(10):652-66.
Thai TT, Jones MK, Nguyen TP, et al. The Prevalence, Correlates and Functions of Non-Suicidal Self-Injury in Vietnamese Adolescents. Psychol Res Behav Manag 2021;14:1915-27.
Donald E, Morisky S, MSPH A, et al. Marie Krousel-Wood,. Predictive Validity of a Medication Adherence Measure in an Outpatient Setting. The Journal of Clinical Hypertension 2008;10(5):348-54．.
Gong YX. Eysenck personality Questionnaire 1992;Revised edition. Changsha: Hunan Map Publishing House:1-50.
Olsen DH, Rusell AS, Sprenkle DH. Circumplex Model of Marital and Family Systems: Vl. Theoretical Update 1983;69-83.
Fei LP, Shen XJ, Zheng YP, et al. Preliminary evaluation of "Family closeness and Adaptability Scale" and "Family Environment Scale". Chinese Journal of Mental Health 1991;5(5):198-202.
Christopher G. Fairburn CG, Shafran R, Cooper Z. Society and the adolescent self-image. Social Forces 1965;3(2):1780-90.
Ji YF, Yu X. (Rosenberg M .The self-Esteem Scale )(SES) Chinese Journal of Mental Health 1999;5(4):318-20.
Campbell-Sills L, Stein MB. Psychometric analysis and refinement of the Connor-davidson Resilience Scale (CD-RISC): Validation of a 10-item measure of resilience. J Trauma Stress 2007;20(6):1019-28.
Dai XY. Manual of commonly used psychological assessment scales. Beijing: People's Military Medical Press 2010;133-37.
Bernstein DP, Stein JA, Newcomb MD, et al. Development and validation of a brief screening version of the Childhood Trauma Questionnaire. Child Abuse & Neglect 2003;27(2):169-90.
Zhao XF, Zhang YL, Li LF. Childhood abuse questionnaire of 435 children. Chinese Journal of Clinical Psychology 2004;12(4):377-79.
Xiao SY. Theoretical basis and research application of Social Support Rating Scale. Journal of Clinical Psychiatry 1999:98-90.
Liu XC, Zhu F, LIU YQ, et al. Reliability and validity test of adolescent life events Scale. Chinese Journal of Clinical Psychology 1991;5(1):33-35.
Zhuo L, XIao SY, He XY, et al. BIS-11reliability and validity test of Chinese version. Chinese Journal of Clinical Psychology 2006;14(4):343-47.
William WK. Zung MD. A rating instrument for anxiety disorders. Psychosomatics 1971;12(6):371-9.
Xu ML, Liu SM. Chen J. Relationship between life events, emotional symptoms and non-suicidal self-injury behavior in adolescents with depressive disorder. Psychiatric journal 2020;33(6):420-22.
Audenaert K, Van Laere K, Dumont F, et al. Decreased frontal serotonin 5-HT 2a receptor binding index in deliberate self-harm patients. Eur J Nucl Med 2001;28(2):175-82.
Zhang AG, Hu XJ, Zhang XL, Correlation between serum inflammatory marker levels and impulsivity traits in depressive disorder patients with non-suicidal self-injury. Journal of Anhui Medical University 2023;58(10):1769-72.
Hysing M, Sivertsen B, Stormark KM, et al. Sleep problems and self-harm in adolescence. Br J Psychiatry 2015;207(4):306-12.
Russell K, Allan S, Beattie L, et al. Sleep problem, suicide and self-harm in university students: A systematic review. Sleep Med Rev 2019;44:58-69.
Wang XX, Wei M, Liu X. The mechanism and influencing factors of adolescents' NSSI: from the perspective of emotion management. Psychological development and education 2017;33(6):759-68.
Organization WH. ICD-11 Classification of Mental and Behavioral Disorders: Clinical Description and Diagnosis [M]. Beijing: People's Medical Publishing House 2021.
Gou CY. Clinical characteristics and influencing factors of non-suicidal self-injury in adolescents with depressive disorders[M]. Anhui Medical University 2022.
Xue ZF, Zhang WJ, Lu QZ. Analysis of influencing factors of non-suicidal self-injury in adolescents with depression. Medical food therapy and health 2021 19(10):3-5.
Stephen P, Lewis1 NLH, Natalie J M, et al. non suicidal self injury youth and source child adolesc psychiatry ment health. Child and Adolescent Psychiatry and Mental Health 2012;13(6):1-9.
Gao YH, Xu XJ, Chen Y. Qualitative study on the psychological experience of non-suicidal self-injury in adolescents with depression. Journal of nursing 2023;38(14)
Sandall J, Tribe RM, Avery L, et al. Short-term and long-term effects of caesarean section on the health of women and children. Lancet 2018;392(10155):1349-57.
Al-Zalabani AH, Al-Jabree AH, Zeidan ZA. Is cesarean section delivery associated with autism spectrum disorder? Neurosciences (Riyadh) 2019;24(1):11-15.
Vanya M, Szucs S, Vetro A, et al. The potential role of oxytocin and perinatal factors in the pathogenesis of autism spectrum disorders - review of the literature. Psychiatry Res 2017;247:288-90.
Ungar M, Theron L. Resilience and mental health: how multisystemic processes contribute to positive outcomes. Lancet Psychiatry 2020;7(5):441-48.
Baumeister RF, Krueger JI, et al. Does High Self Esteem Cause Better P Source Psychol Sci Public Interest Psychological science in the public interest.Sychological Science In The Public Interest 2003; 4(1):1-44.
Nemati H, Sahebihagh MH, Mahmoodi M, et al. Non-Suicidal Self-Injury and Its Relationship with Family Psychological Function and Perceived Social Support among Iranian High School Students. J Res Health Sci 2020;20(1):e00469.

Table 1 Predictors of NSSI risk in adolescents with depression.

Intercept and variables	Prediction model
Intercept and variables	β	z value	OR	P value	95% CI
Sex	0.497	0.333	2.229	0.135	0.856-3.155
Delivery mode	1.074	0.300	12.768	0.000	1.624-5.272
History of peer NSSI	0.658	0.334	3.880	0.049	1.003-3.720
Parental psychiatric history	2.145	0.824	6.781	0.009	1.700-42.916
Score	-0.179	0.203	0.781	0.377	0.562-1.244
Sleep time	-0.746	0.200	13.880	0.000	0.320-0.702
EPQ	0.203	0.225	0.818	0.366	0.789-1.903
FACES II	-0.005	0.008	0.447	0.504	0.980-1.010
SSRS	-0.060	0.011	29.530	0.000	0.921-0.962
CD-RISC	-0.051	0.012	18.314	0.000	0.928-0.973
SES	-0.157	0.031	25.014	0.000	0.804-0.909
ASLEC	0.053	0.009	34.217	0.000	1.036-1.073
SDS	0.039	0.015	6.596	0.010	1.009-1.070

No competing interests reported.

Download PDF

Editorial decision: Revision requested
15 Oct, 2024
Editor assigned by journal
11 Oct, 2024
Submission checks completed at journal
11 Oct, 2024
First submitted to journal
09 Oct, 2024

You are reading this latest preprint version

Construction and verification of nomogram prediction model for non-suicidal self-injury in adolescents with depression

Status:

Version 1

Abstract

Background

Methods

Results

Conclusion

Figures

Introduction

Materials and methods

Results

Discussion

Conclusions

Declarations

References

Table 1

Additional Declarations

Supplementary Files

Status:

Version 1