نقشه برداری خودآسیبی غیرخودکشی در دوران نوجوانی: توسعه و تحلیل عاملی تأییدی از ضربه، پرسشنامه خودآسیبی و افکار خودکشی نوجوانان (ISSIQ-A)

Abstract Non-suicidal self-injury (NSSI) is the deliberate, self-inflicted destruction of body tissue without suicidal intent and an important clinical phenomenon. Rates of NSSI appear to be disproportionately high in adolescents and young adults, and is a risk factor for suicidal ideation and behavior. The present study reports the psychometric properties of the Impulse, Self-harm and Suicide Ideation Questionnaire for Adolescents (ISSIQ-A), a measure designed to comprehensively assess the impulsivity, NSSI behaviors and suicide ideation. An additional module of this questionnaire assesses the functions of NSSI. Results of Confirmatory Factor Analysis (CFA) of the scale on 1722 youths showed items' suitability and confirmed a model of four different dimensions (Impulse, Self-harm, Risk-behavior and Suicide ideation) with good fit and validity. Further analysis showed that youth׳s engagement in self-harm may exert two different functions: to create or alleviate emotional states, and to influence social relationships. Our findings contribute to research and assessment on non-suicidal self-injury, suggesting that the ISSIQ-A is a valid and reliable measure to assess impulse, self-harm and suicidal thoughts, in adolescence.

. Introduction Adolescence is a developmental stage of profound transformations that may be stressful for the youngster, representing a great challenge in the use of coping strategies and general coping style (Stheneur, 2006 and Hasking et al., 2013). Self-harm and suicide are major public health problems in adolescents, with rates of self-harm being highest in the teenage years and suicide being the second most common cause of death in young people worldwide. Borges and Werlang (2006) argued that the changes and conflicts that are normative in adolescence may result in youths resorting to aggressive, impulsive or even suicidal behaviors to cope with their problems. However, systematic research about this phenomenon is still insufficient (Alfonso and Dedrick, 2010), despite recent theoretical and empirical work having significantly advanced in the understanding of this pervasive behavior. Several studies have shown that self-harming behavior is a significant problem (Madge et al., 2008 and Madge et al., 2011) in adolescence, emphasizing the relationship between self-harming behaviors and suicide ideation (Mangnall and Yurkovich, 2008 and Andover and Gibb, 2010), and the importance of impulsivity to self-harm (Hawton, 2002, Claes et al., 2010 and Madge et al., 2011). Despite a number of studies concerning these issues, the underlying mechanisms to these behaviors are still unclear. As a consequence, awareness about this problem and the proper identification of individuals at-risk is hindered (Madge et al., 2008 and Scoliers et al., 2009), constituting also an important obstacle to efficient intervention. Moreover, research concerning the effectiveness of therapeutic protocols, whether in clinical and non-clinical samples is still unexplored. 1.1. Non-suicidal self-injury (NSSI) Harming oneself without the intention of dying is not a new phenomenon. NSSI behaviors, particularly in adolescence, have received a growing attention in clinical and research settings, but also in popular media (Lloyd-Richardson et al., 2007). The use of multiple terms to describe NSSI, such as “deliberate self-harm (DSH),” “parasuicide,” “self-injurious behavior,” “self-mutilation,” and “self-wounding,” in current literature is further complicated by the fact some of these definitions comprise NSSI with and without suicidal intent (Mangnall and Yurkovich, 2008). On the other hand, the presence of suicide ideation in individuals who self-harm in those definitions often vary according to the samples that are studied (Madge et al., 2008 and Fliege et al., 2009). Nevertheless, a systematic review of previous studies point out to similar rates of DSH and NSSI suggesting that these studies may be comparable and refer to similar phenomena (Muehlenkamp et al., 2012). Non-suicidal self-injury can be defined as the intentional and direct injuring of one׳s body tissue without suicidal intent (Herpertz, 1995, Muehlenkamp, 2005 and Klonsky, 2007). This kind of behavior is most prevalent among adolescents and young adults, and typically involves cutting or carving the skin, with a consistent presentation across nations (May et al., 2012). Several studies have shown that 13–25% of adolescents and young adults surveyed in schools have some history of self-injury, and similar findings were found in a review by Muehlenkamp et al. (2012). Others studies, including clinical and community-based samples, revealed that self-harm tends to have its onset in adolescence, commonly occurring during the middle to late adolescence (13–15 years old), which reinforces this developmental stage as a period of difficulties in emotional regulation and risk-taking (Moran et al., 2012). Rates of NSSI appear to be disproportionately high in adolescents and young adults (Ross and Heath, 2002 and Whitlock et al., 2006): approximately 8% of children ages 12–14 (Hilt et al., 2008), 14–15% of adolescents (Ross and Heath, 2002 and Laye-Gindhu and Schonert-Reichl, 2005), and 14–17% of college students (Favazza, 1989 and Whitlock et al., 2006) report having self-injured. In adolescent inpatient samples, rates of NSSI appear to be 80% or higher (Nock and Prinstein, 2004). The most frequent presentations of NSSI include cutting the skin (reported by 70% of participants), scratching, burning, ripping or pulling skin or hair, pinching, hitting or breaking bones. Specifically, self-cutting is the most common method referred in Child and Adolescent Self-harm in Europe (CASE) (Madge et al., 2008), and other studies, present in 70–97% of individuals who self-harm (Suyemoto, 1998); followed by hitting oneself (21–44%); pinching, scratching or biting oneself (Ross and Heath, 2002). Other studies have found methods such as puncturing with pins or needles, severe scratching (Gratz, 2001) and burning oneself with cigarettes (Messer and Fremouw, 2008). Regarding body tissues, areas that are more accessible and easy to hide or conceal, such as arms, wrists, legs and belly, are more frequently reported (Nock et al., 2006). Considering several phenomenological models for self-harm behavior (Suyemoto, 1998, Pelios et al., 1999, Joiner, 2005, Nock et al., 2006 and Williams and Bydalek, 2007), individuals who engage in self-harming behavior seem do to it with specific goals, self-harm has a function or a result that is expected (Nock and Prinstein, 2004). Individuals can experience immediate relief, and biological evidence point out to an actual physiological stress reduction occurring after a self-harm episode (Bunclark, 2000). Those individuals experience daily negative emotions more often than individuals who do not self-harm. The negative emotional states and experiences may be the main reason to engage in self-harming behavior, as a way to relieve emotional distress (Fliege et al., 2009). In addition, similar to several psychopathologies and as recently acknowledged as an independent condition in DSM-5, disruptive early attachment and interpersonal relationships in adolescence can act as risk and maintenance factors to NSSI (Skegg, 2005, Gratz, 2006 and Fliege et al., 2009), and NSSI may arise as a maladaptive coping mechanism to both disruptive emotional states and experiences (Gilbert et al., 2009, Castilho and Gouveia, 2011 and Castilho et al., 2013) and to interpersonal problems. The functional approach by Nock and Prinstein (2004) proposes that NSSI behavior can be classified and treated according to the functional processes involved in the etiology and maintenance of this problematic behavior. The authors have proposed and assessed a model with four different functions divided in automatic reinforcement and social reinforcement. Automatic reinforcement function can serve the purposes of removing or creating feelings (Brown et al., 2002). It can, therefore, act as negative reinforcements (using NSSI as a strategy to alleviate stress or negative emotional states), which is the most frequent function evoked by people who self-harm (“to stop feeling bad”); or positive reinforcements (self-harm as a strategy to create a desirable physiological state – e.g. “To feel something, even if it is pain”). NSSI can also be used to modify or regulate the social environment. In the negative social reinforcement form, individuals tend to use self-harm to escape interpersonal demands (e.g. Avoiding punishment or avoiding doing something undesirable); in the positive social reinforcement form, individuals tend to gain attention or something from others (e.g. to have attention or having someone react to the individual's behavior, even if negatively) (Nock and Prinstein, 2004). In the study by Nock and Prinstein (2004), the most frequent functions of NSSI endorsed by youths was automatic reinforcement, with most youths using NSSI to regulate (reduce or increase) emotional or physiological experiences.

3. Results 3.1. Item analysis and scale dimensionality Preliminary analysis on the scale reliability showed that the internal consistency of the total scale was good (α=0.86). Each subscale also presented acceptable to very good internal consistency: α=0.76 for the Impulse subscale, α=0.91 for Self-harm, and α=0.82 for suicide ideation. However, four reverse-coded items on the impulse subscale presented consistency problems: items #6 (I finish the tasks I begin), #9 (I like to plan things ahead), #13 (I am a careful person) and #15 (It is easy for me to keep focus). Because these items were detrimental to the subscale׳s internal consistency, these items were removed from further analysis. The Impulse subscale was then constituted by 12 items that presented an internal consistency of α=80. The Function subscale was assessed by KR-20 and presented very good internal consistency (α=0.94). In order to assess construct validity, Confirmatory Factor Analysis (CFA) was computed for 2 different models. The model first included 3 latent variables (Impulse, Self-harm, and Suicide Ideation), in which several items presented individual reliability problems. This has led to the creation of an alternative 4-factor model (including Impulse, Self-harm, Suicide Ideation and an additional Risk-behavior factor). In both models, items #7 (It is hard to control my emotions), 10 (I steal or mess with things I shouldn׳t in order to feel better), 11 (It is hard for me to stand still) and 16 (It is hard for me to wait in a line) presented very low factor loadings in the impulse variable (R2<0.20). Because these items were also those that did not contribute significantly to the scales׳ internal consistency and seem to evaluate constructs that are related, but do not directly refer to impulsivity (e.g. hyperkinesia, defiant behaviors), these items were deleted from the analysis and the final Impulse subscale comprised eight items. All items loading freely on their hypothesized factor, except for one item per factor (set to 1 in order to fix the scale of the model). Model comparisons indicated that the initial 3-factor model had a poorer fit (χ2(272)=4104.450, p=0.000; CFI=0.797; RMSEA=0.090, P(rmsea ≤0.05)=0.000; PCFI=0.723; AIC=4260.450) when compared to the 4-factor solution (χ2(269)=3339.959, p=0.000; CFI=0.838; RMSEA=0.081, P(rmsea ≤0.05)=0.000; PCFI=0.751; AIC=3501.959). Therefore, further adjustments were calculated in the 4-factor model, freeing parameters based on the highest modification indices. Correlations were established between the errors of items #8 and #6; #1 and #7; #3 and #4; #3 and #1; #1 and #2 from the self-harm factor, #1 and #2; #5 and #6 from the risk behavior; items #7 and #8 from the impulse factor. After freeing these 8 parameters, the model presented good fit indexes: χ2(261)=2133.025, p=0.000; CFI=0.901; RMSEA=0.065, P(rmsea≤0.05)=0.000; PCFI=0.784; AIC=2311.025. All items presented adequate individual reliability within their parent factor (R2>0.20): Impulse=0.21–0.40; Self-harm=0.43–0.65; Risk behavior=0.26–0.62; Suicidal ideation=0.54–0.67. Thus, all factors presented good Composite Reliability values (≥0.70): Impulse=0.76; Self-harm=0.90; Risk behavior=0.80; Suicidal ideation=0.82. Construct validity was assessed through AVE, with Self-harm and Suicide ideation presenting the highest consistency (AVEsuicide ideation=0.61 and AVEself.harm=0.55) and risk-behavior and impulse presenting less adequate consistency values for exploratory studies (AVErisk-behavior=0.41 and AVEimpulse=0.29) ( Fornell and Larcker, 1981). Self-harm presented moderate or strong correlation with the three dimensions: Suicide ideation, r=0.341; Impulse, r=0.351; and Risk behavior, r=0.815 (p<0.001). For the 31 items Functions of self-harm module, all items presented good internal reliability and adequate factor loading, and therefore all items were kept in further analysis. Based on item׳s content and the model proposed by Nock and Prinstein (2004), an initial 2-factor model was tested, grouping 24 items referring to Automatic Reinforcement (the goal of self-harm was to create an emotional state or to relieve disruptive emotional states) and seven items referring to Social Reinforcement (self-harm was a way to manipulate social interactions, such as calling for help or revenging others). Model fit indexes for the proposed models suggested poor or adequate fit: χ2(433)=1025.088, p=0.000; CFI=0.988; TLI=0.987; RMSEA=0.028, P(rmsea≤0.05)=1.000; and WRMR=1.472. However, modification indices indicated that freeing some parameters would improve the fit of this model. Therefore, adjustments were made by allowing the errors of some items to correlate between each other, as long as they were in the same factor: items #8 and #10, #11 and #12, #5 and #11, #11 and #13, #12 and #13, #5 and #12 and #29 and #30, all belonging to the Automatic Reinforcement factor. The adjusted model presented good model fit indices: χ2(425)=767.041, p=0.000; CFI=0.993; TLI=0.992; RMSEA=0.022, P(rmsea≤0.05)=1.000; and WRMR=1.203. Although WRMR index is slightly above the proposed cutoff value of one, this index has only been studied through computer simulations ( Yu, 2002), lacking further empirical testing with non-normal and categorical outcomes ( Hsu, 2009). Nevertheless, overall model fit of traditional indices indicate very good adjustment. Lastly, model invariance of the ISSIQ-A was tested between boys and girls, by comparison of a free model (unconstrained) to a constrained model where factor loadings and variance/covariances from the two groups. The CFI difference of 0.009 indicate model invariance of the measure for gender. 3.2. Reliability and validity Reliability analysis was calculated for each factor in the final four factor solution and the Functions subscale: Impulse, α=77; Self-Harm, α=90; Risk behavior, α=0.81 and Suicide ideation, α=0.82; Functions of self-harm module presented α=0.93 for the Automatic Reinforcement and α=0.77 for Social Reinforcement subscale. All values are either good or very good indicators of the measure׳s reliability. 3.3. Convergent validity Significant correlations were found between ISSIQ-A and RTSHIA in the expected sense: all correlations were positive, with Risk-behavior subscales of ISSIQ-A and RTSHIA presenting moderate correlations, as well as the Self-harm subscales of RTSHIA and ISSIQ-A, indicating the convergence of the constructs on both subscales. The correlation between Risk behavior and self-harm were weaker in both scales. The Impulse scale of ISSIQ-A also presented moderate, but lower correlation to RTSHIA Risk-behavior subscale and, interestingly, a moderate correlation was also found between Suicide Ideation on ISSIQ-A and Self-Harm in RTSHIA (see Table 1). Table 1. Correlations between ISSIQ-A and RSTHIA (n=1722). RTSHIA-Risk Behavior RTSHIA-Self-Harm ISSIQ-A Impulse 0.458⁎⁎ 0.292⁎⁎ ISSIQ-A Self-Harm 0.265⁎⁎ 0.635⁎⁎ ISSIQ-A -Risk Behavior 0.513⁎⁎ 0.348⁎⁎ ISSIQ-A-Suicidal Ideation 0.175⁎⁎ 0.556⁎⁎ ⁎⁎ p<0.001 (2-tailed). Table options Correlation coefficients were also calculated for the factors found in ISSIQ-A and Early Memories of Warmth and Affection. As expected, negative correlations were found between ISSIQ-A factors and other measures, although the associations were weak. The only exceptions, however, were the moderate negative correlations observed between Suicide Ideation and both memories of warmth and affection (see Table 2). Table 2. Correlations between ISSIQ-A and EMWSS (n=1722). ISSQ-A EMWSS Impulse −0.159⁎⁎ Self-Harm −0.211⁎⁎ Risk Behavior −0.083⁎⁎ Suicidal Ideation −0.459⁎⁎ Automatic Reinforcement −0.264⁎⁎ Social Reinforcement −0.162⁎⁎ ⁎⁎ p<0.001 (2-tailed). Table options 3.4. Age differences in impulsivity, risk-behavior, self-harm, and suicidal ideations As a final step, age differences in ISSIQ-A factors were explored. The study sample was divided into four groups, according to their age: 15 or younger (n=293); 16 years old (n=488), 17 years old (n=492) and 18 or older (n=449). ANOVA analysis revealed differences between groups concerning Impulse and Self-harm. Post-hoc tests indicated that, concerning Impulse these differences were significant between the younger and older groups (15 years old or younger presented significantly higher scores than 18 or older youths). Regarding Self-harm, post-hoc tests showed that the groups of 16, 17 and 18 or older did not differ, but the younger group tended to score significantly higher than the remaining three groups (see Table 3). Significant differences were also found between the functions of self-harm across different age groups: youths with 15 or less years old score significantly higher on Automatic Reinforcement than 18 or older youths and the younger group also score significantly higher than 17 and 18 or older youths on Social Reinforcement function of self-harm. Table 3. Description of ISSIQ-A ANOVA by age groups. Up to 15 years old (n=293) 16 years old (n=488) 17 years old (n=492) 18 years old or older (n=449) F p M S.D. M S.D. M S.D. M S.D. Impulse 1.186 0.555 1.101 0.520 1.087 0.515 1.064 0.533 3.396 0.017 Self-harm 0.253 0.485 0.143 0.357 0.160 0.406 0.113 0.306 8.370 0.000 Risk Behavior 0.212 0.451 0.145 0.344 0.190 0.388 0.170 0.326 2.213 0.074 Suicidal ideation 0.906 0.794 0.827 0.754 0.814 0.726 0.831 0.724 1.021 0.382 Automatic Reinforcement 2.642 4.786 1.982 4.081 1.821 3.874 1.555 3.618 4.469 0.004 Social reinforcement 0.515 1.166 0.305 0.990 0.250 0.739 0.267 0.863 5.781 0.001 Ta