Self-reported frequency of sex as self-injury (SASI) in a national study of Swedish adolescents and association to sociodemographic factors, sexual behaviors, abuse and mental health

Self-injurious behaviors (SIB) can either be direct, such as cutting or burning the skin, or indirect through the use of harmful behavior such as abusive relationships, binge eating or alcohol abuse [3, 4]. Direct self-injury is usually divided into suicidal and non-suicidal self-injury (NSSI) depending on the intention to kill oneself [5]. Earlier definitions of direct self-injurious behaviors have also included more indirect forms of self-injury such as risk-taking, promiscuity and drug abuse [3]. In a study based on 11 European countries the estimated lifetime prevalence of direct self-injurious behavior was 27.6%, occasionally seen in 19.7% and repetitively seen in 7.8%. The behavior was more common among girls [6]. In a review article from 2012, the mean prevalence for NSSI was estimated to 18.0% [7] and according to a Swedish study, 11.1% of girls and 2.3% of boys meet the DSM-5 criteria for NSSI syndrome [8].

Sexual risk-taking behaviors, substance abuse and eating disorders are usually considered to be an indirect form of SIB since they do not cause immediate damage to the body tissue and the effects may not be seen until later [9‐11]. It has been suggested that to be considered as an indirect self-injurious behavior, the behavior should be repetitive, be of concern to clinicians or family members and potentially cause physical damage if continued [11]. Attention has recently been placed on shared factors for the co-occurrence of NSSI and indirect self-injury, such as eating disorders, with common elements seen in using the body to control state of mind and social situations [9, 12].

During interviews, young women who sell sex have described using sex as a way to self-injure, in the same way as cutting or burning the skin [13]. Using sex as affect-regulation was described as follows by one woman who sold sex:

“When I feel bad I contact someone who wants to meet me. I feel so bad than that I’ll do just anything to relieve that pressure. Before the meetings the anxiety is so strong that I barley remember how I got there […] then I shut down. Let someone else take me over and decide. […]. Afterwards I feel like crap. Feel disgusting and empty. Often I am in a lot of pain. […].” [13, p. 23].

Sometimes self-injury through selling sex had even replaced cutting the skin as it was less visible. A further quotation from a young woman selling sex:

“[…] and I was the good one who didn’t self-harm anymore. Everyone was so pleased, but I felt just as bad, I just found other ways […] that weren’t that visible [selling sex] […] things that almost killed me for real.” [13, p. 23].

The self-destructiveness of selling sex and visiting online sex sites often increased in periods of poor mental health and the quitting process was described as challenging since the women found themselves caught in a behavior that was hard to break because of the function of affect regulation [13].

Associations between risky sexual behaviors and NSSI has been seen [13‐15] and adolescents that have displayed risky sexual behaviors are twice as likely to have a history of suicide attempts [16]. Depressive symptoms independently predict risky sexual behavior in adolescents, indicating that sex is being used as a coping strategy for depression [17]. To use sexual intercourse as a way of affect regulation and coping strategy is a behavior that is known from the research field [17‐21]. Using sex as a coping strategy was associated with younger women, more risky sexual behavior with poor condom use, experience of physical abuse during childhood or adolescence and poor communication with the partner [21].

Child sexual abuse is associated with later high-risk sexual behavior such as a greater number of sexual partners, higher frequency of sexually transmitted infections, teenage pregnancy, prostitution and earlier age of sexual debut [22‐24]. Child sexual abuse also increases the risk of later sexual revictimization [19, 20, 22, 23, 25] which seems to be partly mediated by sexual self-esteem, sexual concerns and high risk sexual behavior [25]. The use of sexual intercourse as a way to reduce negative affects has been suggested as a pathway from sexual abuse during childhood or during adolescence, to later revictimization [19, 20]. Symptoms of depression and anxiety have been found to mediate the relationship between using sex as an affect-regulating strategy and sexual assault [18, 20]. The use of sex to reduce negative affects is associated with having more sexual partners, including more partners of casual nature [20]. Emotional dysregulation has been suggested as a direct pathway to revictimization, with risky sexual behavior as one resulting risk factor [26].

Since sex as self-injury (SASI) is a largely unexplored area of research, not yet conceptualized and far from accepted in the research field, there is a need to further explore its occurrence and associations to other behaviors and potential risk-factors.

The study was carried out in the 3rd and last year of Swedish high school during the fall of 2014. The selection of study sample, distribution and collection of the questionnaire was performed by Statistics Sweden (a national administrative agency). To form the study sample, the National School Register for the 2nd year of Swedish high schools for the fall of 2013 was used. By using stratification on the basis of school size and study program a total of 13,903 adolescents from 261 out of 1215 schools were selected for the study. Of the 261 schools selected, 238 were still providing the selected study programs in 2014. A total of 171 schools with 9773 adolescents agreed to participate in the study. Of the 9773 adolescents that had the opportunity to participate in the study, 5873 completed the questionnaire. Thirty-four questionnaires were excluded due to unserious answers or a high amount of missing data. This gave a response rate of 59.7%. A further 89 did not answer the index question about using sex as self-injury, resulting in a total of 5750 participants for the study. Mean age of the participating adolescents was 18.0 years (SD = .6). According to data from 2014, 91.7% of all Swedish 18 years old adolescents were enrolled in the Swedish high school system [29].

The study group was selected with the aim of being representative of the 3rd year of Swedish high schools. However, for a separate study concerning Stockholm, an extra sample from the county of Stockholm was included in the study. The additional Stockholm sample showed a lower response rate (48.7%) compared to the rest of the country (65.3%), came more often from middle-size schools with 191–360 pupils (51.2 vs. 41.6%), giving a small effect size (Cramer’s V = .10, χ² = 63.6, df = 2, p = .000), and were more often studying practical high school programs (33.2 vs 27.7%), giving no effect size (Cramer’s V = .05, χ² = 17.1, df = 1, p = .000). The Stockholm sample was retained in this study to give a larger and more robust study sample.

Information about the study was sent to the head of each school by mail. Students received written information about the study and gave informed consent for participation by filling in the questionnaire. According to the Ethical Review Act of Sweden, active consent is not required from parents of adolescents’ aged 15 years or older [30]. Participants answered the questionnaire in digital format (by computer, in 165 schools) or, where computers were not available, on paper (6 schools). Regardless of distribution method, anonymity was guaranteed. The study was performed during lecture time in the selected schools during September–November 2014. Reminders were given by phone during November 2014 to those schools that had not yet returned data. With regard to the sensitive topics in the questionnaire, both the head of the school, teachers responsible for the lecture and the participating adolescents received an information letter about the study including contact details for help and support if needed after answering the questionnaire.

The questionnaire for the present study was a modified version of the questionnaire used in 2004 and 2009 [27, 28]. The questionnaire used for this study comprised in total 116 main questions, of which 34 were analyzed in the present study.

The index question for this study was new and literally formulated: “Have you ever used sex to purposely hurt yourself?”. To investigate the occurrence of sex as self-injury, questions included were; age at first occurrence, number of occurrences during the past year and in total, age and gender of the sexual encounter on the previous occasion and the perceived pain of using SASI.

Questions about Sociodemographic factors included gender with the options boy, girl and “The classification ‘male’ or ‘female’ does not fit for me”, parents’ occupation and education, financial situation in the family, immigrant background and living situation.

Sexual behavior and sexual risk-taking, were investigated by questions concerning sexual orientation, voluntary sexual experiences, age at first voluntary intercourse, number of sexual partners, use of contraceptives, occurrence of abortion (self or partner) and sexually transmitted infection of chlamydia. To investigate the occurrence of selling sex, the question used was “Have you ever sold sexual services?”.

The question related to sexual abuse was “Have you been exposed to any of the following against your will?”. Included in the options were: someone having exposed him-/herself to you via the Internet or otherwise, someone having touched your genitals/tried to undress you to have sex with you, forced you to masturbate or have vaginal, oral or anal intercourse. Flashing is by definition an abusive act according to Swedish law if it is against the will of the victim, irrespective of whether it occurs in real life or via the Internet, which is why it was included in the analysis for being exposed to ‘any sexual abuse’. Further analyses were made, including only penetrative abuse (oral/anal/vaginal abuse). Follow-up questions for sexual abuse were asked concerning the first exposure, as follows; age of the victim, relationship to the perpetrator and type of sexual abuse. One question was asked concerning the total number of times exposed to sexual abuse. All questions concerning sexual abuse were used in the questionnaires from 2004 and 2009.

Exposure to emotional and physical abuse was measured by the question; “Have you prior to the age of 18 been subjected to any of the following by an adult?”. Emotional abuse was measured through three questions; insulted, threatened to be hit, isolated from friends. Physical abuse was measured by eight questions, ranging from being pushed or shaken, hit with the hands or items, burned or strangled. The answers were ranked into four; never—rarely—sometimes—often. However, when analyzing the question the answers were dichotomized into ‘been exposed’ including the answers rarely, sometimes and often, or ‘never been exposed’. This instrument has not been validated but has been used in the earlier studies from 2004 and 2009.

Contact with healthcare for psychiatric disorders was measured with the question: “Have you ever been in contact with healthcare services for…” giving the following options: Depression/anxiety, Eating disorders, ADHD/ADD or similar, Autism/Asperger, Suicide attempt, Alcohol/Drug abuse. This question was new and formulated for this survey. The occurrence of NSSI was investigated with a general screening question: “Have you ever done something to purposely hurt yourself without intending to die?” This is a question included in the structural interview Self-Injurious Thoughts and Behaviors Interview—SITIB [31].

Trauma symptoms were measured by Trauma Symptom Checklist for Children (TSCC), an instrument designed to identify a broad range of trauma symptoms in children aged 8–17 years [32]. This is a widely-used self-report instrument for measuring trauma symptoms among children and adolescents [33] that has been used for adolescents up to 19 years of age [34‐36]. The instrument comprises 54 items, divided into six subscales; anxiety, depression, post-traumatic stress (PTS), dissociation, anger and sexual concerns. Answers are arranged in the scale of four options Never‐Sometimes‐Often‐Almost all of the time. Cronbach’s alpha coefficient for the subscales has been assessed to be .77 to .89 and .84 for the entire instrument [32]. There is a Swedish translation and validation for the 10–17 age group, giving a Cronbach’s alpha coefficient for the total scale of .94 with the variation of .78 to .83 for the subscales [37]. In the present study the Cronbach’s alpha was .95 for the total scale and .82 for anxiety, .88 for depression, .87 for PTS, .85 for dissociation, .84 for anger and .65 for sexual concerns.

Categorical data was presented using frequencies and cross tabulation and analyzed with Chi square test and Fisher’s Exact test using p value <.05. When comparing means such as age and TSCC, t test for independent groups was used. Percentages presented in the study relate to the number of adolescents answering the question. Missing answers in individual questions were at most 9.7%. Analyses by gender boy/girl were made but since the number of boys was very small, few statistically significant differences were found, indicating an increased risk of type II errors. Results are therefore presented divided by gender boy/girl only when statistical significance with a p value <.05 was seen. In the analyses the answer alternatives concerning living situation were merged from seven to four alternatives (living with both parents or alternating/living with one parent with or without new partner/alone or with siblings or partner/in foster care or institution), financial situation in the family from five to three alternatives (good/poor/do not know), sexual orientation from six to four alternatives (heterosexual/homosexual/bisexual/other or unsure), number of sexual partners from four to three alternatives (one, two to five and more than five). The questions concerning abortion and treatment for chlamydia were dichotomized from four to two alternatives (yes/no) and the question concerning total number of times exposed to sexual abuse was dichotomized from three to two alternatives (exposed one time/exposed more than one time). To make a model with the most important factors associated to SASI, forward stepwise binary logistic regression was performed with SASI as a dependent variable and sex, financial situation in the family, heterosexual sexual orientation, selling sex, all kinds of sexual abuse, penetrative abuse, emotional and physical abuse, trauma symptoms, healthcare for psychiatric disorders and NSSI, as covariates. All statistical analyses were carried out in Statistical Package of the Social Sciences (SPSS) version 22.

Sociodemographic data for adolescents using SASI and reference adolescents not using SASI (non-SASI) are presented in Table 1. The group of adolescents using SASI had a generally poorer family financial situation, fewer lived with both parents and more often alone, with siblings or partner, in foster care or institution. No differences were seen concerning parents being in employment, parents’ education, immigrant background, attending theoretical or practical study program or being enrolled at a school in the county of Stockholm.

Adolescents using SASI, more frequently reported sexual-minority orientation, as seen in Table 2. Only 60% of the adolescents in the SASI group had a heterosexual orientation compared to 88% in the reference group, 5.6% of the index group reported homosexual orientation, 23.8% bisexual orientation and 11.2% had another sexual orientation or were unsure of their sexual orientation.

As seen in Table 2, adolescents using SASI displayed more experiences of voluntary sexual intercourse, had more sexual partners and an earlier sexual debut, at an age of 14.6 (SD = 1.7) years compared to 15.6 (SD = 1.5) years among the non-SASI group (t = −6.11, p < .001).

Adolescents using SASI were slightly less likely to have used contraceptives at last intercourse and slightly more likely to have had an abortion, but no difference was seen for being treated for chlamydia. Selling sex was seen among 11.3% of the adolescents using SASI compared to 0.7% among non-SASI, see Table 2.

As seen in Table 3, 75.0% of the adolescents using SASI had been exposed to some kind of sexual abuse. When divided by gender this was 35.0% in the SASI group for boys compared to 9.2% in the non-SASI group and 82.8% among girls in the SASI group compared to 27.5% in the non-SASI group. Differences were especially prominent concerning penetrative abuse. Of the adolescents that had been exposed to sexual abuse, 73.1% of the SASI group had been exposed more than once compared to 54.3% in the non-SASI group (χ² = 12.2, df = 1, p < .001). The first occurrence of sexual abuse was at the younger age of 13.8 (SD = 2.8) years for the SASI group compared to 14.6 (SD = 2.7) years in the non-SASI group (t = −2.8, p = .005) and was more commonly penetrative sexual abuse [47.3% (SASI) vs 19.7% (non-SASI), χ² = 36.8, df = 1, p < .001]. The perpetrator on the first occasion was more commonly a boyfriend/girlfriend or former boyfriend/girlfriend for adolescents using SASI [25.8% (SASI) vs 11.8% (non-SASI), χ² = 14.6, df = 1, p < .001], and not a stranger [22.6% (SASI) vs 38.8% (non-SASI), χ² = 9.5, df = 1, p = .003]. For 52 (58.4%) of the 89 adolescents reporting their age at the time of sexual abuse and SASI, sexual abuse preceded the use of sex as self-injury. This figure was 81 (91%) when including adolescents reporting the same year for the first experience of sexual abuse and SASI.

Adolescents using SASI were more often exposed to some form of emotional or physical abuse, as seen in Table 3. Within the SASI group, 87.2% had been exposed to some form of emotional abuse compared to 57.1% in the non-SASI group. Exposure to some form of physical abuse was seen among 69.4% in the SASI group compared to 30.3% among peers in the non-SASI group.

As seen in Table 4, trauma symptoms measured by the subscales for anxiety, depression, post-traumatic stress, dissociation, anger and sexual concerns in TSCC were all more common in the adolescents using SASI. NSSI was seen among 65.6% in the SASI group compared to 16.6% in the group of non-SASI, see Table 5. Contact with healthcare services for depression/anxiety, eating disorders, ADHD/ADD or similar, autism/Asperger, suicide attempt, alcohol and drug abuse was sought to a much higher extent in the SASI group compared to non-SASI, as seen in Table 5. Of the adolescents using SASI, 61% had sought help for depression/anxiety, 31.4% for suicide attempt and 28.8% for eating disorders.

A forward stepwise binary logistic regression with SASI as the dependent variable was performed to find a model with the most important variables associated with the behavior. The model included in nine steps. In the final model, the most important factors associated with the behavior were selling sex, some kind of sexual abuse, penetrative sexual abuse, physical abuse, TSCC for dissociation, NSSI, healthcare for depression/anxiety and eating disorders. The value for TSCC and depression was not significant (p = .060) but was left in the statistic model. The variables of sex, financial situation in the family, heterosexual orientation, emotional abuse, trauma symptoms for anxiety, anger, PTS, sexual concerns and healthcare for psychiatric disorders for ADHD/ADD or similar, Autism/Asperger, suicide attempt and alcohol/drug abuse were all significantly associated with SASI in pairwise chi-2 statistics, but were not left in the final model of the binary logistic regression (Table 6).

The results of this study are intended to be representative of adolescents in the 3rd year of Swedish high school. It was decided to perform the study during lecture times rather than send questionnaires by mail, to improve response rate and consequently the level of representation of Swedish adolescents. However, only 59.7% of the eligible adolescents answered the questionnaire which could be compared to the previous studies from 2004 and 2009 that had a response rate of 77.2 and 60.4% respectively. Of the missing answers, approximately 10% are explained by those being absent on a regular school day. The rest of the missing answers are those that chose not to participate which might be due to poor motivation, a feeling of discomfort when answering the questions in the questionnaire, insufficient knowledge of the Swedish language or not having the ability to focus for the time needed to answer the questionnaire. Conclusively, the adolescents that did not answer the questionnaire might have been a more exposed group, thereby making our findings not representative and more likely to be underestimated than exaggerated.

The strongest limitation of the study is a lack of definition of SASI and that we do not know in which way the participants have purposely hurt themselves by using sex. The definition for using SASI is in this study self-defined and more studies are needed to confirm or reject the concept and the reported number of using it.