Self-Harm Content on Social Media and Proximal Risk for Self-Injurious Thoughts and Behaviors Among Adolescents

Abstract

Objective

Exposure to self-harm content may be an important experience on social media that confers risk for self-injurious thoughts and behaviors (SITBs). The current study used an intensive monitoring design to examine the relation between weekly exposure to self-harm content on social media and adolescent SITBs, including suicidal ideation and nonsuicidal self-injurious (NSSI) urges and behaviors.

Method

Adolescents (N = 61; ages 14-17 years) recruited in the United States (49% girls, 62% LGBTQ+, 10% Asian, 20% Black, 16% Latine, 13% Multiracial, 41% White) completed 8 weeks of daily and weekly surveys. Daily surveys included questions about adolescents’ suicidal ideation and NSSI urges and behaviors. Weekly surveys included exposure to self-harm content on social media and perceived daily social media hours. Logistic multilevel modeling was conducted to evaluate whether exposure to self-harm content on social media was associated with weekly SITBs, controlling for social media duration and depression symptoms.

Results

Overall, 50% (n = 31) of adolescents reported seeing self-harm–related content on social media over the study period. There were significant associations between weeks of self-harm social media exposure and weekly NSSI urges and behaviors. There was no association between weekly social media self-harm exposure and suicidal ideation that week. Perceived social media use duration was not associated with SITBs.

Conclusion

Findings indicate that exposure to self-harm content on social media may be a proximal risk factor for NSSI urges and behaviors among adolescents. Findings shed light on one modifiable way in which social media may heighten risk for SITBs among adolescents, lending empirical support to current guidelines about limiting self-harm content on social media.

Diversity & Inclusion Statement

We worked to ensure sex and gender balance in the recruitment of human participants. We worked to ensure race, ethnic, and/or other types of diversity in the recruitment of human participants. We worked to ensure that the study questionnaires were prepared in an inclusive way. One or more of the authors of this paper self-identifies as a member of one or more historically underrepresented racial and/or ethnic groups in science. We actively worked to promote sex and gender balance in our author group. We actively worked to promote inclusion of historically underrepresented racial and/or ethnic groups in science in our author group. While citing references scientifically relevant for this work, we also actively worked to promote sex and gender balance in our reference list. While citing references scientifically relevant for this work, we also actively worked to promote inclusion of historically underrepresented racial and/or ethnic groups in science in our reference list.

Plain language summary

This study examined whether exposure to self-harm content on social media impacts teens’ self-injurious thoughts and behaviors, using intensive monitoring data. There were no direct links to suicidal thoughts; however, teens who reported exposure to self-harm content were more likely to have nonsuicidal self-injury urges and behaviors that week. There was no effect of screen time. Results indicate that self-harm exposure on social media is related to self-harm in teens, highlighting the importance of asking teens about self-harm exposure and policies geared towards content versus screen time.

Suicide is a leading cause of death among adolescents between ages 10 and 19.¹ Rates of self-injurious thoughts and behaviors (SITBs), including suicidal ideation (SI), suicidal behavior, and nonsuicidal self-injury (NSSI), are highly prevalent among adolescents² and have increased in recent years.^1,3 Although studies often focus on the potential role of social media in conferring risk for SITBs among adolescents,⁴ it remains unclear which specific aspects of social media may be most central in these associations. Identifying key components of social media that are linked with SITBs is critical to inform actionable targets for prevention and intervention, in contrast to current policies and practices mainly focused only on reducing social media access or screen time.

Despite the focus on how much adolescents use social media (eg, screen time) and mental health,⁵ the experiences that teens have on social media may be more important than frequency of use.^6,7 In particular, self-harm content may be particularly important to consider in the relation between social media and risk of SITBs. Self-harm–related content can be highly variable in terms of its content or mode. For example, some posts may be visual in nature (eg, images or videos) and portray someone engaging in self-harm methods such as cutting. However, this content also may be posted as text describing how someone may want to engage in self-harm. These unfiltered posts can make such content easily accessible to youth across different platforms, including those that are most popular among youth (eg, TikTok, Instagram, Snapchat).⁸ A recent meta-analysis of research on this topic found medium-to-large effects of any self-harm content exposure or engagement on social media and SITBs among adolescents and adults (9 studies). Further, individuals who were exposed to SITB content were 2 times more likely to have SI and 3 times more likely to have engaged in NSSI.⁵ One cross-sectional study conducted with inpatient adolescents (ages 12-17) found that higher levels of NSSI exposure on social media were associated with higher NSSI engagement, with 87% of participants reporting exposure to NSSI on social media before NSSI engagement.⁹ Few studies have examined these associations longitudinally, with 1 study finding that discussion forums (on social media) were linked to higher likelihood of SI a year later in adolescents and young adults (ages 14-24).¹⁰ Only 1 study to date has examined these associations on a more proximal time scale,¹¹ finding that frequency of exposure to self-harm content (specifically on Instagram) was associated with SI and self-harm in the next month; however, this study was conducted among young adults. Overall, findings highlight the potential importance of examining self-harm content on social media as a potential risk factor for SITBs in adolescents, and the clear need to conduct rigorous research assessing these relations.

Given the fluctuating nature of SI and NSSI over days and weeks,¹² it is important to examine risk factors that occur more proximally to the occurrence of SI and NSSI to inform actionable targets for SITBs. For instance, SI is not static, but changes across days and weeks among adolescents.¹³ Social media use and experiences also fluctuate over time,¹⁴ and experiences such as exposure to self-harm content may vary over time. Intensive monitoring approaches, such as daily diaries and ecological momentary assessment (EMA), repeatedly assess individuals’ experiences in real time¹⁵ and may be particularly well suited to examine the proximal relations between self-harm content on social media and SITBs. Whereas studies using intensive monitoring methods have emerged to examine predictors of SITBs, particularly SI and NSSI,16, 17, 18, 19, 20 this is the first study to our knowledge applying these methods to examine self-harm content on social media related to SI and NSSI (urges and behaviors) among adolescents.

Current Study

The current study examines the relation between exposure to self-harm–related content on social media and both SI and NSSI in adolescents using a longitudinal monitoring design. Specifically, this study evaluated whether weekly reported exposure to self-harm content on social media is associated with the occurrence of SI and NSSI, including both NSSI urges and behaviors, that occurred in the past week, while controlling for teens’ perceptions of social media use duration and average depressive symptoms that past week. It was hypothesized that exposure to weekly social media self-harm content would be associated with a greater likelihood of weekly SI and NSSI behaviors and higher levels of NSSI urges. Overall, this study aimed to investigate the nature and impact of self-harm–related content on social media in adolescent risk for suicide and self-harm, which may better inform how and when social media may confer risk among adolescents.

Method

Participants and Procedures

Participants were recruited from social media as part of an ongoing study examining pathways linking social media and teen mental health outcomes. Eligibility criteria included being between the ages of 14 and 17, being in grades 9 to 12, using social media, having fluency in English, and living in the United States. Due to restrictions with 1 phone application required for the larger study goals of smartphone sensing, only adolescents who used Android phones were eligible to participate. Teens who did not reside in the United States at the time of the study, did not have an Android phone as their primary phone, and/or had a parent/legal guardian who was not fluent in English to be able to consent were not eligible for the study. All study procedures, including the screener and study visits, were conducted online through Zoom meetings and remote data collection methods such as Qualtrics surveys and smartphone-based surveys. This design allowed adolescents across the United States to participate, which broadened access to a range of diverse geographic and socioeconomic backgrounds.

Social media ads (eg, “help us learn about teen social media use and mental health”) were posted on the laboratory Instagram and Facebook (Meta accounts, with links to the screener. Flyers with QR code links to the screener also were posted in public spaces such as cafes, libraries, and malls where teens were likely to visit, based on advice from the Youth Advisory Board of our laboratory. Given algorithms inherent in social media, adolescents most affected by social media or experiencing heightened suicide risk may have been more likely to engage with study advertisements, potentially leading to a sample that is more representative of these high-risk groups. This selection bias should be considered when interpreting the findings.

After completing the online screener, eligible participants were contacted directly by the study team to schedule a virtual informed consent call (eg, Zoom or phone). Participants and their parents or guardians met with the research team to learn about the study procedures and sign consent and assent forms if they were interested (and after eligibility was confirmed). On completion of consent and assent, participants then completed the first study session, which included downloading required study applications (eg, MetricWire) and a clinical interview (Columbia–Suicide Severity Rating Scale [C-SSRS]) with trained study staff. Participants also completed an online baseline questionnaire through Qualtrics that included demographic information and experiences related to social media and mental health.

Adolescents then began the 8-week intensive monitoring portion of the study, which included surveys 3 times/day (morning, midday, evening) and weekly, followed by a final study visit (eg, questionnaires and interview) at the completion of the intensive monitoring portion. The current study included information only from the evening and weekly surveys, in which adolescents reported daily SI and NSSI urges and behaviors, and weekly surveys of self-harm content exposure. Participants were compensated for their time through gift cards, with payment based on adherence of study procedures. Parents were not compensated for any portion of the study, as they were involved only in providing consent for their child’s participation. All procedures were approved by the local institutional review board.

The first study visit was completed by 92% (67 of 73) of eligible participants who met with the study team for the consent call, and 85% (62 of 73) continued to the intensive monitoring portion of the study. One participant did not complete any weekly surveys, resulting in a final sample of 61 adolescents for this study. Further information about the study procedures, recruitment, and attrition is described elsewhere.²¹

Safety Procedures

To ensure participant safety, all participants received crisis resources throughout the study, including in all surveys. Risk alerts were automatically sent to notify the study team if participants reported SI, and participants (and their parents/legal guardians) were contacted to assess and ensure safety as needed. In addition, all participants completed an online safety plan to identify coping strategies, national and local professional resources, social support sources, and means restriction methods to use during crises. During safety check-ins, trained team members reviewed participants’ safety plans, assessed suicide risk, and discussed a plan to ensure commitment to safety with the participant and their parent/guardian, pending severity.

Daily SITBs

Daily evening surveys assessed whether adolescents had active suicidal thoughts that day (“Today, I thought about killing myself”), with responses of any active suicidal thoughts that day coded as no SI (0) or SI present (1). Evening surveys also examined whether or not adolescents performed NSSI that day (“Today, I hurt myself on purpose, but was not trying to kill myself.”), with responses of yes (1) or no (0), and the intensity of the NSSI urge that day (“How intense was your urge to hurt yourself on purpose today without trying to kill yourself [whether you did it or not]?”) with a sliding scale from 0 (not at all) to 100 (extreme). These items have been used in prior EMA research.²² Any occurrence of SI or NSSI behaviors (0 or 1) and maximum NSSI urges reported on evening surveys were used to reflect weekly SI or NSSI.

Daily Depression Symptoms

Participants reported their depressed mood (“Currently, how much are you feeling down, depressed, or hopeless?”) on a scale ranging from 0 (not at all) to 10 (the worst it gets) for each EMA prompt. Depression symptoms were averaged across the week for each participant to create a weekly average of depression symptoms, which was included as a covariate.

Exposure to Self-Harm Content on Social Media

Exposure to self-harm content on social media was measured in weekly surveys. Participants were asked to respond to the question: “On social media, did you post or see others post about suicide or self-harm this week?” (with options of “No”; “Yes, I posted”; and “Yes, others posted”). Participants could select both “Yes, I posted” and “Yes, others posted” if both options applied to them. For this study, exposure to self-harm content on social media was coded as yes (1) if “Yes, others posted” was selected and no (0) if this response was not selected. Given the limited number of responses endorsing direct engagement (15 participants), only exposure to self-harm on social media was included in the current study.

Perceived Social Media Duration

Weekly surveys also asked participants to report average daily social media hours over the past week. Participants were asked to respond to the question: “On a typical day this week, how many hours per day did you spend on social media?” (with answers ranging from “0 minutes” to “24 hours”). The first 4 options gave a 30-minute block in between the duration (0 minutes, 30 minutes, 1 hour, 1.5 hours), and all durations after that were in a 1-hour block (2 hours, 3 hours, until 24 hours). This was included as a covariate in the current study.

Lifetime History of SI and NSSI

The C-SSRS clinical interview²³ was conducted by trained interviewers and used to assess SI and NSSI. In the current study, lifetime SI and NSSI (presence/absence) was used to better characterize the sample.

Analytic Plan

To evaluate whether exposure to self-harm–related social media content was associated with the occurrence of SITBs (SI, NSSI urges, NSSI behaviors) in the past week, logistic and linear 2-level multilevel modeling with random intercepts were conducted using R 4.3.1 lme4 package (https://www.r-project.org/). Weekly-reported daily social media use (hours estimated for daily use) and average weekly depression symptoms were included as covariates for within-person analyses. All primary variables were repeated weekly (level 1), nested within person (level 2). The occurrence of SI and NSSI behaviors in the past week (0 = none; 1 = present) and maximum NSSI urge intensity in the past week were included as primary outcomes. Variability in study outcomes was calculated using intraclass coefficients) to indicate variance at between- and within-person levels.

Results

Participants

The study sample comprised 61 participants with a mean (SD) age of 16.16 (0.97) years; 49% identified as female, and 62% identified as LGBTQ+ (lesbian, gay, bisexual, transgender, queer, and others), including 36% identifying as transgender or nonbinary). Participants indicated racial/ethnic identity as follows: 10% Asian, 20% Black, 16% Latine, 13% multiracial, and 41% White. Table 1 lists all demographic information about the sample as well as study variables. Most participants completed 7 weekly surveys (median = 7.00; mean [SD] = 6.40 [2.18]), with a total of 398 weekly surveys completed across all participants. Of participants, 46% completed weekly surveys on all 8 weeks, and 80% completed at least 5 weekly surveys. Based on the clinical interview, 36 (59%) participants endorsed SI in their lifetime, and 31 (50%) endorsed a lifetime history of NSSI.

Table 1.

Sample Characteristics (N = 61)

Variables	Values
	n	(%)
Gender
Female	30	(49)
Male	31	(51)
Nonbinary, transgender, or genderqueer	22	(36)
Racial/ethnic identity
Asian	6	(10)
Black or African American	12	(20)
Hispanic or Latine	10	(16)
Multiracial	8	(13)
White	25	(41)
	Mean	(SD)
Age, y	16.16	(0.97)
Grade	10.90	(0.97)
Perceived socioeconomic statusa	5.25	(1.78)
	n	(%)
Study variables
Lifetime SI, clinical interview	36	(59)
Lifetime NSSI, clinical interview	31	(51)
Self-harm SM exposure, any	31	(51)
	Mean	(SD)
SM daily duration, h	5.66	(3.32)
	n	(%)
SI, any	29	(47)
NSSI behaviors, any	13	(21)
NSSI urges, any	34	(55)
	Mean	(SD)
NSSI urges, meanb	2.27	(2.01)

Note: NSSI = nonsuicidal self-injury; SI = suicidal ideation; SM = social media.

^aPerceived socioeconomic status is assessed using the MacArthur Perceived Social standing scale, which ranges from 1 (worst off) to 10 (best off).

^bNSSI urges (mean) are for participants who endorsed having any urges (n = 34) on a scale of 0 (none) to 10 (most).

Overall, about 50% (n = 31) of adolescents reported seeing self-harm–related content on social media over the study period, with 18% reporting exposure in 1 week, 14% reporting exposure in 2 or 3 weeks, and 18% reporting exposure in 4 weeks or more. There were no significant differences in frequency of self-harm exposure on social media among adolescents with a prior lifetime history of NSSI (t = 1.92, p = .06) or SI (t = 1.26, p = .21) compared with adolescents without a history of NSSI. However, participants with a history of NSSI were more likely to see any self-harm exposure (compared with none) over the 8-week period than participants without prior NSSI (χ² = 4.74, p = .02), but there were no differences for adolescents with and without lifetime SI (χ² = 2.12, p = .14). Among adolescents who endorsed any exposure to self-harm content over the study period, exposure was reported an average of 3 out of 8 weeks (range 1-7 weeks). Across all participants, social media self-harm exposure was endorsed on 97 weeks (24%) out of the 398 total weeks, with only 15 reports of direct engagement (posting) of suicide or self-harm content among 7 participants.

A total of 29 (48%) participants reported having active SI at any time across the intensive monitoring portion of the study, with 13 (21%) adolescents engaging in NSSI behaviors at least once and 34 (55%) having any NSSI urges during the study period (mean = 2.27 out of 10). Most adolescents who engaged in NSSI (n = 10) also endorsed having SI over the study period, whereas 18 (30%) participants reported SI, but no NSSI, and 30 (49%) participants did not endorse any SI or NSSI over the study period. Adolescents reported a mean (SD) of 5.66 (3.32) daily hours (range 15 minutes to 17 hours) using social media as reported on the weekly survey. Of note, there was significant within-person variability in weekly suicidal thoughts (57% within), NSSI urges (49% within), and NSSI engagement (76% within). There also was variability at the individual level in adolescents’ self-harm exposure on social media (51% within).

Weekly Self-Harm Exposure and SITBs

In terms of the results for primary analyses of weekly self-harm exposure on social media and SITBs, there were significant effects for NSSI only (Table 2). Specifically, there were significant associations between weeks of self-harm–related social media exposure and both NSSI urges (B = .81, SE = 0.30, p = .01) and NSSI engagement (B = 2.15, SE = 0.96, p = .03), such that adolescents were 8.60 times more likely to have NSSI on weeks of self-harm social media exposure. There was no association between weekly social media self-harm exposure and SI that week (B = −.80, SE = 0.59, p = .17), controlling for self-reported social media hours and weekly average depression symptoms. Although weekly depression symptoms were associated with SI and NSSI urges each week, there was no significant association between perceived social media duration and SITBs (SI: B = −.07, SE = 0.08, p = .34; NSSI behavior: B = −.14, SE = 0.14, p = .33; NSSI urges: B = .01, SE = 0.04, p = .72).

Table 2.

Social Media (SM) Self-Harm Exposure and Weekly Self-Injurious Thoughts and Behaviors

Predictors	Weekly SI			Weekly NSSI			Weekly NSSI Urges
	Odds Ratios	CI	p	Odds Ratios	CI	p	Estimates	CI	p
Intercept	0.06	0.01 to 0.25	<.001	0.00	0.00 to 0.04	.001	0.25	−0.38 to 0.88	.43
SM self-harm exposure weekly	0.45	0.14 to 1.41	.17	8.60	1.31 to 56.34	.03	0.81	0.22 to 1.39	.007
SM hours weekly	0.93	0.80 to 1.08	.34	0.87	0.67 to 1.15	.33	0.01	−0.06 to 0.09	.72
Depression weekly	1.50	1.13 to 2.00	.005	1.24	0.74 to 2.06	.41	0.45	0.31 to 0.59	<.001
Random effects
σ2	3.29			3.29			2.93
τ00	6.88			26.11			1.98

Note: NSSI = nonsuicidal self-injury; SI = suicidal ideation.

Discussion

The current study examined the relations between exposure to self-harm content on social media and SITBs, including SI and NSSI (urges and behaviors) using an intensive, longitudinal design and within-person analyses. Overall, results indicate that during weeks adolescents were exposed to self-harm content on social media, they experienced more urges to engage in NSSI and were more likely to engage in NSSI, highlighting the importance of understanding the relation between teens’ specific social media experiences and their relation to SITBs. No association was found between weekly exposure to self-harm content and SI in the same week. There also were no associations between perceived weekly social media use duration and SITBs, including SI or NSSI urges or behaviors.

This is the first study to demonstrate that self-harm content on social media was proximally associated with both NSSI urges and engagement among adolescents with diverse identities. The findings of the current study extend the existing, mainly cross-sectional, research on the relation between self-harm content on social media and risk for SITBs⁴ by examining these constructs on a weekly basis. Further, the within-person analytic approach allows us to disentangle effects of self-harm content at the individual level (eg, weeks when individual teens were exposed to more self-harm content on social media) compared with relying on only nomothetic approaches that assess between-person differences (eg, some teens are just more likely to be exposed to self-harm content and engage in NSSI). Our intensive and within-person approach meaningfully extends past research and points to a proximal association at the individual level, emphasizing the importance of using these approaches and assessing self-harm exposure on social media (and NSSI urges and behaviors) on an ongoing basis among adolescents.

There are several reasons why self-harm content on social media may be associated with weekly NSSI urges and behaviors. First, adolescents are uniquely susceptible to social influence from peers.²⁴ It is possible that exposure to such content on social media may heighten an adolescent’s own thoughts and behaviors about NSSI or increase their perception that other teens are engaging in it more than they are,²⁵ thereby normalizing NSSI. Exposure to self-harm content that glorifies self-harming behaviors or describes them in detail also may put vulnerable adolescents at higher risk of modeling these behaviors,²⁶ particularly given the highly visual nature of social media. Exposure to self-harm content on social media also may be perceived as acutely distressing by some adolescents, leading adolescents to have more urges to engage or even to engage in NSSI to regulate emotional distress.²⁷ Yet, this study did not examine the temporality, specific types of self-harm content (eg, visual, text, individual engaging in self-harm or promoting recovery), or type of exposure (eg, incidental or intentional). In this sense, it is possible that teens actively sought self-harm content when NSSI thoughts or urges emerged or after engaging in NSSI that week or to gain social support or resources from a community struggling with self-harm.²⁶ Future research studies examining the temporality of these associations on a more fine-grained scale (daily or momentary) will be critical to gain insight into how teens are interacting with this content and when it may be helpful or harmful for teens.

The lack of association between weekly self-harm content exposure on social media and SI that week may shed light on the specificity of self-harm content and SITBs. For example, it is possible that content related to NSSI shows up more frequently than content specifically related to SI or suicidal behavior or that only exposure to suicide-related content would be associated with active SI. Further, the lack of association between perceived weekly social media use duration and SITBs may indicate that time spent on social media is potentially less relevant to SITBs, with the experiences that adolescents have during their social media use being more salient. However, participants were asked about their perceived social media use duration without any formal definition of social media sites, apps, or programs, which means that these estimates could vary greatly across adolescents and introduce variability into how our participants defined social media use (or even content they saw on social media). Given bias in self-reported estimates of social media use,²⁸ it is important that research applies objective metrics to examine these potential relationships. However, findings of the current study indicate that perceived social media use may be less important to target to prevent SITBs within clinical practice and policy than exposure to self-harm content or other negative social media experiences.²¹

There are several study limitations that raise important points and that can be used to inform future directions. First, social media algorithms may have impacted the likelihood of participants being exposed to self-harm content, with adolescents who had previously engaged with it potentially being more likely to be exposed to similar content later. Indeed, adolescents with prior NSSI were more likely to see self-harm exposure at any point in the study period than adolescents without NSSI, and results should be interpreted with this in mind. However, the within-person analytic approach allowed us to parse out changes in NSSI urges and engagement as an adolescent’s exposure to self-harm content changed across weeks in the study, slightly mitigating this concern. Whereas SI and NSSI were assessed daily under an EMA paradigm, exposure to self-harm content on social media was assessed only once per week, which was initially designed to reduce participant burden within daily prompts. However, daily or momentary assessments using EMA (ie, multiple surveys within a day) may better elucidate the nuanced, proximal relations between this type of exposure and SITBs.

The current study also was limited by a lack of detail in assessing whether self-harm exposure was related to content addressing SI, SITBs, or NSSI or whether the participants were able to specify the platforms on which they experienced exposure to self-harm content. Further, the repeated self-harm content exposure question did not pinpoint the type of self-harm content to which participants were exposed (ie, content related to support, recovery, and advocacy vs endorsement of self-harming behaviors or self-harming behaviors described and displayed in detail). We also did not gather information about how participants engaged with self-harm–related content on social media, even if after exposure (eg, like, share, post). Thus, future research should apply these methods using a more fine-grained assessment of self-harm content, including both exposure and engagement, to examine the temporality and nature of these associations more fully.

Although our sample was small, it was highly diverse across race, gender, and sexual orientation, which is a population typically underrepresented in research on both social media and SITBs. However, this also may limit generalizability of study findings to other adolescents. In addition, there was a higher proportion of youth with SI and NSSI in their lifetime and over the study period, which is higher than would be expected from a community sample. While our recruitment methods through social media did not intentionally target youth identifying as LGBTQ+ or at higher risk for SITBs, it is possible that social media algorithms drove which adolescents saw our ads and were most interested in them, potentially biasing our sample toward adolescents most affected by social media and SITBs. This warrants careful interpretation of the study findings and its generalizability and underscores the need for transparency in social media–based algorithms, which may enable better understanding of how recruitment via social media affects the self-selection of study samples. Additionally, the inclusion of Android-only users may have biased our sample, especially given that adolescents are more likely to own iPhones according to recent surveys.²⁹ However, Android phones are more accessible and available at a broader price point, potentially enhancing the inclusion of youth from diverse socioeconomic backgrounds and geographic locations across the United States. Future research should examine these associations in a larger, diverse, and more representative sample, as well as in a clinical sample of youth and youth without such history of SI and NSSI to enhance study generalizability. Despite these limitations, our diverse sample is a strength and offers insights into an important population in which to understand risk for SITBs.

The current study was strengthened by a robust, longitudinal, within-person design, which allowed for repeated assessment of exposure to self-harm content on social media, SI, and NSSI urges and behavior. This design may have reduced participant recall bias and improved accuracy of responses compared with studies relying on self-reported surveys assessed over longer time periods. This study provides a deeper look into the relation between exposure to self-harm content on social media and SITBs, with findings indicating that during weeks of self-harm exposure on social media, adolescents were more likely to have both NSSI urges and NSSI engagement. Indeed, the American Psychological Association recently issued recommendations for healthy social media use, which include limiting exposure to and engagement with self-harm content among adolescents.³⁰ Our findings support these recommendations and highlight their potential proximal associations with SITBs. Interventions designed to address and reduce NSSI behaviors or youth self-harm more broadly should incorporate assessment of and conversations about self-harm exposure on social media. Clinicians may benefit from targeted questions about social media exposure and engagement with this content, as adolescents are less likely to approach clinicians about these experiences unless directly asked.³¹ Resources about safe engagement with social media about self-harm should be provided (see #chatsafe)³² by clinicians, parents, and educators, which may further enhance accessibility of this information for young people. Finally, conversations about social media should focus more on reducing self-harm content exposure, as well as engaging teens in healthy ways of interacting with this content, rather than focus only on simplistic or reductionist approaches of eliminating screen time.

CRediT authorship contribution statement

Jessica L. Hamilton: Writing – review & editing, Writing – original draft, Supervision, Resources, Project administration, Methodology, Formal analysis, Data curation, Conceptualization. Srushti Untawale: Writing – review & editing, Writing – original draft, Conceptualization. Maya N. Dalack: Writing – original draft, Data curation. Athena B. Thai: Writing – original draft. Evan M. Kleiman: Writing – review & editing, Formal analysis. Aijia Yao: Writing – original draft.