Engagement with Algorithm-curated Short-form Video Content and Its Association with Non-suicidal Self-harm and Personality Traits

Summary

This cross-sectional study among Saudi adolescents and young adults found high rates of self-harm, anxious traits, and probable borderline personality traits disorder indicators, with significant associations observed for algorithm-curated short-form video use and psychological distress.

Abstract

The rapid expansion of algorithm-curated short-form video (ACSFV) platforms has raised concerns about their potential association with youth mental health, particularly through personalized content streams that may amplify psychological vulnerabilities and exposure to harmful material. This study examined the relationship between ACSFV use and self-harm behaviors, probable borderline personality disorder (BPD) traits indicators, and anxious personality traits among Saudi adolescents and young adults. It also examined associations with depression, anxiety, stress, and sociodemographic factors. A cross-sectional study was conducted among Saudi individuals aged 12–30 years, utilizing self-reported data on social media use. The Taylor Manifest Anxiety Scale-50 (TMAS-50) was used to assess anxious personality traits, and the Self-Harm Inventory (SHI) was used to screen for BPD traits indicators and evaluate the lifetime history of self-harm. The questionnaire also included the Depression, Anxiety, and Stress Scale-21 (DASS-21). Among the 566 participants (74% female, 51.8% aged 19–25 years), daily engagement with ACSFV content was frequent, with 22.5% using it for 3–4 h and 15.3% for more than 4 h per day. The most frequently used platforms were TikTok (84.2%), Snapchat (80.0%), and Instagram (70.0%). Self-harm behaviors (37.5%), BPD indicators (9.5%), and severe / very severe anxious traits (31.0%) were prevalent. Self-harm correlated with age category 15–25 years, depression, and stress. Anxious traits were associated with extensive engagement with ACSFV content in unadjusted analysis, while Facebook use and higher depression and anxiety scores remained significant in adjusted analysis. BPD indicators were independently associated with psychiatric consultations and stress levels. This study highlights high rates of self-harm, anxious traits, and BPD indicators among Saudi adolescents and young adults, emphasizing the complex associations between social media use and psychological distress and the need for culturally tailored interventions, school-based programs, and further research.

Introduction

Self-harm represents a significant public health concern among adolescents and young adults, with substantial variability across populations and settings. In community-based samples, the lifetime prevalence of self-harm behaviors has been estimated at approximately 15–20%, while higher rates are consistently reported in clinical populations, particularly among individuals with mood disorders, anxiety disorders, and personality pathology^1,2. Non-suicidal self-harm is especially prevalent during adolescence, a developmental period characterized by heightened emotional reactivity and vulnerability to psychosocial stressors^3,4.

The psychological mechanisms underlying self-harm are multifactorial and include emotional dysregulation, maladaptive coping strategies, and interpersonal difficulties. Self-harm is often conceptualized as a means of regulating intense negative affect, providing temporary relief from psychological distress, or expressing internal experiences that are difficult to verbalize. Social learning processes may also contribute, particularly in the context of exposure to self-harm-related content, where normalization and behavioral modeling can occur^5,6.

Clinically, self-harm is of particular importance due to its strong association with subsequent suicidal behavior. Individuals who engage in self-harm are at significantly increased risk of future suicide attempts and suicide, making early identification and intervention critical. As such, understanding the behavioral, psychological, and environmental factors associated with self-harm, including emerging influences such as digital media exposure, remains a priority for both clinical practice and public health research^7,8.

Extensive research documents associations between digital media use and mental health difficulties among adolescents and young adults^9,10,11, as well as among those with personality traits marked by isolation and low self-control, either inhibitory or initiatory¹². Yet associations are heterogeneous across platforms and populations¹³. Particular concern has been raised around algorithm-curated short-form video (ACSFV) environments, such as the “For You” feed seen on platforms like TikTok, which is very common among youth¹⁴. ACSFV content provides highly personalized streams designed to maximize engagement^15,16. Emerging research increasingly links the use of algorithm-driven video platforms to a range of psychiatric conditions, underscoring the need to better understand their mental-health implications^17,18,19.

Beyond psychological effects, evidence indicates that specific personality traits, particularly borderline personality disorder (BPD) traits, are associated with distinctive social-media patterns²⁰. Individuals with elevated BPD traits may attribute greater importance to platforms and display usage patterns mirroring relational difficulties in social and clinical contexts²¹. Self-harm is likewise associated with social media, which can offer support yet also expose users to harmful content, heightening vulnerability⁶. This association is complex, with variable presentations that defy simple categorization^6,22.

Various theories explain these associations, including social learning, cognitive-behavioral, and media influence models²³, along with contagion theory, which highlights how self-harm may spread through exposure on social media, especially among adolescents⁵. These frameworks collectively underscore how social media may amplify vulnerabilities, particularly in adolescents, by shaping behaviors, reinforcing negative cognitions, and fostering environments that may perpetuate self-harm risks.

Social learning processes are particularly relevant in the context of digital media, where repeated exposure to self-harm-related content may contribute to behavioral modeling and normalization²⁴. The “copycat” or contagion effect refers to the phenomenon whereby individuals adopt behaviors observed in others, especially when such behaviors are perceived as relatable, rewarding, or socially validated. In social media environments, this effect may be amplified by algorithm-driven content curation, which can repeatedly expose users to similar material, potentially reinforcing maladaptive behavioral scripts. Self-harm-related content, often shared through personal narratives, imagery, or coded language, may reduce perceived stigma, increase identification with affected individuals, and inadvertently normalize self-injurious behaviors. Adolescents, in particular, may be more susceptible to these influences due to developmental factors such as heightened sensitivity to peer validation, identity formation processes, and increased emotional reactivity. However, it is important to note that such exposure does not imply direct causation; rather, it may interact with pre-existing psychological vulnerabilities, including depression and anxiety, to shape behavioral outcomes²⁵.

Given the growing influence of ACSFV environments, this study aimed to examine the associations between engagement with such content, self-harm behaviors, borderline personality traits, and anxious traits among adolescents and young adults in Saudi Arabia, while assessing the relationships with depression, anxiety, stress, and relevant sociodemographic factors. Unlike traditional social media platforms, ACSFV environments deliver highly personalized, continuously adapting streams of content that may intensify exposure to emotionally salient or repetitive material. Despite increasing global use of these platforms, there remains a lack of empirical evidence specifically examining their association with self-harm and related psychological outcomes in youth populations, particularly within non-Western contexts, which this study seeks to address.

Protocol

Design and population

This cross-sectional study targeted Saudi children, adolescents, and young adults aged 10 to 30 years. Data collection was carried out between 01 April 2023 and 31 January 2024. With a population exceeding 33 million in 2023, an annual growth rate of 3.38%, and a median age of 26.6 years²⁶, Saudi Arabia is characterized by a significant youth population. The study was approved by the Taif University Institutional Review Board (Ref. No. HAO-02-T105) and conducted in full compliance with the ethical principles outlined in the Declaration of Helsinki, local institutional guidelines, and relevant regulatory standards. Participant anonymity was strictly preserved, and all data were handled with full confidentiality.

All participants provided informed consent before taking part in the study. For individuals younger than 17 years, the legal age of consent in Saudi Arabia, written parental consent was additionally obtained in accordance with national regulations and ethical requirements. A consent statement was also embedded as a mandatory item at the beginning of the questionnaire.

Participants and sampling

A convenience sampling method was used to recruit eligible individuals across Saudi Arabia. The target sample size (N = 377) was determined to estimate the unknown prevalence of self-harm behavior, with a 95% confidence level, 80% statistical power, and a type I error rate of 0.05, based on the assumption of an infinite population. The sample size was increased by 50% to account for incomplete participation in this national study. Hence, the final target sample size was 566.

Data collection and outcome

Data were collected using a structured questionnaire comprising: (1) sociodemographic and psychiatric history variables; (2) patterns of social-media use, including average daily exposure to algorithm-curated short-form video (ACSFV) content and regular use of various platforms; (3) the Taylor Manifest Anxiety Scale-50 (TMAS-50); (4) the Self-Harm Inventory-22 (SHI-22); and (5) the Depression, Anxiety, and Stress Scale-21 (DASS-21).

Sociodemographic and psychiatric variables were collected using a structured questionnaire and categorized as follows: gender (male, female); age (categorized into <15, 15–18, 19–25, and >25 years); educational level (primary/middle, high/technical, bachelor’s degree, postgraduate); and professional status (no job activity, with job). Psychiatric history variables included prior psychiatric consultation (no/yes), history of psychotherapy (no/yes), history of psychiatric medication use (no/yes), history of psychiatric emergency visits (no/yes), and general practitioner (GP) visits for psychiatric symptoms (no/yes). Patterns of social-media use included self-identification as a regular user of various platforms, as well as average daily consumption of ACSFV content [<1 h to ≥4 h], which was defined as short-form videos or reels regardless of the platform. The TMAS-50 is a validated 50-item true/false measure of manifest anxiety as a stable personality trait (total 0–50, five categories from no anxiety [0–16] to very severe [30–50])²⁷. Although sometimes applied to transient anxiety, TMAS was used here to indicate an anxious personality trait, consistent with its construct²⁸. The SHI-22 is a 22-item yes/no tool assessing lifetime deliberate self-harm; scores ≥1 indicate self-harm, while scores ≥5 suggest probable BPD indicators, with 83.7% accuracy^6,21,22. The DASS-21 is a widely used, validated scale that quantifies depression, anxiety, and stress levels²⁹.

Primary outcomes were: (1) history of self-harm (SHI-22 ≥1), (2) probable BPD indicators (SHI-22 ≥5), and (3) anxious personality trait (TMAS-50 ≥27: severe/very severe). Independent variables included patterns of social-media use (TikTok status and duration; other platforms), sociodemographic characteristics, and DASS-21 depression, anxiety, and stress scores.

Procedure

The researchers assembled a team of senior medical students, who were recruited from various medical schools across Saudi Arabia and trained to explain the study, obtain informed consent prior to participation, and manage and coordinate data collection at the national level. Participants were approached in public places, including shopping malls, parks, and clinical settings. For participants younger than 17, the legal age of consent in Saudi Arabia, an additional parental consent form was included to meet ethical guidelines. The consent statement was further included in the questionnaire as a conditional item before completing the questionnaire. Participants were recruited from public and clinical settings and provided with a barcode linking to an online, self-administered questionnaire on SurveyMonkey. To guarantee data quality and completeness, participants were required to complete all sections of the survey. All responses were securely recorded, with respect for anonymity, privacy, and autonomy.

Statistical methods

Descriptive statistics were generated for demographic data, psychiatric history, and scale scores. Internal consistency of the scales was assessed using Cronbach’s alpha. Spearman’s rho was used to analyze bivariate correlations between SHI, TMAS, and DASS-21 scores. Factors associated with self-harm, probable BPD traits indicators, and anxious personality traits were analyzed using chi-square test, Fisher’s exact test, or independent-samples t-test, as applicable. Unadjusted and adjusted logistic regression models were used to analyze factors independently associated with the three outcomes. Adjusted models used Entry mode to include all variables that showed statistical significance in the unadjusted models. Statistical significance was set at p < 0.05.

Results

Participants’ characteristics and levels of social media use
Among 601 individuals approached, 566 (94.2%) consented to participate in the study, while 35 (5.8%) declined. Most participants were female (74.0%), aged 19–25 years (52.1%), and held a Bachelor's degree (60.6%). Regarding psychiatric history, 19.3% had consulted a psychiatrist, 12.5% received psychotherapy, 12.4% were on psychiatric medication, and 9.2% had sought psychiatric care from a family doctor (Table 1).

Discussion

This study examined the associations between engagement with ACSFV content and self-harm behaviors, anxious traits, and borderline personality features among adolescents and young adults in Saudi Arabia. A high prevalence of self-harm behaviors was observed, and these behaviors were strongly associated with measures of psychological distress, particularly depression and stress. While higher levels of ACSFV engagement were associated with self-harm and anxious traits in unadjusted analyses, these relationships were attenu...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Questionnaire / Scale	Structured sociodemographic and psychiatric history questionnaire	Study investigators	Custom-designed
Questionnaire / Scale	Taylor Manifest Anxiety Scale-50 (TMAS-50)	Taylor JA	50-item version
Questionnaire / Scale	Self-Harm Inventory-22 (SHI-22)	Sansone et al.	22-item version
Questionnaire / Scale	Depression, Anxiety, and Stress Scale-21 (DASS-21)	Lovibond & Lovibond	21-item version
Data Collection Tool	Online self-administered questionnaire platform: SurveyMonkey	Not specified	Momentive
Recruitment Material	Barcode / QR code link	Study investigators	Not applicable
Equipment	Computers / mobile electronic devices	Not specified	Not specified
Statistical Software	SPSS for Windows	IBM Corp.	Version 21
Statistical Method	Cronbach’s alpha analysis	Implemented in SPSS	Version 21