A new social media anxiety study 2025 readers should know about is a 2024 paper on adolescents receiving intensive psychiatric care, and it found that the most problematic, cybervictimisation-laden patterns of use cluster sharply in the teens who are already inside the hospital walls. social media anxiety is the cluster of worry, comparison and physiological arousal that builds in people whose use of platforms like Instagram, TikTok and Snapchat starts to feel out of their control. The inpatient finding is striking. It is also exactly the kind of result that needs careful handling, because the sample was selected on already-severe distress.
This brief summarises what the study reports, where it sits in the 2025 evidence pool, and the limitations that any honest read has to keep in view.
What the study reports
The centerpiece is a 2024 paper, Problematic and Harmful Social Media Use among Adolescents Receiving Intensive Psychiatric Care, published in an open-access international public-health journal and archived at PubMed Central as PMC11507369. The authors compared adolescents enrolled in intensive psychiatric treatment with peers drawn from a general-population sample of similar age, asking how often each group met thresholds for problematic social media use and for being targeted by online aggression. For context across the wider anxiety literature, see also our deeper biggest-findings roundup.
The headline result: inpatient adolescents reported substantially higher rates of problematic social media use than community peers, and they were also more likely to report having been cybervictimised in the past year. Both measures, the behavioural pattern and the exposure to online aggression, moved in the same direction at the same time.
Methodology
The design was cross-sectional and observational. Two adolescent samples were compared:
- A clinical sample of adolescents (roughly age 12 to 18) admitted to intensive psychiatric care, including inpatient and intensive outpatient programmes, at participating Dutch treatment centres.
- A matched general-population comparison sample of adolescents in the same age range, drawn from a national school-based survey.
Problematic social media use was measured with a validated multi-item scale that captures loss of control, preoccupation, conflict, and use as escape, the same family of instruments used across the 2025 meta-analytic literature. Cybervictimisation was measured with a separate self-report scale covering experiences of online harassment, exclusion and threats in the prior twelve months.
The clinical sample size sat in the low hundreds; the general-population comparison was larger. Differences between groups were tested with standard between-group statistics and adjusted for age and sex.
Findings
Three findings carry the brief.
- Inpatient teens reported the heaviest problematic use. A markedly larger share of adolescents in intensive psychiatric care met the threshold for problematic social media use than peers in the community sample. The pattern held after adjustment for age and sex. This is consistent with the broader 2025 message that problematic use, not raw screen time, is the variable that tracks distress most tightly. See our signs and symptoms guide for the items most commonly used in these scales.
- Cybervictimisation tracked clinical status as well. Adolescents in intensive psychiatric care were more likely to report having been targeted online than community peers. Cybervictimisation correlated with problematic use within the clinical sample, suggesting that the two often co-occur in the same young people rather than being independent risks.
- Co-occurrence with internalising symptoms was high. Within the clinical sample, both problematic use and cybervictimisation showed associations with self-reported anxiety and depression, the kind of internalising-symptom profile that brings teens into intensive care in the first place.
Context, how it fits the 2025 evidence pool
The 2024 inpatient paper does not stand alone. It joins a year-on-year build-up of evidence showing that the relationship between social media and adolescent mental health is concentrated in subgroups, not spread evenly across all teens. A separate 2024 analysis in Social Media Use and Serious Psychological Distress Among Adolescents (PMC11120505) reported similar concentration in the small share of adolescents reporting serious psychological distress. The 2023 US Surgeon General’s advisory had already flagged the same gradient, vulnerable adolescents disproportionately exposed to harmful content and online aggression. None of this is new in direction. What the 2024 inpatient study adds is a sharper picture at the top of the severity curve: the teens who actually end up in intensive care look very different on these variables from the general adolescent population.
That nuance matters because much of the public debate has been about the average teen. The 2025 evidence base now points toward a more useful framing: the average effect of social media on anxiety is small; the effect inside vulnerable subgroups is much larger and clinically meaningful. Readers exploring what the evidence does and doesn’t justify will also find our overdiagnosis debate piece useful.
Limitations
At least three limitations bear directly on how the inpatient finding should be read.
- Selection on distress. The clinical sample was, by definition, drawn from adolescents already in intensive psychiatric care. They are not a random slice of teen life; they are a slice selected on the very outcomes the study then measures. That makes the comparison informative, but the inpatient/community contrast cannot tell us whether social media drove these teens into care, whether being in care reshaped their online lives, or both. Our treatment protocols overview covers what evidence-based care for severely affected adolescents actually looks like.
- Cross-sectional design. Use, victimisation and symptoms were all measured at the same time. The data cannot establish temporal order. Anxious teens may turn to social media in ways that intensify problematic use; intensified use may then worsen anxiety; cybervictimisation may sit upstream, downstream or both. The paper is honest about this; readers should be too.
- Self-report and recall. Both problematic-use and cybervictimisation measures rely on adolescents reporting on themselves. Teens in active psychiatric distress may report differently from community peers, not necessarily inaccurately, but with different reference points. The paper’s effect sizes are large enough that this is unlikely to flip the direction of the finding, but it does temper how precisely we can interpret the gap.
To these we should add the wider caveat that the inpatient sample is Dutch and modest in size. Replication in other health systems, and inside US inpatient populations specifically, will strengthen the case.
What it means for readers
For parents of a hospitalised or intensively treated teen, the practical message is direct: problematic social media patterns and exposure to online aggression are common in that group, and they are worth bringing into the treatment conversation rather than treating as an unrelated background activity. Clinicians on the wards already know this; the 2024 inpatient paper gives them a published number to point at when families ask. Mothers and other caregivers navigating these conversations may also find our piece on how parental and adolescent anxiety patterns differ useful for setting expectations.
For the rest of us, readers in the general anxiety conversation, the finding is a quiet correction. The 2025 evidence is not saying every teenager is being harmed at scale. It is saying the harm is concentrated, the cause-and-consequence arrows still cross both ways, and the teens at the top of the severity curve are using these platforms in patterns that look qualitatively different. That is a more honest picture than the one the headlines usually carry, and a more actionable one if you happen to be the parent, partner or clinician of someone in that group.