Adolescents who spend more than three hours a day on Facebook, Twitter, Snapchat, or other social media are at increased risk for depression, anxiety, or other internalizing problems, new research suggests.
The study did not find an association between social media use and externalizing problems such as physical aggression and impulsivity — but there was an association with such problems comorbid with internalizing symptoms.
“Social media has a lot of benefits; for example, it offers users an opportunity to connect with other people on lots of different topics and with people from all over the world, but there appear to be associated harms,” lead author Kira E. Riehm, a PhD student in the Department of Mental Health, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, Maryland, told Medscape Medical News.
“I think that trying to balance the two of those will be really important in the future,” she added.
The study was published online September 11 in JAMA Psychiatry.
The authors note that depression is on the rise among American teens, and rates of deaths by suicide and suicide attempts in this population have increased sharply over the past two decades.
These days, most adolescents have a smart phone, making it easy for them to access social media platforms at any time. But while some experts blame the rapid rise in social media use for increased depression rates, the authors point out that evidence for this in nationally representative samples is scarce.
To date, most studies looking at the impact of social media on mental health are cross sectional, which presents significant limitations because of the possibility for reverse causation, said Riehm.
“It might be that instead of social media leading to more mental health problems, people with mental health problems are more likely to use social media.”
Unlike previous studies, Riehm and colleagues used longitudinal data collected at multiple time points, which helped determine that the social media use came before mental health problems.
The team used a representative sample of 6595 US adolescents who participated in the Population Assessment of Tobacco and Health (PATH) study. Researchers gathered data from this survey in 1-year intervals, starting with wave 1 from September 12, 2013, to December 14, 2014, when study participants were aged 12 to 15 years.
In this wave, researchers collected data on demographic characteristics, body mass index, self-reported lifetime marijuana and alcohol use, and lifetime internalizing and externalizing problems. They adjusted for these potential confounders in their analyses.
During wave 2 (2014-2015), researchers asked respondents how much time they spent per day using social media — sites such as Facebook, Google Plus, YouTube, MySpace, LinkedIn, Twitter, Tumblr, Instagram, Pinterest, and Snapchat.
About 16.8% of adolescents reported no social media use, 31.8% reported 30 minutes or less per day, 30.7% reported more than 30 minutes to 3 hours or less, 12.3% reported more than 3 hours to 6 hours or less, and 8.4% reported more than 6 hours.
Investigators assessed mental health problems during wave 3 (2015-2016) using the Global Appraisal of Individual Needs — Short Screener (GAIN-SS). From responses, they created a single outcome variable with four categories: no or low internalizing and externalizing problems (the reference group), internalizing problems alone, externalizing problems alone, and comorbid internalizing and externalizing problems.
“Internalizing” and “externalizing” problems are common childhood psychopathology classifications, said Riehm. Internalizing refers to inward distress such as anxiety, depression, withdrawal, and loneliness. Externalizing problems manifest in behaviors such as inattention, physical aggression, rule breaking, and impulsivity.
Some 9.1% of study subjects reported internalizing problems alone, 14.0% reported externalizing problems alone, 17.7% reported comorbid internalizing and externalizing problems, and the remaining 59.3% reported no or a low level of problems.
That more than 40% of survey respondents indicated some level of internalizing or externalizing problems “makes sense for this population,” said Riehm.
She noted that the simple four-item questionnaire “will pick up a lot of people who may have had some level of distress, but perhaps not a clinical level of distress that would be characterized as a mental health problem.”
The unadjusted analysis showed that compared with adolescents who did not use social media, its use for more than 30 minutes per day was associated with greater risk of internalizing problems alone and comorbid internalizing and externalizing problems.
In the adjusted model, the associations for the two highest categories of social media use persisted for internalizing problems alone (> 3 to ≤ 6 hours: relative risk ratio [RRR], 1.60; 95% confidence interval [CI], 1.11 – 2.31; > 6 hours: RRR, 1.78; 95% CI, 1.15 – 2.77).
The associations for the three highest categories of social media use persisted for comorbid internalizing and externalizing problems (> 30 minutes to ≤ 3 hours: RRR, 1.59; 95% CI, 1.23 – 2.05; > 3 to ≤ 6 hours: RRR, 2.01; 95% CI, 1.51 – 2.66; > 6 hours: RRR, 2.44; 95% CI, 1.73 – 3.43).
“It was pretty much a linear relationship; the more time you spent on social media, the more likely you were to report internal and comorbid problems,” said Riehm.
In contrast, the association of social media use with externalizing problems alone was inconsistent in the unadjusted analyses, and not significant in the adjusted analysis.
This finding suggests that the association of social media use with comorbid problems is driven mostly by internalizing symptoms and the high comorbidity with externalizing problems, said the authors.
There did not appear to be any sex differences for internalizing, externalizing, or comorbid problems. This is in contrast to previous studies that found the association between social media use and internalizing problems tends to be stronger in girls.
“The reason our study didn’t pick up on that is that we were just measuring time spent on social media,” explained Riehm. “It could be that girls and boys have different interactions on social media that would lead to more risk for women, but we wouldn’t be able to capture that with how we measured social media.”
To estimate to what degree mental health problems could be reduced by spending less time on social media, researchers carried out a population-attributable fraction analysis. Essentially, this involved cumulatively reclassifying those in one category of hours spent on social media into the next lowest category.
The findings here suggested that if adolescents used social media for 30 minutes or less per day instead of for more than 30 minutes per day, there would have been 9.4% fewer high internalizing problem cases and 7.3% fewer high externalizing problem cases.
That doesn’t necessarily mean, though, that reducing social media use would result in fewer mental health problems. “A lot more research would need to be done to determine if real changes in social media use would lead to changes in internalizing symptoms,” said Riehm.
Such research could take the form of a randomized clinical trial. Riehm cited one such trial that has already linked reduced social media use to less depressive symptoms.
However, that study was relatively small and restricted to college students. “Future researchers could consider scaling that up and doing it in a population of adolescents,” said Riehm.
Numerous factors might help explain the association between social media use and internalizing problems. For example, adolescents who use social media extensively may have sleep problems and may be at increased risk of cyberbullying, which is strongly associated with depressive symptoms.
As well, lengthy social media use may negatively impact self-esteem, said Riehm.
“Social media has a lot of idealized presentations of people, and constantly being exposed to that might make you feel like you’re not living up to what other people are doing. That could affect your self-esteem, which in turn could lead to anxiety and depression.”
The current study could not determine whether the association between social media use and internalizing problems is greater for particular platforms, or whether certain user groups are particularly vulnerable, said Riehm.
Families have options when it comes to curtailing social media use. Riehm pointed to the American Academy of Pediatrics’ Family Media Use Plan that helps parents set reasonable rules for daily digital media use.
For example, parents can set screen-free zones in the home, establish screen-free times such as during meals, and initiate “device curfews,” said Riehm.
No More “Likes”?
Social media giants are starting to recognize that their platform features may contribute to mental health problems among young users. Riehm noted that Instagram has been testing the removal of its “likes” function.
“It will be interesting to see if other platforms follow suit and change things as well.”
But social media does have potential benefits for teens. These include learning about current events, being able to communicate over geographic barriers, and social inclusion for those who may otherwise be excluded in their day-to-day lives, for example, LGBTQ youth.
So what’s the optimal amount of social media time for young people? Based on research to date, 30 minutes a day “appears to be a good threshold, at least to start from,” said Riehm, although she stressed that “I can’t say if that’s safe or not.”
A limitation of the study was that the exposure and mental health problems were self-reported.
Commenting for Medscape Medical News, Joshua Weiner, MD, a psychiatrist in McLean, Virginia, whose practice is at least 40% adolescents, said that while he always finds studies like this new one interesting, they tend to include so many variables that the design becomes rather complicated.
“I think that makes it hard to really fully appreciate the findings.”
While the authors point out that suicide rates are on the rise among teens, Weiner noted that the current suicide rate is not the highest historically.
“Although suicide rates in teens have been on the rise over the last roughly 15 years, they’re actually still not as high as they were in the early ’90s” in boys aged 15 to 19 years, when social media didn’t even exist, he said.
He believes these high rates might have been due to the relatively high level of alcohol use at the time, which disinhibited boys.
Weiner noted that from his clinical experience, girls are more affected by social media than are boys. “Girls are coming in with complaints that are clearly linked to social media.”
The rates of internalizing and externalizing problems reported by study subjects are lower than Weiner would have expected. “It’s estimated that in general, about 25% of 13- to 18-year-olds will experience an anxiety disorder, and that doesn’t even include depression,” he said.
The rate of internalizing problems for teens “is very high in general” whether or not social media is contributing, he added.
While the study linked social media use at wave 2 to internalizing problems at wave 3, it doesn’t consider that age itself is a factor in rising rates of mental health issues.
“You might not have had issues with anxiety or depression at 13 years old, but you’re more likely to have these issues when you’re 16, anyway.”
However, he agreed that the study did find a progressive increase in the amount of internalizing problems as kids spent more time on social media.
Weiner is convinced that one of the biggest contributors to internalizing problems is sleep deprivation — but that could be due to staying up to watch TV or some other reason not involving social media.
He noted that the study did not differentiate between social media sites. Watching YouTube might be more “benign” than visiting some other platform, he said.
Medscape Medical News also invited Vinu Ilakkuvan, DrPH, a public health consultant and professorial lecturer at George Washington University Milken Institute School of Public Health, Washington, DC, and lead author of another study of social media use and health risks, to comment.
He called the new study well designed and executed, and an important contribution to the literature. He added that using longitudinal data that accounts for prior history of mental health problems is “a key strength” of the study.
Use of social media and widespread sharing of its content “has run rampant and unchecked,” Ilakkuvan said, adding that “neither regulation nor research has kept pace.”
“Studies like this one are a step in the right direction, particularly if they can help inform programmatic and policy interventions, as well as changes to social media platforms, that could mitigate the negative influence of social media on adolescent health while maintaining or strengthening the potential benefits.”
Riehm was supported by a grant from the National Institute of Mental Health Psychiatric Epidemiology Training Program and by a doctoral foreign study award from the Canadian Institutes of Health Research. Weiner and Ilakkuvan have disclosed no relevant financial relationships.
JAMA Psych. Published online September 11, 2019. Abstract