Facebook and other social media platforms are under fire these days, accused of causing a wide variety of harms to individuals and societies. Among the charges is that spending time on social media can cause depression. A study published in November in the journal JAMA Open Networks attempted to look at this question more closely and has garnered widespread attention from behavioral health experts and the media.
The paper’s lead author is Roy Perlis, a psychiatrist at the Massachusetts General Hospital and Harvard University in Boston and is titled “Association between social media use and self-reported symptoms of depression in US adults.” The authors note that a number of studies have hinted at this association between social media and depression, but most of them have been cross sectional or involved only a small number of participants, making it impossible to draw any cause-and-effect conclusion.
In the Perlis et al study, more than 5000 people with a mean age of 55.8 years and who had very low scores at baseline on the nine-item Patient Health Questionnaire (PHQ-9), indicating that they were not depressed, were surveyed approximately monthly between May 2020 and May 2021, with measures of social media use and repeat PHQ-9 at each month. The investigators found that about nine percent of the participants had a five point or more worsening of their scores on the PHQ-9, indicating that they had become significantly depressed over time. Those who had worsening depression also had the most use of three social media platforms, Snapchat, Facebook, and Tik Tok. For Facebook, participants with worsening depression scores had about a 40% increased use of the social media platform compared to those participants without worsening depression scores. Controlling for measures of social contact and social support did not alter these findings, suggesting that social media use substituting for real-life social contact was not the cause of these findings.
Does It Show Causation?
There are many strengths to this study, including the very large sample size, the use of a well-validated instrument to measure depression, and the longitudinal design. These were all people who were not depressed at study initiation, so the investigators were able to trace the relationship between social media use and the evolution of depression. The question becomes, then, can we say that this study suggests that looking at social media platforms like Facebook can actually cause depression?
Some headlines on internet sites seemed to hint that this might be the case. “Social Media Use Tied to Self-Reported Depressive Symptoms,” read one. “Social media use linked to depression in adults,” read another. While neither of these used the word “cause” in its headline, the impression one gets from words like “tied to” and “linked to” might easily be misinterpreted to mean that a causal relationship had been demonstrated.
In fact, Dr. Perlis and co-authors note in the paper that “Notably, social media use may simply be a marker of underlying vulnerability to depression.” That is, even though they did not score in the depressed range on the PHQ-9 at baseline, it is still possible that those people who went on to develop worsening depression scores nevertheless had a vulnerability or predisposition to depression that both drives the illness and drives one to spend more time on Facebook and other social media platforms. As the authors themselves point out, the study did not control for innumerable confounding factors that could be the cause of this association between depression and social media use (e.g. previous history of depression; current life stress; reasons for going to social media platforms) and therefore causality cannot be established by this study.
How Would We Establish Causality?
Yet the findings are troubling because they leave open the possibility that one way to explain them is that in fact spending time on social media is a causative factor for depression. Is it possible that someone who would not otherwise become depressed does so because they spend an excessive amount of time on Facebook, Snapchat, or Tik Tok? What kind of a study could really answer the causality question here?
Typically, our default study design to answer causality questions is the randomized trial, in which a group of people is randomized to different conditions. In studies to test whether a new medication is effective, for example, we would randomize a group of people who have a particular health condition to either receive the experimental drug or a placebo. Because nothing else is different between the two groups except whether they have received drug or placebo, the assumption is reasonably made that we have controlled for other potentially confounding factors and isolated the true effects of the drug (things aren’t really that simple, but that’s for another commentary). Could we do that with social media use? The basic design would have to be something like starting with a large group of people without depression at baseline and randomizing them to different levels of social media use. We would then follow them over time, assuming that by the randomization process we have controlled for all possible confounding factors except for how much social media is viewed. Then, we would see if people in high social media utilization groups were more likely to develop clinical depression than people in low utilization groups.
It is easy to see why doing such a study would be challenging. First of all, it would be expensive, requiring that either the social media companies pay for it, which would raise issues of conflict of interest and bias, or that external funds from a foundation or federal funder be obtained. It is unclear how much appetite there would be for doing this.
Next, it would be hard to recruit such a group of people because participants would have to agree to accept the level of social media use to which they were assigned. A person who spent little time on Facebook, for instance, might be randomized to a group that must view it at a high level while someone who is used to viewing a lot of social media might get randomized to a group in which they would be asked to stop altogether for an extended period of time. It would also be hard to enforce that people really adhered to the amount of time on social media to which they were assigned.
Finally, if such a study did show that social media use caused depression, what would we do with that information? Note that in the Perlis et al study, 90% of the participants did not develop worsening depression scores during the year of observation. Whom would we tell, then, to refrain from too much social media use? And what demands could we make on social media companies to reduce the risk that their platforms harm mental health?
One solution recently announced by the social media platform Instagram is to offer “take a break” reminders to teenage users. The idea here is that limiting the time people spend on social media could help prevent adverse mental health outcomes. Given that it is voluntary whether a person agrees to get these reminders or to follow their advice and actually get off the site, it is unclear how effective this would be.
We probably cannot depend on randomized trials of social media use being completed any time soon to decide if it does in fact cause psychiatric illness. Remember, however, that we did not discover that cigarette smoking causes lung cancer by doing randomized trials; instead, we depended on large population studies to make that case. It may be that we will have to rely on studies like the Perlis et al one, large cross-sectional population studies, and other novel forms of study design to tease out whether there is a causal relationship here. Right now, it might be plausible to advise anyone with a history of depression or in situations that increase risk for depression to be judicious about how much time they spend on social media.