Friendships May Shield People With Bipolar Disorder From Long-Term Depression

For people living with bipolar disorder, the depressive pole of the illness is often the most punishing and persistent part of the condition. While manic episodes tend to capture public attention, it is depression that accounts for the greatest share of the burden, occupying far more of patients’ time over the course of the disorder and eroding quality of life, functioning, and physical health. A new longitudinal study now adds a striking piece to the puzzle of what shapes that long-term depressive trajectory: the quality of a person’s friendships. Researchers report that how supported patients felt by their friends at the start of the study predicted how severe their depressive symptoms would be years later, an effect that held even after accounting for how depressed they were when the study began.

The research, conducted by a team spanning Northwell Health and the Feinstein Institutes for Medical Research in New York and the Medical University of Graz in Austria, followed 117 patients with bipolar disorder, half of them women, for four and a half years. Participants were assessed approximately twice a year, for up to nine timepoints in total, an unusually dense and extended window for a psychosocial study in this population. At baseline, the researchers measured social support using two items from the World Health Organization’s Quality of Life instrument, the WHOQOL-BREF, one capturing satisfaction with personal relationships and one capturing the support patients received from friends. Depressive severity was then tracked repeatedly over the follow-up period using the Hamilton Depression Rating Scale, a clinician-administered measure that remains one of the most widely used standards in mood disorder research.

The analytical approach was designed to exploit that repeated-measures structure. Rather than comparing simple before-and-after averages, the team used Generalized Estimating Equations, a statistical framework built for clustered longitudinal data in which multiple observations from the same individual are inherently correlated. GEE models allowed the investigators to estimate whether baseline social support was associated with depressive symptom severity across the entire follow-up period while controlling for baseline depressive severity, age, and time. This matters because the central methodological trap in studies of social support and mood is reverse causation: people who are depressed may simply rate their relationships as less supportive. By statistically adjusting for how depressed patients were at the start, the researchers could ask whether friendship support carried information about the future beyond what current symptoms already revealed.

The answer was yes, with an important nuance. Across the whole sample, baseline friend support showed a significant negative main effect on long-term depressive symptom severity: patients who reported stronger support from friends at the outset tended to have lower depressive severity over the subsequent years, independent of their starting symptom levels, their age, and the passage of time. Notably, the second WHOQOL-BREF item, satisfaction with relationships in general, did not emerge as the driving predictor. It was specifically the support derived from friends, a distinct social niche separate from partners and family, that carried the predictive signal, suggesting that not all social ties contribute equally to the long-term course of the illness.

When the researchers stratified their analyses by sex, a further layer of complexity appeared. In male patients, the effect was clear and statistically significant: higher baseline friend support predicted lower depressive symptom severity across the 4.5-year follow-up, even after controlling for baseline depression. In female patients, by contrast, the same association did not reach statistical significance. The authors are careful about how this should be interpreted. The sample, while substantial for a longitudinal bipolar cohort, is modest in size, and splitting it by sex halves the statistical power available for each subgroup analysis. A null result in women therefore does not mean friendship is irrelevant to their outcomes; it means the data could not demonstrate the effect in that subgroup, and the finding requires replication in larger samples before firm conclusions can be drawn about sex-specific mechanisms.

Nevertheless, the sex difference is plausible in light of a broader literature on how men and women construct and use social networks. Research in social epidemiology has repeatedly suggested that men tend to rely more heavily on friendships, and particularly on friendships that provide companionship and shared activity, as their primary context for emotional disclosure, while women often maintain more diversified sources of intimate support, including partners and family. If men with bipolar disorder have fewer alternative channels through which they receive emotional support, then the quality of their friendships may carry disproportionate weight in buffering the stress and social withdrawal that accompany depressive episodes. The new findings offer preliminary longitudinal evidence consistent with that framing, though the study was not designed to test the underlying psychological mechanisms directly.

What makes the result clinically interesting is that friend support is a potentially modifiable factor. Much of what determines the long-term course of bipolar disorder, from genetic vulnerability to the illness’s neuroprogressive features, lies beyond a patient’s control, and pharmacological treatment, while essential, leaves many patients with residual depressive symptoms. Psychosocial interventions, by contrast, can in principle be engineered. If friendship-based support genuinely buffers depressive symptoms over years, then interventions that help patients build and sustain friendships, whether through structured group programs, peer support networks, community mental health services, or therapies that explicitly target social functioning, could become a meaningful complement to medication management. The authors frame their finding as preliminary evidence that friend support may be an important buffer against depressive symptoms, particularly in male patients, and emphasize that replication in larger samples is the necessary next step.

The study also illustrates the value of long follow-up windows in psychiatric research. Cross-sectional studies can show that socially isolated people are more depressed, but they cannot disentangle cause from consequence. By following patients across nine assessment points and modeling depressive severity as a continuous longitudinal outcome, the researchers could show that the association persisted across years rather than reflecting a snapshot in which low mood and low perceived support simply co-occur. The use of a validated quality-of-life instrument for the exposure and a clinician-rated scale for the outcome further strengthens the design, reducing reliance on self-reported mood for both sides of the equation. All participants provided informed consent, and the study was overseen by the ethics committee of the Medical University of Graz.

Cautions remain. Two questionnaire items are a coarse measure of something as multidimensional as social support, which encompasses emotional closeness, practical help, and perceived availability. The cohort was drawn from a single clinical region, and patients willing to participate in repeated research assessments over four and a half years may differ systematically from the broader population of people with bipolar disorder. The GEE models control for baseline severity, age, and time, but other unmeasured variables, from illness duration to medication regimens to personality traits, could partly explain the association. And because the sex-stratified analysis rests on smaller subgroups, the male-specific effect should be treated as a hypothesis-generating result rather than a settled fact.

Even with those caveats, the study adds weight to a growing recognition that the social world is not merely background context for severe mental illness but an active ingredient in its course. For clinicians, the practical message is that asking patients with bipolar disorder about their friendships may be more than a courtesy; it may offer a genuine window into their long-term prognosis and a target for intervention. For patients and families, the findings are a reminder that maintaining friendships through the disruptions of the illness is not a luxury. As the authors conclude, friend support may serve as an important buffer against depressive symptoms, and if larger studies confirm the effect, one of the most accessible protective factors in bipolar disorder may turn out to be the company we keep.

Subject of Research: Longitudinal association between friend-based social support and depressive symptom severity in bipolar disorder

Article Title: Social support from friends predicts lower long-term depressive symptom severity in bipolar disorder: a 4.5-year follow-up study

Article References: Social support from friends predicts lower long-term depressive symptom severity in bipolar disorder: a 4.5-year follow-up study. (n.d.). https://doi.org/10.1186/s12888-026-08673-y

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08673-y

Keywords: bipolar disorder, depression, social support, friendship, longitudinal study, sex differences, WHOQOL-BREF, Hamilton Depression Rating Scale, psychiatry, mental health, psychosocial factors, GEE models