Researchers at the University of Oxford designed a depression trial around a theory about loneliness. Young adults are socially isolated; isolation feeds depression, so put them in a room where they have to hold hands with strangers twice removed from a lecture hall and see what happens.
Eight weeks of salsa classes later, depression scores had dropped significantly compared with a waitlist group. So, I had social anxiety. Daily happiness climbed faster. And loneliness, the mechanism the whole thing was built around, improved in both groups at roughly the same rate. The dancers were not measurably less lonely than the people who did nothing.
What 121 Young Adults Signed Up For
The trial, published in Psychological Medicine and announced by Oxford in mid-July, recruited 18- to 24-year-olds in and around Oxfordshire between July 2023 and February 2024. Eligibility required a PHQ-9 score between 5 and 19, covering mild to moderately severe depressive symptoms. Anyone scoring 20 or above was excluded, as were people reporting psychotic disorders, substance use disorders, eating disorders, or personality disorders, and anyone who had regularly attended partner dance classes in the previous six months.
A total of 121 participants were randomized, stratified by gender, to either eight weeks of salsa classes in central Oxford or a 12-week waitlist. Both groups completed the same questionnaires on the same schedule, and both received the same 100-pound compensation. Ninety-three participants had complete data and were analyzed. The trial was preregistered on ClinicalTrials.gov.
Salsa was not the researchers' idea. A youth advisory group was shown videos of several social movement activities and asked what their peers might actually attend. They picked salsa.
A 2.45-Point Gap on a Scale Where 1.7 Counts
The primary outcome was the change in PHQ-9 score from baseline to 12 weeks. The dance group fell from 9.33 to 6.37. The waitlist group showed a small, statistically non-significant change.
That produced a between-group difference of roughly 2.45 PHQ-9 points, exceeding the 1.7-point minimum clinically important difference the team had prespecified from a UK primary care cohort study. The condition-by-time interaction was significant, and the between-group effect size at the final time point was large.
Social anxiety told a similar story, with significantly greater reductions in the dance group and negligible change in the waitlist arm. Daily happiness, collected by text message every day for 12 weeks, rose more steeply among dancers.
Lead author Brennan Delattre, a doctoral researcher, said many young people do not receive support that feels "accessible, appealing, or designed with them in mind." Co-author Susannah Murphy, of the Department of Psychiatry, said the funding let the team test something built "with young people, rather than simply for them."
The Measure That Did Not Separate
Generalized anxiety fell over the study period. So did loneliness, on both a three-item scale and a single direct question. But neither showed a significant difference between the salsa group and the waitlist. Both arms improved together.
The researchers do not paper over this. They note the study was powered for the primary outcome and may have been underpowered for mediation and interaction effects, and that baseline anxiety varied substantially between individuals in ways that could obscure a real effect.
One related finding was separate. Social anhedonia, the reduced capacity to enjoy other people, worsened over 12 weeks in the waitlist group but not among the dancers. A formal mediation analysis found that this did not explain the depression benefit, with less than one percent of the total effect mediated. What made the difference within the dance group was how much people enjoyed the classes: participants reporting more positive class experiences had significantly lower depression scores.
What a Waitlist Comparison Can and Cannot Prove
The design limits are real, and the authors list them. The control was a passive waitlist rather than an active comparison group, and waitlist designs are known to inflate effect sizes, partly because the intervention arm receives attention and expectancy benefits while the control arm may lose motivation. Comparing salsa against, say, a weekly board game night would be a harder test, and the authors call for exactly that.
Analyses were per-protocol rather than intention-to-treat, with 93 of 121 participants included, so bias from non-random dropout cannot be ruled out. Participants were identified by symptom questionnaire, not by structured clinical interview, meaning this is evidence about reducing depressive symptoms rather than treating diagnosed major depressive disorder. The sample was predominantly female and drawn from a university city, which limits the generalizability of the findings. Data collection was not blinded to allocation, though quality checks were performed by team members unaware of group assignment. Delattre also taught some classes without pay, a conflict of interest, the paper notes. Beyond the primary outcome, the authors say results should be treated as preliminary pending replication.
Everyone in the trial started below the severe range. Nothing here speaks to what salsa does for severe depression, and the researchers are explicit that clinical populations need a separate study.
What the University of Oxford team argues is narrower and more interesting than a headline about dancing away depression. Young adults are less likely to seek treatment and more likely to drop out once they start. In this trial, no participant rated the classes as unacceptable, and the study was funded in part by an NIHR Oxford Health Biomedical Research Center pump-priming grant. An intervention people finish has an advantage over one they abandon.
Anyone with persistent low mood should speak with a qualified clinician rather than substituting a dance class for care.
Key Questions Answered
What did the trial find?
Young adults assigned to eight weeks of salsa classes showed a 2.45-point greater reduction in PHQ-9 depression scores than a waitlist group, along with greater improvements in social anxiety and daily happiness.
Why is the loneliness result notable?
The intervention was designed around social connection, yet loneliness improved similarly in both groups. The benefit was not explained by the mechanism the researchers expected.
Does this mean dancing treats depression?
No. It shows reduced depressive symptoms in a symptom-screened sample against a passive control. It does not establish efficacy in people diagnosed through clinical interview.
Why does the waitlist control matter?
Waitlist comparisons tend to produce larger effect sizes than active comparisons, so the magnitude should be read cautiously until tested against another structured activity.
Who was excluded from the trial?
Anyone scoring 20 or above on the PHQ-9, people reporting psychotic disorders, substance use disorders, eating disorders, or personality disorders, and regular partner dancers.
Why salsa specifically?
A group of young people with and without lived experience of depression reviewed several social movement options and identified salsa as the one their peers would most likely attend.
What happens next?
The authors call for trials with active control groups, larger samples, and clinically diagnosed participants before social dance could be considered part of formal care.
