The Classes Worked Better For The People Who Enjoyed Them, Which Is The Whole Problem With Social Prescribing

The instruction is always the same at the start. Weight on the left foot, step forward on the one, and when the music comes back around, change partners. In a hall in central Oxford, on a weekday evening, a few dozen 18 to 24-year-olds who mostly did not know each other were doing exactly that, week after week, for eight weeks. Some of them had been carrying depressive symptoms severe enough to sit near the top of the mild-to-moderate range on a clinical screening questionnaire.

They were also, though nobody in the room would have been able to tell, half of a randomized controlled trial. The other half were at home, filling in the same questionnaires on the same schedule, waiting their turn.

The results, published in Psychological Medicine, are the kind that make you look twice at the effect size and then look again at the control group. By the final assessment at 12 weeks, the salsa group’s scores on the PHQ-9 (a nine-item depression measure used routinely across UK primary care) had fallen roughly 2.45 points further than the waitlist group’s. That figure clears the minimum clinically important difference of about 1.7 points established in a cohort of 400 UK primary care patients. The standardized between-group effect was large, a Cohen’s d of around 0.92. Which, for an intervention consisting entirely of learning to dance badly in front of strangers, is not nothing.

The choice of salsa was not the researchers’. Before the trial began, a youth advisory group in Oxford, including young people with lived experience of depression and anxiety, was shown videos of various social movement activities and asked what their peers might actually turn up to. Salsa won.

Designed With, Not For

That detail matters. Depression peaks in onset between roughly 15 and 25, and yet the average participant in the largest meta-analysis of antidepressant efficacy was 44. Young adults are less likely to seek treatment, more likely to drop out once started, and tend to do worse in talking therapies than working-age adults. So there’s an awkward mismatch: the age group where the illness most often begins is the age group the evidence base knows least about.

“Too often, there is a gap between the mental health support available and the kinds of activities young people feel motivated to engage with,” says Susannah Murphy, a professor in Oxford’s Department of Psychiatry and one of the study’s authors. The pump-priming grant, she says, made it possible to test an intervention that was “designed with young people, rather than simply for them.”

What the trial actually tested was not exercise, or at least not only exercise. Brennan Delattre, the doctoral researcher who led the study, thinks the structure is doing some of the work. Salsa, she notes, is “physical, social, musical, structured, and often playful, it asks people to engage with others, but within a clear framework: steps, patterns, partner rotation, and a predictable class routine. For some people, this may make social contact feel more manageable than an unstructured social group setting.” You are not obliged to make conversation with the person opposite you. You are obliged to count to eight. There is something merciful about that for a socially anxious 19-year-old, and the social anxiety numbers back it up: the dancers’ scores dropped substantially while the waitlist group’s barely moved.

Six instructors taught the classes, in varying combinations, from the UK, the US, Italy, Romania and Spain. The sessions ran through existing Oxford dance societies, and participants walked in through the same door as everyone else, indistinguishable from ordinary members. No separate therapeutic space, no name badges. Just a class.

What the Numbers Don’t Settle

Now for the caveats. The comparison group was a waitlist, not an active control, and waitlists are a famously generous benchmark: the intervention arm gets attention, structure and the expectation of improvement, while the people on the list may simply lose heart. Something like that may have shown up in the data. Social anhedonia, the reduced capacity to enjoy other people’s company, actually rose in the waitlist arm over 12 weeks while holding steady in the dancers. Read one way, that’s dance protecting something. Read another, it’s a control condition deteriorating.

Loneliness and generalized anxiety fell in both groups by about the same amount, which complicates the simple story about social connection. And participants who volunteer for a dance trial are presumably people who already suspect dancing might help them, a selection bias the authors acknowledge. The sample was mostly women, mostly not clinically diagnosed. Ninety-three people finished. That’s a decent trial, not a definitive one.

Delattre is careful about the claim: “While salsa dancing is not a replacement for other types of mental health care, nor will it be the right fit for everyone, our findings suggest that social dance deserves serious attention as part of a broader menu of wellbeing support for young people.” Nobody is proposing salsa on prescription tomorrow.

Participants rated each class immediately afterwards on enjoyment, connection and happiness, and how they answered predicted how their depression scores moved. The classes worked better for the people who liked them. Which sounds obvious until you notice what it implies for social prescribing: not that dance is the active ingredient, but that pleasure, taken seriously and scheduled weekly with other people, might be. If that holds up against an active control, the interesting question stops being whether salsa works and becomes which other eight-week habits nobody has bothered to run a trial on.

Frequently Asked Questions