The RCT of tango dancing versus mindfulness meditation for depression

giphy1If someone is unhappy with their life, should they take up mindfulness meditation or some other activity?

That is a question that a lot of people confront. For an answer, they might even look to this RCT comparing tango dancing lessons to mindfulness to a waiting-list control.

The authors actually specifically asked whether tango dancing is as effective in reducing stress, anxiety, and depression as mindfulness meditation. Furthermore, they raise the question whether tango dancing increased mindfulness more than meditation itself.

The authors did not actually produce a direct answer to the questions they raised, but what happened is interesting to me, even if my interpretation is counter to the authors.

In the end, if this RCT provides any answers at all, it is “if you’re trying to decide what to do to resolve unhappiness, you might want to select an activity that you enjoy participating in, for itself.” And I really don’t think you need an RCT to guide that decision.

But further investigating this RCT can provide some useful insights about interpreting such studies, because there are many out there like it.

Certainly one take away lesson is that we should pay more attention to the CONSORT flowchart when we began evaluating a report of an RCT than we typically do.

The citation and the access to a PDF of the study is directly below.

Pinniger R, Brown RF, Thorsteinsson EB, McKinley P. Argentine tango dance compared to mindfulness meditation and a waiting-list control: A randomised trial for treating depression. Complementary Therapies in Medicine. 2012 Dec 31;20(6):377-84.

Objectives: To determine whether tango dancing is as effective as mindfulness meditation in reducing symptoms of psychological stress, anxiety and depression, and in promoting well-being. Design: This study employed analysis of covariance (ANCOVA) and multiple regression analysis. Participants: Ninety-seven people with self-declared depression were randomised into tango dance or mindfulness meditation classes, or to control/waiting-list. Setting: classes were conducted in a venue suitable for both activities in the metropolitan area of Sydney, Australia. Interventions: Participants completed six-week programmes (1½h/week of tango or medita-tion). The outcome measures were assessed at pre-test and post-test. Main outcome measures: Depression, Anxiety and Stress Scale; The Self Esteem Scale; Satisfac-tion with Life Scale, and Mindful Attention Awareness Scale. Results: Sixty-six participants completed the program and were included in the statistical anal-ysis. Depression levels were significantly reduced in the tango (effect size d = 0.50, p = .010), and meditation groups (effect size d = 0.54, p = .025), relative to waiting-list controls. Stress levels were significantly reduced only in the tango group (effect size d = 0.45, p = .022). Attending tango classes was a significant predictor for the increased levels of mindfulness R2 = .10, adjusted R2 = .07, F (2,59) = 3.42, p = .039. Conclusion: Mindfulness-meditation and tango dance could be effective complementary adjuncts for the treatment of depression and/or inclusion in stress management programmes. Subsequent trials are called to explore the therapeutic mechanisms involved.

There are a number of deficiencies in this abstract as a representation of the study, but let’s put them aside for now.

The take away message that seems to be intended is that in some respects the tango group and meditation groups similarly reduce depression, and the tango group increased mindfulness. On Twitter, a lot of attention was given to the interpretation that only the tango group reduced stress.

Ratonale for the study.

If you are interested, you can read this passage in its context, but the authors offer an interesting rationale for their study

Toneatto and Nguyen17 suggest that any activity requiring awareness of current experience is likely to interrupt an individual’s thoughts about their past and fears about the future, so potentially lessening association between nega-tive thoughts and possible affective symptoms. Tango is one activity that could achieve this goal, since it is an absorb-ing activity that requires significant skills acquisition,8,16 and an awareness of current experience.17 It also facilitates an involvement in music, exercise, and touch, all of which have previously been reported to be effective in alleviating psychological distress.18—20

The important details of the recruitment

 The paper is actually better than most in describing explicitly what participants were told who were invited to enroll in the study.

Participants were recruited via advertisements asking for volunteers with self-reported stress, anxiety, and/or depression. They were invited to participate in a RCT evaluating tango dance relative to mindfulness meditation or waiting-listed control. They were advised that they could be allocated to one of those three groups.

Participating in an RCT or any research, for that matter, usually involves assuming some burden in terms of having to complete assessments and not drop out except for some good reason. But it also involves the gamble of randomization. Participants may be seeking an experience that they might not otherwise have available, and risking not being assigned to it, but nonetheless feeling obligated to stick with it.

We will soon be seen that most participants in the study were more interested in getting tango lessons than in becoming more mindful. Any hopes of interpretable results for a randomized controlled trial could been doomed from the start by a lack of equipoise- different participant valuation of tango lessons versus mindfulness at the outset, or it could have emerged once participants actually gotten involved in these activities. Regardless, different preferences and differences in the ability of the two interventions to retain participants were ultimately fatal for an interpretable findings, at least in terms of the outcomes of anxiety, depression, and stress.

The CONSORT flowchart

 consort flow tango

Most readers will skip over the CONSORT flowchart for these kind of studies. But it is good that most journals have accepted the requirement that reports of RCTs have such a flowchart. What is revealed about this study is that there were so many dropouts from participants assigned to mindfulness, much more that so than the tango group.

What is also interesting is that the waiting list group stuck around for assessments. I might’ve predicted ahead of time that assignment to this treatment would have the greatest number of dropouts because of the delay in getting any reward.

Inappropriate analyses.

The most appropriate and informative analysis for an RCT includes results for all participants who originally enrolled. That is an intention to treat analysis. Less preferable is what these authors did, an analysis limited to the participants available at follow-up.

In a situation of such different abilities of conditions to retain participants, many researchers would have resorted to some kind of imputation of missing data from what data were available. Such imputation typically assumes that loss to follow-up is random, which is unlikely. Aside from violation of this assumption, there is simply too few participants retained in the mindfulness group to provide a basis of guessing the outcomes of participants lost to the study. So, intent to treat analyses based on imputing outcomes to participants who were no longer available would be invalid.

But what the investigators did instead was similarly invalid. They ignored that a substantial portion of the participants assigned to mindfulness meditation were no longer available and analyzed only the results of those who stuck around. Consider this analogy of a climb up Mount Everest as a way of reducing stress. Most participants would make it, but would you make a decision about the value of attempting to climb Mount Everest based on results only from those who made it?

The voucher

 I did not pay attention to a key statement in the description of recruitment that later took on particular importance in interpreting what happened in the study. The statement was

What we were later told in the discussion section:

Finally, 97% of participants in the study chose to receive a tango dance voucher after the study, rather than a meditation voucher, suggesting the popular appeal of tango dance. This is an important consideration, since people tend to adhere to mindfulness-based programs more than other therapy approaches,22 especially if the experience is intrinsically positive,25 such as tango dance is reported to be.8

So, it appears that the participants recruited to the study had a strong preference for getting tango lessons, not mindfulness meditation. There was a broad dissemination of invitations to participate in the study. We’re not told, but maybe the researchers had to advertise so broadly, because they initially had difficulty obtaining participants. Maybe, –we are not the position to know–mindfulness was readily available in Sydney Australia, but tango lessons were in short supply. Participants enrolled in the study in hope of getting assigned to tango lessons.

Bad abstract

Having read the study, I could now see that the abstract was misleading. It left out the important detail of how many participants were recruited to the study instead focused on the number from whom results were available. Furthermore, I’m suspicious about the participants being described has self identified for depression. Most accurately, participants from were recruited for a study aimed at reducing anxiety, depression, and stress. They did not self identify as depressed, they simply filled out some self-report questionnaires. An even suspicious about depression being implicitly presented has the primary outcome. They could be confirmation bias here: it is the one of the three outcomes for which the results were strongest.

Back to the question of tango dance lessons versus mindfulness meditation

 Both dance lessons  and mindfulness meditation require commitment and practice to obtain benefit. I don’t think that participants enrolled in the study only to reduce their self-report measures. They expect to get something out of the experience. If nothing else, the dance lessons posed the prospect of leaving participants better able to dance. No guaranteed, of course, but it was in the offering.

So, back to a person pondering what to do with their life. The key issue is one of personal preference: do they think they are likely to engage in the activity enough to get any benefit? If the activity was supposed to produce something tangible, like dancing skills, would they be content with that alone? If a participant were only looking to have a bit of fun, would they really need the guidance of an RCT?