What Happens If You Do Nothing: The Number Every Uncontrolled Study Ignores

Every so often this site describes a study as uncontrolled and moves on, as though that settles it. It does settle it, but the reason deserves a number rather than an assertion. Here is the number.

The study nobody runs on purpose

A 2011 meta-analysis in the Journal of Psychosomatic Research asked a question almost nobody asks directly: what happens to people in a tinnitus trial who are assigned to the wait-list and given nothing at all?

The authors went through randomised controlled trials of cognitive behavioural therapy for tinnitus distress, pulled out the wait-list groups, and pooled them. The outcomes were the standard instruments — the Tinnitus Handicap Inventory, the Tinnitus Questionnaire, the Tinnitus Reaction Questionnaire — measured before and after the wait.

Eleven studies. 314 people. No treatment.

Over a waiting period of 6 to 12 weeks, their scores on tinnitus-specific measures fell by 3% to 8%, with a small but statistically significant within-group effect size of Hedges' g = 0.17.

Why that single figure changes how you read everything

Take the shape of a typical claim from a clinic or a device company: we treated fifty patients and their tinnitus scores improved significantly.

That sentence is now readable. Some of that improvement was always going to happen, to those same fifty people, if the clinic had done nothing but book them in and measure them twice. The uncontrolled study cannot separate its treatment from the waiting, because it never measured the waiting.

This is the missing piece under the pattern this site keeps describing — in PEMF, in low-level laser, in the homeopathy literature — where the well-designed trial finds nothing and the uncontrolled one finds benefit. The uncontrolled one is not lying. It is measuring something real. It is just not measuring what it claims to.

What the people who waited were actually experiencing

Calling it "the placebo effect" is too narrow. Wait-list participants received no dummy treatment, so there was no placebo to respond to. At least four things were going on:

• Natural fluctuation. Tinnitus varies. Flare-ups come and go, and a measurement taken on a bad day will usually be followed by a better one.
• Regression to the mean. People enrol in trials when things are bad. Bad is, statistically, a departure from their own average, and departures tend to come back.
• Assessment itself. Being asked careful questions about your tinnitus by someone taking it seriously is not nothing — a point the article on being dismissed by a doctor approaches from the opposite direction.
• Expectation. Being on a wait-list means help is coming, which is its own intervention.

The meta-analysis found the effect was moderated by the methodological quality of the trial and by the participants' age and tinnitus duration — which is what you would expect if it is a mix of real phenomena rather than one clean mechanism.

What this does not mean

It does not mean tinnitus gets better on its own in any useful sense. Three to eight percent over three months is not recovery, and nothing here contradicts what can tinnitus be cured says.

It does not mean waiting is a strategy. The trials this data was extracted from exist because their treatment arms did substantially better than their wait-list arms. That difference is the whole point, and it is what CBT's evidence base is made of.

And it does not mean improvement you experience is imaginary. If your tinnitus bothers you less than it did three months ago, that is real regardless of what caused it. The finding is about what a study can conclude, not about what a person felt.

Three questions this gives you

When you next meet a claim that something helped tinnitus:

1. Was there a group who got nothing? If not, some of the reported improvement is the 3–8% above, and there is no way to know how much.
2. How much bigger was the treatment group's change? Not whether it improved — by how much more.
3. Over what period? The waiting effect was measured at 6 to 12 weeks. A study reporting improvement at three months is reporting over exactly the window in which doing nothing also works a little.

How this site rates evidence sets out the wider standard, and how tinnitus is measured covers what the instruments in this meta-analysis actually capture — including the 13-point threshold on the Tinnitus Functional Index, which is the other number worth carrying around.