Deoxycholic acid has the thickest evidence of any procedure on this site — and still is not 'strong'

Both were randomised, double-blind, placebo-controlled phase 3 trials in adults with moderate or severe submental fat who were dissatisfied with it.

The coprimary endpoints were the same in both: a composite response of improvement by at least one grade (or at least two grades) on both the clinician-reported and the patient-reported scale. Assessment was at 12 weeks after the last treatment.

The word 'composite' matters here. A response was not counted when only the doctor saw improvement, nor when only the participant did. Both had to improve for it to count. Given that on the PRP page the doctors and the participants reached opposite conclusions within one trial, you can see how demanding that condition is.

REFINE-1 (256 on deoxycholic acid, 250 on placebo):

• At least one grade, composite: 70.0% versus 18.6% (P < .001)
• At least two grades, composite: 13.4% versus 0% (P < .001)

REFINE-2 (258 per group):

• At least one grade, composite: 66.5% versus 22.2% (Mantel-Haenszel risk ratio 2.98, 95% CI 2.31–3.85)
• At least two grades, composite: 18.6% versus 3.0% (risk ratio 6.27, 95% CI 2.91–13.52)
• P < .001 for both

Two separate trials produced almost the same numbers. 70.0% and 66.5%; 18.6% and 22.2%. Agreement that close is hard to explain by chance.

Now the things to read carefully.

First, the placebo groups improved too — 18.6% and 22.2%. One in five people injected with something inert met the composite definition of 'improved by a grade'. No figure demonstrates better why uncontrolled trials are dangerous.

Second, most of it was one grade. The two-or-more-grade response was 13.4% and 18.6%. Eight in ten treated people improved by one grade or not even that. 'Seventy per cent improved' and 'the jawline was transformed' are different sentences.

Third, it takes several sessions. In REFINE-1, one-grade improvement on the clinician scale was reached by 55% after two treatments and 75% after four.

So why is this not 'strong evidence'?

This site's rule is that evidence resting entirely on manufacturer-funded trials does not get the top badge. REFINE-1 and REFINE-2 were both run by the company that developed the drug (Kythera Biopharmaceuticals), whose staff appear in the author lists.

The rule is not a statement of suspicion about the results. The designs are excellent and the numbers agree with each other. But we decided that this site's highest badge goes on when the same result has also come from somewhere without a stake in it. Applying the rule without exceptions is the point of having it.

This item is what separates this page from the others in the layer.

REFINE-1 measured submental volume directly by MRI.

• MRI responders: 46.3% on deoxycholic acid versus 5.3% on placebo (P < .001)
• More than eight times the difference

Why that matters.

On most procedure pages on this site the endpoint is a photograph score or a self-rating. Both pass through human eyes and memory. On the PRP page, blinded doctors and the participants reached opposite conclusions inside a single trial.

MRI does not go through that channel. The volume of fat does not change with expectation, lighting or the mood of the day.

And the MRI result pointed the same way as the human assessment in the item above. When two endpoints of different natures point to the same place, the result is far more trustworthy than either alone.

But look at the gap between the numbers too. By human assessment 70.0% responded; by MRI, 46.3%. Same people, same trial.

Asking which is 'the real one' has no simple answer. It depends on where the MRI response threshold was set, and a small drop in volume that changes the shadow along the jawline can look larger to the eye.

What can be written down is this: about two thirds of the improvement people saw was also confirmed on imaging.

The least-answered question in aesthetic medicine is 'how long does it last'. On almost every procedure page on this site we have written 'not established'.

For deoxycholic acid there is a follow-up report tracking phase 3 participants for up to three years.

The report's thrust is that improvement in submental contour was maintained for up to three years.

The conditions to attach to this are large.

There is no control group during the follow-up. You cannot hold a placebo group on placebo for three years. So this data shows what happened to people who were treated, without any comparison against what would have happened had they not been.

Who changed weight over three years, who had other procedures, and who chose to stay in the study to the end all enter the result. That people with good outcomes stay in follow-up more readily is a well-known bias.

So this item sits at 'early evidence' with the comparator marked 'none'. It should not be read with the same weight as the two items above.

Even so, it is rare data for this layer. That a three-year follow-up exists at all is a condition most procedures do not have.