Ulthera (Ultherapy) — a device we do not have, and its evidence is thicker than ours

We have repeated the same sentence in how far the evidence for XERF has come and in the three radiofrequency devices compared — “there was no control group”. Keep writing that and a question follows of its own accord. Which lifting device does have a trial with a control group, then, and what did that trial say?

Within the range we searched, the only one in this class that met that condition was Ulthera. That is why we are writing about someone else’s device. The worth of this column is not in whether Ulthera is good or bad but in what happens to the numbers once a control group is added.

In an earlier column we wrote that “it is not that no clinical trial was submitted because the route was 510(k); what decides whether clinical data are submitted is the size of the indication being claimed”. Ulthera is both the counter-example to that sentence and the evidence for it.

There is a word that must not be dropped when this is quoted. The Ulthera indication statement opens with “… to achieve temporary changes in the physical appearance of the skin”. “Temporary” is in the original. And clearing 510(k) is still not an approval — 21 CFR 807.97 of the US federal regulations states directly that a representation creating an impression of official approval is a misbranding. We have set that rule out in how to read the numbers, and it applies to Ulthera in exactly the same way.

There is a trial the manufacturer (Merz North America) registered and for which it also posted results — registration number NCT04795622. Within the range we searched, this is the only randomised trial in this class with an untreated control group.

The results were these.

The same trial, the same patients. Scored by the physician who performed the procedure, 96.8% improved. Scored by blinded assessors looking only at photographs, 54.2% improved. And 47.0% of the people who had nothing done at all were judged to have improved too. Even the adverse events were almost the same — 36 subjects in the treated group against 34 in the control group — which means a good share of what was reported was not caused by the procedure but simply happened during that period.

Reading this result as “Ulthera does not work” would be wrong. This is one trial, one session, at the 90-day time point, at 5 sites in China, in mild to moderate laxity. The accurate sentence is “under these conditions, superiority over an untreated control was not demonstrated”, not “a lack of effect was proven”. It is equally a fact, though, that this is the only trial in this class with a control group.

A systematic review and meta-analysis published in 2025 gathered the literature in this class in one place. 4,019 records were screened and 42 studies were included, of which only 2 were controlled clinical trials, 9 were observational studies, and the remaining 31 were designs comparing before and after the procedure only. The level of evidence is marked as grade 3.

• Improvement rated by the treating physician 89% (95% confidence interval 81 to 94, n=411)
• Improvement rated by the patient 84% (73 to 91, n=312)
• Satisfaction 84% (61 to 94, n=326) — face and neck 93%, other areas 63%
• Mean pain 4.85/10 (4.35 to 5.35, n=785) — face and neck 4.62, other areas 5.58

One thing here the authors pointed out themselves. Pooling only the studies whose satisfaction questionnaire offered a “neutral” option gives 62% (95% confidence interval 37 to 82), and pooling only the studies that did not offer it gives 91% (46 to 99). The authors put it this way — where the neutral option is absent, some patients who belong in neutral may be misclassified as satisfied, and so the efficacy may have been overestimated. One option on a questionnaire moved satisfaction between 91% and 62%.

We carry over the limitations the authors stated, as they wrote them — “many studies lacked control groups”, which makes robust comparison difficult, and the absence of randomisation and blinding raises concern about measurement bias. In the quality assessment every non-randomised study was judged to carry confounding bias, and for two outcomes, the investigator rating and satisfaction, publication bias was detected as well.

The same meta-analysis gives the device make-up of the 11 studies covering facial areas — Ulthera 63.7%, Ultraformer III (Classys, Korea) 27.3%, Microson (Cosmoplus, Korea) 9.1%. That is, the literature cited together as “the evidence for focused ultrasound” has data from devices other than Ulthera mixed into it. The authors wrote as much themselves — “the interchangeable use of ultrasound technology names in clinical practice may cause confusion”.

In consultation this shows up in two directions. One is material introduced as “an Ulthera paper” that is in fact data from another device; the other is whether the device and its consumable (the transducer cartridge) are genuine. Korean pharmaceutical trade press has reported several times that the manufacturer has run a scheme for verifying genuine product through hologram stickers and a mobile application. Whether that scheme is still run in the same form today, however, we could not verify, and the stories that circulate about the consumables used at particular clinics could not be confirmed at first hand, so we do not carry them here.

The Korean MFDS product licence number we again could not verify. The MFDS electronic civil-application portal and the medical device information portal block automated lookup, so all we can offer is how to look a number up yourself.

The number we looked at longest while writing this was not 54.2% but 47.0%. It is the proportion of people who had no procedure at all whose photographs at least 2 of 3 blinded assessors judged to have improved. Lighting, angle, swelling, expression, the state someone happened to be in that day, and the bias that arises whenever a person is asked to pick one of two photographs are all inside that figure.

So what does an improvement rate obtained from before-and-after photographs alone, without a control group, mean? It means this 47% is mixed into it. And the numbers we have written on our own reference pages are all of that kind.

• XERF — three studies in people, none of the three with a control group
• Density — 16 subjects, no control group
• Oligio Kiss — on class data, no controlled trial of this device itself

We have no basis for saying our devices are better than Ulthera, and we also do not know how far above 47% the improvement rates for our devices sit. Nobody has run that trial. The most honest sentence we can write here is this one — if the device that does have a trial with a control group failed to show superiority in that trial, the numbers for devices that have no controlled trial at all should be read more carefully still.

• “Is this material data for this device?” — the focused ultrasound literature has several devices mixed into it
• “Does this material have a control group?” — that one question decides how much a number weighs
• “Who rated this number?” — the physician who performed the procedure, or a blinded assessor
• “At what month is this number?” — the randomised trial above is at 90 days, the décolleté clearance material at 180 days
• Pain — the meta-analysis mean is 4.85/10. Ask about it together with the method of anaesthesia

The consultation we give at Miso Clinic is the same. We do not offer Ulthera. We have written the evidence for the devices we do hold, and its limits, to the same standard in our clinical reference library, so we suggest putting the two sets of material side by side.

This column is a summary of the material on a device Miso Clinic does not hold. Whether to have a procedure and which device to use are matters to decide in discussion with the medical staff treating you, and we do not offer this device.

Written and reviewed by Lee Chi-Hak, MD, medical director of Miso Clinic in Daegu, South Korea. Every study cited above is given together with the conclusion of the original and its limitations, and where we could not verify something, we have said that we could not verify it.