Looking into XERF in Daegu — filling in the material we had missed

The conclusion, first

“Which place in Daegu is good at XERF” is a question that public material cannot settle. There is no public measure comparing one operator’s skill with another’s, reviews cannot serve as evidence of an effect under the medical advertising law, and a count of procedures performed is not the same thing as skill. So we write a different answer — we set out how far the evidence for this device has come as of now, and we publish what we check in consultation. There are three points. First, the human data is not one paper but three — a multicentre study of 39 people (two sessions) and a single-centre retrospective of 16 people (one session). Second, none of the three has a control group, and in two of them the doctor who performed the treatment scored the result. The authors themselves wrote that the work is “preliminary” and that “no conclusion can be drawn that it is better than an existing single-frequency device”. Third, the 2024 porcine histology paper we had cited was published with “no conflicts of interest” and was corrected in March 2026 to “seven of the authors are employees of the manufacturer”. It is true that the evidence has grown, and it is equally true that you have to look at who produced that evidence.

The three sets of human data now available

The picture we set down in the evidence piece was “one paper in humans (20 people) plus five manufacturer-funded preclinical studies”. In fact it already looked like this at that time.

Reading the numbers on their own leads to misunderstanding. “92.3% improved at 90 days” is a fact, for instance, but the person who gave those scores was the doctor who performed the treatment, and it was not blinded. The authors of the paper wrote themselves that “assessor bias cannot be excluded”, and put their conclusion as “hypothesis-generating rather than definitive evidence”. The p values attached for statistical significance are not values against a placebo or a control group either, but post hoc tests against a benchmark rate of “70%”, and the authors pinned them down as “exploratory”.

Neither of the two new studies concluded that XERF is better than an existing single-frequency radiofrequency device. The Israeli study states in its text that “in the absence of a control group, and without any comparison with a single-frequency device, no conclusion whatever can be drawn about the superiority of dual-frequency technology”.

★ A conflict of interest statement corrected two years later

One of the papers we cited as preclinical material in the evidence piece was a porcine histology study published in 2024. It originally carried this sentence — “The authors declare no conflicts of interest.”

In March 2026 the same journal published a correction. The corrected sentence reads — “J.H., H.G.R., C.P., J.P., K.K., K.M.M.L., S.I.C. are employees of Cynosure Lutronic, the manufacturer of the monopolar radiofrequency device evaluated in this study.” That seven of the authors were with the manufacturer came to light two years afterwards.

Our conclusion does not change — the evidence piece had already classified this paper as “authors employed by the manufacturer, or manufacturer-funded”. What the episode shows, though, is that even a conflict of interest statement printed in a journal can be overturned later. That is why, when we cite material, we set down the authors’ affiliations and the funding source alongside the conclusion. It is the reason the “Limitations” column of the table above carries whether the manufacturer supported the work.

Why “how many shots of XERF” means a different treatment at every clinic

The number that comes up most often in consultation is the shot count. Yet the same number does not mean the same treatment. The two studies newly added here show exactly that — both used XERF and both reported good results, and yet their protocols run flatly against each other.

The same words, “600 shots”, mean the whole face with one large tip on one side and three tips divided between areas on the other. When the tip areas differ, the area actually heated differs several-fold at the same shot count. So in consultation we do not give the shot count on its own; we set down which tip, on which area, for how many shots alongside it. This is written out in detail in what the number “shots” really is and what changes when the EFFECTOR tip changes.

There is still no study in humans that randomised shot count and compared the arms. The numbers above are a record that “this is what was actually used in this study”, not evidence that “this much is the right amount”.

Pain — the two studies reported almost the same value

Without anaesthesia the figures were a mean of 4.2/10 (39 people) and 4.31 ± 1.71 (16 people). That values arrived at in different countries · under different protocols come out close is worth noting. But you have to look at the distribution alongside them — in the 16-person study 38% were mild (3 or below), 50% moderate, and 12% (2 people) had severe pain of 7 or above. A mean in the fours does not mean that everyone felt a four.

The American study describes this as lower than the mean pain reported for the Thermage family (CPT 6.06, FLX 6.9 — the latter even with analgesic premedication). The authors themselves added, though, that “comparison between studies of differing design and conditions carries inherent limitations”. We too do not carry that comparison across as a settled “it hurts less”.

What still has not been confirmed

• There is not a single randomised controlled trial. None of the three sets of human data has a control group.
• There is no evidence for how long it lasts. The longest follow-up is 90 days, and the six-month photograph was presented as the case of a single patient, which the authors stated they had not included in the analysis.
• We were not able to confirm a Korean marketing approval number. The search at the Korean Ministry of Food and Drug Safety (MFDS) is blocked, and it is not on the manufacturer’s official site either. What we did confirm is only the medical device advertising review number. The difference between an approval number and an advertising review number is set out in an approval number and an advertising review number are different things.
• The phrase “FDA approved” is not accurate. What passed in the United States is a 510(k), the indication cleared is not lifting but electrocoagulation and hemostasis, and no clinical trial material was included in the submission. US federal regulation 21 CFR 807.97 defines it as misbranding to present a 510(k) clearance as though it were an approval.
• There is no data directly measuring how deep the heat reached in humans. Every account of depth comes from computational models and animal data, and the authors of the new studies recorded this as a limitation as well.

Five things worth asking in consultation

These are questions you can ask at any clinic. If the answers are specific, and if it is a place that says it does not know when it does not know, that helps you judge.

1. “Which tip, on which area, for how many shots?” — a shot count alone does not let one clinic be compared with another.
2. “How many sessions is this plan premised on?” — the published material mixes one-session and two-session work.
3. “Who gave that effectiveness figure?” — a blinded assessor, the doctor who performed the treatment, or the patient’s own report.
4. “What is the evidence for how long it lasts?” — whether the follow-up really ran that long, or whether it is an estimate.
5. “Is there a reason you would not recommend this for me?” — an explanation that does not state the limits is not material you can judge on.

Confirmed / inferred / not confirmed / material pointing the other way

Frequently asked questions

Who wrote this

References

1. New — prospective multicentre, 39 people — Weiss RA, Wang J, DiBernardo B, Bhatia AC, Cureus 2026;18(3):e104546, PMID 41930066, DOI 10.7759/cureus.104546. Four centres · 39 people (37 women · 2 men) · aged 28–70 (mean 54.4) · Fitzpatrick I–IV · all with two sessions (4 weeks apart) · EFFECTOR 60 alone · 300–677 shots per session (mean 568.8) · total energy mean 56,391 J (the paper states 100.7 J per shot, but dividing the total by the shot count gives 99.1 J, so the two do not agree) · pain without anaesthesia mean 4.2/10. GAIS of 3 or above in 33/39 (84.6%) at 30 days · 36/39 (92.3%) at 90 days. No control group · unblinded, with the treating investigators scoring the GAIS · follow-up 90 days · the manufacturer provided the devices · study support · data management · statistical programming. The p values are not against placebo but post hoc one-sided exact binomial tests against a benchmark rate of 70%, and the authors wrote that they were “exploratory”.
2. New — retrospective 3D measurement, 16 people — Erlich G, Dahan E, Wolf Y, Lasers in Medical Science 2026;41(1), PMID 42611100, DOI 10.1007/s10103-026-04996-0. A retrospective analysis of treatments carried out at the first author’s own clinic in January 2026 · 16 women · median age 54.5 (25–73) · one session · 3 months. Pain VAS 4.31 ± 1.71 (3 or below 38% · 4–6 50% · 7 or above 12%). Brow · eyelid measurements at three months +1.11–1.44 mm (the right lateral canthus value, however, was not significant at p=0.562). No control group · the authors state that, without a comparison against a single-frequency device, no conclusion on superiority is possible · the volume redistribution measure is a surrogate and not a validated standard one. The six-month photograph is the case of a single patient and was not included in the analysis.
3. A report of a small-tip periorbital protocol — Erlich G, Dahan E, Skorochod R, Wolf Y, Cureus 2026;18(1):e101187, PMID 41523730, DOI 10.7759/cureus.101187. We have not counted this paper in the tally of evidence — because the authors themselves wrote that “the observations are anecdotal and are not presented as quantitative outcomes” and that they “should not be interpreted as evidence-based conclusions”. It does, though, confirm the existence of the new small tips (E10 · E5).
4. ★ The conflict of interest correction — Skin Research and Technology 2026;32(3):e70335, PMID 41749463, DOI 10.1111/srt.70335. The original paper is Hong J et al. in the same journal, 2024;30(6):e13821 (porcine histology, DOI 10.1111/srt.13821), the paper cited as preclinical material in our evidence piece. It was originally published with “the authors declare no conflicts of interest” and was corrected to “J.H., H.G.R., C.P., J.P., K.K., K.M.M.L., S.I.C. are employees of Cynosure Lutronic, the manufacturer of the monopolar radiofrequency device evaluated in this study”.
5. Existing — the Korean clinical study of 20 people — Hwang JK, Medical Lasers 2025;14(1):23–30, DOI 10.25289/ML.24.035. 20 people · aged 28–63 · Fitzpatrick III–IV · one session · 12 weeks. Blinded assessors identified the post-treatment photograph correctly 80% of the time, mean GAIS 2.75/4, patient satisfaction 7.84/10, pain VAS 4–5 (no anaesthesia), no adverse events. No control group · single centre · n=20 · one session · 12 weeks, and the journal is indexed in KCI · KoreaMed and not in PubMed/MEDLINE · SCIE.
6. The US 510(k) and the word “approval” — FDA 510(k) K251327 (decision 2025-08-11, Substantially Equivalent, product code GEI, Class II, predicate device Thermage FLX K170758). The summary states in its own text that “no clinical tests were included as part of this submission”, and the US indication wording is electrocoagulation and hemostasis. US federal regulation 21 CFR 807.97 provides that a 510(k) clearance “does not in any way denote official approval, and any representation that creates an impression of official approval is misleading and constitutes misbranding”.
7. The law on medical advertising — Article 56(2) of the Medical Service Act (the Korean statute governing medical advertising). It prohibits advertising through accounts of treatment experience, comparison with other medical institutions, superlative expressions and the like. It is the reason this column carries no expression such as “good at” or “the best”, no comparison with other medical institutions, and no price · discount figures.