When results actually start to appear, and when you should judge them

Let us start by putting everything side by side. Each cell is a time point that a trial actually measured.

What to read from the table is that there is no single rule of “it takes so many weeks”. Toxin is two days; PCL is four years.

This is the fastest treatment. Onset has been measured in a study of 41 Korean patients.

Two things follow from this.

First, lines at rest take longer to smooth out. On day two, 85% responded on frowning while the figure at rest was 51%. The moment when you stand still in front of the mirror and feel “nothing has changed” is genuinely there in the data.

Second, the practice of “reviewing and touching up at 2 weeks” has no evidence behind it. We looked, and there is no trial comparing a 2-week touch-up with a 4-week touch-up, and the 2024 Korean expert consensus statement contains no recommendation on the assessment time point either. What has been measured is that the peak of effect is at 30 days and the peak of satisfaction at 2 to 3 months. Two weeks is not the peak.

We often hear that “it takes 2 weeks to settle in”. While looking for the evidence behind that phrase, we found something interesting. There is no imaging study that measured volume at the 2-week time point at all. The existing studies all jump from “immediately → 30 days” or “immediately → 24 weeks”.

What has been measured is the following.

• Swelling generally resolves within a week. In the manufacturer's patient diary data, swelling after cheek treatment occurred in 85.7%, and 41% resolved in 1 to 3 days and 33% in 4 to 7 days. Bruising lasts longer, with 29.6% at 8 to 14 days.
• Volume is at its maximum immediately and declines from there. In a study using ultra-high-frequency ultrasound (13 patients), dermal thickness went from +11% immediately to +7.4% at 24 weeks, and three-dimensional volume was about 45% retained at 24 weeks relative to immediately after.
• That said, the direction is opposite depending on the product. In a randomised lip study (20 patients), one product fell from 20.56mm immediately to 19.56mm at 30 days, while another rose from 19.02 to 19.62.

So the accurate phrasing is “as the swelling goes down it becomes less full than it was at first”, not “it settles in”. And the figure of 2 weeks itself has never been validated. Our practice of seeing you again at around 2 weeks is a convention, and we say so.

When explaining boosters, the phrase “collagen starts forming from 4 to 6 weeks” is commonly used. We used it too. We could not find evidence confirming this from human biopsies.

The time points at which tissue was actually taken from humans are these.

How you read this matters. “Not visible at 2 weeks” does not mean “it does not work”; it means “it is still too early”. And for PLLA and PDLLA, human biopsies in the 4 to 8 week window do not exist at all. That window is a gap in the data, not an absence of effect. The common explanation of a “peak at 4 weeks” comes from mouse data, not human data.

We should also disclose one piece of conflicting data among the boosters. For PDLLA, a randomised · evaluator-blinded trial in 260 Asian patients reports that it was already superior to the control at 4 weeks and declined rapidly after 24 weeks. A 40-patient study with no control group, by contrast, reports continued improvement from 8 weeks to 24 weeks. The former study is the stronger design.

The radiofrequency study of 20 Korean women conducted with blinded evaluators is the most practical piece of data in this article.

The most interesting thing in this table is that the two values in the 4-week row disagree with each other. By instrument it has already improved, but the human eye cannot see it yet. The physical properties of the skin change first and appearance follows later.

The other data point the same way.

• 39-patient multicentre study: improvement rate 84.6% at 30 days → 92.3% at 90 days
• Two independent studies report a peak at 4 to 6 months. One reports a slight decline at 7 to 10 months
• Ultrasound lifting: quantitative lift is maximal at 90 days and falls to 22 to 33% at one year

So the point at which judgement is reasonable is 3 months, and the peak is later still. Being disappointed at 4 weeks is natural, but on the evidence it is not yet the place from which to judge.

This section is the most surprising part of the article. How differently patients and observers see the same treatment at the same time point has been measured.

The direction flips completely depending on when you look.

• At 90 days patients see less than observers (42% versus 92%). That is why they are disappointed early.
• At one year patients see more than observers (95% versus 43%). That is why they remain satisfied after the effect has entirely worn off.

These two lines explain almost every misunderstanding in the consulting room. Early on it feels like “it did not work”, and later it feels like “it is still fine”. That is why we take photographs. Photographs are more accurate than memory and mirrors.

There is experimental evidence behind the phrase “I see it every day so I cannot tell”.

Visual adaptation. In one experiment, 48 women were split into two groups and repeatedly shown either very full lips or very thin lips. Within minutes, the standard for “attractive” shifted towards whichever they had been exposed to. Ratings of “natural”, however, did not shift.

The mirror effect. In a study of 198 people asked to choose between a photograph of their own face and its left-right reversed version placed side by side, 56.6% preferred the mirror image of their own selfie, while for photographs of other people the majority preferred the true image. You see yourself as a mirror image, while we and your family see you as the true image. In effect you are looking at different faces.

So you cannot catch the change with a mirror every day. Two photographs from the same position, in the same lighting, at the same angle are far more accurate. We strongly recommend having a photograph taken before treatment — it is the only way to determine later whether things really did change.

We would add that a phase of “looking worse at first and then getting better” has never been measured for any treatment. What has been measured is the opposite — filler looks fuller immediately than it really is, and lifting is recorded early on not as ‘worse’ but as ‘not yet’.

To answer first: we could not find a trial comparing the ‘time of onset’ between one session and several. What has been measured in relation to the number of sessions is magnitude, not timing.

• In a microneedling radiofrequency study (24 patients analysed), there was a positive correlation between the number of sessions · total energy and volume change (r=0.676, p<0.001). That said, it was not a comparison of schedules.
• In a study completed by 9 Korean patients (3 sessions 4 weeks apart), wrinkles became significant from 8 weeks (after the second session), and every measure was at its best 6 months after the final session. — meaning things keep improving even after the course has ended.

And satisfaction did not rise with larger amounts either. Satisfaction by toxin dose was 60% at 40 units → 51% at 60 units → 39% at 80 units. Dissatisfaction is linked more often with overcorrection than with undercorrection. We have written about this topic in detail in what is the difference between a natural face and an artificial one?.

1. We take photographs before treatment. Mirrors and memory cannot settle the question. This is the single most practical measure.
2. We give a different time point for each treatment. Toxin two days · radiofrequency 3 months · the collagen-stimulating agents later than that. There is no single answer.
3. You should not judge radiofrequency and ultrasound before 3 months. At 4 weeks there was no significant difference even to a clinician's eye.
4. For toxin, reviewing at around a month rather than at 2 weeks fits the evidence. The peak of effect is at 30 days. That said, if something is clearly under-treated we will see you before then.
5. With filler, it becomes less full than it was at first as the swelling goes down. If you are satisfied immediately afterwards, it may in fact be overcorrected.
6. We do not add more straight away just because it looks insufficient. Dose and satisfaction were not proportional.
7. We distinguish between “it is still early” and “it did not work”. The numbers in this article are the basis for that distinction.

The redness and swelling of the first few days after treatment, and when you should contact us, are covered separately in aftercare following RF (radiofrequency) treatment and the order in which bruising and swelling pass after treatment. Number of sessions and duration are in number of sessions and how long results last.

• What we confirmed — in the radiofrequency study of 20 Korean women, there was no significant improvement on clinician assessment at 4 weeks and it became significant at 12 weeks. In the same study, instrument-measured elasticity improved from 4 weeks.
• What we confirmed — the peak of toxin's effect is at 30 days, and the peak of satisfaction at 2 to 3 months. Two weeks is not the peak.
• What we confirmed — on human biopsy, no new collagen was detected at 2 weeks, with type III confirmed at 2 months and type I at 3 to 4 months.
• What we confirmed — at 90 days, practitioners 92% versus patients 42%; at one year, observers 43% versus patients 95 to 100%. The two views flip depending on when you look.
• What we could not find evidence for — “filler settles in at 2 weeks”, “judge toxin at 2 weeks”, “collagen forms from 4 to 6 weeks” (no human evidence, mouse data), “more sessions make it appear sooner”.

So the conclusion of this article is not “wait a little longer” but “each treatment has a set point at which to judge, and it is better to know it before you start”. Waiting is unsettling when it is vague, but when it is set at 3 months, that is a plan.

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 its design, its size and the limitations the authors themselves recorded, and where we could not find data, we have said that we could not find any.