What Causes Rosacea, and What Actually Helps?

Rosacea is not one thing going wrong. The best-supported account is an overactive innate immune response — a cathelicidin peptide called LL-37 — running alongside blood vessels and nerves that overreact to heat, alcohol and stress. Demodex mites are a likely trigger. Skincare can calm it; only drugs have high-certainty evidence.

In November the gap between a Hongdae street and the café you duck into is about twenty-five degrees, and we see the same face several times a week: flushed across the cheeks and nose, years of it, and the word used is always “sensitive.” A Korean multicentre study of 580 rosacea patients across fourteen hospitals found that 47.4% had relatively low awareness of their own condition. That number matches our appointment book almost exactly.

What is actually going wrong in rosacea-prone skin?

Two systems are overreacting at once, and the immune one is better understood than the vascular one. A 2024 review in Frontiers in Immunology sets out the best-supported chain: UV light or Demodex activates TLR2 on keratinocytes, which raises kallikrein-5; kallikrein-5 cleaves cathelicidin into the peptide LL-37; LL-37 activates mast cells, neutrophils and macrophages, which release enzymes that convert still more kallikrein-5. A loop. The part measured directly in human tissue is that rosacea skin carries more cathelicidin and LL-37 than normal skin. The loop itself is assembled from mouse models, keratinocyte cell lines and small clinical pilots.

Running beside it is the neurovascular half. TRP channels — TRPV1, TRPA1, TRPV4 — sit on nerves, vessels and keratinocytes and respond to heat, alcohol and capsaicin by releasing neuropeptides such as CGRP and PACAP, which dilate vessels and degranulate mast cells. In humans those peptides are elevated, and injected PACAP38 produced flushing and facial swelling. The review is explicit that which half fires first is unresolved.

We have written before about why repeated flushing eventually leaves visible vessels behind. This is the other question: what is driving the flushing in the first place.

Do Demodex mites cause rosacea, or do they just live there?

They are a credible trigger, not a demonstrated cause — and nearly every adult has them. Demodex folliculorum is a normal finding in adult skin. Dermatology references define demodicosis not by the presence of mites but by density, conventionally above five per square centimetre. People with rosacea do carry higher facial densities than controls.

The proposed mechanism is tidy: the mite's chitin exoskeleton raises TLR2 signalling and can activate the inflammasome, which feeds straight into the LL-37 loop above. The clinical argument, though, runs backwards — drugs that kill mites, such as ivermectin and metronidazole, improve the lesions. That is indirect evidence, because those same drugs are anti-inflammatory in their own right.

Which rosacea treatments actually have good evidence?

A Cochrane review of 106 studies and 13,631 participants, searched to July 2014, put only a short list in the top tier. For papules and pustules, high-certainty evidence supports topical azelaic acid, topical ivermectin, oral doxycycline and low-dose oral isotretinoin. For redness, high-certainty evidence supports topical brimonidine, working for up to twelve hours. Moderate certainty: topical metronidazole and oral tetracycline — while topical clindamycin with tretinoin was no better than placebo. Low certainty: low-dose minocycline, and laser or intense pulsed light for visible vessels.

Two honest notes about that list. Almost all of it is prescription-only, and none of it is a decision an esthetics studio makes. And strong evidence does not mean costless: in a manufacturer-funded retrospective review of roughly 1,000 brimonidine users, 29% in the 52-week arm reported treatment-related adverse events and 14.9% stopped because of them, with redness commonly returning eight to twelve hours after application as the drug wore off.

Why does Seoul set it off in November?

Because the triggers with the clearest mechanism are heat, alcohol and stress, and a Seoul winter evening delivers all three. A jjimjilbang, a hot bath, soju, a bowl of malatang and the twenty-five-degree step from street into a heated room are, as far as TRP channels are concerned, the same instruction.

The Korean data complicates the usual advice, though. In that 580-patient study the two most commonly reported aggravating factors were emotional change (51.7%) and stress (48.4%) — less actionable than “avoid spicy food.” Trigger rankings are also not universal: a Tunisian series cited in a review of rosacea in skin of colour found sun exposure leading at 64%.

One more detail that matters here. In a Korean series of 168 patients, 96.4% had the erythematotelangiectatic subtype — flushing and vessels rather than bumps and pustules. Reviews of rosacea in skin of colour note this is also the presentation most easily missed, because the hallmark redness is harder to see and burning or itching may be the clearer clue.

Know your own skin before you buy anything

If this describes you, the instinct is to go and buy the calming line. Hold that for a moment. Ingredients should go in by stages, matched to the condition your skin is in now rather than the condition you want it in. Someone whose body is in the middle of fighting something serious is not helped by a harder training programme, however good training is for a healthy body; what that situation calls for is rest, and effort spent against it is borrowed from a recovery that has not happened. Reactive, inflamed skin is in that position. The first move is removal — fewer actives, no scrubs, no heat — and only then, slowly, one thing back.

And none of this is a diagnosis. Rosacea is diagnosed clinically, with no confirmatory test, which means a doctor makes the call by looking, and several other conditions look similar. Persistent facial redness, eye irritation, or bumps that will not settle belong in a dermatology clinic, not in a trial of a new toner.

The case against some of this

Three things above are weaker than they sound.

  • “The LL-37 loop explains rosacea.” By the review's own account, the feedback loop rests mainly on mouse models and in vitro keratinocyte work; human clinical support is small-scale, including a rapamycin pilot in 18 women, and some neuropeptide claims come from a 13-patient case series. The review states plainly that whether nerves or immune cells fire first is hypothesised, not known. Narrowed: LL-37 is the strongest single thread, not a settled explanation.
  • “Demodex is a trigger.” The mites are a normal finding in adults, and the demodicosis threshold of five per square centimetre is a working convention rather than a biological cliff. Measured density also depends on sampling method — oil scraping, skin biopsy and surface biopsy are all in use. And the treatment evidence is indirect, since anti-mite drugs are independently anti-inflammatory. Narrowed: plausible contributor, unproven cause.
  • “Barrier skincare helps rosacea.” The strongest collection I could find is a 2026 narrative review — not a systematic one — whose supporting trials run from 20 to 82 patients over four to eight weeks, with non-standardised formulations and frequently unstated subtype. Its authors say rosacea-specific randomised trials are lacking, and one split-face trial of a Centella asiatica foam and cream found no difference in erythema between the active and vehicle sides. Two of its three authors declare honoraria from cosmetic manufacturers. Narrowed: skincare's job is to stop making things worse and to keep medical treatment tolerable — not to treat the condition.

Questions we get asked

Is rosacea curable?

The literature describes it as managed rather than cured. It is a chronic, relapsing condition, and the trial evidence is about controlling features — redness, papules, vessels — not eliminating the condition. What your own course looks like is a question for a dermatologist, not for a blog.

Can skincare alone fix rosacea?

The evidence does not support that. Trials of barrier moisturisers in rosacea are small and short, while the interventions with high-certainty evidence are prescription drugs. Good skincare has a real but narrower job: avoiding flares and making medical treatment tolerable enough to stay on.

Why do so many people in Korea not realise they have it?

In the 580-patient multicentre study, 47.4% had relatively low awareness of their condition, and in a separate Korean series 96.4% had the flushing-and-vessels subtype — which reads easily as “sensitive skin” or just a personal colouring. Reviews note rosacea is under-recognised wherever the redness is harder to see.

If you are in Seoul

We are an esthetics studio, not a clinic: we do not diagnose rosacea and we do not treat it. What a 1:1 consultation can do is work out which of the usual triggers are actually yours, strip a routine back to something that stops adding fuel, and say plainly when the next step is a dermatologist rather than another product. If you are here through the cold months, that conversation is worth having before you buy anything.

Sources

  • Advances in the pathogenesis of rosacea (Frontiers in Immunology, 2024) — narrative review; full text read. Used for the kallikrein-5/LL-37 pathway, TRP channel signalling and the Demodex mechanism, including the review's own labelling of which parts are human-tissue findings and which are mouse or in vitro.
  • Interventions for rosacea (Cochrane, CD003262) — systematic review of 106 studies and 13,631 participants with moderate-to-severe rosacea, searched to July 2014, with GRADE certainty ratings. Plain-language summary read.
  • Epidemiological and clinical features of rosacea in Korea (The Journal of Dermatology, 2018) — multicentre cross-sectional study, 14 Korean hospitals, 580 patients, October 2014 to February 2015. Abstract only, read via the institutional repository.
  • Rosacea in skin of color: a comprehensive review (IJDVL) — narrative review; used for the diagnostic difficulty in darker skin, the Korean series of 168 patients (96.4% erythematotelangiectatic) and the Tunisian figure of 64% for sun exposure. Read in full except the final portion of a very long page.
  • DermNet NZ — Demodex, demodicosis, for mites as a normal adult finding, the >5/cm² convention and the sampling methods in use.

Sources for the case-against section

  • Rosacea and Sensitive Skin: Mechanism-Based Use of Cosmeceuticals in Clinical Practice (J Clin Aesthet Dermatol, 2026) — narrative review, not systematic; supporting trials of 20 to 82 patients over four to eight weeks; authors declare honoraria from cosmetic manufacturers. Full text read. Used against my own claim that barrier skincare helps.
  • Dermatological Adverse Events Associated with Topical Brimonidine Gel 0.33% (J Clin Aesthet Dermatol) — retrospective review of about 1,002 subjects across two four-week Phase 3 trials and one 52-week safety study. Funded by Galderma, which markets the product; several authors are company employees. Full text read.
  • The Frontiers review above, used against itself: its own statements that the feedback loop rests on animal and in vitro work and that the causal order is unresolved.
  • Not read: a Parasitology paper comparing two techniques for counting Demodex was identified but could not be opened, so no figures from it are used here.
Next
Next

How Do You Simplify a Skincare Routine That Stopped Working?