Facial redness, honestly.

The National Rosacea Society surveyed 1,066 patients on what set their faces off: sun exposure in 81%, emotional stress 79%, hot weather 75%, wind 57%, heavy exercise 56%, alcohol 52%, hot baths 51%, cold weather 46%, spicy foods 45%, humidity 44%, indoor heat 41%, skin-care products 41%, heated beverages 36% (trigger survey). Sun leads for a reason: ultraviolet damage releases signals that the over-produced cathelicidin peptide turns into vessel dilation, leukocyte recruitment and new vessel growth, the review of UV and the exposome in rosacea making photoprotection the foundation of treatment (UV and rosacea review; cathelicidin study). Moderate, and honestly so: no randomised trial has compared sunscreen with none in rosacea — the one trial that built SPF 15 into a metronidazole cream gave it to both arms (metronidazole-sunscreen trial) — and the grade rests on the mechanism, the survey and a consensus without dissent. Practically: a mineral (zinc or titanium) sunscreen at SPF 30–50 that does not sting, a hat, shade at midday, and a fortnight's diary that finds the two or three triggers that are actually yours rather than a list of twenty to fear; the sun-damage guide grades sunscreen for the ageing it also prevents.

The barrier studies explain the stinging: papulopustular rosacea skin lost more water and was drier and more reactive than acne skin in 463 against 412 patients (barrier comparison), and RNA sequencing found the barrier's structural genes, lipid production and tight junctions disturbed in a pattern the authors found unexpectedly similar to eczema, arguing for barrier-repair as treatment (barrier gene study). The trial: 50 people with rosacea applied a niacinamide-containing moisturiser twice daily for four weeks to the face and one forearm, with the other forearm as control; barrier function and hydration improved on instruments and a chemical probe, the same trends appeared on the face, and both the investigator's evaluation and the patients' self-assessment recorded improvement in the rosacea (niacinamide moisturiser study). Moderate: one small randomised study and a consistent mechanism. The routine that follows from it: lukewarm water, a non-foaming cleanser once or twice a day, a plain moisturiser with ceramides or niacinamide (the ceramides guide and the dry-skin guide grade them), no scrubs, toners, alcohol, menthol, witch hazel or fragrance, and new products patch-tested on the jaw for a week. Every prescription on this page is tolerated better on a repaired barrier, and the redness gels are specifically riskier on a broken one.

Two analyses of the same 82,737 women in the Nurses' Health Study II, with 4,945 cases of rosacea over 14 years, are the best data on diet. Alcohol was associated with developing the disease: compared with never-drinkers, a hazard ratio of 1.12 at one to four grams a day and 1.53 at 30 grams or more, with white wine and liquor most associated and red wine not significantly (alcohol cohort study). Caffeine went the other way: the highest fifth of caffeine intake had a hazard ratio of 0.76 against the lowest, four or more servings of caffeinated coffee a day 0.77 against one or fewer a month, and decaffeinated coffee, tea, soda and chocolate showed no association (caffeine cohort study) — which suggests the "coffee trigger" is the heat of the cup, consistent with heated beverages firing 36% and spicy foods 45% in the trigger survey (trigger survey). Emerging: cohorts and surveys, no trial, and evidence about who develops the disease rather than what treats it. The practical reading: let coffee cool, drink alcohol as the diary allows, and treat spice, saunas and hot baths as the personal experiments they are; the sauna guide has rosacea's entry.

Corrective cosmetics have an outcome study: 1,840 subjects, 95% women, with visible facial conditions — acne in 49%, melasma 17%, rosacea 15% — used a corrective cosmetic daily for four to six weeks, and Skindex-16 scores for symptoms, emotions and functioning all improved significantly, with 96% compliance, better skin comfort and good tolerance (corrective cosmetic study). The National Rosacea Society's own surveys found green- and yellow-tinted bases hide the redness best and pink or orange tones worst (make-up survey). Emerging: uncontrolled, and about appearance and mood rather than vessels. Choose a sheer green primer under a mineral foundation, fragrance-free, removed with the same gentle cleanser, and patch-tested — the barrier row applies to make-up too.

The mite is real: 23 case-control studies of 1,513 patients found Demodex nine times more likely and much denser in rosacea skin, in both the flushing and the bumpy forms (Demodex meta-analysis), and killing it is how ivermectin cream is thought to work. Tea tree oil kills mites in the laboratory, and one randomised, double-blind split-face trial tested it in a gel with permethrin 2.5%: 47 patients with papulopustular rosacea enrolled and 35 finished 12 weeks, mite density fell significantly on the treated side from week five, and papules, pustules and persistent redness improved on that side against the placebo side (permethrin–tea tree trial). Emerging: one trial of a combination that a pharmacy has to make, not of neat oil. Undiluted or high-strength tea tree oil is a contact allergen and an irritant, and on a barrier-damaged face it burns; the honest reading is that the mite hypothesis is why the ivermectin row exists, and the shelf version is a poor substitute for it.

Brimonidine is an alpha-2 agonist — an eye-drop drug for glaucoma — that constricts the small vessels of the face for most of a day. In two identical randomised, double-blind trials, 553 people with moderate-to-severe persistent redness applied the gel or vehicle once daily for four weeks; the primary endpoint, a two-grade improvement on both the Clinician's Erythema Assessment and the patient's self-assessment at hours 3, 6, 9 and 12 on day 29, was reached in 31%, 30%, 26% and 23% against 11%, 10%, 10% and 9% in the first study and 25%, 25%, 18% and 22% against 9%, 9%, 11% and 10% in the second, with separation from vehicle as early as 30 minutes after the first application (phase 3 trials; prescribing information). A one-year open-label study accumulated 345 subject-years of use with adverse events highest at the start and falling (one-year study), and the Cochrane review rates the evidence high-certainty for temporarily reducing persistent erythema (Cochrane-group review). Strong, for exactly what it claims: a switch, not a treatment. The catch is the rebound — the label warns of redness returning worse than baseline and spreading to areas previously unaffected, the label's own adverse-event table lists erythema in 4% and flushing in 3% against 1% and 0%, and the expert review of "paradoxical erythema" puts reversible worsening at 10–20% of users, about 80% improving without it (paradoxical erythema review; adverse-event review). The rules: a pea-sized amount, a patch test on the jaw for three days, an intact barrier underneath, and the first full-face use on a day with nothing at stake.

Oxymetazoline is the nasal-decongestant drug reformulated as a face cream; it works on alpha-1 receptors and constricts more gently. The two REVEAL trials randomised 440 and 445 people with moderate-to-severe persistent redness to cream or vehicle once daily for 29 days: the composite success — a two-grade improvement on both the clinician's and the patient's scale — at 3, 6, 9 and 12 hours on day 29 was 12%, 16%, 18% and 15% against 6%, 6%, 8% and 6% in the first trial and 14%, 13%, 16% and 12% against 7%, 5%, 9% and 6% in the second, statistically superior at every point, with discontinuation for adverse events at 1.8% (first REVEAL trial; second REVEAL trial; prescribing information). The 52-week open-label study of 440 patients found 36.7% and 43.4% achieving the composite improvement at three and six hours after a dose at week 52, treatment-related adverse events in 8.2%, discontinuation for them in 3.2%, no meaningful change in blanching, lesions or telangiectasia, and a rebound after stopping in under 1% (52-week study). Strong evidence for a modest effect: the Cochrane review graded it moderate-certainty when only the abstracts were available (Cochrane-group review), and the absolute gain is a few people in twenty. It was approved in the United States in 2017 and is not authorised in the European Union, so for most readers of this site it is the drug their brimonidine rebound is compared against, not one they can buy.

Ivermectin is an anti-parasitic that also calms inflammation, and the mite hypothesis is why it was tried. In two identically designed randomised, double-blind trials, 1,371 adults with moderate-to-severe papulopustular rosacea applied the cream or vehicle once daily for 12 weeks: the proportion "clear" or "almost clear" on the Investigator's Global Assessment was 38.4% and 40.1% against 11.6% and 18.8%, with greater lesion reductions, better satisfaction and quality of life, and separation from vehicle from week four (pivotal trials; prescribing information). Head to head, 962 patients were randomised to ivermectin once daily or metronidazole 0.75% twice daily for 16 weeks; ivermectin reduced inflammatory lesions by 83.0% against 73.7% and was superior on the global assessment with high patient satisfaction (ivermectin versus metronidazole trial). The Cochrane review rates the evidence high-certainty for papules and pustules (Cochrane-group review). Strong. It does little for the fixed redness or the vessels, can sting in the first weeks on a damaged barrier, and needs three months before a verdict — the mite takes time to die and the inflammation to follow.

Azelaic acid is a dicarboxylic acid from grain that suppresses the cathelicidin-protease pathway and the inflammation it drives. Two multicentre, double-blind, randomised trials enrolled 329 and 335 patients with moderate papulopustular rosacea and applied the 15% gel or vehicle twice daily for 12 weeks; therapeutic success — clear, minimal or mild on the investigator's global assessment — was reached by 61% and 62% against 40% and 48%, with greater lesion and erythema reductions (phase 3 gel trials). A phase 3 trial of the foam at 48 US sites in 961 participants confirmed superiority on lesions, global assessment and erythema (foam trial). Against the older standard, 251 patients randomised to azelaic acid 15% or metronidazole 0.75% gel twice daily for 15 weeks: lesions fell 72.7% against 55.8%, and azelaic acid kept improving through week 15 while metronidazole plateaued at week 8 (azelaic acid versus metronidazole trial). High-certainty evidence in the Cochrane review (Cochrane-group review). Strong. It stings and itches for the first one to two weeks in a good share of users — the barrier row first, a moisturiser underneath, once daily before twice — and the dark-spots guide grades the same molecule for the pigment it also lightens.

Tetracyclines treat rosacea by damping inflammation, not by killing bacteria, and 40 mg of doxycycline in a modified-release capsule does that at blood levels below the antimicrobial threshold. In two parallel phase 3 trials, 269 patients received the capsule and 268 placebo once daily for 16 weeks; from a baseline of about 20 inflammatory lesions, the count fell by 11.8 and 9.5 on doxycycline against 5.9 and 4.3 on placebo, with no more adverse events than placebo (phase 3 trials). The Cochrane review grades the evidence high-certainty on physician assessment in its 2015 update and moderate-to-high in 2019 (Cochrane abridged review; Cochrane-group review). The alternatives: in the DOMINO trial, 80 patients were randomised to doxycycline 40 mg or minocycline 100 mg for 16 weeks; lesion reductions were comparable (13 against 14 fewer), global-assessment success favoured minocycline (60% against 18%) and remission lasted longer, at the cost of minocycline's known rarer but more serious risks (DOMINO trial); and a 205-patient phase 2 trial of an extended-release minocycline 40 mg found success in 66% against 33% on doxycycline 40 mg and 11.5% on placebo (minocycline extended-release trial). Strong. The course is weeks, not years; a cream — ivermectin or azelaic acid — holds the result; the sun sensitivity is real and the sunscreen row is not optional; and the 100 mg antibiotic dose is no better for rosacea and worse for the microbiome.

The acne drug at a fraction of the acne dose shrinks the sebaceous glands and calms the inflammation that rosacea's creams cannot. In a five-armed, double-blind randomised study at 35 German centres, 573 patients with papulopustular or phymatous rosacea received isotretinoin at 0.1, 0.3 or 0.5 mg/kg, doxycycline or placebo for 12 weeks; 0.3 mg/kg was the most effective dose, superior to placebo and non-inferior to doxycycline with a 90% reduction in lesions against 83% (five-arm trial). In a multicentre placebo-controlled trial of 156 patients with difficult-to-treat disease and at least eight lesions, 0.25 mg/kg for four months produced a 90% reduction in papules and pustules in 57.4% against 10.4% on placebo (low-dose isotretinoin trial). The 2025 meta-analysis of 16 studies in 1,445 patients found large reductions in lesions and erythema, both still down 70% and 47% sixteen weeks after stopping, a relapse rate of 35% at five and a half months, worsening in 0.4% and serious adverse events in 0.4% (isotretinoin meta-analysis); the Cochrane review grades the evidence moderate-to-high (Cochrane-group review). Strong evidence for an off-label use, and the drug that is teratogenic at any dose, dries every mucous membrane, needs blood tests and belongs to a dermatologist. It relapses in a third within six months; it is the rung for the disease that has defeated the rest, and the one early phymatous change is offered before the laser.

Topical metronidazole was rosacea's standard cream for thirty years and remains the gentlest. A randomised, double-blind trial of 120 patients with moderate-to-severe rosacea found metronidazole 1% cream with SPF 15 significantly better than its sunscreen vehicle on lesions, erythema and telangiectasia scores at 12 weeks, well tolerated (metronidazole-sunscreen trial), and the Cochrane group pooled the older trials to moderate-certainty evidence against placebo (Cochrane abridged review). The comparisons put it behind: ivermectin once daily beat metronidazole 0.75% twice daily on lesion reduction in 962 patients (ivermectin versus metronidazole trial), and azelaic acid 15% beat it on lesions and kept improving after it plateaued in 251 (azelaic acid versus metronidazole trial). Moderate. Its place is the face that cannot tolerate the other two, the maintenance phase after a course of doxycycline, and the budget.

Two newer topicals have trial programmes behind them. Benzoyl peroxide, the acne standard, is usually too irritating for rosacea; a formulation that traps it in silica microcapsules for slow release was tested in two 12-week randomised, double-blind trials of 733 adults with moderate-to-severe rosacea: clear or almost clear at week 12 in 43.5% and 50.1% against 16.1% and 25.9% on vehicle, lesion counts down 17.4 and 20.3 against 9.5 and 13.3, with tolerability similar to vehicle (encapsulated benzoyl peroxide trials); in the 40-week extension, success reached 66.5% in those switched from vehicle and 67.6% in those continuing, with 1.4% stopping for adverse events (extension study). Minocycline 1.5% foam reduced lesions by 17.57 and 18.54 against 15.65 and 14.88 on vehicle in two phase 3 trials (minocycline foam trials), and a meta-analysis of five randomised trials in 2,453 participants found a modest relative gain in global-assessment success, risk ratio 1.31 (topical minocycline meta-analysis). Moderate: each rests on a single manufacturer's programme without independent replication, the benzoyl peroxide effect is large and the minocycline effect small, and neither is authorised in the European Union — so for most readers they are what a US dermatologist would add, not what a European one can prescribe.

The nerves that fire the flush use serotonin among their signals, and the antidepressant that raises it was tried on the patients no cream had helped. In the PRRE trial, patients with moderate-to-severe refractory erythema were randomised to paroxetine 25 mg or placebo daily for 12 weeks; 97 completed. Clinical Erythema Assessment success at week 12 was 42.9% against 20.8%, flushing success (a reduction of two points or more) 44.9% against 25.0%, burning improved in 46.9% against 18.8%, and depression scores improved, with dizziness, lethargy, nausea, dyspepsia and muscle tremor the reported adverse events (paroxetine trial). Emerging: one trial, one dose, twelve weeks, and a drug with real side effects and a discontinuation syndrome that needs tapering. It is a dermatologist's third-line choice for the flushing-and-burning face that has failed triggers, the barrier, the alpha-agonists and light, and a reasonable one for the patient whose flushing and anxiety feed each other.

Beta-blockers blunt the adrenaline that fires the flush and are the old remedy for blushing. The 2020 systematic review found nine studies of carvedilol, propranolol, nadolol and beta-blockers generally — one randomised trial, one cohort, one case-control, three case reports and three series — in which carvedilol and propranolol produced a large reduction in erythema and flushing with rapid onset of control, bradycardia and hypotension being the common adverse events and most designs retrospective, small and subjectively measured (beta-blocker review). The original report: five patients with severe frequent flushing or persistent erythema and burning added carvedilol titrated to 12.5 mg twice daily for at least six months, and the five-point Clinician Erythema Assessment fell from 3.4 to 0.4 (carvedilol report). Emerging: consistent, small and mostly uncontrolled. A drug for the person whose flushing is triggered by emotion and exertion and whose blood pressure and pulse can spare it, prescribed by someone who checks both.

Toxin placed superficially in the dermis is thought to block the nerve release of the peptides that dilate the vessels and drive the inflammation, without paralysing the muscle below. The scoping review and meta-analysis found seven studies in 167 rosacea patients treated by intradermal injection; two randomised trials pooled to a standardised mean difference of 1.68 in erythema at three months, and the seven single-arm studies showed significant improvement at one, two and three months (toxin meta-analysis). The 2026 systematic review restricted to randomised trials found three, both parallel-group and split-face designs showing significantly reduced erythema and flushing with improved satisfaction, adverse events mild, transient and self-limited — facial tightness, superficial stiffness, injection bruising — and the evidence limited by small samples, suboptimal designs and short follow-up (toxin systematic review). Emerging, and expensive per year. The crow's-feet guide grades the same drug in the muscle where its evidence is strong; the injector who does the micro-droplet technique regularly is the one who does not weaken your smile.

Tranexamic acid, the clot-stabilising drug the dark-spots guide grades for melasma, also damps the vessel growth and plasmin-driven inflammation of rosacea. In the randomised trial, 70 patients with papulopustular rosacea received traditional therapy with or without oral tranexamic acid for eight weeks and were followed four more; erythema, investigator and patient global scores, dryness and rosacea quality of life were significantly better with the drug, the aesthetic improvement score higher, water loss lower and hydration higher, without significant side effects and with the largest gains in dry-skinned patients (oral tranexamic acid trial). Topically, 45 women with erythematotelangiectatic rosacea had three sessions of 10% tranexamic acid with microneedling on one side and the solution alone on the other; both improved, the microneedled side more on clinical and dermoscopic scores at three months (topical tranexamic acid study). Emerging: an open-label add-on trial and a split-face study, the oral drug carrying a clotting risk in anyone predisposed, the topical version cheap and gentle. A reasonable adjunct where a dermatologist offers it; not a substitute for the strong rows.

Hydroxychloroquine calms the innate immune signalling — the same pathways cathelicidin drives — and is the drug of lupus, rosacea's closest look-alike. A multicentre randomised, double-blind, double-dummy pilot study compared it with doxycycline in rosacea (pilot trial), and in a 2025 randomised, double-blind trial, 73 patients with moderate-to-severe rosacea received hydroxychloroquine 200 mg twice daily or placebo alongside doxycycline 100 mg daily; among 58 completers the combination gave significantly higher clinical efficacy at week eight, with an erythema improvement rate of 89.7% (combination trial). Emerging: two trials from one country, modest sizes, and a drug that needs a baseline eye examination and retinal monitoring beyond a few months. A specialist's option for erythema that has failed the tetracycline, particularly where the flushing has an autoimmune flavour.

The gut–skin link in rosacea has one striking study. In Genoa, 113 consecutive rosacea patients (82 women, mean age 52) and 60 matched controls had lactulose and glucose breath tests; small-intestinal bacterial overgrowth was found in 52 of 113 patients against 3 of 60 controls. Those positive were randomised to rifaximin 1,200 mg a day for ten days or placebo, eradication was checked a month later, and two dermatologists — unblinded to the treatment — graded the skin: eradication induced an almost complete regression of the cutaneous lesions, maintained for at least nine months (bacterial overgrowth study), and the same group reported a three-year follow-up (three-year follow-up). Emerging: a single centre with an unusually high prevalence, unblinded assessors, and results other clinics have reproduced only partly. The reading for a patient: bloating, diarrhoea or the irritable bowel alongside a bumpy rosacea is worth a breath test and, if positive, a course that treats both — and a normal gut is not made better by a course of antibiotics for the skin.