Rosacea has real treatments. None of them cure it, and each one targets a single feature

Rosacea is treatable and not curable, and the treatments do not work on the condition as a whole — they work on one of its features. A drug that reliably reduces facial redness does nothing for the papules; a drug that clears the papules does not touch the visible vessels. That is the single most useful thing to understand about the evidence, and it is why "what works for rosacea" has no one answer.

What rosacea is, and why the classification changed

Rosacea is a chronic inflammatory condition of the cheeks, nose, chin, forehead and often the eyes, characterised by recurrent flushing or transient redness, persistent redness, papules, pustules and visible dilated vessels [s2]. Thickening of the skin, especially of the nose, occurs in some people [s1].

For most of the past two decades it was divided into four subtypes and one variant — erythematotelangiectatic, papulopustular, phymatous, ocular and granulomatous [s2]. That scheme has been abandoned in favour of a phenotype approach, because several of its defining features were not specific to any one subtype, phyma was excluded as a primary feature, and grouping multiple features into a subtype conflated things that behave independently [s2]. The change is not cosmetic: it reorganises the evidence around what a given patient actually has.

The randomised evidence

A Cochrane review included 106 randomised trials covering 13,631 participants, evaluating 67 distinct comparisons [s1]. Sample sizes of 30 to 100 and durations of two to three months were the norm; participants were more often women, with a mean age of 48.6 years, and most had papulopustular rosacea [s1]. Risk of bias was unclear in 57 of the 106 studies, high in 37 and low in 12 [s1]. Twenty-two studies provided no usable or retrievable data at all [s1].

An updated review organised by phenotype, searching to March 2018, expanded that to 152 studies and 20,944 participants, with 46 of the studies new to the update, and applied GRADE certainty ratings throughout [s2].

For persistent redness. Topical brimonidine has high-certainty evidence [s2]. In two trials it was more effective than vehicle at reducing erythema at every time point across 12 hours, with participant-assessed risk ratios at three hours of 2.21 (95% confidence interval 1.52 to 3.22) and 2.00 (1.33 to 3.01), confirmed by physician assessment, and with no rebound or worsening after treatment stopped [s1]. Topical oxymetazoline carries moderate-certainty evidence for the same indication [s2]. Both are vasoconstrictors: they reduce redness temporarily, which is what the evidence describes and what the guidance says — the updated review is explicit that brimonidine reduces persistent erythema temporarily [s2].

For papules and pustules. Topical azelaic acid and topical ivermectin have high-certainty evidence [s2]. In the Cochrane pooling, azelaic acid beat placebo on participant assessment with a risk ratio of 1.46 (1.30 to 1.63) across four trials [s1]. Two trials of topical ivermectin showed statistically significant and clinically important improvement over placebo, with participant-assessed risk ratios of 1.78 (1.50 to 2.11) and 1.92 (1.59 to 2.32) [s1]. One study found ivermectin slightly more effective than topical metronidazole, on high-quality evidence for quality of life and for both participant and physician assessments [s1].

Metronidazole itself is effective against placebo — three trials gave a physician-assessed risk ratio of 1.98 (1.29 to 3.02), moderate quality [s1] — but the direct comparison against azelaic acid is unresolved: three studies produced contradictory results on which of the two is better [s1]. The phenotype review rates topical metronidazole and topical minocycline at moderate certainty [s2].

Oral options for papules and pustules. Doxycycline was significantly more effective than placebo on physician assessment in two trials (risk ratios 1.59, 1.02 to 2.47; and 2.37, 1.12 to 4.99), rated high quality, with no statistically significant difference in effectiveness between 100 mg and 40 mg [s1]. The phenotype review gives moderate-to-high certainty to doxycycline 40 mg in modified release form and to isotretinoin, and reports that oral minocycline is equally effective as doxycycline 40 mg modified release at moderate certainty [s2]. Low-dose isotretinoin at 0.25 mg per kilogram greatly reduces papules and pustules against placebo, on high-certainty evidence [s2]. Tetracycline and low-dose minocycline are supported only by low-certainty evidence [s2]; the Cochrane review's moderate rating for tetracycline rested on two old studies of short duration [s1].

For visible vessels. Laser and intense pulsed light have low-to-moderate certainty evidence for erythema with mainly telangiectasia [s2]. This is the one feature with no strong drug option.

For the eyes. Ocular rosacea has the thinnest evidence of any phenotype. Oral omega-3 fatty acids carry moderate-certainty evidence for improving dry-eye symptoms and tear gland function [s2]. Ciclosporin ophthalmic emulsion and doxycycline are supported at low certainty [s2]; the Cochrane review found ciclosporin ophthalmic emulsion effective with improved quality of life on low-quality evidence [s1].

What did not work

Topical clindamycin phosphate combined with tretinoin was not considered effective against placebo on moderate-quality evidence [s1]. That is a useful negative in a field where combination products are common.

The gaps that persist

Only 11 of the 106 Cochrane trials assessed change in quality of life as an outcome, in a condition whose principal burden is visible and social [s1]. Time to improvement and duration of remission were incompletely reported or not reported at all in most studies [s1]. Trials lasting two to three months are the norm in a condition people live with for decades [s1].

Adverse events were addressed in almost all studies but usually with limited data; in most comparisons there was no statistically significant difference in the number of adverse events, and most were mild and transient [s1].

The practical shape of the evidence, then, is a set of well-supported single-feature treatments with short follow-up, no cure, and almost no data on what any of it does to how people feel about their faces over years.

This article is informational and is not medical advice.

Sources

• [s1] Interventions for rosacea — Cochrane Database of Systematic Reviews, 28 April 2015. https://doi.org/10.1002/14651858.CD003262.pub5
• [s2] Interventions for rosacea based on the phenotype approach: an updated systematic review including GRADE assessments — British Journal of Dermatology, published online 26 December 2018. https://doi.org/10.1111/bjd.17590