Sensitivity

Skin that reacts easily and stings readily.

Start here

What it is: Stinging, reddening or itching from things that are fine for other people. It is less a diagnosis than a name for what you feel.
Why it happens: Either the barrier has weakened so irritants get in easily, or nerves are responding more readily than usual, or both. Sometimes a condition such as eczema or rosacea sits behind it.
Where to start: Removing comes before adding. Cut down how many products you use, and start by checking for fragrance allergens and essential oils.

There are no numbers and no citations in this box. What has been established, and how far, is below.

With sensitive skin, what irritates matters before what you add.

This site holds two examples. In a 1997 trial, gluconolactone-treated sites lost significantly less barrier function after a challenge than vehicle — and significantly less than glycolic acid. Two things called 'acids' are not the same thing.

The example on the other side is tocopherol. In a 1999 trial, 33% developed contact dermatitis — a conspicuously high rate among the ingredients on this site.

Evidence level

Moderate evidenceBarrier lipid

Panthenol

In a three-week study of 23 people on deliberately damaged skin, the reduction in transepidermal water loss was larger than control (−168.36 vs −123.38 g/m²/h, P = .023). What that control was, the abstract does not say.

Moderate evidenceSoothing

Centella asiatica

Centella Asiatica Extract

In a split-face trial of 30 people after laser resurfacing, the side treated with Centella extract gel showed a lower erythema index (coefficient −0.03, P = .046) and better crusting and wound appearance in the first days. Skin biophysics, though, did not differ between the two sides.

Moderate evidenceHumectant

Glycerin

In a double-blind crossover study of 51 people with dry skin, the finished emollient reduced transepidermal water loss more than the vehicle and more than the vehicle plus glycerol. Moisturising is the work of a formula, not of one ingredient.

Moderate evidenceHumectant

Urea

5% and 40% are effectively different products. At low concentrations it holds water; at high concentrations it dissolves keratin. A 40% cream beat 12% ammonium lactate at two weeks, and in a 167-person trial of a urea-containing cream the primary endpoint was missed (P = .088) while the other measures were significant.

Moderate evidenceAcid

Gluconolactone (PHA)

Gluconolactone

A 1997 trial put four acids side by side and then challenged the skin. Sites treated with gluconolactone lost significantly less barrier function than vehicle, than untreated skin, and than glycolic acid. And the Korean AHA warning text does not apply to this ingredient.

Moderate evidenceSoothing

Colloidal oatmeal

An ingredient class — not a single line on an ingredient list

In a double-blind trial of 90 children with atopic dermatitis, an over-the-counter 1% colloidal oatmeal cream was non-inferior to a prescription barrier cream. But all three clinical trials we found came from the company that sells it. And none of them included an untreated arm.

Moderate evidenceBarrier lipid

Petrolatum

A 1997 study proposed using petrolatum as the standard reference substance in irritation testing. And in a 2000 trial comparing 'skin-identical lipids' with pure petrolatum, there was no indication the lipids did better. Recorded here as a case where the expensive ingredient did not beat the cheap one.

Moderate evidenceHumectant

Ectoine

Ectoin

Two randomised trials exist, both in atopic dermatitis. One beat its vehicle but tested a cream that also contained hyaluronic acid; the other tied with a different active cream. Both ran four weeks.

Early researchSoothing

Licochalcone A

In an 80-person randomised, investigator-blinded trial, a moisturiser containing licochalcone A was compared against 0.02% triamcinolone acetonide cream. The steroid improved things faster; the moisturiser did better on redness control and hydration. But it was not licochalcone A alone, there was no vehicle-only arm, and after two weeks everyone moved onto the test moisturiser.

Early researchSoothing

Aloe vera

An ingredient class — not a single line on an ingredient list

In a double-blind trial of 12 skin-graft donor sites, epithelialisation was significantly faster: 11.5 days against 13.67. But in the same trial, pain was not significantly reduced. And we found no trial of aloe for cosmetic outcomes — wrinkles, pigment, elasticity.

Ingredient dataSoothing

Alpha-bisabolol

Bisabolol

The Korean listed amount is 0.5%, the lowest of the nine whitening actives. But we could find no human trial comparing this ingredient on its own against vehicle. What turns up are studies of multi-ingredient formulas.

Ingredient dataOther

The 25 labelled fragrance allergens

An ingredient class — not a single line on an ingredient list

Korean regulation names 25 fragrance allergens and requires that, above a set threshold, they be written out by name rather than folded into the word 'fragrance'. The thresholds are above 0.01% in rinse-off products and above 0.001% in leave-on products. All 25 are transcribed below with their CAS numbers.

Not establishedBarrier lipid

Ceramides

An ingredient class — not a single line on an ingredient list

In 39 children with atopic dermatitis randomised three ways, a ceramide barrier cream, another prescription cream and an over-the-counter petrolatum ointment did not differ, and the petrolatum was at least 47 times more cost-effective. Applying emollients to newborns to prevent eczema also failed in a Cochrane review.

Clinical procedures related to this concern

Procedures are not listed alongside topical ingredients. They are not something you use at home.

Moderate evidenceVascular laser (pulsed dye laser)Device-based
Early researchMesotherapyInjection