Hard water and skin: what the papers actually say

Hard water and eczema: what is actually established

Pooled across seven observational studies and 385,901 participants, children in harder water areas had higher odds of eczema: odds ratio 1.28, 95% confidence interval 1.09 to 1.50. The reviewers graded their own certainty in that figure as very low, and the pooled estimate carries 63 percent heterogeneity. We print the certainty rating next to the number every time, because a number without it is a different claim.

The association also does not behave the way a biological effect usually does. In the same review it is absent in infants (odds ratio 1.29, 95% CI 0.89 to 1.86), present in primary school children (1.43, 1.09 to 1.87), and absent and pointing the wrong way in teenagers (0.90, 0.65 to 1.25).

The most informative test was following the same children forward rather than looking at them once. The EAT cohort produced the famous cross-sectional finding at three months of age. Followed to thirty-six months, the overall association disappeared: adjusted hazard ratio 1.07, 95% CI 0.92 to 1.24.

And taken at face value, the effect is small. In the largest study ever run on this question, 52,950 Danish children, hard water accounted for about 2 percent of infant atopic dermatitis. Being born in autumn mattered more.

The famous number, and what it measures

In a forearm wash test, washing with very hard water (394 to 404 mg/L calcium carbonate) left 2.8 times more sodium lauryl sulfate on the skin than washing with deionised water. The paper gives the spread as plus or minus 0.6, which is a standard error and not a confidence interval, and it reports no p value for that specific comparison.The study was funded by Harvey Water Softeners, the manufacturer of the ion-exchange water softener it tested, which also supplied the test water and advised on the science.

The comparison matters. It was very hard water against deionised water, which has essentially no minerals in it and which nobody has coming out of a tap. There was no ordinary tap water arm, so the study cannot tell you what happens at 150 or 200 milligrams per litre, which is where most households actually sit.

Those forearm sites were also not simply washed and left. They were washed and then sealed under a 12 millimetre patch with a water-resistant dressing for 48 hours. That is a model designed to make a small effect measurable. It is not what happens when you wash your face, and the study never claims it is.

One finding inside it is worth more attention than it gets: who you are made no difference to how much residue was deposited. No effect of eczema status, no effect of filaggrin genotype. The water chemistry happens to everyone equally. What differed was how skin reacted to the residue afterwards.

Does hard water raise your skin pH?

The everyday version of this story is that hard water is alkaline, alkaline water raises your skin's pH, and a raised pH damages the barrier. The first link in that chain is not what the research found.

The one experiment that measured skin surface pH alongside water hardness found no significant difference between hard water and deionised water. The water that moved skin pH most was the softened water containing chlorine, because it had the highest alkalinity of any water tested. The authors are explicit that alkalinity, not calcium, is what moved pH.

Ion exchange softening does not change your water's pH at all. In that study hard water and softened water both measured pH 7.4, with essentially identical alkalinity. Softening swaps calcium and magnesium for sodium; it does not make water less alkaline. Anyone selling a softener on a pH argument is selling something the chemistry does not support.

A second, independent study happened to measure both supplies and found the relationship pointing the other way: its soft water was more alkaline than its hard water, pH 7.8 against 7.2. Its authors note this themselves, as a limitation that "could have obscured stronger differences". It is not industry funded.

What does raise skin pH, measurably, is the cleanser. In forty infants, an alkaline soap at pH 9.5 raised skin pH by 0.45 from a baseline of 6.60, plain tap water raised it by 0.19, and a pH 5.5 syndet raised it only 0.09 above the water control.

We should be equally careful in the other direction. Nobody has ever measured skin pH after washing in hard water without a surfactant, so strictly speaking the question has never been isolated. We searched for that study specifically and did not find one.

The mechanism was tested once, in a randomised trial

Everything above rests on a mechanism: hard water leaves more cleanser residue on skin, and that residue irritates. In 2018 the authors of the deposition study wrote that an intervention study was required to test the hypothesis. That trial was run. Households in hard water areas were randomised, before the babies were born, to have an ion-exchange softener installed or not, and the infants' skin was measured at four weeks, three months and six months.

The same infrared method that produced the 2.8-fold figure found no difference in surfactant residue at any visit. Nor was there a difference in water loss through the skin, in skin pH, in hydration, in cholesterol sulphate, in natural moisturising factors, or in four of five cytokines.

Those results have never appeared in a journal. The trial itself was published in 2021, and its methods section says, of the mechanistic measurements: "Additional mechanistic outcomes were also evaluated in a sub-study and the analyses of these will be reported separately." They have not been. They exist in a doctoral thesis awarded in November 2024.

The trial's own author is careful about what that null can carry, and so are we. It was a pilot, deliberately small, and underpowered to detect differences between arms. Few infants developed eczema and it was mild. And one confounder matters: the thesis notes these high-risk families may already have avoided harsh surfactant wash products. If the families had removed the surfactant, the water had less to react with.

The filaggrin finding, and the number under it

When the overall association vanished on longitudinal follow-up, one subgroup kept it: children carrying a loss-of-function mutation in filaggrin, a gene that helps build the skin barrier. In them the adjusted hazard ratio for hard water exposure was 2.72, 95% CI 2.03 to 3.66, with a statistically significant interaction. That is a large number and it is the strongest positive finding in this literature.

Two facts belong next to it, and both come from a systematic review co-authored by the same investigator. First, in that review's words, "interaction of FLG genotype and water hardness has been investigated in two papers within the same cohort." Two analyses, one cohort. It has never been tested in an independent hard water cohort. Second, the number of children who both carry the mutation and were exposed to the hardest water isseventy-five.

That review's own conclusion about this whole class of finding: "Evidence on FLG-environment interactions in AD aetiology is limited. However, many of the studies lacked large enough sample sizes to assess these interactions fully."

So the honest description is all of these at once: a real and statistically significant interaction, found twice in one cohort, resting on seventy-five exposed children, in a field the researchers themselves call limited and underpowered. Filaggrin status is not routinely tested, and knowing yours would not change what to do, because the prevention trial's removal of hard water did not produce a measurable barrier difference in anyone.

Does a water softener help eczema?

For eczema that already exists, this has been answered properly. The Softened Water Eczema Trial randomised 336 children aged 6 months to 16 years with moderate to severe eczema, all in hard water areas averaging about 310 milligrams per litre. On the blinded primary outcome there was no benefit: a mean difference of 0.66 on the severity score, 95% CI -1.37 to 2.69, slightly favouring usual care. All three blinded secondary outcomes were null too.

Both groups improved by roughly the same amount, about 20 and 22 percent. That is what ordinary care plus regression to the mean looks like, and it is why testimonials from people who installed a softener and got better are not evidence that the softener did it.

Parents in the softener group did report better outcomes, and we are telling you because a softener advert probably will. Three of four parent-reported measures improved significantly. The trialists' own explanation is response bias: nobody in this trial was blinded, because softened water makes visibly more suds and a sham unit was impossible. Where the assessor was blinded, nothing happened.

Softeners were also dominated economically, meaning more expensive and no more effective: about 28,000 pounds per quality-adjusted life year in the base case, and roughly 3,250 pounds for a typical family over a device's life. Britain's national guidance already reflects all of this. NICE tells clinicians to explain that "there is no evidence of any benefit in taking fewer baths or showers, or in using ion exchange water softeners or silk clothing", a recommendation amended as recently as September 2025.

Prevention is a different question and it is genuinely open. The pilot trial found visible eczema at six months in 33 percent of the softener group against 48 percent of controls, a difference of -15 percent with a 95% confidence interval running from -38 to +8.3, which includes no effect at all. That trial was powered only to test whether such a study is feasible and performed no statistical testing at all. A definitive answer would need more than 860 participants and about two years of follow-up. Its chief investigator told us that trial is in preparation.

What this leaves you with

Less than the internet suggests, and that is the finding rather than a failure to find one.

If you have eczema, the water is not the lever the evidence points at, and the one device sold for it has been tested and did not help. That conversation belongs with a clinician, not a plumber.

If your skin feels tight and dry after washing and you live somewhere hard, the measured effect was never water alone. It was an interaction between minerals in the water and the surfactant in a cleanser, and the reaction happened to the residue left behind. Here is our reasoning, labelled as reasoning rather than as evidence: change either side of an interaction and the interaction changes, and one side costs about ten pounds while the other costs thousands and has already failed a trial. But nobody has run that study. We are not going to dress an untested inference up as a finding, and you will notice there is nothing to buy on this page.

What would settle it is a trial that randomises cleanser type, soap against synthetic detergent against sulfate-free, crossed with water hardness, powered on the interaction rather than the main effects, measuring residue directly. It does not exist.