Medical-Grade Honey and Wounds: What Is Known
Honey dressings have one genuinely strong result in the trial record: Cochrane’s review of 26 trials found high quality evidence that they healed partial thickness burns about four and a half days faster than conventional dressings, and a smaller trial found infected surgical wounds healing faster under honey than under antiseptic washes and gauze. The honey behind those results deserves understanding in its own right — a manufactured, sterilised product, irradiated to destroy the bacterial spores ordinary honey carries, standardised for antibacterial strength, supplied as an impregnated dressing and regulated as a medical device. It is not the jar in a kitchen cupboard, and every trial below is a trial of the manufactured product. Wounds themselves are managed by a GP, a practice nurse or a tissue-viability service; this page is about what the research found.
Plate I · ArtefactRecord
What is medical-grade honey?
Four things, and none of them describes food honey.
It is sterilised. Raw honey is not sterile, and the standard method of making it so is gamma irradiation. Postmes and colleagues (1995) irradiated six batches with cobalt-60 at 6–25 kGy. Batches spiked with roughly a million spores of Clostridium botulinum or Bacillus subtilis per 50 g were sterile after 25 kGy; at up to 5,000 spores per 50 g — the authors’ stated upper limit of natural contamination — 18 kGy sufficed. Irradiation is not free of consequence: antibacterial activity was unaltered at 25 kGy, but amylase activity fell to 19–43% of baseline in five of the six batches. The honey on a dressing has been changed on purpose.
It is standardised. Manuka is the honey that has, in the Cochrane reviewers’ phrase, “been of particular interest”, because its antibacterial activity is “independent of the effect of honey’s peroxide activity and osmolarity”. The manuka evidence page reads that record on its own terms.
It is a dressing, not a spoonful. In the trials Cochrane pooled, honey came as an impregnated gauze dressing in six, an impregnated calcium alginate dressing in two, and otherwise under a film dressing or gauze.
And it is regulated as a medical device — placed on the market for wound care as a device, not a food and not a licensed medicine. In Great Britain that route is the Medical Devices Regulations 2002 (SI 2002/618, as amended): general medical devices under Part II, classified I, IIa, IIb or III, UKCA-marked to show conformity. A jar of honey sold to eat is food, under a different regime entirely.
What did the Cochrane review find?
Two clear wins, and a lot of low-quality territory around them. Jull and colleagues (2015) identified 26 eligible randomised or quasi-randomised trials, 3,011 participants — eleven in burns, ten in chronic wounds, three in minor acute wounds, two in mixed populations. One habit of reading pays off throughout: Cochrane grades its evidence high, moderate, low or very low, and the grade expresses how confident the reviewers are that the estimate would survive further research — “low quality” means a new trial could plausibly overturn it. With that key, the results sort themselves. Two comparisons came out clearly in honey’s favour:
- Partial thickness burns. High quality evidence from 2 trials (n = 992) that honey dressings healed partial thickness burns more quickly than conventional dressings — a weighted mean difference of −4.68 days (95% CI −5.09 to −4.28), the average effect across trials weighted by their size. Whether adverse events or infection differed was unclear (very low and low quality evidence). One caveat matters, and the reviewers raise it themselves: it turns on what “conventional” meant. In one of the two trials the control dressing was a polyurethane film dressing; in the other the control group received a range of interventions — polyurethane film, paraffin gauze, sterile linen, framycetin-impregnated tulle, or the burn left exposed. On their own account, “practitioners must consider whether the comparator dressings in these trials are clinically meaningful to them”.
- Infected post-operative wounds. One trial (n = 50, moderate quality evidence): honey healed these more quickly than antiseptic washes followed by gauze, with fewer adverse events, risk ratio for healing 1.69 (95% CI 1.10 to 2.61) — about 1.69 times the chance of healing within the trial window. One trial, fifty people: a genuine result, resting on a narrow base.
Against silver sulfadiazine the picture divides, and it divides both ways: very low quality evidence from 4 trials (n = 332) of faster healing with honey (WMD −5.12 days, 95% CI −9.51 to −0.73); high quality evidence from 6 trials (n = 462) of no difference in the risk of healing within six weeks (RR 1.00, 95% CI 0.98 to 1.02); and high quality evidence from 5 trials (n = 412) of significantly fewer adverse events with honey than SSD (RR 0.29, 95% CI 0.20 to 0.42). The reviewers hedge that last one themselves: it was generally unclear whether events had been counted per person or double-counted, so they were “not confident of the accuracy of the adverse events data”. One low quality trial (n = 50) found early excision and grafting faster than honey by 13.6 days.
Venous leg ulcers are the review’s most instructive negative, because the clearest thing found there is a harm. Whether honey healed them was unclear (RR 1.15, 95% CI 0.96 to 1.38, 2 trials, n = 476, low quality evidence, downgraded for risk of bias and imprecision). But the larger trial (n = 368) recorded significantly more adverse events in the honey-treated group than the control group (RR 1.28, 95% CI 1.05 to 1.56, low quality evidence) — pain in 47/187 against 18/181, and deterioration of the ulcer in 19/187 against 9/181. The smaller trial counted only events it attributed to the wound agent and recorded none — an approach the reviewers note is open to ascertainment bias in an open-label study, where knowing who got what can shape which events get written down.
Two trials in a mixed population of acute and chronic wounds (n = 140, different comparators, low quality evidence) found honey healed more quickly than silver sulfadiazine or sugar dressings. For the remaining wound types the effects were unclear at low or very low quality: minor acute wounds (3 trials, n = 213), diabetic foot ulcers (2 trials, n = 93), cutaneous leishmaniasis (1 trial, n = 100), mixed chronic wounds (2 trials, n = 150). Single very low quality trials favoured honey in pressure ulcers (n = 40) and Fournier’s gangrene (n = 30).
The reviewers’ own summary, verbatim: beyond partial thickness burns and infected post-operative wounds, “any evidence for differences in the effects of honey and comparators is of low or very low quality and does not form a robust basis for decision making”.
One disclosure sits beside that, printed in the review’s own declarations of interest: two of the review’s authors were among the investigators on the Honey as Adjuvant Leg ulcer Treatment (HALT) trial — one of the included trials, and the same venous leg ulcer trial that produced the adverse event result above — and the research unit that employed them “received a small, unconditional cash contribution from a manufacturer of honey dressings for the conduct of the HALT trial”.
A newer meta-analysis has not moved the picture much. Tang, Chen and Ran (2024) pooled eight studies, 906 people with chronic wounds: faster healing time (MD −17.13 days, 95% CI −26.37 to −7.89), no significant difference in healing rate (RR 2.00, 95% CI 0.78 to 5.10), evidence rated very low quality, and honey may increase pain during treatment.
Fig. I · The Cochrane ledger, comparison by comparisonDiagram
What the review pooled
26 randomised or quasi-randomised trials 3,011 participants
Eleven in burns, ten in chronic wounds, three in minor acute wounds, two in mixed populations.
Clearly in honey’s favour
Two comparisons came out clearly in honey’s favour.
Honey dressings healed them more quickly — weighted mean difference −4.68 days (95% CI −5.09 to −4.28)
Whether adverse events or infection differed was unclear — very low and low quality evidence. The control dressings ran from a polyurethane film dressing to paraffin gauze, sterile linen, framycetin-impregnated tulle, or the burn left exposed; the reviewers say “practitioners must consider whether the comparator dressings in these trials are clinically meaningful to them”.
Healed more quickly, with fewer adverse events — risk ratio for healing 1.69 (95% CI 1.10 to 2.61)
One trial, fifty people.
The other comparators
Against silver sulfadiazine the picture divides, and it divides both ways. One further trial set honey against early excision and grafting.
Faster healing with honey — WMD −5.12 days (95% CI −9.51 to −0.73)
No difference in the risk of healing — RR 1.00 (95% CI 0.98 to 1.02)
Significantly fewer with honey — RR 0.29 (95% CI 0.20 to 0.42)
The reviewers hedge this one themselves: it was generally unclear whether events had been counted per person or double-counted, so they were “not confident of the accuracy of the adverse events data”.
Early excision and grafting faster than honey — by 13.6 days
Venous leg ulcers
The clearest thing the review found there is a harm.
Whether honey healed them was unclear — RR 1.15 (95% CI 0.96 to 1.38)
Downgraded for risk of bias and imprecision.
Significantly more in the honey-treated group than the control group — RR 1.28 (95% CI 1.05 to 1.56)
Pain in 47/187 against 18/181; deterioration of the ulcer in 19/187 against 9/181. The smaller trial counted only events it attributed to the wound agent and recorded none — an approach the reviewers note is open to ascertainment bias in an open-label study.
Declared in the reviewTwo of the review’s authors were among the investigators on the Honey as Adjuvant Leg ulcer Treatment (HALT) trial — the same venous leg ulcer trial that produced the adverse event result above — and the research unit that employed them “received a small, unconditional cash contribution from a manufacturer of honey dressings for the conduct of the HALT trial”.
Everywhere else
For the remaining wound types the effects were unclear at low or very low quality.
Healed more quickly than silver sulfadiazine or sugar dressings
Effects unclear
Effects unclear
Effects unclear
Effects unclear
Favoured honey
Favoured honey
Quality is the reviewers’ own wording as this page reports it: high, moderate, low, very low. A solid stamp frame marks high and moderate, a dashed frame marks low and very low — the words carry the reading, the frames only repeat them.
The reviewers’ own conclusion
Beyond partial thickness burns and infected post-operative wounds, “any evidence for differences in the effects of honey and comparators is of low or very low quality and does not form a robust basis for decision making”.
Does it matter which honey is on the dressing?
No trial in Cochrane’s review was built to answer it. Cochrane pooled honey dressings as a single class, and every comparison it ran set honey against something that was not honey — conventional dressings, silver sulfadiazine, antiseptics, hydrogel, early excision and grafting. Monofloral honey — “aloe, jarrah, jamun, jambhul or manuka” — was used in ten of the 26 trials, and the type of honey was not specified in the rest. Manuka is one named source inside a minority of the evidence, and there is no manuka-against-other-honey comparison in the review to pool. The two comparisons that came out in honey’s favour are results about honey dressings; they say nothing about one floral source over another.
Manuka’s reputation rests on the laboratory property quoted above, and the laboratory record holds a surprise of its own. Bucekova and colleagues (2020) measured minimum inhibitory concentrations against Staphylococcus aureus and Pseudomonas aeruginosa — the lowest concentration that stops the bacteria growing, so lower means stronger — for 36 commercial honeys, three from local beekeepers and three medical-grade products. The medical-grade products came in at 9% to 22.5%; the three beekeeper honeys ran 7% to 12%, an effect the authors call “even higher than that of medical-grade honeys”. More than 40% of the commercial honeys, meanwhile, were indistinguishable from artificial honey — a plain sugar solution. Two lessons in one dataset: honey’s laboratory strength varies enormously jar to jar, and a dish result is a different kind of fact from a trial result. An inhibitory concentration describes what happens to bacteria in culture, not to a wound in a person — and an unirradiated jar, whatever its laboratory strength, still carries the spores described above, which is why none of this is a reason to put food honey on broken skin.
Is there evidence for honey and eczema?
Barely any, and what exists is negative or uncontrolled — a finding in itself, because eczema is among the most searched questions about medical-grade honey and among the least studied.
The one randomised trial found nothing, and it was well built for its size. Fingleton and colleagues (2014) recruited 15 adults with a doctor’s diagnosis of eczema and bilateral limb lesions, randomised by coin toss which side got medical-grade kanuka honey and which aqueous cream, treated overnight for two weeks under a non-adherent dressing, and used a blinded assessor — each person serving as their own control, which strips out the person-to-person variation that swamps small trials. The change in SCORAD lesion score differed by −0.1 (95% CI −1.5 to 1.4); Three Item Severity by −0.1 (95% CI −0.9 to 0.6); self-rated itch by −7.3 (95% CI −27.7 to 13.2). None approached significance. Their conclusion: “Topical application of medical-grade Kanuka honey does not appear to be effective in the management of eczema.” The authors state the limits themselves — 15 people, a sample sized for variance estimation rather than to detect a clinically important difference, incomplete blinding because honey looks like honey, and a comparator (aqueous cream) not recommended in eczema and acting as a negative control — and the study was sponsored by a manufacturer of medical-grade kanuka honey.
The study cited on the other side teaches a different lesson: what an uncontrolled result can and cannot show. Alangari and colleagues (2017) asked 14 adults with bilateral lesions to apply manuka honey to one site overnight for seven nights and leave the other untreated: treated lesions improved significantly from their own baseline (mean TIS −2 points, 95% CI −2.75 to −1.25) while the untreated lesions did not (−0.7, 95% CI −1.71 to 0.28) — but the paper reports no direct statistical comparison between the two sites, and improvement from baseline, without randomisation, blinding or placebo, cannot separate the honey from everything else that changes over a week. Skin staphylococci did not change, and the accompanying mechanistic work was done in cell culture, not in people. The authors’ closing line is the fair summary: “This needs to be confirmed by randomized and controlled clinical trials.”
Two studies of roughly fifteen people, one null and one uncontrolled, is not a body of evidence. The page on honey and skin covers the wider cosmetic literature; on eczema the record is simply thin.
What does UK guidance actually say?
For home wound care, nothing involving honey. NHS advice for burns and scalds is cool running water for 15 to 30 minutes, cling film laid over the cooled area, and — in its own list of things not to do — “do not put any creams, oils or butter on burns or scalds”. NHS advice for cuts and grazes is to rinse with water or sterile wipes and cover with a sterile dressing. No NHS page located for this review recommends any honey for home wound care.
Why sterilisation is not a technicality sits in other NHS guidance: honey is not given to babies under 12 months because it can contain bacteria whose toxins cause infant botulism. Spores that are a documented hazard in an infant gut are not something to put in an open wound — which is why dressing honey is irradiated first.
Allergy to honey and bee products exists, and the trial record is close to silent on it by design: at least four of the trials Cochrane included screened such people out before randomisation, among them the larger venous leg ulcer trial (n = 368), whose exclusion criteria list “allergy to honey or calcium alginate”. Anyone with a known honey allergy is therefore outside this evidence entirely, not shown to be safe within it.
How the food regulator scored honey completes the picture, and the split is worth knowing. The GB nutrition and health claims register — the authoritative list of what may be said about food — carries eight honey entries, every one non-authorised, including “natural antimicrobial action”, the wording closest to anything discussed here. Wound dressings are judged under the Medical Devices Regulations 2002 instead: a different system asking different questions. The two regimes do not read across in either direction — a dressing’s device certification says nothing about food honey, and the food register’s verdicts say nothing about dressings.
Read as a whole, the record is precise about where it is strong: a sterilised manufactured product with high quality evidence in one burn type — against comparators the reviewers say practitioners must weigh for themselves — a single 50-person moderate quality trial in one post-operative comparison, and weak or absent evidence everywhere else, eczema included. The rest of the bee-shelf record is read the same way across the evidence section. If a wound needs attention, ask a GP, a practice nurse or a tissue-viability service.
Asked, answered.
Is medical-grade honey the same as the honey in my kitchen?
No. The honey used in wound care is a manufactured product: sterilised by gamma irradiation, standardised for antibacterial strength, and supplied as an impregnated dressing placed on the market as a medical device rather than as a food. Ordinary honey is none of those things — it is not sterile, its antibacterial strength is not standardised, and it is regulated as a food. The two are different products that happen to share a name.
What did the Cochrane review actually conclude?
Two clear positives, and thin evidence everywhere else. Across 26 trials and 3,011 participants, Jull and colleagues (2015) found honey dressings healed partial thickness burns faster than conventional dressings, and infected post-operative wounds faster than antiseptics and gauze. For every other wound type — venous leg ulcers, diabetic foot ulcers, minor wounds — the reviewers judged the evidence "of low or very low quality" and said it "does not form a robust basis for decision making".
Does honey help eczema?
On the one randomised trial that exists, no. Fifteen adults applied medical-grade kanuka honey to one side and aqueous cream to the other for two weeks, and the difference in lesion severity was −0.1 (95% CI −1.5 to 1.4). A separate uncontrolled study of 14 adults reported improvement, but it had no randomisation, no blinding and no placebo, and its own authors called for controlled trials before drawing conclusions.
Why is honey used on wounds irradiated?
Because raw honey is not sterile. It can carry bacterial spores — the reason NHS guidance says honey is not for babies under 12 months. Postmes and colleagues (1995) found that honey deliberately spiked with about a million Clostridium botulinum or Bacillus subtilis spores per 50 g was sterile after 25 kGy of cobalt-60 gamma radiation, and that 18 kGy sufficed at the upper limit of natural contamination.
Is any of this an authorised health claim in Great Britain?
No — and the reason is instructive. The GB nutrition and health claims register, which governs what may be said about food, contains eight honey entries and every one is non-authorised. Wound dressings are judged under a different system entirely, the Medical Devices Regulations 2002. The two regimes do not read across: a honey dressing on a hospital formulary has passed a device conformity route, which says nothing about a jar of honey sold as food.
Sources
- Jull, Cullum, Dumville, Westby, Deshpande & Walker, Honey as a topical treatment for wounds, Cochrane Database of Systematic Reviews 2015 (CD005083.pub4, full text)
- Postmes, van den Bogaard & Hazen, The sterilization of honey with cobalt 60 gamma radiation, Experientia 1995
- Fingleton, Helm, Tofield, Weatherall & Beasley, randomised controlled trial of topical kanuka honey for eczema, JRSM Open 2014
- Alangari et al., manuka honey applied to atopic dermatitis lesions, clinical and mechanistic studies, Immunity Inflammation and Disease 2017
- Tang, Chen & Ran, honey dressings for chronic wounds, updated systematic review and meta-analysis of 8 studies, Nutrients 2024
- Bucekova et al., antibacterial activity of commercial and medical-grade honeys, Foods 2020
- MHRA — Medical devices: conformity assessment and the UKCA mark (UK MDR 2002), gov.uk
- Great Britain nutrition and health claims (NHC) register — gov.uk
- NHS — Burns and scalds (self-care and treatment)
- NHS — Foods to avoid giving babies and young children