Does Local Honey Help Hayfever? The Trials

No — the one trial that put genuinely local honey to the test found nothing, and the finding is worth understanding rather than just knowing. A randomised, double-blind, placebo-controlled trial of 36 adults with allergic rhinoconjunctivitis (Rajan and colleagues, 2002) compared locally collected, unpasteurised, unfiltered honey against nationally collected filtered and pasteurised honey and against corn syrup with synthetic honey flavouring. Its result, in one sentence: “Neither honey group experienced relief from their symptoms in excess of that seen in the placebo group.”

Unusually for a bee-product belief, this one has been tested — three times, with disagreeing results, which makes it a small masterclass in reading trials. Weighing all three, the 2018 International Consensus Statement on Allergy and Rhinology graded the aggregate evidence B, summarised the benefit as “Unclear, as studies have shown differing results. Honey may be able to modulate symptoms and decrease need for antihistamines”, and reached a verdict the folk claim rarely hears: “No recommendation due to inconclusive evidence.”

Fig. I · Two pollens, one jarDiagram

The two kinds, drawn in one frame

Wind-borne · anemophilous

The pollen of hay fever

“with the wind dispersing the grains through the air”

Grasses, trees, and weeds. “anemophilous pollens are considered the most relevant allergens in the group of respiratory allergies”.

Insect-borne · entomophilous

The pollen bees carry into the jar

“carried out by insects”

Heavy: “are not airborne and hence should not be able to induce allergic sensitivity”.

Both grains sit in one frame. Height above the line is the page’s distinction — airborne, or not. No diameter is drawn: this page gives none for either grain.

What one survey found in the jar

19 natural honeys · Shanxi, northern China

61 pollen types from 37families

• 56 insect-pollinated
• 5 wind-pollinated

A single regional survey is not a universal law, and wind-borne pollen does reach honey — as a small minority of what is there.

Filled mark — one insect-pollinated pollen type, 56 of 61Hollow mark — one wind-pollinated pollen type, 5 of 61

The pollen in the jar is mostly not the pollen in the air.

Plate I · PaintingRecord

What happened when local honey was actually tested?

Thirty-six people complaining of allergic rhinoconjunctivitis, recruited at a hospital research centre and scratch-tested at entry for common aeroallergens, were randomly assigned to three groups of twelve: locally collected, unpasteurised, unfiltered honey — the closest any study has come to the folk claim; nationally collected, filtered, pasteurised honey; or the corn-syrup placebo. Each took one tablespoonful a day, keeping a diary of ten symptoms and of days needing medication. A 2021 review’s tabulation records a 30-week schedule.

Neither honey arm separated from the placebo. The authors’ conclusion: “This study does not confirm the widely held belief that honey relieves the symptoms of allergic rhinoconjunctivitis.”

Two limits sit beside that null, and they teach a general lesson about small trials. The 2021 review notes about a third of volunteers withdrew, the honey becoming unpleasant to eat daily, and suggests the dose may have been too small. And 36 people split three ways is small enough to hide a modest effect: a null in a small trial is weak evidence of absence, not proof of it. It is still the only direct test of local honey anyone has run.

What about the trial that reported honey working?

Asha’ari and colleagues (2013) ran a randomised placebo-controlled trial in two Malaysian referral centres, published as “Ingestion of honey improves the symptoms of allergic rhinitis”. Its abstract is widely quoted; its tables are not — and the tables are where the lesson is.

Forty adults, mean age 35.7, split evenly. All took 10 mg of loratadine daily for four weeks; twenty also took Tualang honey at 1 g per kilogram of body weight daily, twenty a honey-flavoured corn-syrup placebo. Symptoms were scored at weeks 0, 4 and 8.

The published table of mean total symptom scores shows no significant difference between groups at any point: p = 0.464 at baseline, p = 1.00 at week 4, p = 0.428 at week 8, where honey scored 11.9 (SD 5.66) against placebo’s 13.1 (SD 4.28). The positive finding is within-group — the honey arm kept improving between weeks 4 and 8, the placebo arm stalling once the antihistamine stopped. Here is the distinction worth carrying away: a before-and-after change inside one arm is far weaker evidence than a difference between arms, because almost anything — the season, regression to the mean, expectation — can move a single group over time. The between-arm difference is what randomisation exists to test, and this trial did not produce one.

The consensus statement read it the other way, and the disagreement is checkable. Its text records: “One double-blind trial and 1 RCT showed a significant decrease in total symptom scores in the treatment group compared to control.” Its evidence table sums this trial up as “Improvement in overall and individual AR symptoms with honey” — both between-group readings. The published table of mean total symptom scores does not carry them: at week 8, the point the claim rests on, p = 0.428. A reader who follows the citation will find the disagreement sitting in the paper’s own tables rather than in its abstract.

One point runs in this trial’s favour, and it belongs on the record. The consensus statement notes that 50 to 80 g of honey a day has been reported as the intake needed before any effect of honey would be expected to appear, and that Asha’ari’s is the only one of the three trials to dose at that level, at 1 g per kilogram of body weight. That argument cuts in both directions: it is the same argument the 2021 review makes against Rajan’s tablespoon, and it means the best-dosed test anyone has run is the one whose between-group tables came back null. It also cuts against the intake itself. At 1 g per kilogram a 70 kg adult is eating around 70 g of honey a day, and the same consensus statement pairs that level with a caution about blood glucose in prediabetes and diabetes, quoted in full below.

The honey was Tualang honey from Malaysia, local to nobody in Britain, and the results “are only applicable to the batches of honey used here”. Most participants had persistent, non-seasonal allergic rhinitis, not seasonal pollen allergy, and 97.5% skin-prick tested positive to house dust mite. The authors call the sample small.

What about the Finnish birch pollen honey study?

Saarinen and colleagues (2011) produced the one clearly positive result here, and it repays close reading — the design is the story.

Forty-four patients with physician-diagnosed birch pollen allergy took either birch pollen honey — honey with birch pollen added — or regular honey, daily in incremental amounts, from November 2008 to March 2009. Seventeen further patients on their usual allergy medication formed a control group; fifty completed. Through the birch pollen season that April and May, the birch-pollen-honey group recorded a 60% lower total symptom score than medication-only controls (p < 0.01), twice as many asymptomatic days (p < 0.01), 70% fewer severe-symptom days (p < 0.001), and used 50% fewer antihistamines (p < 0.001).

Now the design. Those numbers come from a comparison against a separate group that took no honey, knew it, and was not randomised alongside the honey arms — an open comparison, and expectation alone can move a symptom diary. The randomised, like-for-like comparison was spiked honey against plain honey, and there “the differences between the BPH and RH groups were not significant”; only antihistamine use separated them (p < 0.05). The authors’ own gloss on that comparison is that the birch-pollen-honey patients “had marginally better control compared to those on RH”. Their Conclusions section, whole and theirs: “Patients who preseasonally used BPH had significantly better control of their symptoms than did those on conventional medication only, and they had marginally better control compared to those on RH. The results should be regarded as preliminary, but they indicate that BPH could serve as a complementary therapy for birch pollen allergy.” The consensus statement grades the study level 2b, below the two double-blind trials.

The honey was also loaded, deliberately, with the exact allergen the patients reacted to — not local honey but a preparation nobody sells.

Fig. II · The three trials on one shelfDiagram

Eaten honey in allergic rhinitis — every arm, and the vessel it used

◆

◆Local

National

Placebo

Rajan, 2002

Tualang

Placebo

Asha’ari, 2013

Birch-spiked

Plain

No honey

Saarinen, 2011

Rajan and colleagues, 2002

Randomised, double-blind, placebo-controlled

36 adults with allergic rhinoconjunctivitis · three groups of twelve · one tablespoonful a day · a 30-week schedule, in a 2021 review’s tabulation

• LocalTwelve patients. Locally collected, unpasteurised, unfiltered honey — the only arm of the three trials that used honey local to the patients.
• NationalTwelve patients. Nationally collected, filtered, pasteurised honey.
• PlaceboTwelve patients. Corn syrup with synthetic honey flavouring.

ComparedTwo honeys against a placebo, between groups. The only direct test of local honey.

Result“Neither honey group experienced relief from their symptoms in excess of that seen in the placebo group.”

Beside itThat 2021 review notes about a third of volunteers withdrew, the honey becoming unpleasant to eat daily, and suggests the dose may have been too small. And 36 people split three ways is small enough to hide a modest effect.

Asha’ari and colleagues, 2013

Randomised, placebo-controlled — two Malaysian referral centres

40 adults, mean age 35.7, split evenly · symptoms scored at weeks 0, 4 and 8

• TualangTwenty patients. Tualang honey at 1 g per kilogram of body weight daily — honey from Malaysia, local to nobody in Britain.
• PlaceboTwenty patients. A honey-flavoured corn-syrup placebo.

Both arms: 10 mg of loratadine daily for four weeks

ComparedHoney against a placebo, between groups, on top of an antihistamine.

ResultNo significant difference between groups at any point: p = 0.464 at baseline, p = 1.00 at week 4, p = 0.428 at week 8, where honey scored 11.9 (SD 5.66) against placebo’s 13.1 (SD 4.28).

Beside itPublished as “Ingestion of honey improves the symptoms of allergic rhinitis”. The positive finding is within-group — a change inside one arm, which is far weaker than a difference between arms.

Saarinen and colleagues, 2011

Randomised controlled pilot study — graded level 2b

44 patients with physician-diagnosed birch pollen allergy · daily in incremental amounts, November 2008 to March 2009 · fifty completed

• Birch-spikedBirch pollen honey — honey with birch pollen added; not local honey but a preparation nobody sells. The forty-four honey patients are nowhere divided between this arm and the plain-honey arm.
• PlainRegular honey. Its share of the forty-four is not given.
• No honeyNo vessel at all: seventeen further patients on their usual allergy medication, open and not randomised alongside the honey arms.

ComparedSpiked honey against plain honey, randomised. The 60% figure belongs to the other comparison — against the separate medication-only group.

Result“the differences between the BPH and RH groups were not significant” — only antihistamine use separated them, p < 0.05. Against medication-only controls the birch-pollen-honey group recorded a 60% lower total symptom score, p < 0.01.

Beside itThe authors call the results preliminary. The honey was loaded, deliberately, with the exact allergen the patients reacted to.

◆The only arm in any of the three that used honey local to the patients. Two later trials used honey that was not local to anyone.

Why does the local-honey theory struggle before any trial is run?

Because the pollen that causes hay fever and the pollen that ends up in honey largely come from different plants — a piece of botany worth knowing in its own right.

Pollination is either entomophilous, carried out by insects, or anemophilous, with the wind dispersing the grains through the air. Airborne allergy is a story about the second kind: “anemophilous pollens are considered the most relevant allergens in the group of respiratory allergies”, and “the main anemophilous plant species are grasses, trees, and weeds” — grasses worldwide, birch, hazel and oak across northern Europe, ragweed in North America (Torres-Borrego and Sánchez-Solís, 2023).

Honey is the other kind. Bees forage nectar from insect-pollinated flowers, and the pollen carried into the jar reflects that. One melissopalynological survey of 19 natural honeys from Shanxi, northern China, identified 61 pollen types from 37 families — 56 insect-pollinated, 5 wind-pollinated. A single regional survey is not a universal law, and wind-borne pollen does reach honey — as a small minority of what is there.

The consensus statement draws the conclusion directly: heavy, insect-borne pollens “are not airborne and hence should not be able to induce allergic sensitivity”. A jar carries what the bees brought, in whatever quantity — which is why the folk theory struggles before anyone is enrolled.

Isn’t this the same idea as immunotherapy?

It rhymes with immunotherapy, which is exactly why the differences are instructive. Allergen immunotherapy is a supervised medical treatment: the NHS describes being given “small amounts of pollen as an injection or tablet to slowly build up your immunity to pollen”, on GP referral. Subcutaneous immunotherapy “involves the injection of increasing doses of an extract of the allergen in question, followed by repeated injections of the top or maintenance dose for periods of 3 to 5 years” — measured, escalated on a schedule, and so dose-dependent that “reduction of the effective maintenance dose by 90% to 95% causes partial or complete loss of efficacy”.

A jar offers none of that: not the relevant allergen reliably, not a known quantity, not a schedule, not supervision. Whether unfiltered honey carries more pollen than filtered is a real question about processing — see what “raw” means — but more uncharacterised, mostly wrong-species pollen is not a smaller version of immunotherapy.

Can honey or pollen itself cause an allergic reaction?

Yes, rarely — and here the literature runs the other way. The consensus statement lists among the harms of eating honey that “some patients could have an allergic reaction to honey intake, and in rare instances, anaphylaxis”. The same list adds that “use of this therapy in prediabetics and diabetics would likely need to be avoided out of concern for elevated blood glucose levels” — the sentence that belongs beside the 50 to 80 g a day the same statement reports as the intake level, and beside Asha’ari’s 1 g per kilogram. Eaten bee pollen carries the same allergens as the airborne pollen it is made from, and the published anaphylaxis cases concentrate in people with pollen allergy — the bee pollen page sets them out. That is the very population the hay fever idea addresses.

One NHS rule is absolute and covers every honey, however local or raw: “Do not give your child honey until they’re over 1 year old”, because honey occasionally contains bacteria that can produce toxins in a baby’s intestines, causing infant botulism.

Hay fever itself is managed by a pharmacist or a GP, and anyone managing diabetes should raise a change in sugar intake with their diabetes team.

Is a hay fever claim for honey allowed in the UK?

No — and the register’s silence is its own data point. Under retained food law, “only authorised claims in the Great Britain NHC register may be used in Great Britain” — the register for England, Scotland and Wales. No health claim for honey, or any bee product, is authorised on it; the eight honey entries, and the reasoning regulators gave, are set out on the honey evidence page. Nothing on hay fever or allergy appears there at all.

So — does local honey help hay fever?

Tested once directly, and it did not. Tested twice more with honey that was not local: once with a null between-group result published as a positive, once with a spiked preparation in an open comparison. The mechanism points the wrong way — the pollen in the jar is mostly not the pollen in the air — and the consensus grades the whole file inconclusive. What remains true of local honey is what was always true: it tastes of the place it came from.