The proposal that the immune system requires exposure to organisms it co-evolved with in order to develop proper regulation, and that losing those organisms, rather than losing childhood infections, explains the rise in allergy and autoimmune disease. It is the successor to the hygiene hypothesis and it corrects a specific error in it.
The hygiene hypothesis, treated separately, proposed that reduced childhood infection leaves the immune system underdeveloped. It came from David Strachan's 1989 observation that children with more older siblings had less hay fever.
That framing has two problems the old friends account addresses.
Childhood infections are the wrong organisms. Measles, mumps and influenza are evolutionarily recent, requiring dense populations to sustain transmission, and only became common after agriculture. They cannot have shaped immune development over the timescales that matter.
And the hypothesis carries an unfortunate practical implication: that hygiene is harmful. Sanitation, clean water and handwashing prevent enormous amounts of disease, and public messaging suggesting otherwise is actively dangerous.
Graham Rook set out the revision from 2003.
The organisms that matter are ones present throughout mammalian and human evolution: helminths, or parasitic worms; saprophytic mycobacteria and other microbes in soil, untreated water and animals; and the commensal organisms of the gut, skin and airways.
These were universally present until very recently. Because they were always there, the developing immune system came to rely on signals from them, in the same way that a developing visual system relies on receiving light.
The mechanism proposed is regulatory rather than stimulatory. These organisms induce regulatory T cells, which suppress inappropriate immune responses. Helminths in particular actively promote this, because a worm that provokes a strong immune response gets expelled, so worms evolved to dampen host immunity for their own reasons.
Losing them leaves regulation underdeveloped. The result is an immune system that responds to things it should ignore: harmless proteins, producing allergy, and the body's own tissue, producing autoimmunity.
The epidemiology fits reasonably well. Allergy and autoimmune disease are far more common in industrialised countries and have risen as helminth infection and rural exposure declined. Within countries the gradient follows urbanisation.
Farm studies are the strongest observational evidence. Children raised on traditional farms, particularly with livestock and unprocessed milk, have substantially lower rates of asthma and allergy, and the protection correlates with exposure to microbial diversity rather than with hygiene practices as such.
Helminth infection is associated with lower allergy prevalence in several populations, and deworming programmes have in some studies been followed by increases in allergic sensitisation, which is an awkward finding for public health and is reported honestly by the people running them.
Migration studies show that people moving from low-prevalence to high-prevalence countries acquire the higher risk, and their children more so, which rules out genetics as the explanation.
The intervention trials have largely disappointed. Deliberate helminth therapy, infecting patients with hookworm or pig whipworm to treat inflammatory bowel disease, multiple sclerosis or allergy, produced encouraging early results in small open studies and has generally failed in larger randomised trials.
That is the same pattern seen in the gut-brain literature and it should be read the same way: promising uncontrolled results that do not survive proper controls.
The mechanism is described at a level that resists testing. Microbial diversity is not one exposure, and identifying which organisms, at what age, through what route, has not been achieved. Without that the hypothesis cannot be tested sharply.
Confounding is severe. Farm children differ in diet, vitamin D exposure, physical activity, antibiotic use, air pollution and stress. Urbanisation changes everything at once.
And the rise in allergy has other candidate explanations that are not exclusive: antibiotic use in infancy, caesarean delivery bypassing maternal microbial transfer, dietary change, reduced vitamin D, indoor air, and altered timing of allergen introduction. The last is now established in a way the others are not: introducing peanut early substantially reduces peanut allergy, which was demonstrated in the LEAP trial and reversed clinical guidance.
The hypothesis does not recommend abandoning hygiene, and Rook has been explicit about this.
Handwashing, clean water and food safety prevent transmission of pathogens and have no established role in the loss of regulatory exposure. The organisms implicated come from soil, animals, untreated environments and the maternal microbiome, none of which sanitation targets.
The realistic implications are narrower: restoring specific exposures rather than reducing hygiene, identifying the molecules through which helminths suppress immunity and developing them as drugs rather than using live parasites, and attending to how infants acquire their initial microbial community.
It is classified as a hypothesis because the epidemiological pattern is strong, the proposed mechanism is plausible and partly demonstrated in animals, and the interventions predicted by it have not worked in humans.
