BLOG

Microbes and Microbiota: Benefits and Risks

'Dread Reckoning' And 'Blind Spots' for Oral Transmission of H5N1

Advocates of pasteurization of the microbiota of both donor breastmilk and bovine milk appear mired in dogma, ‘opinions believed to be true or irrevocable’ (Hofmaenner and Singer 2022).

When dogma is founded on belief or myth rather than objective, rigorous scientific inquiry including hypothesis testing, difficulties arise from ‘blind spots’ of individuals and organizations, even professional organizations. In his book Blind Spots: When Medicine Gets It Wrong and What It Means for Our Health (Makary 2024), Dr. Marty Makary quotes Harvard economist Henry Rosovsky: “Never underestimate the difficulty of changing false beliefs by facts.”

From March 2024 up to the present, communications in the media and journal papers have largely reflected ‘dread reckoning’ or overestimations in risk, introduced in the 2012 article entitled ‘Dread Reckoning: H5N1 Bird Flu May Be Less Deadly to Humans Than Previously Thought--or Not’ (Branswell, 2012). Explanations of the relevance of this language and the rigor of the Risk Analysis Quality Test of the Society for Risk Analysis are provided in the appended text box after the ‘Learn More: References’ button at the bottom of this post. ‘Dread Reckoning’ for H5N1 is continuing in 2025, failing to address existing risk analysis quality components developed by risk practitioners.

Miscommunications about avian influenza H5N1 begin with common use of the term HPAI, Highly Pathogenic Avian Influenza. I refer to the virus as ‘avian influenza H5N1’ or simply ‘H5N1’. Exposure to high loads of H5N1 can cause severe and fatal pneumonia and is highly pathogenic in some birds and unprotected poultry workers; H5N1 is mildly pathogenic in dairy cows and dairy workers (Graziosi et al. 2024).

Just as Dr. Makary pointed out the glaring ‘blind spot’ in medicine that there was no ‘downside’ to antibiotics (other than ‘carpet-bombing the microbiota’), pasteurization has evoked a serious ‘blind spot’ common among many journalists, scientists, and industry and professional organizations that there is no ‘downside’ to pasteurizing donor breastmilk and bovine milk, ignoring ‘carpet-bombing’ of the milk microbiota and documented ‘downsides’ that adversely affect human health.

Pasteurization is perceived as a silver bullet, killing pathogens and causing no adverse effects. So many decades have passed mired in the ‘pasteurization blind spot’ that pro-pasteurization bias is deeply entrenched. That systematic bias requires that those with a ‘stake’ in decision making about pasteurizing milks (stakeholders) cultivate healthy skepticism, particularly willingness to fact-check claims with the actual studies and insist upon evidence-based policies and regulations.

When studies are cited in foodborne risk communications, take care to spot half-truths, invalid conclusions not supported by the evidence provided, and distortions of scientific evidence that misinform and intentionally deceive the unwary. For example, see Figure 1 of the Koski study and decide for yourself if the burden of raw milk illness is increasing or if the authors’ interpretation of the outbreak data might reflect pro-pasteurization bias.

Also notice what is left out of communications about milkborne risk. Often what is left out is a relevant reference to valid scientific evidence, particularly references to rigorous peer-reviewed studies. None of the media reports in the past year have acknowledged this fact: neither raw nor pasteurized milks are ‘risk-free’.

Communications of pro-pasteurization advocates commonly leave out evidence, limitations and uncertainties, and assumptions, all inconvenient truths, in order to perpetuate pro-pasteurization myths and maintain status quo to preserve the dominance of pro-pasteurization ‘blind spots’ a bit longer.

The claims made in the media and scientific journals are often not supported by the current body of scientific evidence on benefits and risks of raw and pasteurized milks (Coleman et al. 2021; Dietert et al. 2022). Be on alert for exclusion or dismissal of evidence that does not fit a certain cultural or political belief system or worldview (confirmation bias). Beware of entrenched beliefs masquerading as scientific facts that in reality are not supported by scientific evidence.

Below are some relevant facts that are often left out of public communications about milk benefits and risks, each with a reference that you are encouraged to review and fact-check.

1.      Neither pasteurized nor raw milk is ‘risk-free’.

a.      See Table below for summary of the burden of illness from CDC outbreak data for 2005-2020 and the peer-reviewed manuscript on this dataset (Stephenson, Coleman, and Azzolina 2024).

b. FDA/FSIS assessed pasteurized milk as high risk of severe listeriosis per annum and unpasteurized milk as high risk per serving (FDA/FSIS, 2003); however no listeriosis outbreaks associated with raw milk were confirmed in the CDC dataset for 2005-2020 (Stephenson et al., 2024) .

c.      More deaths are reported from leafy greens (23), pasteurized milk (4), and oysters (2) than the one confirmed death of an adult with severe underlying illness associated with consumption with raw milk (Davis et al. 2016). Note that another death in an adult with severe underlying illness reported in this time period could not be attributed to raw milk consumption.

d.       Higher burdens of illness are reported from leafy greens (16,434 illnesses), oysters (2,408), pasteurized milk (2,111), and many other foods than for raw milk (1,696).

2.      Pasteurized milk is a highly processed food that is linked to adverse health effects.

a.       Significantly higher outbreaks, hospitalizations, and deaths were associated with listeriosis in pasteurized dairy from 2007-2020 of listeriosis compared to raw dairy (Sebastianski et al. 2022).

b.       Stillbirths, miscarriage, premature delivery were reported for pasteurized dairy, not raw dairy (Sebastianski et al. 2022).

c.        Heating milk (boiling and pasteurization) denatures milk proteins, increasing allergenicity and contributing to inflammatory disease (Abbring et al. 2019; 2020).

d.       Allergy to and intolerance of pasteurized milk is estimated to have a substantial public health burden in the US, with 30-37% intolerance of thermally treated milk from recent US surveys (Warren et al. 2022; Warren et al. 2024). Approximately 15 million US consumers (4.7% or nearly 1 in 20) are affected by thermally-treated/pasteurized milk.

e.       Industrial processing of milk (heating, filtration, pressure, drying, freezing) causes denaturing, aggregation, and loss of function of cow milk antigens (α-lactalbumin, β-lactoglobulin, serum albumin, caseins, bovine serum albumins, and others), reductions in concentrations of bioactives (immunoglobulins, cytokines, peptides, lipophilic components, and microbiota), and impairs immunologic tolerance mechanisms (Jensen et al., 2022).

3.      Approximately 15 million raw milk consumers benefit from access.

a.       A recent government survey estimated 4.4% of US population consumes raw milk (Lando et al., 2022).

b.       Multiple sources report consumption of raw milk is increasing, not decreasing (NielsenIQ figures cited by (Aleccia 2024; Lyubomirova 2024); Figure below included in Coleman manuscript under review in Risk Analysis).

c.       CDC reported 1,696 raw milk illnesses for 2005-2020; Stephenson et al., 2024) but not inflammatory disease (Dietert et al. 2022).

d.       Raw milk was tolerated by children with allergy to pasteurized milk, and pasteurized milk induced adverse effects (Abbring et al. 2019).

e.        Raw milk has a dense and diverse microbiota, similar to the breastmilk microbiota, both inducing benefits to gut microbiota, immune system function, and suppressing growth of pathogens (Coleman et al. 2021; Dietert et al. 2022; Coleman et al. 2023; Coleman, submitted).

f.        Raw milk was associated with increased functional richness of the gut microbiota (notably genus Lactobacillus and Lactococcus and the short chain fatty acid valerate) and participants with higher than median anxiety scores showed significant score reduction for stress and anxiety (Butler et al., 2020).

4.      Illness associated with raw milk is not increasing in the US or any US state based on recent CDC data.

a.       No significant increase was reported for illnesses associated with raw milk outbreaks from 1998-2018 or numbers of outbreaks from 2005-2018 (Koski et al. 2022, Figure 1 ) or 2005-2020 (Stephenson et al., 2024, Figures 12 and 13 below).

5.      Regarding children, consuming raw milk complete with intact natural microbes (microbiota) is beneficial to health.

a.       Just as children benefit from raw breastmilk and its protective microbiota, children (and adults) also benefit from raw cow milk complete with its protective microbiota that enhance health of gut, immune, nervous, and respiratory systems (Coleman et al., 2021; Dietert et al., 2022).

b.       No child has died in the US from consuming raw milk in recent decades based on CDC data for 2005-2020 (Stephenson et al., 2024).

c.        Children with allergies to pasteurized milk tolerate raw milk consumption with no adverse effects (Abbring, 2019).

d.       Children consuming raw milk in multiple large studies developed no diarrheal illness, significantly fewer respiratory and ear infections, protection from inflammatory disease including atopy, asthma, and eczema, and improved immunologic and lung function later in life (Perkin and Stranchan, 2006; Depner et al., 2013; Loss et al. 2015; von Mutius 2016; Wyss et al., 2018; Brick et al., 2020; Dietert et al. 2022).

6.      Claims that raw milk is ‘inherently dangerous’ and that ‘risks exceed benefits’ are unfounded and not supported by the body of scientific evidence. 

a.    Recent trends in unpasteurized fluid milk outbreaks, legalization, and consumption in the United States (Whitehead and Lake, 2018): 
The study is based on CDC data from 2005-2016 documenting 1,903 illnesses associated with pasteurized milk and 1,735 illnesses associated with raw milk. The rate of raw milk related outbreaks is decreasing, meanwhile the consumption of raw milk is increasing. The authors concluded that, “Controlling for growth in population and consumption, the outbreak rate has effectively decreased by 74% since 2005.” The study suggested that the improving food safety record is the result of expanded safety training for raw milk dairy producers.  

b.      Examining Evidence of Benefits and Risks for Pasteurizing Donor Breastmilk (Coleman et al., 2021). This application of evidence mapping, a formal benefit-risk methodology with demonstrated utility as a bridge between opposing world views, structured evidence regarding the controversial issues underpinning human donor milk bank policies requiring pasteurization of donor breastmilk. Seventy-one studies were cited documenting evidence for benefits and risks of raw breastmilk for infants. Evidence of benefits was clear, convincing, and conclusive, while evidence for infectious disease risks to infants consuming raw breastmilk was limited. Supporting studies provided evidence of plausible mechanisms of benefits of ‘seeding and feeding’ the infant gut ecosystem, notably providing ‘colonization resistance’ or protection against pathogens and stimulating balanced development of infant immune systems. The study recommends broad societal deliberations of the body of evidence to resolve misunderstandings and mixed messages that invoke fear and dread so that evidence-based policies might maximize benefits and minimize risks to NICU infants.

c.        Nourishing the Human Holobiont to Reduce the Risk of Non-Communicable Diseases: A Cow’s Milk Evidence Map Example (Dietert et al., 2022). (Dietert et al., 2022). 
This study builds on the previous applications of evidence mapping as a bridge, structuring evidence that might support wider deliberation between those with opposing world views about benefits and risks of raw and pasteurized donor milks from cows. The study cited 135 studies illuminating raw bovine milk as a superfood complete with its natural microbiota intact and chosen by humans for 200 million years. A significant focus of the study is the impact of raw bovine milk on the immune system, specifically on the microimmunosome that significantly determines risk of the primary global health threat, non-communicable and inflammatory diseases (NCDs; allergy, asthma, eczyma, obesity). As noted for the breastmilk evidence map, consistent evidence of raw bovine milk benefits (promotion of gut, immune, and lung health; protection from infectious diseases and NCDs) and limited evidence of risks of infectious disease were documented in large human studies, many focusing on children.
Notably, prior discussions of the evidence focused primarily on raw milk outbreaks. This study included NCDs and significant findings from microbial risk assessments: i) Listeria monocytogenes was not considered a main hazard in raw milk and could be mitigated by cold chain and other factors; and ii) risk associated with Shiga toxigenic E. coli had subsided to negligible levels in Italy, and that models for campylobacteriosis, listeriosis, and salmonellosis overestimated risk. The study recommends that the evidence map for raw bovine milk similarly support broad societal deliberations of the evidence to begin a paradigm shift away from 20th century ideas about microbes as germs that will kill us and towards the advancing understanding of our microbiota as our partners in health and raw milk as a superfood ‘seeding and feeding’ human and microbial cells in the gut. Indeed, “restoring health may be tied to restoring diverse microbes to the industrial diet dominated by [highly] processed foods, including pasteurized milks”.

d.       Trends in Burdens of Disease by Transmission Source (USA, 2005–2020) and Hazard Identification for Foods: Focus on Milkborne Disease (Stephenson et al., 2024).
The study assessed trends for CDC data from all transmission sources for 2005-2020 and conducted the first element of microbial risk assessment, Hazard Identification, for food-pathogen pairs. The burden of illness and mortality was dominated by person-to-person transmission. Foodborne disease accounted for 21% of the disease burden with no increasing trend. Foods representing the greatest hazards were identified for campylobacteriosis (pasteurized and raw milk), illness from Shiga toxigenic E. coli (leafy green vegetables and beef), listeriosis (melons and pasteurized solid dairy products), and salmonellosis (poultry and leafy vegetables). Fatal foodborne disease was dominated by fruits, vegetables, peanut butter, and pasteurized dairy. No increasing trend of raw milk illnesses were observed overall for any state, nor did rates of illness increase after legislation to allow greater access to raw milk. The authors question the common claim that raw milk is an “inherently dangerous food”. The available epidemiologic and microbiologic evidence conflicts with assumptions of zero risk for pasteurized milk and increasing trends in the burden of illness for raw milk.

7.      Scientific evidence fails to support the hypothesis that avian influenza H5N1 transmits to humans by ingestion.

a.       Avian influenza H5N1 is not a foodborne pathogen that causes stomach flu in humans, but it can cause respiratory infection, pneumonia, and pink eye (inflammation, conjunctivitis) (Jones and Adida 2011; Lockhart, Mucida, and Parsa 2022; AbuBakar et al. 2023; Graziosi et al. 2024).

b.       The reservoir for avian influenza H5N1 is birds, NOT dairy cattle.

c.        All lines of evidence for assessing influenza transmission (Killingley and Nguyen-Van-Tam 2013) fail to support the hypothesis about oral transmission of avian influenza in humans as documented below.

i.       Avian influenza H5N1 is not highly pathogenic or highly virulent for dairy cows or dairy workers (Graziosi et al. 2024).

ii.       Mild eye inflammation was reported for 41 dairy workers in five states (36 in CA, 2 in MI, 1 each in CO, NV, and TX) exposed to infected cows (CDC 2025), despite limited transmission to dairy herds in 16 states and higher transmission to herds in California.

iii.       No influenza disease transmission was observed for non-human primates inoculated with a high oral dose of H5N1; in contrast, nasal inoculation at the same dose caused mild illness and inoculation into the deep lung caused severe and fatal illness in non-human primates (Rosenke et al. 2024). The most reliable animal model for extrapolation to humans, the non-human primate, has effective innate barriers to oral transmission of avian influenza H5N1.

iv.       None of the ferret, mice, and cat studies among the available 44 inoculation studies for H5N1 are suitable for extrapolation to oral exposures to the human gastrointestinal tract (Coleman, under review in the journal Risk Analysis).

v.       No epidemiologic evidence documents oral transmission of H5N1 to raw milk consumers. No oral infections or pneumonia were documented in dairy workers or consumers.

vi.       More than 263,000 gallons of H5N1-positive raw milk (~4.6 million servings) circulated in the CA retail market last November (Coleman, manuscript under review), with no human influenza cases reported (CDC, 2025).

vii.       Weak pathogenicity and low virulence of bovine H5N1 strains are consistent with evidence that the virus strains adhere weakly to human receptors (Santos et al. 2025).

viii.       Validated models of transmission for respiratory and ocular exposure exist for influenza A (Jones and Adida, 2011), but no mechanistic models exist for oral transmission of H5N1.

b.      Transmission of H5N1 virus between dairy herds in 2024 is consistent with the common practice of movement of animals to farms other than those where they were born (Nguyen et al. 2025), practices uncommon in smaller dairies licensed to sell raw milk direct to consumers.

c.       Mandatory testing before animal movement contributed to reduced transmission to other dairy herds in US states (Nguyen et al., 2025).

d.      No evidence exists for the hypothesis that milk is a vehicle for H5N1 disease transmission between animals, between herds, or to human consumers.

8.      Evidence of the presence of viral or bacterial pathogens in milk or on surfaces is insufficient to estimate risk of illness in raw milk consumers with attendant uncertainties, consistent with risk analysis principles and established guidance on risk analysis quality (RAQT of the SRA, 2020).

a.       Presence of a pathogen is an invalid predictor of risk of human illness.International consensus defines four required elements for assessing microbial risk: Hazard Identification; Exposure Assessment; Dose-Response Assessment; and Risk Characterization (Codex Alimentarius Commission, 1999).

b.       Estimating risk requires cohesive knowledge well characterized body of knowledge by which to reliably scale doses and responses in animals to humans (Swearengen, 2005; Swearengen, 2018; Zheng, Song, and Zheng 2024).of: i) mechanisms of infection in humans; ii) ecological and mathematical relationships between microbial load (ingested or inhaled or contacted on surfaces) and transmission of human illness by route (oral, inhalation, contact); and iii) doses causing no illness, mild illness, and severe illness by route (McClellan et al. 2018; Coleman et al. 2018).

c.       Extrapolation from animal models requires a well-characterized body of knowledge by which to reliably scale doses and responses in animals to humans (Swearengen, 2005; Swearengen, 2018; Zheng, Song, and Zheng 2024).

i.       Ferrets, mice, and cats are not equivalent to humans, and host extrapolation would require data on anatomical, physiological, immunological, and behavioral differences between humans and laboratory animals

ii.       Nor are non-human primates equivalent to humans.

iii.       If extrapolation is necessary, non-human primates are more representative of human systems than other laboratory animals based on a long history of risk analysis.

iv.       Animal models that can accurately replicate both fatal and non-fatal human infections, as well as consistent time courses and severities, are relevant for extrapolation to human disease transmission.

d.        Quality risk analysis studies commonly document no disease transmission at low pathogen exposures (ingested or inhaled doses or delivered pathogen load) and increasing likelihood and severity of illness at increasingly higher loads, influenced by our microbial partners in health, our microbiota (Coleman et al., 2008; McClellan et al. 2018; M. Coleman et al. 2018; Lathrop et al., 2024; Waller et al., 2024).