In March of this year, I wrote a post discussing the limitations and misleading conclusions of a research study in the Journal of Veterinary Internal Medicine (JVIM) which claimed that feeding puppies a raw diet could reduce the risk of allergies later in life. This article was produced by the DogRisk research group, which has consistently supported raw diets in scientific publications and through other platforms, including the bogus documentary The Truth about Pet Cancer. My main concerns with the article were:

• The potential for ideological and funding bias
• The reliance on a convenience sample of online owner surveys as the main source of data
• The inconsistency of many of the findings with the underlying hypotheses and conclusions

Because such an article in such a prestigious journal is likely to create the false impression of strong evidence in favor of feeding a raw diet, I felt it was worthwhile to submit a letter to the editor of JVIM detailing my concerns about the report. I was joined in this by Dr. Jennifer Larson, a board-certified veterinary nutrition specialist. 

The process of getting this letter printed in the journal was surprisingly prolonged and difficult. The letter was initially submitted in January, and has finally appeared today, seven months later, which is an unusually long time. The main problem appears to have been concerns on the part of the JVIM legal team that the journal might be sued over my characterization of one of the funders, Dr. Joseph Mercola. Regular readers will know that Dr. Mercola, and the leader of the veterinary side of his website, Dr. Karen Becker, are stalwart promoters of pseudoscience and alternative medicine. Dr. Mercola has been warned and sanctioned several times by the FDA and the FTC for illegal medical claims, but it was quite a struggle to get JVIM to accept this factual information as part of the letter. 

Despite this, and the compulsory insertion of much excessive qualifying language, the letter has finally appeared in the journal. Though it repeats much of what I wrote in my blog post, I will reproduce the full content here:

The report Puppyhood diet as a factor in the development of owner-reported allergy/atopy skin signs in adult dogs in Finland, which appeared in the September/October issue of the Journal of Veterinary Internal Medicine (JVIM), suggests that “puppyhood exposure to raw animal-based foods,” “human meal leftovers” and other “real foods” might have a protective influence against canine atopic dermatitis (CAD) while “heat-processed foods” might increase later occurrence of CAD. We are concerned that the results of this study likely represent residual bias in data collection and analysis rather than the actual influence of specific dietary components on the risk of CAD.

Common recognized sources of bias in research studies include a priori beliefs on the part of researchers, recall and selection bias in survey respondents,¹ funding bias,² and a large number of researcher degrees of freedom in the design, conduct, and analysis of the study.³All of these are present in this study.

Dr. Hielm-Björkman and the Dog Risk research group have expressed strong beliefs about the health benefits of raw diets and the dangers of conventional commercial pet foods in previous research reports and popular media.^4, 5

The main source of data in this study is an online survey available only in Finnish. Previous publications regarding validation of this data source have identified low response rates to validation questions and important proportions of duplicate, automated, and discordant responses.^6, 7 Asking owners to remember in detail what they fed their puppy between 2 and 8 months of age and then trying to associate that with health outcomes years later is, in our view, a questionable strategy. Self-reporting of diet and health information are unreliable in humans, and we believe that it is unlikely to be more reliable among dog owners.¹ Dog owners consistently misperceive even straightforward measures such as body condition score despite formal training in this assessment, so their assessment of signs of CAD is, in our view, likely to be equally unreliable.⁸ The survey responses are expressions of the perceptions and beliefs of the owners who participated, not necessarily the true nutritional and environmental exposures nor health outcomes experienced by the dogs.

Dr. Hielm-Björkman has previously cited funding bias as a problem in the study of raw diets, stating that raw foods are “not really researched in universities. Most universities get sponsored by these big billion-dollar companies, and you don’t really want to step on their toes, I guess.”⁵ In light of the recognized problem of funding bias it could be relevant that disclosed funding sources for this research include companies selling raw pet foods.

The authors also report accepting funding from Dr. Joseph Mercola. Dr. Mercola’s organization advocates for raw pet diets and argues against conventional pet foods on its web site.^9–11 We believe that he is also a consistent promoter of unapproved medical practices and medical product claims, as evidenced by multiple warning letters from the Food and Drug Administration and a lawsuit by the Federal Trade Commission related to therapeutic claims that violate the U.S. Food, Drug, and Cosmetic Act.^12–16 Such an affiliation is at least as relevant to assessing potential bias as accepting funding from an organization with a commercial interest in raw or conventional pet foods. Bias in scientific research is as likely to arise from ideological as financial factors.

We believe that the results of the analysis and how they are reported also suggest the influence of bias. Many comparisons are listed only in supplemental materials, and those reported in the main article tend to be those which support the authors’ hypotheses. This creates the appearance of consistency when, in fact, the associations identified are often inconsistent and do not logically support the claims in the conclusions. For example, why would raw tripe and organ meats be protective against CAD but raw red meat, eggs, and poultry not be? If cooking is the key risk factor, why would cook vegetables be protective and raw vegetables would not or why would both cooked and raw eggs be protective while neither cooked nor raw poultry is associated with the likelihood of CAD? If exposure to bacteria is the main variable, why is eating dirt, sticks or carcasses protective but eating clay and grass is not and drinking from puddles is actually associated with increased risk? If excessive processing is the issue, why was there no association with eating processed meats or canned foods and only a marginal association with dry food when it was the only food offered?

Dogs with CAD were reportedly more likely to be eating no raw food at all than controls, and dogs without CAD were more likely to be fed 20% or 90% raw, but there were no statistically significant differences at any other ratio of the 2 foods. Similarly, allergic dogs were more likely to be fed 80% dry than controls, but there was no significant difference in CAD risk if they were fed more than 80% dry. Control dogs were more likely to be fed 50% or less than 10% dry, but there was no difference at intermediate ratios. It is easier to see cherry picking and researcher degrees of freedom than a consistent dose-response in these results.

While the authors do acknowledge the potential for recall bias and misclassification, and they do state that their findings cannot prove causal relationships, the overall message of the paper is that uncooked foods and human meal leftovers likely have health benefits, and this is inconsistent with the methodological limitations and potential for residual bias of this study.

Given the importance of CAD, in terms of both prevalence and the negative impacts on pet and owner quality of life, we agree that identification of protective factors is warranted. However, considering the significant risks of illnesses and death that have been consistently associated with the feeding of raw animal products, rigorous and well-designed research methodology, objectivity, and full reporting of results are needed to explore if any benefits exist for this practice, and certainly are required before this practice can be confidently recommended.

CONFLICT OF INTEREST DECLARATION
Jennifer Larsen declares the following conflicts:

Investigator in clinical trials and other research partly or fully sponsored by Royal Canin, Nature’s Variety Instinct, and Nestle Purina PetCare.

Develops educational materials for Mark Morris Institute and HealthyPet magazine.

Serves as advisory group consultant for Elanco Animal Health.

Participates in continuing education events, as a speaker & as an attendee, sponsored/organized by Royal Canin, Nestle Purina PetCare, Nature’s Variety Instinct, and Hill’s Pet Nutrition.

A resident of the Nutrition Service received funds through the Hill’s Pet Nutrition Resident Clinical Study Grants program.

The Veterinary Medical Teaching Hospital at the University of California, Davis receives funds from Nestlé Purina PetCare to partially support a nutrition technician.

Brennen McKenzie declares no conflict of interest.

As is usual and appropriate, the authors had an opportunity to provide a response to the letter. The response didn’t really contain any meaningful rebuttal of the concerns we expressed, merely a series of assertions that they weren’t valid, relevant, or important. Here is the full content of the authors’ rebuttal:

We read Drs McKenzie and Larsen letter with interest, and we thank the Editor for the opportunity to discuss the scientific questions through this forum.

Dr McKenzie (skeptvet¹) and Larsen in their letter to the editor indicate that the DogRisk research group has a priori notions about the benefits of raw feeding. We want to point out that we only have three veterinarians and only three dog owners in our research group of eight, only two feeding raw, and that we take pride in discarding our hypotheses every time we prove them wrong. Regarding raw food we have time after time, through published articles and 10 university student theses in Finnish, seen that there are health benefits when owners use a raw diet for their dogs, compared to a dry. We acknowledge the variety of raw and dry dog diets available and therefore we also know that the reasons behind the health benefits are hard to estimate. We have analyzed gene-expression² and metabolomics³ in a raw-dry diet-intervention study. We look at processing, the absence of heat, macronutrient profile (proteins, fats, and carbohydrates, content and origin), bacterial load, among other factors. We now work with 10 hypotheses of why the raw food diet repeatedly comes up as healthier in our studies. In a world where our dogs suffer increasingly from noncommunicable diseases that are similar to those of humans (atopy, allergies, IBD, diabetes, certain cancers and similar) we can also make veterinary medicine relevant for human medicine by reporting what we see in dogs, that share most of the other environmental factors with us, their human owners. To this end, diet is an excellent variable to investigate, as dog owners tend to keep their dogs on the same diets for extended periods of time, sometimes a whole lifetime. Also, because of diets being so homogenous, owners do not tend to forget or misrepresent it, as they might do about their own diet. These are our motivators, not proving a preconceived hypothesis.

The second issue that Dr McKenzie and Larsen were concerned about was survey problems. All relevant concerns are mentioned in the limitations part of the article, as they themselves point out. In contrast to what was said in the letter, we want to point out that our research is, and has been in all our papers, completely transparent. We always mention the reasons for omitting cases: for example, robot answers, dogs not reported to eat enough to stay alive, too young controls as they still could develop the disease etc. As expected from an academic research group, our survey is validated,⁴ including the dependent variable of this paper (allergy/atopy). We are of the same opinion with the authors of the letter that owners misperceive especially body condition score; our data show that only 13% of owners have reported that their dogs are overweight (12%) or obese (1%). But we also know that they are good at perceiving the clinical signs their dogs have: itching, ear, eye or skin infections, anal gland problems, teeth and gum problems etc., as those are the reasons that they contact their veterinarians for appointments. The owners in our survey also had the possibility to tick other skin related disorders (eg, hot spots, demodicosis, seborrhea etc.) than allergy/atopy. And as they did not, there are not so many other diagnoses than allergy/atopy they could have had, as flea allergic dermatitis is not a common entity in Finland.

Further, the authors of the letter to the editor addressed an issue we again take pride in, analysis of the data. We totally disagree with the notion of “cherry picking” data and instead again want to highlight that by putting all results in either the main article or supplemental material for other researchers to see, we can best forward this area of research. Regarding selection bias, when using backward stepwise regression, the computer starts with all variables and excludes the ones with the lowest coefficient of determination. The result is a computer-generated final model, including the significant variables. Only the machine chooses, not humans so no bias is even possible. It is common practice to discuss the significant results and trends while the pre-analyses are put into the supplemental data (S1-2). Figure 2A,B in our article show that there are more non-atopic dogs than atopic dogs when the dogs have been eating more than 10% raw food and on the contrary, more atopic dogs, when they have been eating more than 80% of their diets as dry. Not all 10% intervals will show significance between the two diets as there will not be enough cases at all intervals and it is also normal that some of 20 intervals (here one; eating 60% dry) will be out of line, for the same reason. That it is so consistent, is remarkable.

At the start of our response, we mentioned some of our hypotheses for why raw diets come up as healthy, but that we lack an answer to the “why?”. The authors to the letter also had many “If-why?” questions. At this time, we do not have the answers, but we hope that we will be able to answer them within a couple of years.

Another issue raised was funding bias. Also here, we have been completely transparent as can be seen in our long list of funders. We have been able to attract funding, for example, from state funding bodies, foundations, private people and companies by crowdfunding, raw food companies etc. and we are equally thankful to all. However, the big traditional dry feed companies have not been interested in funding our research, despite enquires. On our website, we disclose and thank all our sponsors (Moomin trolls and all) while letter author McKenzie instead has paid advertisements from Hills’ and Mars, for example, featuring the same Royal Canin advertisement on atopy diets both as “sponsored content” and in his “education center,” on his Veterinary Practice News homepage.⁵ Dr JA Larsen discloses her close co-operation with Mars, Hills’ and Nestle’ Purina in the Conflict of Interest section in all of her published articles.

Finally, we would like to point out that we have no “raw food agenda” but if we find that the “raw” is of value, we feel that we have an ethical obligation to the community to report it. We will therefore now recommend that people who feed dry food supplement the diet with at least 20% raw. If the health benefits come from beneficial bacteria, we will recommend that they should be added to the dry food, etc. Also, contrary to the letter authors’ comments, we have and will continue to report on raw feeding not being dangerous for neither dogs nor family. In our risk-analysis study, we found only three verified cases of food pathogen transmission from raw dog food to humans in 16?475 households feeding raw and in 98?353 pet years at risk.⁶ This study has been replicated with similar results.⁷ As a university-based independent research group, we are driven by a concern for the growing disease load in our dogs. We hope our research will benefit both canine and human health and therefore we report everything we find, as we find it.

Here is how I view these responses:

• “We only have three veterinarians and only three dog owners in our research group of eight, only two feeding raw”
  
  If I interpret this correctly that all eight members of the group own dogs and only two of them feed raw, I would say that is interesting, but not really relevant to the issue of whether the group in general has a bias in favor of raw diets that influences their work. Our personal practices don’t always perfectly mirror our beliefs.

• The next point they make is that their studies have consistently found benefits to feeding raw and that they have numerous hypotheses for why, though they cannot yet claim a specific reason for these benefits. Again, the fact that their work consistently finds what they believe and expect to find doesn’t say much about whether those beliefs, of the evidence they have generated, are correct. It would be much more convincing if other research groups with other perspectives and agendas found the same results. 
  
  It is always the case in science that when much of the evidence for a claim comes from a single source, that is a reason to be skeptical of that evidence, and there is no reason not to apply that principle to the subject of raw diets or the DogRisk group.

• The group next acknowledges the limitations of an owner survey as a data source while simultaneously claiming that the particular evidence and conclusions they have drawn from such a survey are correct. This is just an assertion of belief, not evidence that the limitations Dr. Larson and I raised are not influencing the findings.
  

• They then respond to our pointing out the inconsistency of the findings and the emphasis only on those that support their claims while those that do not are relegated to the supplemental material. This response is, again, just to assert that there is nothing about these findings or reporting that is problematic or should undermine confidence in their conclusions. That is for readers of the article to decide, but simply responding to the identification of a potential problem by saying “that’s not a problem” isn’t a very compelling argument.
  
  In particular, they acknowledged that they cannot explain some of the inconsistencies we pointed out in the findings, but they don’t see these inconsistencies as a reason to question their underlying conclusions about the benefits of raw diets. That response certainly has all the hallmarks of cognitive dissonance.
  

• With regard to the issue of funding bias, the authors respond by arguing that it isn’t a problem since they disclose their funding sources and then move on to the tu quoque fallacy, which is the strategy of arguing that if someone else has a potential financial conflict of interest, that negates the potential significance of their own. 
  
  Apart from being irrelevant, this claim fails with regard to my own supposed financial interests. Since I receive no funding of any kind from the pet food industry, the only way they could make a claim of funding bias was by pointing out that the Veterinary Practice News puts ads for pet food on some of the pages where they print my column on evidence-based medicine. 
  
  I not only have no influence on advertisers the magazine works with and receive no income from ad in the magazine, I am actually unaware of what ads appear in the magazine since I rarely even read my own columns once they have been published. How this could possibly influence my positions on raw diets is unclear, so this is a bit of a desperate tactic.
  

• They close their rebuttal by simply reasserting their claims that raw diets have benefits and making the truly astounding claim that they have virtually no risks because they haven’t seen any in their survey-based research. Given the robust literature showing the risks of infectious disease for humans and animals associated with raw pet foods, this is a completely incredible claim, and nothing could possibly illustrate the biases and blindness to contrary evidence of this group more clearly.

References from Original Letter

1.         Ravelli MN, Schoeller DA. Traditional Self-Reported Dietary Instruments Are Prone to Inaccuracies and New Approaches Are Needed. Front Nutr. 2020;7:90. doi:10.3389/fnut.2020.00090

2.         Resnik DB, Elliott KC. Taking financial relationships into account when assessing research. Account Res. 2013;20(3):184-205. doi:10.1080/08989621.2013.788383

3.         Simmons JP, Nelson LD, Simonsohn U. False-positive psychology: undisclosed flexibility in data collection and analysis allows presenting anything as significant. Psychol Sci. 2011;22(11):1359-1366. doi:10.1177/0956797611417632

4.         Fredriksson-Ahomaa M, Heikkilä T, Pernu N, Kovanen S, Hielm-Björkman A, Kivistö R. Raw Meat-Based Diets in Dogs and Cats. Vet Sci. 2017;4(4):33. doi:10.3390/vetsci4030033

5.         Habib R, Becker K. The Dog Cancer Series: Rethinking the Cancer Epidemic Vol. 1- Chapter 4 (Transcript).; 2018:73.

6.         Hemida M, Vuori KA, Salin S, Moore R, Anturaniemi J, Hielm-Björkman A. Identification of modifiable pre- and postnatal dietary and environmental exposures associated with owner-reported canine atopic dermatitis in Finland using a web-based questionnaire. PLoS One. 2020;15(5):e0225675. doi:10.1371/journal.pone.0225675

7.         Roine J, Uusitalo L, Hielm-Björkman A. Validating and reliability testing the descriptive data and three different disease diagnoses of the internet-based DOGRISK questionnaire. BMC Vet Res. 2016;12(1):30. doi:10.1186/s12917-016-0658-z

8.         Eastland-Jones RC, German AJ, Holden SL, Biourge V, Pickavance LC. Owner misperception of canine body condition persists despite use of a body condition score chart. J Nutr Sci. 2014;3:e45. doi:10.1017/jns.2014.25

Introduction
Given the media frenzy the article has kicked off, it is unlikely very many people are unaware of a recent study reported in the journal PLOS One that claims to show vegan diets are healthier for dogs and cats than diets containing meat. 

Knight A, Huang E, Rai N, Brown H. Vegan versus meat-based dog food: Guardian-reported indicators of health. PLoS One. 2022 Apr 13;17(4):e0265662.

Dramatic headlines have appeared in many major news sources proclaiming that vegan diets are “linked to better health” or “are healthier and safer” than conventional diets. Getting beyond the headlines, one may even read that this study demonstrates raw diets to be healthier than conventional foods, though that is less emphasized in both the original article and the media coverage.

I’ve written about vegan diets for dogs before, and indeed this article cites my column on the subject, though only to dismiss my claim that vegan diets “should not be recommended” as “without evidence.” This is not entirely accurate, since my article did cite the limited evidence available, and I did not actually recommend against vegan diets for dogs, though I did conclude they were a bad choice for cats. My actual claims were:

There is no evidence vegetarian diets have health benefits for dogs and cats, and no real reason to believe they should, based on the physiology and nutritional requirements of these species.

Dogs are omnivores that are able to eat both plant and animal foods, and in theory, they should be able to thrive on vegetarian or vegan diets. However, these diets must be carefully formulated, and many commercial vegetarian dog foods do not appear to be nutritionally adequate. There is also little reliable research evidence showing dogs can remain healthy when fed only a vegan diet. Given the unexpected health problems seen with theoretically adequate grain-free diets, we should be cautious about the potential risks of vegetarian formulations for dogs until there is better and more evidence showing their long-term health effects.

Cats are clearly obligate carnivores with nutritional requirements that are unlikely to be effectively met by vegan diets. Such diets offer only risks and no benefits for cats and should be avoided.

In any case, I am sympathetic to many of the potential environmental and health benefits of plant-based diets, and I have been a vegetarian, though not a vegan, for almost 20 years  (technically, an ovo/lacto/pescatarian, though realistically I don’t like fish and so eat very little of it). However, the evidence for benefits of plant-based diets for dogs and cats is far weaker than that for humans, which is itself often nuanced and not always conclusive. And I believe in following the evidence wherever it leads, even when it doesn’t support my personal beliefs or practices.

So, does this paper change the game in terms of showing real health benefits to raw and/or vegan diets? Spoiler, but not really! Let’s look at the actual findings a bit closer.

The Study- Results
The study reported the following statistically significant differences between “conventional,” raw, and vegan diets (leaving aside the fuzzy definitions of these, the ubiquitous feeding of unidentified treats to all pets, and the lack of clarity about how strictly feeding practices corresponded to each category).

• Dogs fed raw and vegan diets were less likely to have had multiple veterinary visits in the year observed.
• Dogs fed raw and vegan diets were less likely to have been given medications
• Dogs fed raw and vegan diets were less likely to have been transitioned onto a therapeutic diet
• Dogs fed raw and vegan diets were less likely to have an owner-reported veterinary assessment of poor health status
• Dogs fed raw and vegan diets were less likely to have an owner reported assessment of poor health
• Dogs fed raw and vegan diets were less likely to be reported to be “unwell,” and when unwell dogs on raw diets had fewer reported disorders than dogs on conventional diets.
• There were differences in the occurrence of specific health conditions between diet groups as illustrated in this figure and table:
• Dogs fed raw and vegan diets were less likely to have had multiple veterinary visits in the year observed.
• Dogs fed raw and vegan diets were less likely to have been given medications
• Dogs fed raw and vegan diets were less likely to have been transitioned onto a therapeutic diet
• Dogs fed raw and vegan diets were less likely to have an owner-reported veterinary assessment of poor health status
• Dogs fed raw and vegan diets were less likely to have an owner reported assessment of poor health
• Dogs fed raw and vegan diets were less likely to be reported to be “unwell,” and when unwell dogs on raw diets had fewer reported disorders than dogs on conventional diets.
• There were differences in the occurrence of specific health conditions between diet groups as illustrated in this figure and table:

All-in-all, these results would seem to be pretty bad news for conventional diets (whatever that means) and pretty good news for feeders of raw and vegan diets. Anyone reading this, however, can probably sense the “but” coming, so let’s get into the caveats.

The Study-Limitations
The biggest limitation (and boy is it a big one!) is that all the data were collected from online surveys of pet owners. Everything from the diet fed to the health status and even the reported veterinary assessment of health all relied entirely on the responses of pet owners who found the survey online and took the trouble to complete it. None of the facts, not the feeding practices nor the health of the pets, were verified objectively in any way. Right from the start, then, that makes this a study of what some small subset of pet owners believe about their pets’ diet and health, not about the actual feeding and health of these animals.

It is not difficult to find reasons to be concerned that these responses may not reflect reality, or even the opinions of other groups of pet owners. The respondents were overwhelmingly female (92%) and European (86%), which is a pretty narrow population to survey. They were, of course, also in a socioeconomic class inclined to participate in online pet health surveys, which is also not representative of many other pet owners.

More concerning, 22% of the population of owners were themselves vegan. This is a group likely to have strong beliefs and biases about plant-based vs meat-based nutrition and health, so it would be, frankly, shocking if they did not report that plant-based diets were healthier than other diets. 13% of these owners fed their pets vegan diets and 33% fed raw meat, indicating nearly half fed unconventional diets not usually recommended by veterinarians or veterinary nutrition specialists. Such a group is clearly a population biased in favor of the kind of outcomes reported in this study, and given the study only shows us the opinions of owners, not actual objective data about health and diet, the outcomes are simply a fancy way of reporting what people in this group think. 

Other studies in other populations have reported lower rates of veganism (5.8%) and of feeding vegan diets to pets (1-3%), which emphasizes that this study surveys a very specific, narrow group of owners. A review of previous owner surveys also shows more nuanced and variable feeding practices reported in these studies, again suggesting this paper may not be applicable outside of the specific population included in the survey.

In terms of potential sources of bias beyond the survey population, it is worth pointing out that the lead author is a consistent advocate for plant-based diets, for humans and pets, on environmental and ethical grounds. The study was also funded by a plant-based diet advocacy group. As I have discussed in detail in previous articles on conflict of interest, these facts do not indicate the research is fraudulent or inaccurate, nor are they a justification for ignoring the claims, arguments, and evidence provided in this paper. They are, however, a reason to consider carefully the potential for uncontrolled unconscious bias in the design, conduct, analysis, and reporting of the study. Given that the study itself was essentially a survey of subjective opinions, methods to control such bias are minimal, so the results have to be viewed in that context.

The authors, to their credit, do acknowledge some of these limitations. They call out the fact that, for example, fewer reported veterinary visits in dogs fed raw diets may be due to the fact that feeders of raw diets are often skeptical of conventional veterinary medicine and less likely to seek care rather than to any actual difference in health status. The same logic, of course, applies to the frequency of vet visits for dogs fed vegan diets, and to the reported use of medicine and therapeutic diets by owners who have a clear preference for unconventional health practices to begin with.

Conclusions
So what does this study mean? Overall, it means that the particular population of pet owners surveyed believes that feeding raw and plant-based diets are associated with better health in their pets. They also believe that their veterinarians think their pets are healthier (though whether these vets actually believe this is unknown). And these owners report less use of veterinary medical services, though whether this means their pets are healthier or simply that they try harder to avoid taking their pets to the vet is also unknown.

Like previous studies relying on owner surveys and both conducted and funded by folks with strong a priori opinions about diet and health, this is a useful insight into such beliefs. It is not compelling or probative evidence for actual health effects of different feeding strategies. Sadly, the media coverage of the paper rarely recognizes or emphasizes this.

As the authors themselves suggest, though with little evident enthusiasm, controlled studies with objective measures of outcome and more defined and verified feeding practices are required to draw any meaningful, actionable conclusions about the healthiest feeding strategy for our pets. I am neither for nor against vegan diets for dogs, and I am even open to reversing my objection to feeding vegan to cats or raw diets to cats or dogs if strong evidence is generated that these are safe or beneficial practices. However, regardless of the difficulties in funding and conducting the necessary research, we are not justified in making confident claims about the health impact of raw, vegan, or conventional diets without it.

I have written many times about the controversial subject of raw diets for dogs. The bottom line, based on existing evidence, is there are no proven health benefits to these diets and several well-documented risks, including infectious disease and nutritional inadequacy.¹

Proponents of these diets offer varied theoretical arguments for why they should be healthier than commercial cooked foods. Many are simply fallacious claims about what is “natural” or “ancestral.” However, there are some plausible, but unproven hypotheses, about the potential negative health impacts of conventional diets or the possible health benefits of feeding uncooked meat.

Concerns about Maillard reaction products (MRPs) in cooked foods, for example, are reasonable given the potential carcinogenicity of some of these compounds at some amounts in some species. However, it is far from evident these compounds actually do cause harm in the form and amount found in cooked dog foods, and research has identified potential benefits as well as risks to their consumption, so the assumption that cooked foods must be unhealthy because they contain MRPs is unreasonable.²

Similarly, the hypothesis that exposure to microorganisms during development can reduce the subsequent risk of allergies and other immune-system diseases is plausible, and there is some supportive evidence, though there is also evidence against this “hygiene hypothesis.”^3-6 However, this does not justify the assertion that exposure to such organisms through eating raw meat, dirt, or other substances, has more benefits than risks.

A recent paper published in the Journal of Veterinary Internal Medicine (JVIM) purports to support claims for health benefits from raw diets.⁷ 

Hemida MBM, Salin S, Vuori KA, et al. Puppyhood diet as a factor in the development of owner?reported allergy/atopy skin signs in adult dogs in Finland. J Vet Intern Med. 2021;35(5):2374-2383.

The authors suggest exposure to raw diets in puppies may reduce the risk of later development of canine atopic dermatitis (CAD), and exposure to conventional commercial diets may increase CAD occurrence in adulthood. There are, however, numerous limitations and red flags associated with this study, and it is not strong evidence.

The authors of the paper have a long history of promoting raw diets and various alternative therapies, and they have a clear bias on the subject.^8-16 Of course, researchers often have confident beliefs, rather than tentative hypotheses, when they design and conduct scientific studies. It is well-established that “researcher degrees of freedom,” the amount of leeway investigators have in designing studies and in collecting and analyzing data, is associated with the likelihood of false-positive results.¹⁷

Given researchers with clear a priori beliefs about raw diets designed the survey, defined the variables of interest, and had total freedom to create and conduct the regression analysis, the results likely reflect the beliefs of the researchers more than the underlying reality of nature.

One example of the clear bias behind this study is in the terminology chosen for different diet categories. They divide foods into “non-processed meats/ingredients” and “ultra-processed carbohydrate-rich foods.” These are artificial categories reflecting the ideological bias of the researchers, not the reality of the kinds of foods dogs are actually fed.

“Non-processed” is meaningless unless the dogs are eating whole prey and plants they find for themselves. Likewise, the term “ultra-processed” is also an inappropriate extrapolation of terms used to describe human snack foods, which are not meaningfully similar to commercial pet foods. The label is a value judgment that implies a negative health impact not actually proven to exist.

Commercial canned and dry foods are also not accurately described as “carbohydrate-rich.” The carbohydrate content of commercial dry diets varies dramatically, as does the type of carbohydrate (from simple sugars to complex fibers). The nutritional value and potential health effects of commercial diets will vary greatly based on these differences, and lumping them all together in this artificial category is not a sound scientific approach.

Another potential influence on this study is funding bias. It is, again, well-established that funding sources can influence the outcomes of research, which is why disclosure of funding sources is required by scientific journals.¹⁸ In a web documentary called The Dog Cancer Series, one of the authors of this study herself cites funding bias as a factor in the literature concerning raw diets:

“[Raw food is] not really researched in universities. Most universities get sponsored by these big billion-dollar companies, and you don’t really want to step on their toes, I guess. But, I think that’s not really ethical,” says Dr. Anna Hielm-Bjorkman.¹⁹

It is relevant, then, to point out this study received financial support from raw diet manufacturers, as well as Dr. Joseph Mercola, an infamous proponent of alternative medicine and anti-vaccine activist repeatedly warned by the Food and Drug Administration (FDA) for promoting COVID-19 misinformation and selling unproven treatments.^20,21 Given the potential for ideological and funding bias in this research project, the lack of rigorous controls for possible bias in this study is concerning.

The main source of data is an online survey available only in Finnish. This is a very narrow sample population, and it is unlikely to be representative of the views or practices of dog owners generally.

More importantly, the responses are simply expressions of the perceptions and beliefs of the owners who participated, not necessarily the nutritional and environmental exposures nor health outcomes actually experienced by the dogs. Self-reporting of diet and health information has been shown to be unreliable in humans, and it is unlikely to be more reliable among dog owners.^22,23

Previous reports evaluating this source of data do not inspire confidence.^2,15 The researchers validated responses by emailing respondents and looking for consistency in their answers to selected questions. In a previous study, only about 30 percent of the respondents replied to these emails, and almost 13 percent of responses to this survey tool were discarded as “robot” answers. Also, 26 percent of respondents provided inconsistent or discordant answers regarding diet and CAD symptoms.¹⁵

The current study reports 4.2 percent of answers were discarded as duplicates or robot responses, but provides no information on discordant responses. The report does indicate, however, 31 percent of the responses were discarded because an analysis of the foods reported indicated significant underfeeding.⁷

It seems more likely owners were misremembering their feeding practices rather than 30 percent of owners were dramatically underfeeding their dogs.

Asking owners to remember in great detail what they fed their puppy between two to eight months of age, and then trying to associate this with health outcomes years later, is a questionable strategy. With no direct or objective verification, it seems very likely the dietary and health information reported in this survey is unreliable.

The associations discovered during statistical analysis support this interpretation because they are often inconsistent and make no obvious sense in terms of the authors’ hypothesis that raw foods are protective against CAD. For example, why would raw tripe and organ meats be protective, but raw red meat, eggs, and poultry would not be?

If cooking is the main risk factor, why would cooked vegetables be protective and raw vegetables would not? Why would both cooked and raw eggs be protective while neither cooked nor raw poultry is associated with the likelihood of CAD? If exposure to bacteria is the main variable, why is eating dirt, sticks, or carcasses protective, but eating clay and grass is not, and drinking from puddles is actually associated with increased risk?

If excessive processing is the issue, why was there no association with eating processed meats or canned foods, and only a marginal association with dry food when it was the only food offered?

Additionally, the paper reports associations between the proportion of various foods fed and adult CAD risk that are hard to explain.

Dogs with CAD were more likely to be eating no raw food at all than controls, and dogs without CAD were more likely to be fed 20 percent or 90 percent raw, but there were no differences at any other ratio of the two foods. Similarly, allergic dogs were more likely to be fed 80 percent dry than controls, but there was no significant difference if they were fed more than 80 percent dry.

Control dogs were more likely to be fed 50 percent or less than 10 percent dry, but there was no difference at intermediate ratios. It is easier to see cherry picking and researcher degrees of freedom than a consistent dose-response in these results.

Identifying causal associations between nutrition and health outcomes is inherently difficult. Available feeding strategies involve many differences in nutrient form and content other than the narrow aspects we may choose to focus on, such as cooked versus raw.

It is worth conducting research to investigate such relationships, but strong evidence will always be scarce and difficult to produce. However, productive research efforts must involve meaningful attempts to control for bias and to obtain fundamentally reliable data, even when conclusions may always have to be tentative.

The authors of this study continue to publish papers suggesting health benefits to raw diets and risks to conventional diets based on a single online survey tool and a convenience sample of Finnish-speaking dog owners. This is never going to be a solid data set upon which useful conclusions can be based, and the research produced from it are more likely to be an expression of the beliefs and perceptions of the respondents and the researchers, rather than a sound representation of the complex relationships between nutrition and health.

Brennen McKenzie, MA, MSc, VMD, cVMA, discovered evidence-based veterinary medicine after attending the University of Pennsylvania School of Veterinary Medicine and working as a small animal general practice veterinarian. He has served as president of the Evidence-Based Veterinary Medicine Association and reaches out to the public through his SkeptVet blog, the Science-Based Medicine blog, and more. He is certified in medical acupuncture for veterinarians. Columnists’ opinions do not necessarily reflect those of Veterinary Practice News.

References

1. McKenzie B. Debating Raw Diets. Vet Pract News. January 2019:30-31. https://www.veterinarypracticenews.com/debating-raw-diets-january-2019/.
2. Roine J, Uusitalo L, Hielm-Björkman A. Validating and reliability testing the descriptive data and three different disease diagnoses of the internet-based DOGRISK questionnaire. BMC Vet Res. 2016;12(1):30. doi:10.1186/s12917-016-0658-z
3. Brooks C, Pearce N, Douwes J. The hygiene hypothesis in allergy and asthma. Curr Opin Allergy Clin Immunol. 2013;13(1):70-77. doi:10.1097/ACI.0b013e32835ad0d2
4. Kemp A, Björkstén B. Immune deviation and the hygiene hypothesis: A review of the epidemiological evidence. Pediatr Allergy Immunol. 2003;14(2):74-80. doi:10.1034/j.1399-3038.2003.00017.x
5. Tizard IR, Jones SW. The Microbiota Regulates Immunity and Immunologic Diseases in Dogs and Cats. Vet Clin North Am Small Anim Pract. 2018;48(2):307-322. doi:10.1016/j.cvsm.2017.10.008
6. Lynch S V, Wood RA, Boushey H, et al. Effects of early-life exposure to allergens and bacteria on recurrent wheeze and atopy in urban children. J Allergy Clin Immunol. 2014;134(3):593-601.e12. doi:10.1016/j.jaci.2014.04.018
7. Hemida MBM, Salin S, Vuori KA, et al. Puppyhood diet as a factor in the development of owner?reported allergy/atopy skin signs in adult dogs in Finland. J Vet Intern Med. 2021;35(5):2374-2383. doi:10.1111/jvim.16211
8. Hielm-Björkman A, Reunanen V, Meri P, Tulamo R-M. Panax Ginseng in combination with brewers’ yeast (Gerivet) as a stimulant for geriatric dogs: a controlled-randomized blinded study. J Vet Pharmacol Ther. 2007;30(4):295-304. doi:10.1111/j.1365-2885.2007.00876.x
9. Teixeira LR, Luna SPL, Matsubara LM, et al. Owner assessment of chronic pain intensity and results of gait analysis of dogs with hip dysplasia treated with acupuncture. J Am Vet Med Assoc. 2016;249(9):1031-1039. doi:10.2460/javma.249.9.1031
10. Hielm-Björkman A, Tulamo R-M, Salonen H, Raekallio M. Evaluating complementary therapies for canine osteoarthritis–Part II: a homeopathic combination preparation (Zeel). Evid Based Complement Alternat Med. 2009;6(4):465-471. doi:10.1093/ecam/nem143
11. Rosendahl S, Anturaniemi J, Vuori KA, Moore R, Hemida M, Hielm-Björkman A. Diet and dog characteristics affect major and trace elements in hair and blood of healthy dogs. Vet Res Commun. November 2021. doi:10.1007/s11259-021-09854-8
12. Hakanen E, Lehtimäki J, Salmela E, et al. Urban environment predisposes dogs and their owners to allergic symptoms. Sci Rep. 2018;8(1):1585. doi:10.1038/s41598-018-19953-3
13. Anturaniemi J, Barrouin-Melo SM, Zaldivar-López S, Sinkko H, Hielm-Björkman A. Owners’ perception of acquiring infections through raw pet food: a comprehensive internet-based survey. Vet Rec. 2019;185(21):658. doi:10.1136/vr.105122
14. Fredriksson-Ahomaa M, Heikkilä T, Pernu N, Kovanen S, Hielm-Björkman A, Kivistö R. Raw Meat-Based Diets in Dogs and Cats. Vet Sci. 2017;4(3). doi:10.3390/vetsci4030033
15. Hemida M, Vuori KA, Salin S, Moore R, Anturaniemi J, Hielm-Björkman A. Identification of modifiable pre- and postnatal dietary and environmental exposures associated with owner-reported canine atopic dermatitis in Finland using a web-based questionnaire. PLoS One. 2020;15(5):e0225675. doi:10.1371/journal.pone.0225675
16. Hemida MBM, Salin S, Vuori KA, et al. Puppyhood diet as a factor in the development of owner?reported allergy/atopy skin signs in adult dogs in Finland. J Vet Intern Med. 2021;35(5):2374-2383. doi:10.1111/jvim.16211
17. Simmons JP, Nelson LD, Simonsohn U. False-positive psychology: undisclosed flexibility in data collection and analysis allows presenting anything as significant. Psychol Sci. 2011;22(11):1359-1366. doi:10.1177/0956797611417632
18. Resnik DB, Elliott KC. Taking financial relationships into account when assessing research. Account Res. 2013;20(3):184-205. doi:10.1080/08989621.2013.788383
19. Habib R, Becker K. The Dog Cancer Series: Rethinking the Cancer Epidemic Vol. 1- Chapter 4 (Transcript).; 2018:73.
20. Satija N, Sun L. A major funder of the anti-vaccine movement has made millions selling natural health products. Washington Post. December 20, 2019.
21. Science A for. FDA warns Mercola: Stop selling fake COVID remedies and cures.
22. Fadnes LT, Taube A, Tylleskär T. How to identify information bias due to self-reporting in epidemiological research. Internet J Epidemiol. 2009;70(2). http://www.ispub.com/journal/the_internet_journal_of_epidemiology/volume_7_number_2_25/article_printable/how-to-identify-information-bias-due-to-self-reporting-in-epidemiological-research.html. Accessed December 23, 2021.
23.  Ravelli MN, Schoeller DA. Traditional Self-Reported Dietary Instruments Are Prone to Inaccuracies and New Approaches Are Needed. Front Nutr. 2020;7:90. doi:10.3389/fnut.2020.00090