Grain-Free Dog Foods Cleared in Major Two-Year Study of Canine Heart Disease

A sweeping longitudinal analysis of nearly 50,000 companion dogs has found no evidence that grain-free diets increase the risk of new cardiac diagnoses, according to research conducted within the Dog Aging Project (DAP). The study, which combined owner-reported dietary data with multiple statistical modelling approaches and a rigorous validation of what owners actually feed their pets, concluded that grain-free diet status was not predictive of newly reported heart conditions over a two-year observation window. The findings offer preliminary reassurance in a debate that has unsettled veterinarians and dog owners since 2018, when the US Food and Drug Administration first announced an investigation into reports correlating grain-free diets with dilated cardiomyopathy, a serious condition in which the heart muscle weakens and enlarges.

The controversy began in earnest during the 2010s, when grain-free formulations exploded across the American pet food market. Sales at one boutique retail chain tripled between 2011 and 2015, and by 2019 grain-free products accounted for 43 percent of all extruded diets sold there. By 2017, diets built on peas, chickpeas, lentils, sweet potatoes and tapioca made up more than half of the dog food market. When the FDA announced its investigation the following year, it triggered a wave of research across academia and industry, spanning randomized controlled trials and observational studies in both client-owned and research colony dogs. Yet although that body of work focused narrowly on dilated cardiomyopathy, many individual veterinarians began broadly advising owners to avoid grain-free diets for any cardiac concern, including murmurs and mitral valve disease, and even for breeds predisposed to heart disease, purely out of caution.

A central weakness in this literature has long been its reliance on owner-reported diet histories. Previous studies repeatedly identified owner recall of primary diet type, treats, supplements and other food items as a substantial methodological limitation, with reports of grain-free status frequently overestimated, misattributed or incomplete. The new analysis, led by researchers affiliated with the Dog Aging Project, set out to quantify exactly how inaccurate owner reporting of grain-free status actually is, and then to determine whether that inaccuracy could distort the apparent relationship between diet and heart disease in a large longitudinal cohort. The DAP follows more than 50,000 companion dogs of all breeds and ages, collecting annual data on health, lifestyle and environmental factors that influence canine longevity.

To validate owner reports, the team took advantage of a natural experiment within the survey structure. From January 2020 to June 2023, the DAP’s original diet survey asked owners a simple yes-or-no question about whether their dog’s primary diet was grain-free. In July 2023, a far more comprehensive questionnaire was introduced, asking for the full brand name and formula of the diet, along with how long it had been fed. Owners who reported feeding the same diet for at least twelve months could therefore be assessed twice: once by their own grain-free label, and once by independent verification. A coder individually checked 13,049 such responses against online ingredient lists, classifying any diet containing rice, corn, wheat, barley or oats as grain-inclusive.

The validation revealed striking misclassification. Thirty-five percent of owners reported feeding grain-free diets, yet when a grain-free claim was made, it was correct only 49 percent of the time, essentially a coin flip. By contrast, owners who reported grain-inclusive diets were far more reliable, with a negative predictive value of 92 percent, and overall accuracy across all reports was 77 percent. The researchers suspect naming conventions drive much of the error: diets with titles such as a meat-and-potato recipe, which carry no grain in the name but contain grains in the ingredient list, were frequently mislabeled as grain-free by owners who judged by packaging rather than ingredients.

Crucially, the team did not stop at documenting the error. They ran simulation studies with synthetic populations of 10,000 dogs, introducing reporting inaccuracy rates of 25, 50 and 75 percent across a range of hypothetical true effect sizes, from protective to severely harmful. The simulations demonstrated that this particular pattern of misclassification, in which grain-free diets are inconsistently reported while grain-inclusive diets are reliably reported, would attenuate true associations, pulling observed odds ratios toward 1, but would not create false associations or reverse their direction. In other words, if grain-free diets truly caused heart disease, the analysis should still have detected a harmful trend, merely a weakened one. Based on the degree of attenuation observed, the researchers estimated that their measured odds ratio of 1.05 corresponds to a true odds ratio of roughly 1.07, a negligible effect.

The longitudinal analysis itself drew on the first three years of DAP diet data, covering 47,444 dogs enrolled before the end of 2023, including 7,786 dogs whose owners completed an end-of-life survey capturing diagnoses reported after the last annual follow-up. Confounders were identified using a directed acyclic graph, a formal causal inference tool, rather than ad hoc selection. The unadjusted odds ratio for grain-free diet on new cardiac diagnoses was 0.98, with a p value of 0.56. After adjusting for breed status, sex, neuter status, body size and survey year, the adjusted odds ratio was 1.05, with a confidence interval spanning 0.91 to 1.21 and a p value of 0.52. Mixed-effects models and a diet-switching analysis produced consistent results. Across every modelling approach, grain-free diet status showed no effect on the risk of incident cardiac disease.

What did matter were other factors. Single-breed dogs were more likely than mixed-breed dogs to receive new cardiac diagnoses, with an odds ratio of 1.23, and neutered animals were diagnosed substantially more often than intact ones, with an odds ratio of 1.68, the strongest association in the model. Larger dogs showed smaller odds ratios, and diagnoses became more common in later survey years. The researchers argue that neuter status, which may reflect biological protection from sex hormones or simply longer lifespans and greater time at risk, deserves far more attention in future diet and cardiac research, noting that previous studies may have been affected by failing to account for it.

The most common diagnoses in the cohort were murmurs, accounting for 57 percent of reported cardiac conditions, followed by congestive heart failure, valve disease and arrhythmia, each around 8 to 9 percent. Because murmurs and valve disease dominate the cardiac burden in companion dogs, the findings are particularly relevant to practitioners who have been advising owners to transition dogs with valve disease away from grain-free diets as a precaution. The authors caution, however, that the study was not designed to detect effects on individual cardiac diseases, nor does it examine whether diet influences survival after a diagnosis, questions that ongoing work aims to address.

Important limitations remain. The original DAP diet survey did not capture individual ingredients, so the analysis cannot speak to specific components such as peas, lentils or potatoes, which prior research has flagged as potential concerns when fed in high proportions, possibly through impaired amino acid digestibility and taurine status. Moreover, the data collection period of 2020 to 2023 coincides with years of intense public scrutiny, during which many manufacturers may have reformulated grain-free products to remove suspect ingredients. If industry reformulation has already addressed the problem, the absence of an association in this cohort may reflect safer formulations rather than proof that all grain-free diets were always harmless. The team also notes possible recall bias regarding diet duration and the possibility that diets were reformulated between the time they were fed and the time their ingredients were verified in 2025.

Even with those caveats, the study’s strengths are considerable: prospective data collection, three years of dietary exposure, inclusion of dogs that died during the study window, confounder selection guided by formal causal tools, and consistency across multiple analytical methods. The conclusion is preliminarily reassuring for the millions of dog owners who have worried, or been warned, that grain-free kibble might be silently damaging their pet’s heart. At the population level, grain-free status appears to have no effect on new cardiac diagnoses, and the imperfections in owner reporting, while real, were of a kind that could only have masked a true effect, not invented a clean bill of health. As the researchers emphasize, considerably more research is needed to determine whether specific ingredients affect specific cardiac diseases, but the blanket avoidance of all grain-free diets now lacks support from the largest longitudinal dataset yet brought to bear on the question.

Subject of Research: The relationship between grain-free diets and new cardiac diagnoses in companion dogs, and the accuracy of owner-reported diet data, in the Dog Aging Project cohort.

Article Title: Multiple Modelling Methods Suggest That Grain‐Free Diets Are Not Predictive of New Owner‐Reported Cardiac Diagnoses Over Two Years in the Dog Aging Project, Despite Owner Misclassification of Diet

Article References: O'Brien, J., Bishop, M., Dog Aging Project Consortium, Tolbert, M. K., & Ruple, A. (2026). Multiple Modelling Methods Suggest That Grain‐Free Diets Are Not Predictive of New Owner‐Reported Cardiac Diagnoses Over Two Years in the Dog Aging Project, Despite Owner Misclassification of Diet. Veterinary Medicine and Science, 12(5), Article e71213. https://doi.org/10.1002/vms3.71213

Image Credits: AI Generated

DOI: 10.1002/vms3.71213

Keywords: grain-free diets, canine cardiac disease, Dog Aging Project, dilated cardiomyopathy, diet misclassification, owner-reported data, veterinary nutrition, longitudinal study, logistic regression, causal inference, dog health, FDA investigation