Impacts of a vegan diet on body weight and related health disorders in 1,305 cats: an exploratory cross-sectional survey

Overweight and obesity are common in domestic cats, and are associated with a range of adverse health outcomes. To investigate the impact of a vegan diet on body weight and related health disorders, we conducted a cross-sectional online survey and analyzed carer-reported veterinary assessments concerning 1,305 cats. We compared the data regarding cats fed conventional meat-based (90.3% - 1,178) and vegan (9.7% - 127) diets for at least one year. The body weight related disorders assessed included overweight, underweight, osteoarthritis/arthritis, intervertebral disc disease, other mobility disorders and other musculoskeletal problems. Cats fed vegan diets showed lower odds of being classified as overweight compared to those fed a meat-based diet. In unadjusted analyses, odds of being overweight were 24% lower. These were 34% lower after adjustment for age, sex, neuter status, activity and feeding practices. However, these estimates were not statistically significant and confidence intervals were wide due to relatively low observed numbers of vegan-fed cats. Lower odds of being underweight were also observed but were not statistically significant. No vegan-fed cats suffered from any of the other body weight-related disorders assessed, which may have been due to the relatively low numbers of vegan-fed cats included in the survey. This low prevalence precluded statistical comparison for these secondary outcomes. Overall, cats fed vegan diets had lower rates of body weight-related disorders within this sample. This result persisted when incorporating cats fed a raw meat diet into the analysis. However, given the small number of cases combined with the moderately sized vegan-fed group, these findings should be interpreted cautiously and considered exploratory.

Domestic cats (Felis catus) are commonly maintained on commercially produced diets. Overweight and obesity are among the most frequently diagnosed disorders in cats, and are associated with reduced quality of life and multiple comorbidities, including osteoarthritis, insulin resistance and diabetes mellitus, urinary tract diseases and skin diseases (Saavedra et al., 2024; Teng et al., 2018). Prevalence estimates vary widely with population and assessment method, but studies using veterinary assessment commonly report that 30-60% of cats are overweight or obese (Blanchard et al., 2025; Courcier et al., 2010; APOP, 2022; Montoya et al., 2025; Silva et al., 2026; Wall et al., 2019).

Obesity is a multi-factorial disease influenced by age, neuter status, sex, activity level and feeding practices (Courcier et al., 2010; Wall et al., 2019). Diet and feeding management remain central because they shape energy intake, energy density, and feeding behavior. Prospective and cross-sectional studies have identified that feeding patterns such as predominantly dry feeding and restricted lifestyle factors, such as being indoors, are associated with higher overweight risk in some cohorts (Courcier et al., 2012; Öhlund et al., 2018). Given that early detection supports prevention and intervention, clinical guidance emphasizes routine monitoring of body weight and the use of standardized body condition scoring (BCS), during veterinary examinations (Cline et al., 2021; Quimby et al., 2021). Wang et al. (2026) have synthesized recent knowledge on cat obesity in a comprehensive review that further underscores these aforementioned points. The authors also draw parallels between the prevention/treatment of overweight and obesity in humans, and that in cats. However, they highlight at least two significant differences: 1) bariatric surgery is typically deemed too risky or inappropriate for overweight cats, and 2) to date, there are no official weight loss medications available for cats. The authors therefore further emphasize the importance of nutrition-based solutions.

Modern commercial cat foods are typically formulated to be ‘complete and balanced’ for a given life stage, meaning they are designed to meet recognized nutrient profiles, or are supported by feeding trials, often through the inclusion of synthetically supplemented essential nutrients (e.g., taurine) to prevent deficiencies, especially due to the degradation of nutritional quality during processing (Cline et al., 2021; Quimby et al., 2021; Wu, 2024). Most commercially available cat food is meat-based. However, in parallel with broader societal shifts towards plant-based consumption, driven by ethical, environmental, and health considerations, vegan diets formulated for cats have become increasingly available with continued growth forecasted (Mordor Intelligence, 2025). Akin to conventional meat-based diets, these diets are typically formulated to meet established nutritional guidelines, partially through supplementation with synthetic or fermentation-derived essential dietary nutrients. They differ from conventional meat-based diets in ingredient composition and nutrient sourcing, in that they contain no animal-based ingredients.

Domestic cats are obligate carnivores. In an evolutionary context, obligate carnivory describes the physiological, metabolic, and anatomical adaptations associated with a prey-based ecological niche (Dodd et al., 2021; Laflamme, 2020; Verbrugghe and Hesta, 2017). However, as mentioned, for domesticated and pet cats in modern times, the essential nutrients acquired from prey can instead be synthetic or fermentation-derived. Thus, the nutritional adequacy of cat food depends on correct formulation—including considerations of digestibility and bioavailability—rather than on ingredient origin (Laflamme, 2020; Davies et al., 2019; Verbrugghe and Hesta, 2017; Fascetti, 2015; Glasgow et al., 2002).

A key facet of bioavailability is digestibility. While studies concerning cats are few relative to dogs, a recent review has highlighted the comparable digestibility of vegan diets for cats, relative to meat-based diets (Knight, 2026). For instance, this review features studies that demonstrate acceptable or even superior nutritional biomarkers and protein digestibility in cats fed plant-based protein (e.g., Golder et al., 2020; Morris et al., 2023). Another systematic review has also highlighted inadequate formulations of sampled commercially available ‘complete’ cat food across a whole range of cat food types—such as dry, moist, meat-based (conventional), meat-based (raw), and vegan. The study found no evidence to suggest these inadequacies are more likely in vegan cat food (Mace et al., 2026). A survey of pet food manufacturers also demonstrated that vegan or vegetarian pet food manufacturers may take more precautions when formulating pet food to ensure sufficient availability of nutrients (Knight and Light, 2021). Thus, the presence of animal-derived tissues does not inherently guarantee nutritional adequacy. As further evidence, one diet trial documented taurine-deficiency cardiomyopathy in cats fed a whole-rabbit raw diet. The diet was analyzed as containing adequate taurine, underscoring the importance of bioavailability, digestibility, and formulation, rather than ingredient origin (Davies et al., 2019; Fascetti, 2015; Glasgow et al., 2002).

Studies to date have reported broadly similar, and in some cases more favorable, health indicators in cats fed plant-based diets, compared with those fed meat-based diets (Dodd et al., 2021; Knight et al., 2023; Domínguez-Oliva et al., 2023; Wakefield et al., 2006). However, with limited exceptions, these studies have mostly been survey-based with a reliance on carer-reported outcomes, rather than clinical studies based on more objective assessments. Additionally, some concerns have been raised in other studies. For instance, Leon et al. (1992) found hypokalemic episodic polymyopathy (a muscle condition) in vegetarian-fed cats, while Fantinati et al. (2021) found dysorexia (abnormal appetite), lethargy, weight loss, and muscle waste in two plant-based cats. However, both of these studies likewise had significant limitations. The former utilized diets that were designed to be hypokalemic (deficient in potassium), while the latter comprised a very small sample size and was based on plant-based diets that were not nutritionally sound. The aforementioned broader-scope examination of nutritional soundness of cat food in 2026 did not find vegan cat food to be more likely to be nutritionally inadequate than meat-based cat food, when considered overall (Mace et al., 2026).

The aforementioned survey-based studies (Dodd et al., 2021; Knight et al., 2023; Wakefield et al., 2006) have also largely assessed overall health indicators rather than specific body weight or obesity-related outcomes. Given the high prevalence of overweight and obesity in domestic cats, and the central role of diet in regulating energy intake, examining whether diet type is associated with differences in overweight and obesity risk and related disorders is warranted. While existing research has described the differences between meat-based and vegan diets in terms of macronutrient composition, energy density, palatability, and general health outcomes, the extent to which such differences in formulation translate into meaningful variation in body weight related disorders remains unclear (Dodd et al., 2021; Knight et al., 2023).

This study therefore investigated the association between diet type (conventional meat-based versus vegan) and body weight-related disorders in domestic cats, using indicators derived from carer-reported veterinary health assessments. The key research question was whether cats fed vegan diets differed in their likelihood of experiencing body weight-related disorders, compared with those fed conventional meat-based diets, building on earlier research that assessed general health outcomes associated with these diets (Knight et al., 2023). The null hypothesis was that there would be no difference between the dietary groups in the prevalence of the body weight-related disorders assessed. Related studies have reported results from other sections of the same cat-focused survey, relating to general health outcomes (Knight et al., 2023), palatability (Knight and Satchell, 2021) and consumer acceptance of vegan and other sustainable pet diets (Mace et al., 2025).

Study design and survey

Data were collected using an online survey of companion animal carers, previously described in detail (Knight and Satchell, 2021; Knight et al., 2023, 2024; Mace et al., 2025). Briefly, the survey was administered using the Jisc Online Surveys platform (https://onlinesurveys.jisc.ac.uk) and collected information from cat and dog carers; however, the present study reports an analysis restricted to cats only, using the same survey instrument and inclusion criteria reported within our broader cat health analysis (Knight et al., 2023). The survey was open from May to December, 2020. It was distributed via social media, especially dog and cat interest groups. Paid Facebook advertisements were also utilized to help broaden the survey’s reach and number of respondents. These paid advertisements were not targeted at particular demographic groups, apart from a requirement for dogs or cats to be mentioned within their social media activity or profiles. Some purposive sampling targeted unconventional diet special interest groups to try and ensure sufficient representation of carers who feed unconventional diets to their cats for statistical analysis. In all cases, every effort was made to keep phraseology neutral when promoting the survey to minimize biasing the type of owner attracted and the type of results generated.

Carers were asked to provide demographic information about themselves in addition to demographic information and the health status of one resident cat that had lived in the household for at least one year. Where cats were fed a prescription or therapeutic diet at the time of survey completion, carers were asked to base responses on the diet in use prior to the therapeutic diet. This enabled the survey to still collect data on cats who would transition onto a therapeutic diet. Carers reported the dietary main ingredient source, selecting one option from predefined categories (conventional meat-based, raw meat, lab-grown meat, insect-, fungal- or algae-based, vegetarian, vegan, or other). Vegetarian diets were defined as excluding meat but permitting eggs or milk, whereas vegan diets excluded all animal-derived products. Reported dietary classification was not independently verified. Information was also collected on treats and dietary supplements. While we did collect detailed information on meal type (including whether food was homemade), this was simplified to three options (dry, wet, or equal mix) for the purpose of statistical analysis, as detailed in section 2.3. Brand of the main food fed was not collected. We also did not assess detailed nutritional composition and compliance with nutritional guidelines [e.g., from the European Pet Food Industry Federation (FEDIAF) or Association of American Feed Control Officials (AAFCO)] (Cline et al., 2021; Quimby et al., 2021). Carers were asked to report on their main veterinarian’s assessment of their cat’s medical condition over the last year. Carers were not asked to specify the methods used by their veterinarians to arrive at this assessment.

Methodological choice

This study employed survey-based and carer-reported data—including carer-reported veterinary data—to gain insights into cat health. Such methods are in widespread use for gaining insights into companion animal health (e.g., see studies by O'Halloran et al., 2024; Wallis et al., 2024; Stallwood et al., 2019; Wiles et al., 2017). Veterinary epidemiology methodology recognizes numerous advantages of survey-based and carer-reported health indicators (Schmid et al., 2026). Carers spend the most time with their companion animals and know them best. Indeed, veterinarians likewise rely on carer reports of cat health and behavior on a daily basis. Surveys are also efficient and accessible, and they can lead to significantly larger sample sizes that are more diverse (e.g., international) in nature than some alternative methods. Moreover, they generate useful information expediently. Surveys of companion animal carers serve as one component of a broader cumulative evidence base. Schmid et al. (2026) first summarized these advantages. Then, albeit focusing primarily on dogs, they conducted a comprehensive analysis of both owner-reported and veterinary data. They concluded that owner-reported data are reliable. Thus, the advantages of surveys and carer-reported veterinary data should also be considered alongside any disadvantages of the methodological choice—in addition to mechanisms employed to minimize any such disadvantages. For instance, in this case, to minimize recall bias, carers were only asked about the last 12 months in part to minimize the recall time period. Method weaknesses are further discussed in section 4.5.

Moreover, carers feeding their cats vegan diets currently remain a very niche, small, and challenging-to-access population. Despite best efforts, it was conceivable that numbers may both be underrepresented and too low in some cases for the most sophisticated statistical analyses. However, in accordance with Hackshaw (2008) among others, unavoidable methodological constraints, should not mean study of such groups is avoided, provided methods and limitations are transparent. Indeed, such study can foster more interest and future participation, leading to more robust future research. Hackshaw (2008) remarks further how there needs to be a balancing of long-term studies and short-term studies that are quicker and more accessible; the latter could help shape the former.

Statistical analysis

The primary research compared outcomes associated with vegan and conventional meat-based diets when considering six body weight-related disorders (dependent variables) derived from carer-reported veterinary health assessments. Respondents could select one or more primary health disorder categories, including body weight problems, mobility problems, and other musculoskeletal disease. Where a body weight problem was reported, respondents were asked a follow-up question to specify whether the animal was ‘overweight’ or ‘underweight’. From this, two binary outcome variables were derived:

1. Overweight

2. Underweight

Where a mobility problem was reported, respondents were asked a follow-up question to specify whether this involved osteoarthritis/arthritis, intervertebral disc disease, or another mobility problem. In addition, respondents could independently report other musculoskeletal disease as a primary category. From this structure, four additional binary outcome variables were derived:

3. Osteoarthritis/arthritis (impairing mobility)

4. Intervertebral disc disease (impairing mobility)

5. Other mobility problems

6. Other musculoskeletal problems

All six outcomes were coded as binary variables (‘no’ or ‘yes’), reflecting whether the condition was reported. To account for the low case numbers with respect to most of these outcomes, the analysis was conducted in three steps and all findings should be treated as exploratory.

In the descriptive step 1, raw observed disease frequencies of body weight-related illness indicators in each diet group were descriptively analyzed and visualized. Steps 2 and 3 consisted of estimating the diet effect on each outcome parameter utilizing logistic regression models, while controlling for one or more of the following ten control variables to account for potential third variable effects: animal age (continuous), animal sex and neuter status (four categories, reference category ‘Female, spayed’), animal primary location (binary, reference ‘Mostly indoor’), feeding frequency (binary, reference ‘Food is always available’), meal type (three categories, reference ‘Dry food’), meals weighed or not (binary, reference ‘No’), competition with other animals during meals (binary, reference ‘No’), feeding of treats (binary, reference ‘No’), animal daily active time (binary, reference ‘0–30 minutes’), frequency of body weight assessment (binary, reference ‘Daily to weekly’). Many of the control variables were deliberately binarized to account for the limited numbers of cases on which the regression models were estimated. The reference categories chosen were normally the most common categories.

In the exploratory step 2, the diet effect on each outcome parameter was estimated once in a logistic regression model without accounting for any control variable (the ‘raw observed diet effect’) and then in ten additional regression models, each controlling for exactly one of the control variables. These ‘one control variable diet effects’ were estimated since most outcome parameters did not have sufficiently large case numbers to simultaneously account for all third variables. The step 2 models were not estimated for outcome parameters with zero cases in one of the diet groups.

In exploratory step 3, one further ‘all control variables regression model’ was estimated for each outcome parameter with at least 50 cases across both diet groups. This applied to only (1) overweight, among the six body weight-related disorders listed previously. While these estimated diet effects should still be interpreted exploratively due to the observational nature of the study design, the simultaneous adjustment for all potential third variables allowed for the derivation of more robust diet effect estimates.

Regression estimates were reported using odds ratios, their 95% confidence intervals, and p-values. Consistent with recent recommendations in statistical practice and our related prior study of general health outcomes (Knight et al., 2023), our interpretations followed the American Statistical Association’s statement on the use of p-values (Nuzzo, 2014; Wasserstein and Lazar, 2016). Accordingly, we used p-values and confidence intervals with a significance level of 0.05 to evaluate the (un)certainty of the effects, but not as an evaluation of the ‘relevance’ of effect sizes. Instead, the main focus was on effect strength, measured based on the coefficient value rather than the p-value. In line with this philosophy, the results were interpreted using the same set of heuristic categories as in our previous study (Knight et al., 2023), as: ‘association’ (when a significant effect ∼10% or stronger), ‘strong tendency’ (when a non-significant effect ∼25% or stronger), ‘tendency’ (when a non-significant effect ∼10 - 25%) or ‘marginal’ (when a non-significant effect <10%). As in our prior study (Knight et al., 2023) we defined these qualitative descriptors and corresponding thresholds as heuristic categories to aid interpretation of effect sizes within this exploratory analysis.

As a measure for the goodness of fit of the step 3 model, we calculated the area under the curve (AUC) value on a randomly drawn 20% hold-out test set after re-estimating the model on the remaining 80% training set. The calculated AUC value for the overweight model was 0.78, reflecting that the model was able to reproduce the structures observed in the data acceptably (Hosmer et al., 2013).

Additionally, we performed two sensitivity analyses: (1) To check the robustness of the Step 2 regression model accounting for the age difference of the two diet groups, we also estimated the age-controlled diet difference based on 2:1 propensity score matching (based on nearest neighbor matching), using function “matchit” from R package “MatchIt” (Ho et al., 2011). (2) We checked if the inclusion of cats fed raw meat into the meat-based group would alter the results. All steps of the analysis were repeated based on redefining the “meat-based” group as “meat-based (conventional or raw)”. This group comprised 1,242 cats (1,178 fed conventionally + 64 fed raw meat).

All analyses were conducted using the statistical open-source software R (version 4.5.0, R Core Team, 2025) (R: A Language and Environment for Statistical Computing, 2025). Logistic regression models were estimated with function “glm” from R base package “stats”. AUC values were calculated with function “calc_auc” from package “plotROC” (Sachs, 2014).

The present analysis focuses on body weight-related disorders in a subset of 1,418 cats and their carers within a larger survey of 4,057 dog and cat carers. We previously reported general health outcomes for this cat subset in a broader analysis (Knight et al., 2023).

Cat carer demographics

1,418 cat carers met the inclusion criteria (18+ years old and reporting on a single cat resident in their household for at least one year). Of these, 1,397 reported the main diet their cat was maintained on. Within this subset, 1,305 cats were maintained on the two primary diet groups of interest (conventional meat-based or vegan) and were included in subsequent analyses. The remaining animals were excluded due to being maintained on alternative or unclear diet types (e.g., raw meat, insect-based, vegetarian, laboratory-grown meat, mixed, unspecified or unclear diets).

These 1,305 cat carers were predominantly female (90% - 1,170/1,305), with most residing in the UK (71% - 923/1,305) or Europe (20% - 262/1,305) and representing a wide range of age groups (18–19 years: 1%; 20–29 years: 14%; 30–39 years: 19%; 40–49 years: 17%; 50–59 years: 21%; 60-69: 20%; 70+ years: 9%). Most respondents were the primary decision-makers for their cat’s diet (94% - 1,229/1,305), and health and nutrition was the most important factor influencing current diet choice (84% - 1,086/1,288), as not all respondents answered this question. Respondents were generally well educated, with half holding at least an undergraduate degree (659/1,305). Full demographic details were reported previously (Knight et al., 2023).

Cat demographics

The mean age of the studied cats was 8.0 years. Cats fed vegan diets were younger on average than those fed conventional meat-based diets (6.2 compared to 8.2 years, respectively). Most cats were neutered, with an approximately equal sex distribution (52% female, 48% male). Full cat demographic details were reported previously (Knight et al., 2023).

The primary location of one cat was unknown. Most of the remaining cats were primarily located indoors (57% - 747/1,304), with the remainder spending time in mixed environments (39% - 509/1,304) and only 48 primarily outdoors (4%). Reported daily activity levels data were available for 1,179/1,305 cats and varied across the sample, with most cats active for 1–3 hours per day (38% - 447), followed by 30–60 minutes per day (27% - 313). Smaller proportions were active for less than 30 minutes (15% - 182), or for more than 3 hours (20% - 237).

As noted, 1,305 cats were maintained on conventional meat-based or vegan diets and included in the subsequent analyses of these two main diet groups. This comprised 90% (1,178) cats fed conventional meat-based diets and 10% (127) fed vegan diets. Feeding practices varied, with cats receiving a range of meal types and feeding frequencies. Wet food (39% - 515/1,305) and mixed diets (33% - 429/1,305) were most common overall, although cats fed vegan diets were predominantly fed dry food (79% - 100/127), in contrast to those fed meat-based diets, which were more commonly fed wet food (43% - 502/1,178) or mixed diets (35% - 415/1,178). Feeding frequency also varied, with 37% (486/1,305) of cats having food always available and 35% (454/1,305) fed twice daily. ‘Ad libitum’ (i.e., at will – with food always available) feeding was more common among cats fed vegan diets, at 53% (67/127), compared to those fed meat-based diets (36% - 419/1,178). Food portions were infrequently weighed (16% - 214/1,305). Additionally, treat provision was common, with 75% (980/1,305) of cats receiving treats. This was more frequent among cats fed meat-based diets (77% - 904/1,178) than vegan diets (60% - 76/127). Competition with other animals for food at mealtimes was reported for 16% (199/1,280, as not all respondents answered this question of cats).

Obesity-related indicators of illness

Raw observed distributions

Observed frequencies of body weight and related disorders among cats fed conventional meat-based or vegan diets are supplied in Tables 1, 2 and visualized in Figure 1. Supplementary Figure 1 further illustrates that these frequencies remain almost identical when incorporating cats fed a raw meat diet into the analysis. Carers were less likely to consider their cats to have ideal body weights than their veterinarians (Table 1). However, both carers and veterinarians reportedly considered the majority of cats in both diet groups to have ideal weights, although this was slightly more common when cats were fed vegan diets. Cats whose weights were not ideal were more likely to be overweight than underweight. Within the relatively small sample of vegan-fed cats (n=127), none were reported as suffering from the additional body weight-related disorders assessed (Table 2).

Exploratory regression analysis on ‘one control variable’ models

These observed differences were further explored using regression analyses as described in the Methodology. The exploratory regression analyses estimating the vegan diet effect for overweight and underweight are visualized in Figure 2 and summarized in Table 3. The raw observed diet effect indicated lower odds of both overweight and underweight among cats fed vegan diets compared to those fed meat-based diets. Because the other four body weight-related disorders had zero cases in the vegan diet group (Table 2), the ‘one control variable’ diet effects could not be estimated for those outcomes.

Without controlling for third variables, the ‘raw observed diet effect’ regression models showed relevant differences between dietary groups (Figure 2; Table 3). Cats fed a vegan diet showed 24% lower odds of being overweight compared to cats fed a meat-based diet. This effect was not statistically significant (p = 0.6124). Because the effect was stronger than 10%, but not statistically significant, it can be considered a tendency. Cats fed a vegan diet also showed 66% lower odds of being underweight. This effect was not statistically significant either (p = 0.2889). Because the effect was stronger than 25%, but not statistically significant, it can be considered a strong tendency.

As a sensitivity analysis, to account for the age difference between the two diet groups (average age 6.2 years of the vegan-fed cats compared to 8.2 years of the meat-fed cats), we re-estimated the age-controlled diet differences regarding the risk of being overweight and underweight based on propensity score matching. The reduction in odds was roughly similar to the regression estimates with 21% lower odds of being overweight and 50% lower odds of being underweight.

While the adjusted dietary group differences in the additional ‘one control variable’ regression models showed quite varied effect strengths (see Figure 2), all odds ratio estimates showed lower overweight and underweight risks for cats fed a vegan diet. Supplementary Figure 2 further illustrates that these results remained almost identical when incorporating cats fed a raw meat diet into the analysis.

Exploratory regression model controlling for all potential third variables

An additional regression model simultaneously controlling for all potential control variables was estimated for outcome parameters with a total of at least 50 cases. Among the body weight-related disorders considered, this criterion was met only for overweight cats. The resulting diet effect estimate is visualized in Figure 3. After adjustment for all relevant third variables in the regression models, differences became more pronounced. Cats fed a vegan diet showed 34% lower odds of being overweight compared with those fed meat-based diets. This effect was not statistically significant (p=0.5308). Because the effect was stronger than 25%, but not statistically significant, it can be considered a strong tendency. Supplementary Figure 3 further illustrates that this result differs by just 1% (i.e., 35% lower odds) when incorporating cats fed a raw meat diet into the analysis.

Key findings

In this observational study, we compared body weight-related disorders in cats fed vegan and conventional meat-based diets, using carer-reported veterinary assessments. Cats fed vegan diets showed lower odds of being classified as overweight, compared to those fed conventional meat-based diets, although uncertainty was substantial and effects were not statistically significant. Results remained nearly identical when incorporating cats fed raw meat diets into the analysis.

These results are consistent with the direction reported in other cross-sectional, carer-reported studies of plant-based feeding in cats. For example, Dodd et al. (2021) reported higher proportions of ideal body condition (71%) and lower proportions of overweight cats (20%) when fed vegan diets (termed ‘plant-based’ in their study) compared to cats fed meat-based diets (54% ideal; 34% overweight). In the present study, a greater proportion of cats fed vegan diets were reportedly classified by veterinarians as ideal and fewer as overweight compared to those fed meat-based diets (96% vs 94% ideal, 3% vs 4% overweight) (Table 1), although differences were smaller in magnitude relative to those found by Dodd et al. (2021). It should also be noted that Dodd et al. used carer-reported BCSs (supported by reference illustrations), whereas the present study used carer-reported veterinary assessment. The differences in methodology may partially explain these differences in magnitude.

Overweight prevalence

Estimates of feline overweight vary substantially depending on method and population sampled. Veterinary-assessed BCS studies in general practice and large record-based cohorts (Courcier et al., 2010; Montoya et al., 2025; Öhlund et al., 2018) commonly report substantially higher prevalence than the 4% observed for reported veterinary assessments of overweight in this study’s sample. For example, 39% of cats were classified as overweight or obese in a first opinion practice study in Glasgow in 2008 (Courcier et al., 2010), while a nationally distributed study of cats attending charity clinics in Great Britain reported a prevalence of 11.5% in 2012 (Courcier et al., 2012).

High prevalence estimates are also reported in studies using veterinary records or structured BCSs. For example, a USA study reported 47% overweight and 14% obese (Montoya et al., 2025), and a Swedish study found 45% overweight in a medical-record cohort compared with 22% when assessed by carers using a questionnaire (Öhlund et al., 2018). Industry-reported estimates are even higher, with a 2024 survey of veterinary surgeons suggesting that 43% of cats seen in practice were overweight or obese, which is just over ten-fold higher than the prevalence observed in the present sample (UK Pet Food, 2025).

These differences likely reflect variation in measurement approaches. Veterinary or structured BCS assessments typically identify higher prevalence than carer-reported or non-standardized assessments, which may underestimate overweight status. This is relevant for interpretation of the present findings, as the low prevalence reduces statistical power and makes diet-weight relationships harder to detect reliably, particularly for secondary outcomes that were only assessed in subsets of cats. Nevertheless, it is noteworthy that, in this study, carers actually reported higher rates of overweight in their cats than their veterinarians apparently did. This could be related to more health-conscious/anxious carers being potentially more likely to respond to cat health surveys and perhaps more likely to fear worse health outcomes—especially regarding the most well known and prevalent conditions, such as overweight. Nevertheless, the potential underestimation of overweight in this study does constitute a key potential limitation of this study in terms of the generalizability of its results, as discussed further in section 4.5.

Interpreting the association between vegan diets and overweight odds

Across the regression models, the vegan diet group consistently showed lower odds of overweight compared to the meat-based group, including in the model adjusted for multiple potential confounders. While these estimates were not statistically significant, the consistency in direction across models suggests a potential association. However, given the observational design and limited number of overweight cases, these findings should be interpreted cautiously. The wide confidence intervals indicate substantial uncertainty, and the results should therefore be considered exploratory and hypothesis-generating, rather than confirmatory.

While no compositional data were collected in this study, several hypothetical mechanisms warrant consideration for future research in terms of potential explanations of why vegan-fed diets may be less prone to overweight. For instance, lower caloric, lower fat, higher fiber, and higher L-carnitine content in cat food have been identified as potentially favorable attributes for either treating overweight or reducing its risk (Wang et al., 2026). It is possible that some vegan formulations may differ in these attributes compared to the same weight of conventional cat food, although this was not assessed. This would, in the first three cases, be due to non-animal sources of food often comprising more fiber than animal sources, and potentially lower calories or fat. The fourth case of L-carnitine requires some further explanation. Neither FEDIAF (2025) nor AAFCO (2026) require this vitamin-like amino acid derivative in any compulsory quantity in cat food; hence, conventional cat food normally does not supplement this nor guarantee a certain quantity of L-carnitine. This is despite L-carnitine being critical for energy metabolism and for retaining lean body tissue (Wang et al., 2026). Pet food institutions might assume ample quantities of L-carnitine in conventional cat food due to the compound’s presence in meat and its relative robustness when heated (Goula et al., 2018), with cats also being able to endogenously synthesize the compound from other nutrients (Panickar et al., 2021). In contrast, however, there is evidence of some nutritionally sound vegan cat foods being supplemented with L-carnitine (e.g., see Benevo, 2026). Hence, it is possible that cats fed nutritionally sound vegan diets may actually be less prone to L-carnitine deficiencies, and so potentially have greater protections against overweight risk through retaining more lean body tissue. Collectively therefore, these potential attributes of nutritionally sound vegan cat food—lower caloric/fat, higher fiber, and higher L-carnitine content—could explain a potentially lower susceptibility toward overweight in cats fed vegan diets.

However, diet-associated differences in body weight are likely multifactorial and may also reflect feeding practices and management, rather than diet composition alone. Several non-dietary factors are consistently associated with overweight risk, including age, neuter status, sex, activity level, and feeding practices (Courcier et al., 2012). In particular, observational studies have reported higher overweight risk in cats fed predominantly dry food or fed ad libitum (Öhlund et al., 2018).

Notably, cats fed vegan diets in this sample were more commonly fed dry food and ad libitum, factors previously associated with increased overweight risk. While these factors were controlled for in the Step 3 model, the very low event count (four overweight cases among the vegan-fed cats) prevents reliable interpretation of whether diet type independently modifies these associations. Nevertheless, it is also possible that associations of increased risk of overweight from feeding dry food and ad libitum may become less pronounced if cats are fed vegan diets. Although cats fed vegan diets were younger on average than those fed meat-based diets, this difference was accounted for in the adjusted analyses. Indeed, Figure 2 illustrates that when controlling for age exclusively, the odds of overweight were still reduced. However, given the observational design, this finding should be interpreted cautiously and cannot be assumed to reflect a direct causal effect of diet alone. It also highlights the need for studies that incorporate objective measures of energy intake and body condition, rather than relying on broad diet categories (Öhlund et al., 2018).

The low prevalence of body weight-related disorders limited statistical power, particularly for analyses of secondary outcomes. Although rates of conditions such as osteoarthritis and other mobility-related disorders were descriptively lower in cats fed vegan diets, these differences were not statistically significant. These conditions are known to be exacerbated by excess body weight, and may therefore be indirectly associated with body condition (Laflamme, 2006; Teng et al., 2018). For example, increased body fat was identified as a risk factor for a higher load of whole body osteoarthritis in a Swedish University study of 72 cats (Ley et al., 2025). As such, the lower observed rates may be consistent with the reduced prevalence of overweight reported in this group, but given the small number of cases, these findings should be interpreted cautiously and cannot be taken as evidence of a protective effect.

Discrepancy between carer-reported weight status and reported veterinary assessments

Also noteworthy was the marked discrepancy in assessments of body weight reported by vets, and carers’ own opinions. Around 94% (meat-based) and 96% (vegan) of veterinarians reportedly considered cats’ weights to be ideal, depending on the diet fed, compared to carers’ own opinions of 70% (meat-based) and 76% (vegan) (Table 1). As previously mentioned, the reverse pattern is more typically true: carers are generally less likely than veterinarians to recognize overweight in their cats. However, these measures likely capture distinct constructs: carer concerns relating to non-ideal weight probably reflected perceived adiposity, whereas reported veterinary concerns probably indicated clinically recognized, and explicitly communicated health problems associated with overweight (most commonly) or underweight.

This interpretation is consistent with evidence that weight and body condition score (BCS) are often under-recorded, and obesity may potentially be under-communicated or inconsistently discussed, in UK veterinary practices (Phillips et al., 2017; Taylor et al., 2022). For example, in a UK study of 129,076 cat visits across 361 practices, BCS, weight loss and overweight were recorded in only 22.5%, 10% and 7% of visits, respectively (Taylor et al., 2022). Clinical recommendations include routine monitoring and recording of body weight and BCS at veterinary examinations, implying that weight should be monitored early rather than only once it is framed as a ‘problem’ (Cline et al., 2021; Quimby et al., 2021). It is clear that overweight and obesity are not consistently discussed in routine veterinary consultations. Observational studies show that weight is addressed in only a minority of feline consultations. For example, in a Canada-based thematic analysis in 2006, 12 veterinarians discussed obesity in only 25 of 123 cat consultations (Phillips et al., 2017). Hence, veterinary concern in our study may only have been captured if it was clearly communicated and remembered by carers, which may also partly explain why a substantially lower prevalence of non-ideal weight was identified through veterinary-reported outcomes. True levels of veterinary concern were probably higher. In the present study, carers were also not asked about the methods used by veterinarians to arrive at their assessments reported. BCS assessment is a common method used (Cline et al., 2021). However, carers may have been reporting less formal opinions by veterinarians in some cases, which may have further led to an underreporting of overweight. These factors may further partially or fully explain the apparent discrepancy with carers’ own opinions, alongside the potential for cat health surveys to attract the more health-anxious or committed carer.

Limitations and further research

Several limitations should be considered. As this study was observational and relied on carer-reported information, diet-associated differences in body weight may be confounded by unmeasured carer-related factors such as lifestyle management, that are difficult to capture accurately in questionnaires. Further factors were omitted, such as whether the diet was homemade and the brand of food used, with there being a precedent of a correlation between homemade diets and reduced odds of feline overweight (e.g., Wall et al., 2019). However, we did attempt to control for 10 key factors, such as feeding frequency, competition with other pets and exercise levels. While these latter results can be viewed as more robust exploratory estimates, the respective fully adjusted regression model did not meet the traditional 10-events-per-parameter rule of thumb for logistic regression (Peduzzi et al., 1996), and the results are still influenced by relevant statistical uncertainty due to the resulting limited statistical power. Outcome measurement is also a key limitation. All outcomes were caregiver-reported, including veterinary assessments, introducing potential recall and reporting bias. Carer-reported body weight assessment is not equivalent to a standardized veterinary-assessed BCS and is prone to misclassification, with evidence that carers tend to underestimate overweight and show disagreement with veterinary assessments as body condition increases (Blanchard et al., 2023). The current study showed greater levels of overweight within carers’ own assessments, than those reported of veterinarians, providing some degree of reassurance on this point. However, as discussed above, true levels of veterinary concern were probably higher than reported. This is for reasons including undeclared methods of carer-reported veterinary assessment (potentially less formal methods than BCSs), attraction of more health-conscious survey respondents, poor recall from carers, and recognized underreporting of overweight by veterinarians. A potential genuine underestimation of overweight cannot be ruled out, and so may affect the reliability and generalizability of the observed potential impacts of diets on weight.

This analysis of body weight-related disorders was restricted to neurological and musculoskeletal conditions, and excluded other commonly reported obesity-associated comorbidities such as endocrine, urinary, cardiac, and dermatological disorders (Saavedra et al., 2024; Teng et al., 2018). As a result, our results do not fully capture the broader health impacts of being overweight, across all diet groups.

There are further potential biases. Social desirability bias may be in play whereby vegan carers may consciously or subconsciously be more motivated than carers of meat-based cats to report favorable health outcomes—although the same could be true of carers passionate about feeding meat-based diets, and there is no evidence to suggest carers feeding vegan diets may be more prone to this. Carers of vegan-fed cats may also be more committed to cat health, which may lead to other cat health protective behaviors not reported here (e.g., more veterinary visits, more health checks at home) that could potentially confound the results. Finally, the possibility of survivorship bias cannot be excluded. There may be vegan-fed cats who develop health problems and are switched back to meat-based diets (therapeutic or not). If so, this could positively skew the reported health outcomes of the remaining vegan-fed cats. However, it is possible the same phenomenon could also affect some meat-fed cats who are subsequently switched to diets perceived as healthier, including vegan diets.

Further research should aim to further address these limitations and to resolve current uncertainty ideally through prospective study designs incorporating direct extraction of veterinary-recorded body weights and standardized BCSs. This should ideally be completed alongside direct measurement of diet composition, feeding amounts, treat energy and overall energy intake. This would help clarify diet-specific effects. Expanding outcome definitions to include a broader range of obesity-associated comorbidities (for example, diabetes mellitus, lower urinary tract disease) would also provide a more comprehensive assessment of health impacts and improve compatibility with the wider feline obesity literature. Larger animal numbers, particularly within the vegan diet group, would enable more comprehensive statistical testing, and potentially, the detection of additional trends or effects.

Nevertheless, it is of value to share non-significant preliminary results such as these (Nuzzo, 2014; Song et al., 2010; Scargle, 2000), especially in niche and novel fields of inquiry. This study can serve a key role in the shaping of future studies and in hypothesis generation.

The findings from this study were non-significant and were accompanied by wide confidence intervals; thus, the study should be used for hypothesis generation and to inform future research, rather than to confirm or refute a dietary effect. Among the 1,305 cats we studied, those fed vegan diets had lower rates of carer-reported body weight-related disorders when compared to cats fed conventional meat-based diets. Vegan-fed cats had 34% lower odds of being assessed by veterinarians as overweight, after controlling for age, sex, neuter status, activity levels, and feeding practices. They were also less likely to be assessed as underweight. The other body weight-related disorders assessed were osteoarthritis/arthritis, intervertebral disc disease, other mobility disorders, and other musculoskeletal problems. No vegan-fed cats suffered from any of these conditions, though this may relate to the small number of vegan-fed cats relative to those fed meat-based diets. Findings remained almost identical when incorporating cats fed a raw meat diet into the analyses. Further research using larger samples, and more objective measures of body condition and energy intake, could further clarify the relationship between diet type and body weight-related disorders in domestic cats.