Introduction

Age is the single greatest predictor of disease risk for most causes of mortality in both human and dog populations [1–3]. However, for many dog diseases, body size is a comparably important predictor of risk [1, 4, 5]. Companion dogs show considerable variation in longevity across size classes [6–8]. Between species of mammals, larger ones tend to live longer than smaller ones, while within species, smaller individuals tend to live longer than larger individuals [9, 10]. Accordingly, dogs from larger size classes tend to have a shorter lifespan. Additionally, different size classes of dogs tend to manifest with different diseases, and ultimately to die from different causes. For instance, larger breed dogs more often die as a consequence of musculoskeletal and gastrointestinal causes whereas smaller dogs die more frequently of endocrine causes [1, 2].

Jin et al. [11] studied a multimorbidity index (i.e., a count of the number of conditions a dog has experienced) and found that although the index increased steadily with age, the size of the dog had no significant effect on the index. This suggests that over time, dogs in larger size classes are not accumulating more conditions, but rather different conditions. An understanding of which conditions manifest differently across age and size could inform our understanding of size-related longevity differences.

The Dog Aging Project (DAP) provides a unique opportunity to investigate how the lifetime prevalence of different conditions varies across age and size in a large community-based population of companion dogs. Additionally, we explore whether the adjustment for the dog’s sex, breed status (purebred vs. mixed-breed status), and the census division defined by the dog’s residence address impacts this relationship [12–17]. Specifically, we use data from the “DAP Pack”, a collection of over 25,000 survey respondents from across the US.

Materials and methods

Results

There were 27,541 dog owners that completed the HLES and whose dogs are included in this report. The dogs ranged in age from puppies to very senior dogs, with a median age of 7 years (IQR 4 to 11). The dogs were equally distributed by sex (50% male) and purebred versus mixed-breed status (49% purebred). A total of 238 breeds were represented, and the frequency of each breed is provided in S1 Table. The respondents were distributed across the US, with the following percentages according to census divisions in decreasing order of representation: Pacific 24%, South Atlantic 18%, East North Central 15%, Mountain 10%, Mid-Atlantic 9%, West South Central 8%, New England 6%, West North Central 6%, East South Central 3%. Owners most commonly described the locations where they lived with their dogs as suburban (62%), with 17% residing in urban locations and 21% in rural locations.

We present the frequency and proportion of owner-reported lifetime prevalence of categorical diagnoses by age categories in Table 1, and by weight categories in Table 2. Disease categories are listed in descending order of overall frequency.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Proportion with disease history by age category.

https://doi.org/10.1371/journal.pone.0295840.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Proportion with disease history by weight category.

https://doi.org/10.1371/journal.pone.0295840.t002

We now provide results for each disease category individually, with the categories that were positively associated with dog size discussed first, followed by those that were either negatively or not associated with dog size. From the dispersion test, we found no significant evidence of over-dispersion in the fitted models for every condition. Adjustment for sex, breed status and geographic region did not have any notable impact on the associations, and this was true for all disease categories we studied. In the Supporting Information (S1 and S2 Appendices), we presented the associations between weight and each disease prevalence across age, using subsets of all purebred and all mixed breed dogs.

Skin conditions

A total of 7915 (28.7%) dog owners reported that their dogs had a history of conditions within the skin disease category. The proportion of dogs with a reported history of skin conditions increased steadily with age, ranging from 7% in puppies to 37% in senior dogs (Table 1). History of skin disease was reported less for smaller vs. larger dogs, ranging from 26% in dogs <10 kg to 33% in dogs over 40 kg (Table 2). In the Poisson regression model with the main effects of age and weight, each SD increment of age (4 years) was associated with a 29% greater relative lifetime prevalence of a reported history of skin conditions. Lifetime prevalence also increased with the owner-reported weight of the dog, with each SD increment (13 kgs) associated with 12% higher relative lifetime prevalence of skin condition history (Table 3). In Fig 1, we illustrate these patterns using a categorical representation of dog size. We see strongly increasing trends over age that are similar across size classes. Toy dogs have the lowest rates across age, and a history of skin conditions is reported progressively more often for each successively larger size category regardless of age at the time of the survey. Older dogs were reported to have skin conditions more often by a similar relative margin across different size classes, which was in alignment with the results in Table 3 that there was no significant interaction between age and weight.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Results from Model 2 with continuous age by weight category (Part 1: Conditions positively associated with weight from Model 1).

Data from the Dog Aging Project for N = 27,541 dogs included in the 2020 data release.

https://doi.org/10.1371/journal.pone.0295840.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Association of age and weight with lifetime prevalence (Part 1: Conditions positively associated with weight from Model 1).

https://doi.org/10.1371/journal.pone.0295840.t003

Ocular conditions

A total of 3625 (13.2%) dog owners reported that their dogs had a history of conditions within the ocular disease category. A reported history of ocular conditions increased steadily with age group, from 5% in puppies to 28% among senior dogs (Table 1). In contrast, the proportion with a positive history was lower among larger dogs, ranging from 17% in the smallest dogs to just 10% in the dogs over 40 kg (Table 2). This was reflected in the models where the significant interaction between age and weight is seen as a steeper slope in smaller dogs (Table 4). Graphically we can see in Fig 2 that below adulthood the size categories have roughly the same lifetime prevalence but then the smaller dogs increase much more quickly in reported history of ocular conditions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Results from Model 2 with continuous age by weight category (Part 2: Conditions negatively associated or not associated with weight from Model 1).

Data from the Dog Aging Project for N = 27,541 dogs included in the 2020 data release.

https://doi.org/10.1371/journal.pone.0295840.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Association of age and weight with lifetime prevalence (Part 2: Conditions negatively associated or not associated with weight from Model 1).

https://doi.org/10.1371/journal.pone.0295840.t004

Discussion

As found in other retrospective studies of the domestic dog [1, 11, 23–27], we found that owner-reported diagnoses (grouped by organ system or pathophysiologic process) disproportionately affect dogs of different sizes. Some conditions were reported less commonly for larger dogs and these included ocular, cardiac, liver/pancreas, and respiratory conditions. The proportion reporting a history of urinary conditions did not vary by weight. Many conditions were reported more commonly with increasing weight category, including skin, orthopedic, gastrointestinal, ENT, cancer, neurologic, and endocrine conditions. The infectious disease category showed a distinct pattern that the smallest dogs (<10 kg) had much lower lifetime prevalence than the other categories and there were no increasing patterns across age. With the exception of infectious diseases, the proportion of dogs with a reported history of each disease category increased with increasing age group, as expected. For different disease categories, size had various effects on the relative lifetime prevalence of reporting a history of disease as well as on the slope of increasing lifetime prevalence across age.

Higher growth rates in larger dogs have been implicated in increased oxidative damage during early life that may predispose to certain diseases such as skin diseases, orthopedic conditions, cancer, and cardiac disorders [28]. Skin is a common site for evidence of oxidative damage; increased production of free radicals including reactive oxygen species can overwhelm the body’s mechanisms for reducing their damaging effects and manifest as skin lesions and other pathology [29]. In mouse models, increased oxidative stress induced by exposure to diisodecyl phthalate exacerbated symptoms of allergic dermatitis [30]. A retrospective study of 721 dogs by Dreschel [31] measuring associations between behavioral health and physical health found that dogs with severe fear and separation anxiety disorders had both significantly higher lifetime prevalence and severity of skin diseases [31]. The author suggests that these associations may be due to physiological stress responses in dogs with behavioral disorders and resulting changes in hormone regulation, immunity, and disease risk.

Faster growth rates and larger body size have also been associated with increased risk for orthopedic diseases [6]. Growth patterns in larger dog breeds involve rapid weight gain throughout the period of skeletal development and maturity, which has been linked to increased risk of developmental musculoskeletal and orthopedic diseases including hip dysplasia, osteoarthritis, and osteochondrosis [6, 32, 33]. In our cohort, osteoarthritis was spread similarly among dogs with body weight greater than ten kilograms. The morbidity associated with osteoarthritis may vary. Orthopedic pain from any of these diagnoses can significantly reduce quality of life (QOL) [34]. Poor QOL has been identified as one of the most influential factors for owners choosing to euthanize their dogs and thereby shorten their lifespan to prevent pain and distress [35]. Most medications commonly used to alleviate pain from osteoarthritis are dosed by weight and are thus more likely to be cost-prohibitive for owners with larger dogs; such factors may also lead to differential effects of osteoarthritis by size on QOL and decisions about euthanasia. Additionally, mobility problems due to orthopedic disease are more easily accommodated in dogs of smaller size, as owners are more likely to be capable of lifting or carrying smaller dogs and/or assisting their movement with slings, harnesses, and other devices. It is important to note that the analyses presented here included size only, and did not analyze subpopulations of specific breeds. Because members of a breed are similar in size, it is possible that common breeds within this cohort contributed disproportionately to the findings for a given size.

Across all body sizes, several retrospective studies of large cohorts of companion dogs in North America and Europe have consistently found that cancer [1, 24, 26, 27, 35, 36] is the most common cause of mortality in pet dogs. Multiple studies that have considered body size in relation to causes of morbidity and mortality have reported that cancer diagnoses are more common in larger dogs than smaller dogs [1, 37–39]. Several theories about the contributing factors to reduced longevity in larger dogs are consistent with increased risk of cancer in larger dogs including correlations between increased serum IGF-1 levels in larger dogs and downstream effects on growth, oxidative damage, and cancer risk [6, 37, 40, 41]. Artificial selection to produce the extreme variation in growth rates among breeds has resulted in wide variation in serum IGF-1 levels before and after skeletal maturity. For example, Great Danes gained weight at 17 times the rate of miniature poodles in a study following the first 21 weeks of life [42]. Correspondingly higher levels of mean basal IGF-1 detected in the plasma of Great Danes compared to miniature poodles from age 13 weeks to 27 weeks in a related study suggested that Great Danes may enter a period of physiologic gigantism postnatally [43]. Plasma levels of IGF-1 remain higher in larger breed dogs compared to smaller breed dogs past maturity, and this inverse correlation between IGF-1 levels and longevity is consistent with patterns found in many other species [6].

Studies investigating the relationship between average breed growth rates and prevalence of life-limiting disorders linked to oxidative damage have revealed correlations between breed size and risk of developing cardiac pathology. Increased risk of cardiac disease and associated morbidity and mortality have been reported at both extremes of the body size spectrum in dogs, though the prevalence of specific cardiac pathophysiologies varies widely from small dogs to giant breed dogs [1, 26, 27]. Artificial selection for extreme size (both small and giant) may have contributed to the distribution of cardiac disease risk in dogs, as many genes associated with body size in dogs also contribute to cardiac development and structure [44]. Telomere length in peripheral blood mononuclear cells was positively correlated with lifespan and inversely correlated with risk of mortality due to cardiac disease in a study of 15 breeds of dog, with the shortest age-adjusted telomere length in the Great Dane [45]. In our cohort, reported lifetime prevalence of cardiac disease increases more rapidly with age in smaller dogs but the associated contributions of these diseases to mortality is unknown.

In our cohort, larger dogs were also more likely to have diagnoses in the neurological category. The most commonly reported specific condition was seizures (including epilepsy) and it was distributed fairly evenly across body sizes (Table 10 in S3 Appendix). Fleming et al. [1] found that larger dogs were relatively spared from death due to neurological diseases. Our results are not inconsistent with these findings as our study does not address cause of death but rather disease history, and many dogs with seizures do not die as a direct result of their neurological disease; the majority of dogs with epilepsy (the most common cause of recurrent seizures in dogs) can be treated successfully with conventional drugs [46] and, similarly, the majority of companion dogs with epilepsy die of a cause not directly related to epilepsy [47]. Additionally, neurological diseases found more commonly in smaller dogs (e.g., intervertebral disc disease) tend to present later in life than epilepsy such that neurological disease in smaller dogs would be less frequent in our cohort than in cohorts examined retrospectively for causes of mortality. This observation aligns with the pattern that the risk of certain diseases in dogs can be associated with their size in various ways, either by increasing, decreasing, or being independent of their size.

Previously, Fleming et al. [1] reported that larger breeds are spared from death by endocrine disease. In our cohort, the reported lifetime prevalence of a diagnosis in the endocrine category was more likely in larger dogs. This contrast between previous reported mortality versus lifetime prevalence in our study may be driven almost entirely by the most commonly reported specific endocrine disorder, hypothyroidism (Table 13 in S3 Appendix). Hypothyroidism was more prevalent in larger dogs, and this diagnosis alone was responsible for most of the trend by size for the endocrine category. While untreated hypothyroidism can cause significant morbidity and even mortality, it is relatively easy and inexpensive to treat in dogs [48, 49]. Cushing’s disease and diabetes mellitus, two frequently diagnosed diseases that are more prevalent in smaller dogs in our cohort, carry greater morbidity and mortality risk than hypothyroidism if untreated, and treatment places a greater financial burden for the owner compared to typical treatments for hypothyroidism [50–53]. It is noteworthy that Fleming et al. [1] used data from dogs seen in referral institutions while the dogs in our study were recruited directly from the community. It is therefore likely that referral bias [54] could also explain the difference in these results, as treatment of hypothyroidism is usually straightforward and does not require referral [55].

For the ENT category, the most commonly reported diagnosis, ear infection, was drastically skewed toward larger dogs. Ear infections in dogs are commonly associated with allergic skin disease and food allergies and thus increased lifetime prevalence in larger dogs would be consistent with the aforementioned theories about increased oxidative damage and skin manifestations. While ear infections can sometimes be easily treated, chronic/recurrent ear infections are also common and can drastically affect QOL [56–59]. Recurrent ear infections can be expensive to monitor and treat, with surgery to remove the entire ear canal(s) as the only curative option for some dogs. Because significant financial resources may be necessary to avoid negative impacts on QOL due to ear infections, cost may be a factor guiding euthanasia decisions and could thus influence lifespan [59, 60]. The next most frequent diagnoses in the ENT category were hearing loss and deafness, which both have frequencies skewed toward smaller dogs. In general, pet dogs do not receive treatment for hearing loss or deafness, nor do they receive adaptive devices such as hearing aids and thus financial limitations of the owner do not tend to change the prognosis for the patient with chronic ear infections. Hearing loss is typically not associated with pain, and dogs are often capable of maintaining a good QOL despite partial or complete deafness.

Diagnoses within the respiratory category were more common overall among smaller dogs. The most common specific diagnoses were chronic cough and tracheal collapse, both of which are known to be more common among smaller dogs [61, 62]. Additionally, components of the brachycephalic syndrome (hypoplastic trachea, redundant soft palate and stenotic nares) were among the top ten diagnoses in the respiratory category and were more often noted among smaller dogs [63]. Currently a number of small breeds with extreme brachycephalic phenotype, such as French Bulldogs and Pugs, are popular in the US and consequently within the DAP Pack [63, 64].

Our study has several strengths and limitations that should be noted. Strengths include the large sample size of this study, which allows us to estimate patterns with high power across the whole age and size spectrum. Additionally, we have a very diverse sample of dogs distributed across the entire United States. Since the sample is not veterinary-hospital or clinic-based it may be more representative of the general population of dogs. Conversely, while our observations can suggest which conditions manifest differently across age and size, they do not prove any causal relationships due to the cross-sectional nature of the analysis. Over time, longitudinal data will be collected on these dogs, and we will be able to examine disease incidence. In addition, recall bias may occur when owners fill out the survey. It is possible that owners may not remember past events at all or incorrectly at the time of the survey. Response bias may also occur when owners who lack understanding of disease categories provide inaccurate answers. Additionally, we acknowledge a low level of potential errors in reporting the dog’s exact weight in pounds in the survey. There is also potential for measurement error because weight is not a precise measure of a dog’s size and can be affected by factors such as the normal or overweight status. Future studies will compare HLES data to Veterinary Electronic Medical Records data to measure the accuracy of owner-reported responses and confirm the trends seen in HLES. Lifetime prevalence of conditions was expected to go up regardless of an increasing age-specific prevalence because it was calculated in a cumulative way. In addition, the lifetime prevalence for conditions which rapidly lead to mortality will be underestimated in this sample. Finally, the sample is not random but self-selected so that data are subject to self-selection bias. Owners who are more exposed to this survey and who tend to participate in surveys are more likely to nominate their dogs and complete the survey, possibly leading to a biased sample. The restriction of allowing only one dog per household to enroll in the DAP could also introduce bias, favoring either healthy pets or those with notable medical histories. The DAP endeavors to create a representative sample of the companion canine population in the US. However, computer access is a necessary condition for survey completion, as well as the ability to complete the surveys in English.

In our analyses, we did not control for individual breeds for several reasons. First, a significant portion of our study population consisted of mixed breed dogs, and previous studies have demonstrated the unreliability of owner-reported ancestry for such dogs [65]. In addition, adjusting for individual breed would generate an assessment of the impact of body condition (overweight or underweight) and skeletal variation within each breed on lifetime disease prevalence, which is not the focus of our study. Furthermore, in the absence of a specific understanding of the genetic origins of a given disease, studies of breed and disease risk may lead to inaccurate impressions of the frequency and magnitude of breed-based risk. Studies have also shown that whether certain genetic disorders are more common among purebred vs. mixed breed dogs depends on the condition [14]. For these reasons, we instead conducted separate analyses using subsets of all purebred and all mixed breed dogs and compared the results in the Supporting Information (S1 and S2 Appendices). Furthermore, we compared the results of the association of age and weight with lifetime prevalence from the entire set to those from a sample excluding the 10 most common breeds, one breed at a time. In this analysis, we observed only 5 among 1,040 cases where the significance of the association shifted from significant to non-significant. Future studies will examine the genetic architecture of diseases in this specific population.

In this study we have quantified the owner-reported lifetime prevalence of a history of conditions within several different disease categories as a function of dog age and size, with and without adjustment for sex, geographic location, and purebred versus mixed-breed status. These are lifetime disease prevalence data rather than cause-of-death data. However, these results nonetheless provide insights into the disease categories that may contribute to reduced lifespan in larger dogs and suggest multiple avenues for further exploration. More focused efforts to look at individual conditions within categories may yield additional insights. Within and across categories, the co-occurrence of different disease subtypes may also be an important factor to evaluate. Of course, as prospective data become available the longitudinal associations of these conditions with subsequent morbidity and mortality will be evaluated.

Supporting information

Acknowledgments

The authors thank Dog Aging Project participants, their dogs, and community veterinarians for their important contributions.

Dog Aging Project Consortium

Joshua M. Akey⁸, Brooke Benton⁶, Elhanan Borenstein^{9, 10, 11}, Marta G. Castelhano^{12, 13}, Amanda E. Coleman¹⁴, Kyle Crowder^{15, 16}, Matthew D. Dunbar¹⁶, Virginia R. Fajt¹⁷, Annette L. Fitzpatrick^{18, 19, 20}, Unity Jeffery²¹, Erica C Jonlin^{6, 22}, Matt Kaeberlein⁶, Kathleen F. Kerr¹, Jonathan M. Levine⁵, Jing Ma²³, Audrey Ruple²⁴, Stephen M. Schwartz^{19, 25}, Sandi Shrager¹, Noah Snyder-Mackler^{26, 27, 28}, M. Katherine Tolbert⁵, Silvan R. Urfer⁶, Benjamin S. Wilfond^{29, 30}

⁸Lewis-Sigler Institute for Integrative Genomics, Princeton University, Princeton, NJ, USA. ⁹Department of Clinical Microbiology and Immunology, Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel. ¹⁰Blavatnik School of Computer Science, Tel Aviv University, Tel Aviv, Israel. ¹¹Santa Fe Institute, Santa Fe, NM, USA. ¹²Cornell Veterinary Biobank, College of Veterinary Medicine, Cornell University, Ithaca, NY, USA. ¹³Department of Clinical Sciences, College of Veterinary Medicine, Cornell University, Ithaca, NY, USA. ¹⁴Department of Small Animal Medicine and Surgery, College of Veterinary Medicine, University of Georgia, Athens, GA, USA. ¹⁵Department of Sociology, University of Washington, Seattle, WA, USA. ¹⁶Center for Studies in Demography and Ecology, University of Washington, Seattle, WA, USA. ¹⁷Department of Veterinary Physiology and Pharmacology, Texas A&M University College of Veterinary Medicine & Biomedical Sciences, College Station, TX, USA. ¹⁸Department of Family Medicine, University of Washington, Seattle, WA, USA. ¹⁹Department of Epidemiology, University of Washington, Seattle, WA, USA. ²⁰Department of Global Health, University of Washington, Seattle, WA, USA. ²¹Department of Veterinary Pathobiology, Texas A&M University College of Veterinary Medicine & Biomedical Sciences, College Station, TX, USA. ²²Institute for Stem Cell and Regenerative Medicine, University of Washington, Seattle, WA, USA. ²³Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA, USA. ²⁴Department of Population Health Sciences, Virginia-Maryland College of Veterinary Medicine, Virginia Tech, Blacksburg, VA, USA. ²⁵Epidemiology Program, Fred Hutchinson Cancer Research Center, Seattle, WA, USA. ²⁶School of Life Sciences, Arizona State University, Tempe, AZ, USA. ²⁷Center for Evolution and Medicine, Arizona State University, Tempe, AZ, USA. ²⁸School for Human Evolution and Social Change, Arizona State University, Tempe, AZ, USA. ²⁹Treuman Katz Center for Pediatric Bioethics, Seattle Children’s Research Institute, Seattle, WA, USA. ³⁰Department of Pediatrics, Division of Bioethics and Palliative Care, University of Washington School of Medicine, Seattle, WA, USA.