The concept of frailty was introduced in 1968 and has become a major aspect of the study of biological aging. Frailty is a biological syndrome characterized by a decrease in the physiological reserves1 and an increase in vulnerability to stressors, resulting from dysregulation of multiple physiologic systems.2 Frail people have a substantially increased risk of falls, disability, requirement of long-term care, and all-cause death.3 Delaying the onset of frailty (ie, slowing the decrease in physiological reserves) may increase healthspan (the time an individual is able to maintain good health with a short period of morbidity before death).4 To date, there is no consensus regarding a single tool to evaluate frailty, and > 20 methods are in use.5 Currently, the index of frailty and the phenotype of frailty are the 2 most widely used tools to assess frailty in humans. However, those measures should not be considered as alternatives but rather complementary, depending on target populations and purposes of implementation.6 In contrast to the quantitative index method, the qualitative phenotypic method may exclude the presence of diseases or at least disabilities (ie, an individual without diseases or at least disabilities can be identified as frail by use of this method) and can be used to prevent consequences of development of a phenotype of frailty, such as hospitalizations, disability, institutionalization, or death.6,7
Many experimental systems involving animals, such as worms, flies, mice, and nonhuman primates, are used in aging research.8,9 Recently, the domestic dog (Canis familiaris) has been suggested as a promising experimental animal for the study of aging in humans for several reasons.10 First, there are numerous similarities between human and dog genomes, and certain genetically related diseases are common between the 2 species.11 Second, among land mammals other than humans, domesticated dogs have the most phenotypic diversity and known naturally occurring diseases,12 including age-related diseases (eg, cancer and heart and renal diseases13) and cognitive disorders (including Alzheimer-like pathological processes).14 Third, the median lifespan of domestic dogs has been estimated to be 10 to 12 years (depending on the breed),15 which is approximately a seventh of that of humans. Fourth, in 2002, approximately 47% of US households owned a domestic dog (often considered as a family member), which may foster improved adherence to research protocols. Fifth, given that domestic dogs and humans share common environments and often the same lifestyles, environmental factors associated with aging in dogs can be studied and those findings may be applicable to humans.
The study of factors related to healthspan by use of animal models is challenging because of the difficulty in defining the notion of good health among those animals.16 Some authors have suggested that frailty should be assessed among nonhuman animals to better understand the pathophysiology of aging.17 In studies involving mice, results have been promising, indicating that the concept of frailty seems transposable from humans to other animals.18 In this context, the purpose of the study reported here was to define an FRP in aged dogs that approximated the phenotype of frailty as it has been clinically defined in the elderly human population and to investigate its association with time to all-cause death, independent of age and other factors associated with survival among dogs.
Materials and Methods
Dogs
Guide dogs from the Parisian School Of Guide Dogs For Blind Or Visually Impaired People were used in the study. The dogs were born between January 1, 1995, and December 31, 2002. All guide dogs of this school are followed by the staff members from birth to death, with a known date of death for deceased dogs; however, postmortem examinations are not performed. Between 8 to 10 years of age, each active guide dog (ie, a dog that is still used as a guide for blind or visually impaired individuals) has to undergo a 1-hour-long CGA (scheduled once in its life), including completion of a geriatric health evaluation scoresheeta by a veterinarian and analysis of a blood sample for quantification of routine biochemical markers in dogs' sera (total cholesterol, total protein, glucose, creatinine, and urea concentrations and alkaline phosphatase and alanine aminotransferase activities). Dogs were included in the study if there were no missing data for items of the geriatric health evaluation scoresheet that were to be used as possible approximates of components of frailty and if a blood sample was available for testing.
Data collection
Demographic data that were collected included breed, sex, date of birth, date of death, type of housing (house vs apartment), presence of other animals in the housing (yes vs no), dog taken on public transportation (yes vs no), type of visual impairment of the owner (blind vs visually impaired), and professional occupation of the owner (with vs without professional occupation).
The geriatric health evaluation scoresheet assessed the status of 13 body systems, including the cutaneous, digestive, respiratory, cardiovascular, urinary, genital, ocular, musculoskeletal, and neurologic systems; oral cavity; general condition; behavior; and oncological status. The scoresheet had 38 clinical items for assessment. Each item was scored by the veterinarian on a scale of 1 (normal state) to a maximum of 3, 4, or 5 (detrimental state). On the day of the CGA, 1 sample of 2 mL of blood was collected from each dog and placed in a dry tube containing a serum gel separator. Each sample was centrifuged at 3,000 × g for 10 minutes at room temperature (approx 21°C). The collected serum was then immediately analyzed to determine urea, creatinine, total cholesterol, total protein, and glucose concentrations and alkaline phosphatase and alanine aminotransferase activities.
Definition of a FRP in dogs
A frailty index-like score could have been calculated by summing the scores for each system deficit noted on the geriatric health evaluation scoresheet.19 However, according to recommendations by Cesari et al,6 we used a phenotypic method for several reasons. First, the target population was comprised of nondisabled dogs that were still active at the time of recruitment. Second, aged dogs were selected as a potential animal model for aging in humans, and a frailty phenotype seemed to “depict an … age-related condition of special interest for system biology.”6 In human gerontology, this phenotype of frailty is based on 5 components20: chronic undernutrition (assessed by unintentional weight loss), exhaustion (self-reported), low physical activity level (measured through a weighted score of number of kilocalories expended per week), poor mobility (assessed as time to walk a distance of 15 feet), and weakness (grip strength).
Although the geriatric health evaluation scoresheet for dogs was not initially designed to evaluate frailty, we sought to define an FRP that would most closely approximate the core clinical signs of frailty with the available data. This approach has already been applied in humans.21 For the study dogs, there was no direct measurement of weakness or mobility; these components were therefore approximated by use of existing items in the geriatric health evaluation scoresheet. To select the geriatric health evaluation scoresheet items related to each of the 5 components (chronic undernutrition, exhaustion, low physical activity level, poor mobility, and weakness [1 item/component]) with which to define the FRP, a 2-step procedure was used. First, items that could be used as possible approximates of each component of frailty (content validity) were qualitatively preselected. When there were several potential items to represent a component, the prevalence of the items with increasing age (construct validity) was quantitatively determined: the most appropriate item for which prevalence increased with age was selected.
Although veterinarians had assigned scores to the items in the geriatric health evaluation scoresheet, those scores were converted to categories of present and absent for purposes of the study. For the components of weakness, exhaustion, and low physical activity level, there was only 1 applicable item. Weakness was assessed on the basis of muscle mass (normal [absent] vs moderate muscle wasting, muscle atrophy, or cachexia [present]). Exhaustion was assessed on the basis of effort tolerance (good tolerance [absent] vs fatigability or marked breathlessness [present]). Low physical activity level was assessed on the basis of perceived activity level (normal [absent] vs moderate or low [present]). For the component of chronic undernutrition, preselection identified 3 items: body condition (normal [absent] vs obese or too thin [present]), appetite (unchanged [absent] vs increased or decreased [present]), and quality and density of hair (normal [absent] vs sparse hair or marked alopecia [present]).22 For the component of poor mobility, preselection identified 2 items: gait (normal [absent] vs stiffness, lameness, or ataxia [present]) and joint pain (no signs of pain [absent] vs moderate or marked signs of pain [present]).
For muscle mass, effort tolerance, activity level, and the multiple preselected items for chronic undernutrition and poor mobility, we determined whether the prevalence of each item increased with age. Dogs' ages were categorized into 3 classes on the basis of the quartiles (< first quartile, first to third quartile, and > third quartile), and prevalences for the items were calculated in each of the 3 age categories. The χ2 (or Fisher exact) and Mann-Whitney rank sum tests for categorical and continuous variables, respectively, were applied. Because the prevalence of nonnormal findings increased with age, quality and density of hair was selected to represent the component of chronic undernutrition and gait was selected to represent the component of poor mobility. The prevalence of the selected item for weakness also increased with age. Finally, a dog was considered to have an FRP when > 2 of the 5 components were present; a dog was considered to not have an FRP when none or only 1 of the 5 components was present.
Statistical analysis
All statistical analyses were performed by a software program.b A χ2 (or Fisher exact, when necessary) test and Mann-Whitney U test were used to compare the distribution of categorical and quantitative variables, respectively, between dogs that had ≥ 2 components of the FRP and dogs that had none or 1 component of the FRP. The median time from CGA to death was estimated with the Kaplan-Meier method. The associations between the presence of components of the FRP and time to death were investigated with Kaplan-Meier curves, log-rank tests, and univariate and multivariate Cox proportional hazards models. For each dog, the study entry date was the date of the CGA and the outcome was death (all causes); the cutoff date for survival analyses was July 16, 2013. For dogs that were still active at the cutoff date, the censor date was the cutoff date. For dogs that were still alive but retired at the cutoff date, the censor date was 15 days before the cutoff date (ie, June 30, 2013) because staff members of the Parisian School Of Guide Dogs For Blind Or Visually Impaired People collect data about retired dogs once every 2 weeks. Of note, no dog enrolled in the study was lost to follow-up.
Because the FRP was hypothesized to be an indicator of biological age, it was necessary to systematically adjust for chronological age to assess the association between the FRP components and time to death independent of the age of the dogs. Furthermore, in multivariate analyses, the associations between the FRP components and time to death were adjusted for 1 additional potential confounder (besides age) in separated Cox models (ie, 1 age-adjusted model/potential confounder). These potential confounders were sex and those variables associated with time to death with a value of P < 0.20 in univariate analyses. Disorders or impairments in isolated systems that were components of frailty were not included as candidate covariates. When included into models, quantitative variables (eg, age or biological variables) were recoded by use of restricted cubic spline functions with 3 knots located at the 5th, 50th, and 95th percentiles of the distribution23 to minimize residual confounding bias. The proportional hazard assumptions were checked graphically (with log-log transformations of survival functions) and by including a linear time-dependent interaction term; for time-dependent interaction terms, all values of P were > 0.69, indicating adequacy of the proportional hazard assumption. For final analyses, a value of P ≤ 0.05 was considered significant.
Results
Among the 159 guide dogs that were born between January 1, 1995, and December 31, 2002, and that underwent the CGA, 116 met the inclusion criteria and were therefore included in the study (Table 1). All dogs were neutered or spayed, and half of them were female. Breeds represented included Golden Retriever, Labrador Retriever, and Golden Retriever and Labrador Retriever crossbreeds; there were 19 (16%) other breeds. Median age at the time of the CGA was 9.0 years (range, 5 to 13 years). Considering the distribution of the breeds of the 116 dogs, application of a previously published mathematical formula24 revealed that the median age of the dogs at the time of the CGA corresponded to an age of approximately 62 years for humans.
Baseline characteristics of 116 neutered guide dogs (born between January 1, 1995, and December 31, 2002) at the time of a CGA performed once from November 2003 through November 2012 summarized as a group and on the basis of the number of components of an FRP (≥ 2 vs none or 1) determined at the time of the CGA.
| Characteristic | All dogs (n = 116) | Dogs with 0 or 1 FRP component (n = 106) | Dogs with ≥ 2 FRP components (n = 10) | P value* |
|---|---|---|---|---|
| Sex (No. of dogs [%]) | 0.05 | |||
| Female | 58 (50) | 50 (47) | 8 (80) | |
| Male | 58 (50) | 56 (53) | 2 (20) | |
| Breed (No. of dogs [%]) | 0.42 | |||
| Golden Retriever | 48 (41) | 41 (39) | 7 (70) | |
| Labrador Retriever | 27 (23) | 26 (25) | 1 (10) | |
| Golden Retriever and | 22 (19) | 21 (20) | 1 (10) | |
| Labrador Retriever crossbreed | ||||
| Other | 19 (16) | 18 (17) | 1 (10) | |
| Age† (y) | 9.0 (8.5–9.3) | 8.9 (8.5–9.3) | 9.2 (8.7–9.9) | 0.12 |
| Alanine aminotransferase† (UI/L) | 39 (34–52) | 38 (30–52) | 42 (36–55) | 0.39 |
| Alkaline phosphatase† (UI/L) | 86 (64–155) | 87 (64–148) | 79 (50–194) | 0.68 |
| Urea† (g/L) | 0.28 (0.24–0.33) | 0.28 (0.24–0.33) | 0.30 (0.20–0.34) | 0.85 |
| Clinical finding (No. of dogs [%]) | ||||
| ≥ 3 cutaneous nodules | 17 (15) | 12 (11) | 5 (50) | < 0.01 |
| Oral disorder | 70 (60) | 62 (68) | 8 (80) | 0.31 |
| Cardiovascular disorder | 7 (6) | 5 (5) | 2 (20) | 0.11 |
| Malignant neoplasia | 7 (6) | 5 (5) | 2 (20) | 0.11 |
| Visual impairment | 79 (68) | 70 (66) | 9 (90) | 0.17 |
| Neurologic dysfunction | 6 (5) | 4 (4) | 2 (20) | 0.08 |
| Behavioral impairment | 13 (11) | 12 (11) | 1 (11) | 1.00 |
During the CGA, veterinarians completed a geriatric health evaluation scoresheet for each dog. An FRP for dogs that would most closely approximate the core clinical signs of frailty in humans was determined from the available data. To select the geriatric health evaluation scoresheet items related to each of the 5 components (chronic undernutrition, exhaustion, low physical activity level, poor mobility, and weakness [1 item/component]) with which to define the FRP, a 2-step procedure was used. First, items that could be used as possible approximates of each component of frailty (content validity) were qualitatively preselected. When there were several potential items to represent a component, the prevalence of the items with increasing age (construct validity) was quantitatively determined from which the most appropriate item was selected. The components of chronic undernutrition, exhaustion, low physical activity level, poor mobility, and weakness were represented by quality and density of hair, effort tolerance, activity level, gait, and muscle mass, respectively. Clinical findings identified at the time of the CGA included oral disorders (presence of tartar on teeth, gingivitis, or periodontitis), cardiovascular disorders (bradycardia, tachycardia, arrhythmia, murmur, or congestive heart failure), visual impairment (ulceration of the cornea, iris disorder, or cataract), neurologic dysfunction (decreased awareness, hearing impairment, decreased proprioception, incontinence, or balance impairments), and behavioral impairments (disorientation, decreased socio-environmental interactions, loss of house training and commands, sleep disorders, vocalizations, destruction, or aggressiveness).
For each variable, a value of P for the comparison between dogs with none or 1 versus ≥ 2 FRP components at the time of the CGA was determined.
Data are reported as median value (interquartile range [25th to 75th percentile]).
Notable clinical conditions identified through use of the geriatric health evaluation scoresheet included cutaneous nodules, oral disorders (presence of tartar on teeth, gingivitis, or periodontitis), cardiovascular disorders (bradycardia, tachycardia, arrhythmia, murmur, or congestive heart failure), malignant neoplasia, visual impairment (ulceration of the cornea, iris disorder, or cataract), neurologic dysfunction (decreased proprioception, hearing impairment, decreased awareness, incontinence, or balance impairments), or behavioral impairment (loss of house training and command responses, sleep disorders, vocalization, destructive behavior, disorientation, decreased socio-environmental interactions, or aggressiveness).
The median follow-up time from CGA until either death or censoring was 3.3 years (interquartile range [25th to 75th percentile], 2.6 to 4.7 years), providing a total follow-up time of 414 dog-years. For 66% (76/116) of dogs that died, the median time from CGA to death was 4.4 years (interquartile range, 2.9 to 5.4 years).
Information from the geriatric health evaluation scoresheet completed for each dog was also reviewed to determine how many of the 5 FRP components (chronic undernutrition, weakness, exhaustion, poor mobility, and low physical activity level) were present. Of the 116 dogs, 10 (8.6%) had ≥ 2 components of the FRP; 5 (4.3%) of those dogs had ≥ 3 components. These dogs were predominantly female and more likely to have ≥ 3 cutaneous nodules, compared with dogs that had none or 1 component of the FRP (Table 1). These dogs were typically older (P = 0.12) and had more neurologic disorders (P = 0.08), malignant neoplasia (P = 0.11), and cardiovascular disorders (P = 0.11), compared with dogs that had none or 1 component of the FRP.
In the univariate analyses, the poor mobility (HR, 2.4; 95% CI, 1.4 to 4.1; P < 0.01) and low physical activity (HR, 2.3; 95% CI, 1.0 to 5.1; P = 0.05) components, individually, were significantly associated with time to death, whereas the chronic undernutrition (HR, 0.8; 95°% CI, 0.2 to 3.2; P = 0.73), weakness (HR, 1.5; 95°% CI, 0.5 to 4.7; P = 0.52), and exhaustion (HR, 1.2; 95°% CI, 0.5 to 2.6; P = 0.70) components, individually, were not significantly associated with time to death. The presence of ≥ 2 components of the FRP was significantly associated with time to death (Figure 1; log rank P < 0.01). The corresponding unadjusted HR was 6.0 (95°% CI, 2.5 to 14.2). After adjustment for age, presence of ≥ 2 components of the FRP remained significantly associated with time to death, with an aHR of 3.9 (95% CI, 1.4 to 10.9; P < 0.01). Further adjustments for biomarkers, such as serum activities of alanine aminotransferase and alkaline phosphatase and concentration of urea, as well as for number of cutaneous nodules (≥ 3 vs none, 1, or 2), oral disorders (presence of tartar on teeth, gingivitis, or periodontitis), visual impairment (ulceration of the cornea, iris disorder, or cataract), neurologic dysfunction (decreased awareness, hearing impairment, decreased proprioception, incontinence, or balance impairments), breed (Labrador vs other breeds), and sex in separated age-adjusted Cox models resulted in significant aHRs quantifying the association between having 2 or more components of the FRP and time to death ranging from 2.9 (95% CI, 1.0 to 8.8; P = 0.05) to 4.5 (95% CI, 1.6 to 12.5; P < 0.01). Of note, adjustments for the presence of malignant neoplasia, behavioral impairment (disorientation, decreased socio-environmental interactions, loss of house training and command responses, sleep disorder, vocalization, destruction, or aggressiveness), or cardiovascular disorders, besides age, resulted in aHRs ≥ 3.9 (P < 0.01).
Kaplan-Meier estimates of time to death after a CGA among 116 aged neutered guide dogs (born between January 1, 1995, and December 31, 2002) according to the number of components of an FRP (≥ 2 [dashed line] vs none or 1 [solid line]) determined at the time of the CGA. During the CGA (performed once from November 2003 through November 2012), veterinarians completed a geriatric health evaluation scoresheet for each dog. Those data were used to define an FRP for dogs that would most closely approximate the core clinical signs of frailty (chronic undernutrition, exhaustion, low physical activity level, poor mobility, and weakness) in humans. For each of the 5 components comprising the FRP, scoresheet items that could be used as possible approximates of each component of frailty (content validity) were first qualitatively preselected. When there were several potential items to represent a component, the prevalence of the items with increasing age (construct validity) was quantitatively determined on the basis of which the most appropriate item was selected (1 item/component). The FRP components of chronic undernutrition, exhaustion, low physical activity level, poor mobility, and weakness were represented by quality and density of hair, effort tolerance, activity level, gait, and muscle mass, respectively. For each dog, the study entry date was the date of the CGA and the outcome was death (all causes); the cutoff date for survival analyses was July 16, 2013. For dogs that were still active at the cutoff date, the censor date was the cutoff date. For dogs that were still alive but retired at the cutoff date, the censor date was 15 days before the cutoff date (ie, June 30, 2013).
Citation: American Journal of Veterinary Research 77, 12; 10.2460/ajvr.77.12.1357
Discussion
To our knowledge, the study reported here is the first to attempt to evaluate frailty in dogs. To date, most of the studies related to frailty in animals other than humans involved mice,18,25 including 1 study that used the phenotypic approach by including 4 (exhaustion, low physical activity level, poor mobility, and weakness) of the 5 components included in the human FRP.18 The aim of the present study was to define an FRP in dogs on the basis of data obtained from a geriatric health evaluation scoresheet completed by veterinarians during a CGA. Study findings indicated that independent of age, health status, or subclinical and clinical diseases, dogs that had ≥ 2 components of the FRP at the time of the CGA were 3 times as likely to die during the follow-up period as were dogs that had 1 or none of the FRP components.
Of the 116 recruited dogs, 4.3% had ≥ 3 of the 5 components of the FRP. This prevalence rate was similar to that of humans in a frailty study20; most of the recruited dogs were 8 to 10 years old, which corresponds to an age of approximately 62 years for humans.24 Furthermore, as previously reported for humans,26 dogs that had ≥ 2 components of the FRP were predominantly female. Dogs with ≥ 2 components of the FRP were also typically older and more frequently had neurologic disorders, malignant neoplasia, and cardiovascular disorders. In humans, cardiovascular disorders are associated with frailty.27 The study of this report may have lacked statistical power to investigate the association between cardiovascular disorders and the presence of an FRP in dogs because of the genetic origin of many forms of cardiovascular diseases such as dilated cardiomyopathy, tri-cuspid valve dysplasia, or ventricular septal defect28 and because the 116 selected study dogs were of similar breeds, which decreases the interindividual variability of clinically apparent cardiovascular disorders. Altogether, the results of the present study indicated that the components evaluated in humans to identify frail individuals may also be pertinent to the identification of frail dogs. More generally, the study findings suggested that the frailty syndrome of humans is a geriatric syndrome that can also be observed in dogs.
Of the 5 components of the FRP in dogs, the poor mobility and low physical activity level components were individually and significantly associated with time to death. However, the concept of frailty is thought to be intimately related to the systemic loss of complexity associated with aging29; therefore, the whole picture reveals much more than the sum of its parts.30 As a consequence, prevention of the onset of frailty should be implemented through systemic interventions instead of problem-specific interventions. Recent studies31,32 of such systemic interventions in mice have had promising results, opening this field of research to dogs.
Dysregulations in the neuroendocrine, immune, and nervous systems are hypothesized to be involved in human frailty development.33 The findings of the present study suggested that frailty in dogs also involves dysregulations in multiple physiologic systems, thereby sharing a common pathway with the human frailty syndrome. First, in humans and in various other animals such as worms, flies, and mice, there is some evidence that reduced IGF-1 and insulin signaling may favorably impact longevity.34 No direct association between reduced plasma IGF-1 concentrations and longevity in dogs has yet been identified. However, among dog breeds, lifespan is inversely associated with body size,24 and an IGF-1 allele was shown to be a major locus governing this trait, especially in small and giant breeds.35 Altogether, these results suggest that IGF-1 might contribute to aging in dogs.36 Second, in humans, hypovitaminosis D is associated with a decrease in muscle function and strength37 as well as with the presence or occurrence of frailty.38 In dogs, recent studies39,40 have revealed that hypovitaminosis D appears to be associated with age-related diseases. Furthermore, sarcopenia, a major component of the development of frailty41 in which IGF-1 as well as vitamin D may have a role,42 has been observed in aged dogs.43 Third, aging in dogs is associated with cognitive decline (deficits in learning complex tasks and decline in memory) with neurodegenerative changes (eg, cortical atrophy, ventricular widening, white matter volume reduction, and neurogenesis abatement) similar to those that develop in humans.44 In humans, cognitive decline is associated with frailty; frail individuals have lower cognitive performance, compared with that of non-frail individuals.45 Altogether, these findings support the hypothesis that, similar to humans, dogs may develop signs of frailty with increasing age. However, as cognitive decline is thought to be one of the core indicators of frailty in humans, the cognitive dysfunction syndrome (also known as DISH [disorientation, reduced social interactions, changes in wake and sleep patterns, and house soiling] or as its expanding definition DISHAAL [disorientation; alterations in interactions with owners, other pets, or the environment; sleep-wake cycle disturbances; house soiling; changes in activity; perceived anxiety; and learning or memory deficits]16) in dogs should be assessed in future studies to investigate frailty, as it is recommended for elderly dogs.46
Results of several studies in mice indicate that frailty can be assessed by application of various criteria (eg, muscle strength and travelling and standing activities), calculation of a frailty index (determined from 31 invasive and noninvasive health-related variables,25,47), or by assessment of 4 of the 5 frailty components described by Fried et al.20 Although the use of mice as a laboratory animal model of aging in humans is popular48 and seems promising for frailty research, mice (and other experimental animals kept in captivity), unlike dogs, do not share the same environment (eg, housing) or live under similar conditions (eg, nutrition, exercise, lifestyle, and medical follow-up) as humans, all of which are implicated in the human aging process.49 On the basis of that perspective and the results of the study reported here, we believe that the study of frailty in dogs conjointly with mice studies is clinically relevant and may provide insights in aging research.
In the present study, the 5 components of the FRP for dogs were adapted from a geriatric health evaluation scoresheet, the purpose of which was not to assess frailty. This approach may have had some limitations. According to the definition of a frailty phenotype,20 the chronic undernutrition component is evaluated by an unintentional weight loss of ≥ 4.5 kg in the preceding year or, at follow up, ≥ 5% of body weight in the preceding year. For dogs, only the second criterion would be pertinent, but in the present study, the weight of the dogs was not recorded at the time of CGA. Because the skin has a high metabolism, any nutritive deficiency in dogs has a direct impact on the quality of the hair.22 From the 3 preselected items to represent chronic undernutrition in the FRP, quality and density of hair (normal, sparse hair, or marked alopecia) was finally selected. Furthermore, the mobility component of the human frailty phenotype is determined on the basis of the time required to walk a distance of 15 feet for the slowest 20% of the population. Gait speed was not measured in the dogs of the present study; however, we assumed that dogs with gait disorders such as stiffness, lameness, or ataxia (as recorded during the CGA) would have had a slower gait (had it been measured), compared with gait speed of dogs without such mobility disorders. Of the 2 items preselected to represent poor mobility in the FRP, gait (normal, marked stiffness or lameness, or ataxia) was finally selected. In future studies, gait could be evaluated more accurately,50 although implementation of that type of evaluation in clinics may be difficult for some practitioners. Finally, the weakness component of the human frailty phenotype is assessed through a grip strength test. In the absence of any performance test in the guide dogs of the present study, we used an item related to muscle mass (normal, moderate muscle wasting, muscle atrophy, or cachexia) because of the existing association between the physical strength and availability of muscle mass. In future studies investigating frailty in dogs, a muscle condition score should be used to more accurately evaluate lean body mass.51
As previously done in humans,21 we used the term FRP rather than frailty, specifically to emphasize that the intent was not to accurately estimate the true prevalence of frailty in dogs, but rather to transpose a concept from one species (Homo sapiens) to another (C familiaris) by use of a priori-selected items that were closely related to the original 5 components of the human frailty phenotype. Consequently, we cannot rule out lack of accuracy in defining each of the 5 components of the canine FRP as well as a lack of accuracy in assessing the presence versus absence of an FRP component in the study dogs. However, these potential misclassification errors did not depend on survival time, which would have led to nondifferential bias in estimated HRs toward the null. Nevertheless, despite these potential (nondifferential) misclassification errors, the FRP, an approximation of the frailty phenotype defined by Fried et al,7 had a strong association with time to death, suggesting a basis for further research in this area.
Longevity in dogs is multifactorial, wherein life course exposures including reproductive history (in bitches, the number of years of lifetime ovary exposure was shown to be positively associated with longevity52) as well as genetics within and between countries15,53 have major roles. Therefore, because the dogs used in the present study were born and raised in France, were all neutered or spayed at the time of the CGA, and were working guide dogs, inference about the incidence rate of death from this study sample cannot be made. However, the purpose of the present study was not to estimate this rate but rather to assess the association between an FRP and time to death in dogs. Such a lack of representativeness would have led to selection bias away from the null (ie, would have created an association between having ≥ 2 FRP components and time to death) if the association between having ≥ 2 FRP components and time to death only exists in neutered or spayed French guide dogs. That situation is, to our knowledge, not supported by any pathophysiologic hypothesis. Future studies to investigate the concept of frailty in dogs are needed to confirm this hypothesis.
In the present study, the component of chronic undernutrition of the FRP was based on the quality and density of the hair of the dogs. In humans, chronic undernutrition leading to a decrease in loss of muscle mass may be associated with obesity.54 Obesity has been shown to be associated with frailty55 through limitations in mobility, comorbid conditions (eg, diabetes mellitus), and biological pathways (including hypothyroidism41) associated with weight gain.56 In dogs, hypothyroidism is associated with weight gain57 and the quality and the density of hair.58 These findings favor the more accurate study of undernutrition and obesity (eg, through use of a body condition score16 and assessment of thyroid hormones concentrations) in the association with frailty in aged dogs.
The low number of dogs in the present study that had ≥ 2 components of the FRP prevented adjustment for ≥ 1 potential confounder besides age. This limitation had an impact in that causal inferences cannot be drawn from the results. However, the multiple adjustments for demographic, biological, subclinical, or clinical data besides age resulted in significant HRs quantifying the association between having ≥ 2 components of the FRP (vs none or 1 FRP component) and time to death.
Among the 116 dogs included in the present study, 97 (84%) were of 2 breeds (Golden Retriever, Labrador Retriever, or mixes of each of these 2 breeds); thus, the sample did not represent the overall distribution of breeds among domestic dogs. However, this lack of representativeness was considered a strength of the study because among a group of 116 dogs, we would not have been able to control for genetic variation. By inclusion of genetically similar dogs, we controlled for this potentially confounding effect.
The study described in this report was intended to develop a frailty index or phenotype for companion dogs on the basis of an existing human frailty index or phenotype. Given that there are specific functional and physiological changes in aging companion dogs, we cannot rule out that a dog-specific index or phenotype of frailty would be more valuable. Dogs indeed have a unique set of frailties, disabilities, and diseases. Although many similarities between the human and canine genome do exist,11 by trying to shoehorn canine physiology into a human framework, one runs the risk of ignoring pathways that might be most relevant to the biology of aging in dogs.
To our knowledge, the present study is the first study to investigate the relevance of the concept of frailty in a nonexperimental animal model, namely domestic dogs not raised or kept for research. Results of this study have suggested that frailty should be evaluated in aged dogs because findings may indicate a worse prognosis. Furthermore, the concept of frailty as previously hypothesized by Walston et al33 a decade ago and shown in experimental animal models as well as the identification of frail individuals seems transposable to dogs, a species that not only shares humans' environment but also develops several naturally occurring age-related diseases similar to those of humans. Future studies are needed to continue to develop this concept of frailty in dogs. If the results of the present study are supported by further research data, observational studies and clinical trials in aged dogs (in which half the individuals die within a 4-year period) that investigate interventions to delay the onset of frailty or disability will be of major importance for human research in gerontology and geriatrics.
ABBREVIATIONS
| aHR | Adjusted hazard ratio |
| CGA | Clinical geriatric assessment |
| CI | Confidence interval |
| FRP | Frailty-related phenotype |
| HR | Hazard ratio |
| IGF | Insulin-like growth factor |
Footnotes
Scoresheet available upon request from Dr. Claude Muller (camuller@free.fr).
SAS, version 9.3, SAS Institute Inc, Cary, NC.
References
1. Rockwood K, Rockwood MR, Mitnitski A. Physiological redundancy in older adults in relation to the change with age in the slope of a frailty index. J Am Geriatr Soc 2010; 58: 318–323.
2. Kuchel GA. Frailty, allostatic load, and the future of predictive gerontology. J Am Geriatr Soc 2009; 57: 1704–1706.
3. Clegg A, Young J, Iliffe S, et al. Frailty in elderly people. Lancet 2013; 381: 752–762.
4. Larrick JW, Mendelsohn A. Applied Healthspan engineering. Rejuvenation Res 2010; 13: 265–280.
5. Theou O, Brothers TD, Mitnitski A, et al. Operationalization of frailty using eight commonly used scales and comparison of their ability to predict all-cause mortality. J Am Geriatr Soc 2013; 61: 1537–1551.
6. Cesari M, Gambassi G, van Kan GA, et al. The frailty phenotype and the frailty index: different instruments for different purposes. Age Ageing 2014; 43: 10–12.
7. Fried LP, Ferrucci L, Darer J, et al. Untangling the concepts of disability, frailty, and comorbidity: implications for improved targeting and care. J Gerontol A Biol Sci Med Sci 2004; 59: 255–263.
8. Kim SK. Common aging pathways in worms, flies, mice and humans. J Exp Biol 2007; 210: 1607–1612.
9. Languille S, Blanc S, Blin O, et al. The grey mouse lemur: a non-human primate model for ageing studies. Ageing Res Rev 2012; 11: 150–162.
10. Waters DJ. Aging research 2011: exploring the pet dog paradigm. ILAR J 2011; 52:97–105.
11. Parker HG, Shearin AL, Ostrander EA. Man's best friend becomes biology's best in show: genome analyses in the domestic dog. Annu Rev Genet 2010; 44: 309–336.
12. Rowell JL, McCarthy DO, Alvarez CE. Dog models of naturally occurring cancer. Trends Mol Med 2011; 17: 380–388.
13. Hamlin RL. Geriatric heart diseases in dogs. Vet Clin North Am Small Anim Pract 2005; 35: 597–615.
14. Bosch MN, Pugliese M, Gimeno-Bayon J, et al. Dogs with cognitive dysfunction syndrome: a natural model of Alzheimer's disease. Curr Alzheimer Res 2012; 9: 298–314.
15. O'Neill DG, Church DB, McGreevy PD, et al. Longevity and mortality of owned dogs in England. Vet J 2013; 198: 638–643.
16. Bellows J, Colitz CM, Daristotle L, et al. Defining healthy aging in older dogs and differentiating healthy aging from disease. J Am Vet Med Assoc 2015; 246: 77–89.
17. Kirkland JL, Peterson C. Healthspan, translation, and new outcomes for animal studies of aging. J Gerontol A Biol Sci Med Sci 2009; 64: 209–212.
18. Liu H, Graber TG, Ferguson-Stegall L, et al. Clinically relevant frailty index for mice. J Gerontol A Biol Sci Med Sci 2014; 69: 1485–1491.
19. Rockwood K, Song X, MacKnight C, et al. A global clinical measure of fitness and frailty in elderly people. CMAJ 2005; 173: 489–495.
20. Fried LP, Tangen CM, Walston J, et al. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci 2001; 56: M146–M156.
21. Desquilbet L, Jacobson LP, Fried LP, et al. HIV-1 infection is associated with an earlier occurrence of a phenotype related to frailty. J Gerontol A Biol Sci Med Sci 2007; 62: 1279–1286.
22. Laflamme DP. Nutrition for aging cats and dogs and the importance of body condition. Vet Clin North Am Small Anim Pract 2005; 35: 713–742.
23. Desquilbet L, Mariotti F. Dose-response analyses using restricted cubic spline functions in public health research. Stat Med 2010; 29: 1037–1057.
24. Patronek GJ, Waters DJ, Glickman LT. Comparative longevity of pet dogs and humans: implications for gerontology research. J Gerontol A Biol Sci Med Sci 1997; 52: B171–B178.
25. Parks RJ, Fares E, Macdonald JK, et al. A procedure for creating a frailty index based on deficit accumulation in aging mice. J Gerontol A Biol Sci Med Sci 2012; 67: 217–227.
26. Yang Y, Lee LC. Dynamics and heterogeneity in the process of human frailty and aging: evidence from the U.S. older adult population. J Gerontol B Psychol Sci Soc Sci 2010; 65B:246–255.
27. Afilalo J, Karunananthan S, Eisenberg MJ, et al. Role of frailty in patients with cardiovascular disease. Am J Cardiol 2009; 103: 1616–1621.
28. Meurs KM. Genetics of cardiac disease in the small animal patient. Vet Clin North Am Small Anim Pract 2010; 40: 701–715.
29. Lipsitz LA. Physiological complexity, aging, and the path to frailty. Sci Aging Knowledge Environ 2004; 16: pe16.
30. Fried LP, Xue QL, Cappola AR, et al. Nonlinear multisystem physiological dysregulation associated with frailty in older women: implications for etiology and treatment. J Gerontol A Biol Sci Med Sci 2009; 64: 1049–1057.
31. Arum O, Rasche ZA, Rickman DJ, et al. Prevention of neuromusculoskeletal frailty in slow-aging ames dwarf mice: longitudinal investigation of interaction of longevity genes and caloric restriction. PLoS ONE 2013; 8: e72255.
32. Graber TG, Ferguson-Stegall L, Liu H, et al. Voluntary aerobic exercise reverses frailty in old mice. J Gerontol A Biol Sci Med Sci 2015; 70: 1045–1058.
33. Walston J, Hadley EC, Ferrucci L, et al. Research agenda for frailty in older adults: toward a better understanding of physiology and etiology: summary from the American Geriatrics Society/National Institute on Aging Research Conference on Frailty in Older Adults. J Am Geriatr Soc 2006; 54: 991–1001.
34. Ziv E, Hu D. Genetic variation in insulin/IGF-1 signaling pathways and longevity. Ageing Res Rev 2011; 10: 201–204.
35. Sutter NB, Bustamante CD, Chase K, et al. A single IGF1 allele is a major determinant of small size in dogs. Science 2007; 316: 112–115.
36. Greer KA, Hughes LM, Masternak MM. Connecting serum IGF-1, body size, and age in the domestic dog. Age (Dordr) 2011; 33: 475–483.
37. Gerdhem P, Ringsberg KA, Obrant KJ, et al. Association between 25-hydroxy vitamin D levels, physical activity, muscle strength and fractures in the prospective population-based OPRA Study of Elderly Women. Osteoporos Int 2005; 16: 1425–1431.
38. Wong YY, McCaul KA, Yeap BB, et al. Low vitamin D status is an independent predictor of increased frailty and all-cause mortality in older men: the Health in Men Study. J Clin Endocrinol Metab 2013; 98: 3821–3828.
39. Kraus MS, Rassnick KM, Wakshlag JJ, et al. Relation of vitamin D status to congestive heart failure and cardiovascular events in dogs. J Vet Intern Med 2014; 28: 109–115.
40. Wakshlag JJ, Rassnick KM, Malone EK, et al. Cross-sectional study to investigate the association between vitamin D status and cutaneous mast cell tumours in Labrador Retrievers. Br J Nutr 2011; 106 (Suppl 1):S60–63.
41. Morley JE, Malmstrom TK. Frailty, sarcopenia, and hormones. Endocrinol Metab Clin North Am 2013; 42: 391–405.
42. Waters DL, Baumgartner RN, Garry PJ, et al. Advantages of dietary, exercise-related, and therapeutic interventions to prevent and treat sarcopenia in adult patients: an update. Clin Interv Aging 2010; 5: 259–270.
43. Freeman LM. Cachexia and sarcopenia: emerging syndromes of importance in dogs and cats. J Vet Intern Med 2012; 26: 3–17.
44. Head E. A Canine model of human aging and Alzheimer's disease. Biochim Biophys Acta 2013; 1832: 1384–1389.
45. Canevelli M, Cesari M, van Kan GA. Frailty and cognitive decline: how do they relate? Curr Opin Clin Nutr Metab Care 2015; 18: 43–50.
46. Heckler MCT, Tranquilim MV, Svicero DJ, et al. Clinical feasibility of cognitive testing in dogs (Canis lupus familiaris). J Vet Behav 2014; 9: 6–12.
47. Kane AE, Hilmer SN, Boyer D, et al. Impact of longevity interventions on a validated mouse clinical frailty index. J Gerontol A Biol Sci Med Sci 2016; 71: 333–339.
48. Yuan R, Peters LL, Paigen B. Mice as a mammalian model for research on the genetics of aging. ILAR J 2011; 52: 4–15.
49. Cesari M, Vellas B, Gambassi G. The stress of aging. Exp Gerontol 2013; 48: 451–456.
50. Barthelemy I, Barrey E, Thibaud JL, et al. Gait analysis using accelerometry in dystrophin-deficient dogs. Neuromuscul Disord 2009; 19: 788–796.
51. Hutchinson D, Sutherland-Smith J, Watson AL, et al. Assessment of methods of evaluating sarcopenia in old dogs. Am J Vet Res 2012; 73: 1794–1800.
52. Waters DJ, Kengeri SS, Maras AH, et al. Probing the perils of dichotomous binning: how categorizing female dogs as spayed or intact can misinform our assumptions about the lifelong health consequences of ovariohysterectomy. Theriogenology 2011; 76: 1496–1500.
53. Proschowsky HF, Rugbjerg H, Ersboll AK. Mortality of purebred and mixed-breed dogs in Denmark. Prev Vet Med 2003; 58: 63–74.
54. Roubenoff R. Sarcopenic obesity: does muscle loss cause fat gain? Lessons from rheumatoid arthritis and osteoarthritis. Ann N Y Acad Sci 2000; 904: 553–557.
55. Blaum CS, Xue QL, Michelon E, et al. The association between obesity and the frailty syndrome in older women: the Women's Health and Aging Studies. J Am Geriatr Soc 2005; 53: 927–934.
56. Kumar HK, Verma A, Muthukrishnan J, et al. Obesity and hypothyroidism: symbiotic coexistence. Arch Intern Med 2008; 168: 2168.
57. Dixon RM, Reid SW, Mooney CT. Epidemiological, clinical, haematological and biochemical characteristics of canine hypothyroidism. Vet Rec 1999; 145: 481–487.
58. Credille KM, Slater MR, Moriello KA, et al. The effects of thyroid hormones on the skin of Beagle dogs. J Vet Intern Med 2001; 15: 539–546.
