Muscle loss is common in animals with chronic diseases (eg, chronic kidney disease, congestive heart failure, and cancer) or an acute injury or illness and during aging. Muscle loss associated with disease is called cachexia, whereas muscle loss associated with aging in the absence of disease is called sarcopenia.1 Because older animals are more likely to develop chronic diseases, sarcopenia and cachexia can occur concurrently. Given that there is a lack of reserve for endogenous protein, cachexia and sarcopenia result in loss of functional tissue, which impacts strength, immune function, wound healing, and mortality rates.1
Identifying cachexia and sarcopenia at their earliest stages is important to enable the most effective treatment. In humans, most definitions for cachexia rely on weight loss. However, weight loss occurs later than muscle loss, so identifying muscle loss is important. Muscle mass can be estimated in humans by measurements of lean body mass obtained with DEXA. This works well in healthy humans, although adjustments for age, sex, and race are important for accurate results.2 However, there are inherent assumptions with DEXA that negatively impact its accuracy for ill people and companion animals. An additional challenge is that DEXA requires that companion animals be anesthetized.3 Computed tomography has been used as a more direct method to measure muscle mass in humans and dogs, but CT still requires that companion animals be anesthetized or sedated and exposes subjects to some radiation. Therefore, more clinically relevant measures of muscle mass are desirable.
Accuracy of the MCS has been validated in cats,4 and MCS is recommended by the American Animal Hospital Association5 and World Small Animal Veterinary Association6 as part of a nutritional assessment that should be performed on all dogs and cats at every veterinary visit. An MCS system has been developed for cats and dogs,7 but it has not yet been validated for use in dogs.
Quantitative magnetic resonance is a technique that does not require that dogs and cats be anesthetized, and results of QMR are significantly correlated with DEXA results and total body water (determined by use of deuterium) in healthy Beagles.8 However, even in a healthy population of dogs, QMR and DEXA both underestimated lean body mass by 13.4% and 7.3%, respectively, compared with the value determined by use of total body water.8 Precision and accuracy of QMR and DEXA for ill animals have not been determined.
Recently, ultrasonography of muscle has been used to assess muscle loss in humans with cachexia and sarcopenia.9,10 An ultrasonographic method for assessment of epaxial muscles has been validated for use on healthy dogs11 and cats.12 However, to be of clinical use for dogs, this method must be evaluated in dogs with cachexia and sarcopenia.
Therefore, the objectives of the study reported here were to evaluate the repeatability and reproducibility of a 4-point MCS system for use in dogs; to assess the convergent validity between the MCS, ultrasonographic images, and QMR images in dogs; and to identify cutoff values for ultrasonographic measurements of muscle that could be used to identify dogs with cachexia and sarcopenia.
Materials and Methods
Animals
Forty dogs owned by a pet nutrition companya were selected for the study. Dogs represented a wide range for age, BCS, and MCS. Dogs were 3 to 16 years old and categorized as young (< 6 years), middle aged (6 to 10 years), and elderly (> 10 years). Dogs were categorized as underweight (BCS < 4/9), ideal to slightly overweight (BCS 4/9 to 6/9), or overweight (> 6/9). At least 4 dogs were selected to represent each age-by-BCS category. Dogs with chronic medical conditions (eg, chronic kidney disease, cardiac disease, cancer, and hepatic disease) were eligible for inclusion in the study. The study was performed at the facility of the pet nutrition company, and the study protocol was reviewed and approved by the institutional animal care and use committee of the same company.
Experimental procedures
A prospective cross-sectional study was conducted. On the day of the study, physical examination was performed on each dog to confirm health status. For all dogs, assessments included body weight, BCS, MCS, QMR, thoracic radiography, ultrasonographic measurement of epaxial muscle, and forelimb circumference.
Body weight and BCS—Body weight was recorded to the nearest 0.01 kg by use of a calibrated scale. The BCS was measured by use of a 9-point scoring system.13
MCS—Four raters (2 investigators [LMF and KEM], 1 additional veterinarian, and 1 animal technician) received training on using the MCS system to assign an MCS of normal muscle mass or mild, moderate, or severe muscle loss.7 Each of the 4 raters then evaluated and scored all 40 dogs 3 separate times during a 2- to 4-day period. Order of evaluation of the dogs was randomly assigned by use of an online random number generatorb among raters and for each scoring session for each rater.
QMR—The QMR scanning was performed in triplicate on each dog, as described elsewhere.8 Briefly, each unsedated dog was placed in a polymethylmethacrylate crate that allowed for limited movement. The crate was placed within the magnet bore of the QMR unit, which was designed for scanning of animals that weighed < 50 kg. Dogs were positioned at the magnet isocenter such that the long axis of each dog was perpendicular to the long axis of the magnet bore. Data were collected as a standard 3-minute fat-and-water acquisition for each scan in accordance with the manufacturer's protocol.
Ultrasonographic measurement of muscle—Ultrasonographic measurements were obtained at the level of T13 by use of an ultrasound machinec with a 5- to 12-MHz linear transducer,d as described elsewhere.11,12 Unsedated dogs were manually restrained in a standing position, the hair was shaved over the dorsum at the level of T13, and ultrasound gel was applied to the shaved area. The transducer was located at T13 with the beam angle perpendicular to the axis of the vertebral column; minimal pressure was used to avoid distorting the shape of the tissues. Measurements were obtained at the level of the articulation of the transverse process of the vertebra with the rib (ie, costotransverse joint). Three measurements (a separate placement of the probe for each measurement) of the maximal height of the right epaxial muscle were obtained, and the mean of the 3 measurements was used as the final value for muscle height. Height measurements were obtained with distance calipers of the ultrasound machine, with the first cursor placed at the junction of the T13 lamina and transverse process at the bone-muscle interface and the second cursor placed dorsolaterally at the muscle–subcutaneous tissue interface, which provided the smallest possible short-axis dimension. Muscle height was measured on all dogs by 1 investigator (JF).
Radiography—Lateral thoracic radiographs were obtained of all dogs. Length of T4 was measured and used to calculate a ratio of the muscle height at T13 to T4 length (ie, VEMS). This ratio was used to standardize muscle measurements across dogs of various shapes and sizes. To investigate a technique that could provide a metric for patient size without the use of a thoracic radiograph, forelimb circumference was measured at the midpoint between the carpus and elbow joint. This technique has been validated for use in cats as an alternative to the use of T4 length and is defined as the FLEMS.12
Statistical analysis
Data distributions were tested with the Shapiro-Wilk test. Because many of the data were not normally distributed, data were reported as median and range. For analyses, MCS categories were converted to numeric values as follows: 3 = normal muscle mass, 2 = mild muscle loss, 1 = moderate muscle loss, and 0 = severe muscle loss. Relationships between MCS, VEMS, FLEMS, percentage of lean mass determined by use of QMR, and body weight were tested with Spearman correlation analysis. Interrater agreement (reproducibility) and intrarater agreement (repeatability) were assessed with the κ statistic. Agreement based on the κ coefficient was interpreted as fair (κ = 0.21 to 0.40), moderate (κ = 0.41 to 0.60), substantial (κ = 0.61 to 0.80), or almost perfect (κ = 0.81 to 1.00). An ROC analysis was used to estimate the optimal cutoff point of VEMS and FLEMS for use in diagnosing mild muscle loss on the basis of the MCS. The ROC curve was calculated by use of logistic regression to estimate the probability that a dog would be classified as having mild muscle loss versus a normal muscle mass as a function of VEMS or FLEMS. An optimal cutoff point was estimated by use of the Youden index ([sensitivity + specificity] −1).14 All analyses were performed with commercially available software.e,f Values of P ≤ 0.05 were considered significant.
Results
Forty dogs were enrolled in the study. Median age was 13.9 years (range, 2.8 to 16.0 years). There were 23 males (all castrated) and 17 females (16 spayed). Dogs comprised 20 Beagles, 14 Labrador Retrievers, 2 Coonhounds, 2 Manchester Terriers, 1 English Setter, and 1 Smooth Fox Terrier. Dogs had a variety of chronic medical conditions, including osteoarthritis, cardiac disease, hepatic disease, chronic kidney disease, epilepsy, and cancer. Median BCS was 6/9 (range, 3/9 to 8/9); 27 (68%) dogs had a BCS > 5/9. The MCS (determined on the basis of the mean from the 2 investigators [LMF and KEM]) was characterized as normal muscle mass for 8 dogs, mild muscle loss for 9 dogs, moderate muscle loss for 10 dogs, and severe muscle loss for 13 dogs. Nineteen of the 32 (59%) dogs with muscle loss had a BCS > 5/9. There was a significant inverse correlation between BCS and age (r = −0.55; 95% CI, −0.28 to −0.79; P < 0.001) and MCS and age (r = −0.60; 95% CI, −0.33 to – 0.79; P < 0.001), such that with increasing age, BCS decreased and MCS worsened. The BCS was significantly correlated (r = 0.72; 95% CI, 0.81 to 1.00; P < 0.001) with MCS.
Interrater agreement (reproducibility) of MCS among the 4 raters yielded an overall κ coefficient of 0.50 (95% CI, 0.42 to 0.59; P < 0.001). There were differences in the κ coefficient depending on the degree of muscle loss as follows: normal muscle mass (κ = 0.67; 95% CI, 0.54 to 0.80; P < 0.001), mild muscle loss (κ = 0.31; 95% CI, 0.18 to 0.44; P < 0.001), moderate muscle loss (κ = 0.35; 95% CI, 0.22 to 0.48; P < 0.001), and severe muscle loss (κ = 0.74; 95% CI, 0.61 to 0.87; P < 0.001). Overall mean intrarater κ coefficient (repeatability) for the 4 raters was 0.67 (range, 0.59 to 0.77), which indicated moderate agreement. Range of the intrarater κ coefficient was 0.70 to 0.89 (P < 0.001) for dogs with normal muscle mass, 0.39 to 0.72 (P < 0.001) for dogs with mild muscle loss, 0.48 to 0.72 (P < 0.001) for dogs with moderate muscle loss, and 0.64 to 0.83 (P < 0.001) for dogs with severe muscle loss.
Ultrasonography was used to measure epaxial muscle height. The VEMS and FLEMS were calculated to address the fact that dogs were of various shapes and sizes; the median value for the VEMS was 1.05 (range, 0.65 to 1.79), and the median value for the FLEMS was 1.43 (range, 1.08 to 2.27). Correlations between body weight and VEMS and body weight and FLEMS were assessed to determine whether either method effectively resulted in normalization of muscle height among dogs of various body weights; a lack of correlation indicated good normalization among dogs of various body weights. There was no evidence of an association between VEMS and body weight (r = 0.13; 95% CI, −0.20 to 0.41; P = 0.42) or between FLEMS and body weight (r = 0.06; 95% CI, −0.27 to 0.37; P = 0.72).
The VEMS and FLEMS were significantly correlated (r = 0.84; 95% CI, 0.67 to 0.93; P < 0.001). Age was significantly correlated with VEMS (r = −0.63; 95% CI, −0.36 to −0.83; P < 0.001) and FLEMS (r = −0.56; 95% CI, −0.27 to −0.76; P < 0.001). The BCS was significantly correlated with VEMS (r = 0.46; 95% CI, −0.17 to 0.70; P = 0.003) and FLEMS (r = 0.31; 95% CI, −0.01 to 0.58; P = 0.05). The MCS was significantly correlated with VEMS (r = 0.69; 95% CI, 0.49 to 0.85; P < 0.001; Figure 1) and FLEMS (r = 0.55; 95% CI, 0.34 to 0.79; P < 0.001; Figure 2). The area under the ROC curve for VEMS was 0.94 (95% CI, 0.83 to 1.00; P = 0.002). Use of a cutoff value of 1.124 for the diagnosis of mild muscle loss yielded a sensitivity of 0.889 (95% CI, 0.518 to 0.997) and specificity of 1.000 (95% CI, 0.631 to 1.000). The area under the ROC curve for FLEMS was 0.83 (95% CI, 0.64 to 1.00; P = 0.02). Use of a cutoff value of 1.666 for the diagnosis of mild muscle loss yielded a sensitivity of 0.778 (95% CI, 0.400 to 0.972) and specificity of 0.625 (95% CI, 0.245 to 0.915).
Dot plot of MCS versus VEMS (ratio of the muscle height at T13 to T4 length) for 40 dogs. Dogs were scored as normal muscle mass or mild, moderate, or severe muscle loss.7 For the purpose of statistical analysis, MCS categories were converted to numeric values as follows: 3 = normal muscle mass, 2 = mild muscle loss, 1 = moderate muscle loss, and 0 = severe muscle loss. Each symbol represents results for 1 dog. The MCS and VEMS were significantly correlated (r = 0.69; P < 0.001).
Citation: American Journal of Veterinary Research 80, 6; 10.2460/ajvr.80.6.595
Dot plot of MCS versus FLEMS (ratio of the muscle height at T13 to the forelimb circumference measured at the midpoint between the carpus and elbow joint) for 40 dogs. Each symbol represents results for 1 dog. The MCS and FLEMS were significantly correlated (r = 0.55; P < 0.001). See Figure 1 for remainder of key.
Citation: American Journal of Veterinary Research 80, 6; 10.2460/ajvr.80.6.595
Percentage of lean tissue determined by use of QMR ranged from 53.1% to 87.2% (median, 65.9%) and had significant, but only fair, correlations with age (r = −0.33; 95% CI, −0.63 to 0.03; P = 0.04), VEMS (r = 0.33; 95% CI, 0.02 to 0.59; P = 0.04), and FLEMS (r = 0.35; 95% CI, 0.01 to 0.59; P = 0.03). Percentage of lean tissue determined by use of QMR was not significantly correlated (r = 0.30; 95% CI, −0.06 to 0.52; P = 0.06) with MCS (Figure 3). In addition, percentage of lean tissue determined by use of QMR was not significantly correlated (r = −0.05; 95% CI, −0.40 to 0.28; P = 0.75) with BCS.
Dot plot of MCS versus the percentage of lean tissue measured by use of QMR for 40 dogs. Each symbol represents results for 1 dog. The MCS and percentage of lean tissue were not significantly correlated (r = 0.30; P = 0.06). See Figure 1 for remainder of key.
Citation: American Journal of Veterinary Research 80, 6; 10.2460/ajvr.80.6.595
Discussion
In the study reported here, there was substantial repeatability of the MCS, but reproducibility was only moderate. Highest repeatability and reproducibility were for dogs with normal muscle mass and severe muscle loss, with lower agreement for the intermediate scores (ie, mild and moderate muscle loss). These findings are similar to those of a study4 conducted to validate the use of MCS in cats. Combining the mild and moderate categories into a single group may provide some advantages, but this could make it more difficult to identify dogs with muscle loss at an early stage. Additional studies are needed to identify methods to increase reproducibility of the MCS. Because repeatability for some of the raters in the present study was relatively high, even for the intermediate scores, additional training could increase reproducibility. Although some MCS systems have used numbers, our clinical impression is that numbers used in MCSs are often confused (eg, some clinicians use an MCS of 3 as severe muscle loss, whereas others use an MCS of 3 as normal muscle mass). Therefore, in the study reported here, we used the World Small Animal Veterinary Association MCS system that has descriptive terms (ie, normal muscle mass or mild, moderate, or severe muscle loss) instead of numbers. To avoid confusion with a numeric MCS, we recommend use of the World Small Animal Veterinary Association MCS system as the standard for assessment of MCS in every patient at every visit.7
Although both the MCS and QMR measurements were significantly correlated with results for the ultrasonographic methods of muscle assessment (ie, VEMS and FLEMS), the correlation between QMR measurements and VEMS and FLEMS was weak (r = 0.33 and 0.35, respectively). There was not a significant correlation between MCS and QMR measurements. The lack of strong correlations (or any correlation) was not surprising because MCS, VEMS or FLEMS, and QMR measurements are not used to assess the same body compartments. The MCS is a subjective assessment of overall muscle mass, whereas VEMS and FLEMS are based on measurement of epaxial muscle height. In contrast, QMR measurements estimate the percentage of total body tissue that is lean body mass. Lean body mass comprises both muscle and visceral organs. Therefore, these methods are all assessing related but different aspects of muscle in a dog. The MCS (qualitative) and ultrasonographic measurements (quantitative) more specifically assess muscle, whereas QMR measurements estimate lean body mass. Therefore, although these methods are correlated, they cannot be used interchangeably, and it is important to evaluate the specific goal for the measurement. For example, in clinical patients, MCS provides a quick assessment of overall muscle condition that can be used for every patient at every visit. Conversely, a more quantitative method of muscle assessment, such as VEMS, would likely be needed for research conducted to evaluate interventions specifically impacting skeletal muscle mass. When the body compartment of interest is lean body mass, QMR might be a more appropriate technique, although there are a number of conditions in which it is likely that QMR would be less accurate (eg, animals with altered hydration of lean body mass, animals with very low body fat, or animals that are extremely small). The wide variation in QMR results for the present study may have reflected variation in lean body mass of the dogs, but they could have also been attributable to limitations of QMR measurements. Although the ease of use and wealth of information provided make QMR promising for use in measurements of body composition, it must be validated for use in ill and injured dogs and cannot be considered a criterion-referenced standard.
The 2 methods of muscle ultrasonography, VEMS and FLEMS, were significantly correlated with each other but were not significantly associated with body weight. This finding suggested that both T4 length and forelimb circumference effectively normalized epaxial muscle height among dogs of various body weights. This is similar to results for studies of healthy dogs11 and cats,12 but extends these results to dogs with more variation in BCS and MCS. Sensitivity and specificity of VEMS were numerically higher than those of FLEMS. Therefore, although additional testing is needed, the VEMS may be preferable for use in research, especially when small changes in muscle are likely. However, this must be balanced with cost and potential stress of thoracic radiography. Thoracic radiographs are commonly obtained for animals with some chronic diseases (eg, congestive heart failure), but they may not be routinely obtained for animals with other chronic diseases. The ROC analysis was used to estimate the optimal cutoff to diagnose mild muscle loss. Sensitivity and specificity would be higher for diagnosing moderate muscle loss (or any muscle loss), but moderate muscle loss is relatively easy to identify clinically and thus would be of less value. The proposed cutoff values for VEMS (1.124) and FLEMS (1.666) for diagnosing mild muscle loss require prospective evaluation to determine their validity for identifying dogs with mild muscle loss. It also will be important to assess the sensitivity of VEMS and FLEMS as an outcome measure in clinical trials of interventions targeting muscle.
Body condition score, which is an assessment of fat, and MCS, which is an assessment of muscle, are not directly related. For example, a dog can be obese and have severe muscle loss or a dog can be thin but have normal muscle condition. In the present study, 19 of 32 (59%) dogs with muscle loss were overweight (BCS > 5/9). However, BCS was significantly associated with MCS, VEMS, and FLEMS. Despite these correlations, it is important to assess both BCS and MCS because both overweight and underweight animals can have muscle loss.
The study reported here had a number of limitations. Although several methods of assessing muscle were compared, there was no criterion-referenced standard. The only true criterion-referenced standard is carcass analysis, with substantial limitations for all other methods (eg, DEXA, QMR, and bioelectric impedance). Even CT, which can be used to provide an extremely accurate measure of epaxial muscle thickness, cannot be used to provide a measure of total body muscle without additional research to develop equations for determining total body muscle from measurements obtained for a single region. Epaxial muscle measurement with CT could have provided a useful comparison for muscle ultrasonography in dogs with muscle loss, as was reported in a previous study.15 However, general anesthesia was not an option for the dogs of the present study, so we elected to compare multiple, clinically relevant methods for assessment of muscle.
The population of dogs was also quite heterogenous, with variations in BCS, age, and health status, which may have affected reproducibility and repeatability. However, the advantage of this heterogenous population was that it was more reflective of clinical patients. Additional studies of MCS, VEMS, FLEMS, and QMR measurements in more homogenous populations (eg, dogs with a single underlying disease or geriatric dogs without underlying diseases) are warranted to further evaluate these muscle assessment methods.
Additional studies also are needed on changes in muscle for animals with cachexia and sarcopenia. It is unclear whether loss of muscle is consistent along the entire length of the epaxial muscles or whether different breeds have variable structure of these muscles. Because VEMS and FLEMS are assessed at a single location (ie, T13), this may not capture all of the changes that occur in muscle size during aging or illness. In addition, cachexia and sarcopenia involve qualitative changes in muscle as well as quantitative changes, such as conversion of type I to type II fibers, infiltration of muscles with lipid, and loss of α motor neuron input to muscles.16 Therefore, it will also be important to study qualitative changes to the muscles in dogs with cachexia and sarcopenia.
Despite these limitations, results of the present study suggested that an MCS system can be used with substantial repeatability and moderate reproducibility as a subjective method for assessing muscle in dogs. Two normalized ultrasonographic measurements, VEMS and FLEMS, can be used for a more quantitative assessment of muscle. Cutoff values for VEMS and FLEMS were identified in the present study for the diagnosis of mild muscle loss. Prospective studies of these cutoff values in dogs as well as the use of VEMS and FLEMS as outcome measures in clinical trials are warranted.
Acknowledgments
Supported by Nestlé Purina PetCare Co.
Drs. Freeman, Michel, and Fages received reimbursement of travel expenses associated with conducting this study. Dr. Freeman has received research funding or provided sponsored lectures or consulting services for Royal Canin, Nestlé Purina PetCare, Aratana Therapeutics, and Hill's Pet Nutrition. Dr. Freeman also serves on an Advisory Council for Aratana Therapeutics. Dr. Michel has received research funding from Royal Canin and Nestlé Purina PetCare Co and serves on an Advisory Council for Nestlé Purina PetCare Co. Drs. Zanghi and Vester Boler are employees of Nestlé Purina Research.
The authors thank Jim Ambrose, Dr. Melanie Barnes, Kacie Dahms, and Laura Hart for technical assistance.
ABBREVIATIONS
| BCS | Body condition score |
| CI | Confidence interval |
| DEXA | Dual-energy x-ray absorptiometry |
| FLEMS | Forelimb epaxial muscle score |
| MCS | Muscle condition score |
| QMR | Quantitative MRI |
| ROC | Receiver operator characteristic |
| VEMS | Vertebral epaxial muscle score |
Footnotes
Nestlé Purina PetCare Co, St Louis, Mo.
Random integer generator, Randomness Integrity Services Ltd. Dublin, Ireland.
Sonoscape S8 Expert, National Ultrasound, Duluth, Ga.
L752 5- to 12-MHz linear probe, National Ultrasound, Duluth, Ga.
Systat, version 13.0, Systat Inc, San Jose, Calif.
SPSS, version 24.0, IBM Corp, Armonk, NY.
References
1. Freeman LM. Cachexia and sarcopenia: emerging syndromes of importance in dogs and cats. J Vet Intern Med 2012;26:3–17.
2. Petak S, Barbu CG, Yu EW, et al. The Official Positions of the International Society for Clinical Densitometry: body composition analysis reporting. J Clin Densitom 2013;16:508–519.
3. Freeman LM, Kehayias JJ, Roubenoff R. Letter to the editor. J Vet Intern Med 1996;10:99–100.
4. Michel KE, Anderson W, Cupp C, et al. Correlation of a feline muscle mass score with body composition determined by dual-energy x-ray absorptiometry. Br J Nutr 2011;106 (suppl 1):S57–S59.
5. Baldwin K, Bartges J, Buffington T, et al. AAHA nutritional assessment guidelines for dogs and cats. J Am Anim Hosp Assoc 2010;46:285–296.
6. Freeman L, Becvarova I, Cave N, et al. WSAVA nutritional assessment Guidelines. J Small Anim Pract 2011;52:385–396.
7. World Small Animal Veterinary Association Global Nutrition Committee. Muscle condition score charts for dogs and cats. Available at: www.wsava.org/Guidelines/Global-Nutrition-Guidelines. Accessed Nov 24, 2018.
8. Zanghi BM, Cupp CJ, Pan Y, et al. Noninvasive measurements of body composition and body water via quantitative magnetic resonance, deuterium water, and dual-energy x-ray absorptiometry in awake and sedated dogs. Am J Vet Res 2013;74:733–743.
9. Sergi G, Trevisan C, Veronese N, et al. Imaging of sarcopenia. Eur J Radiol 2016;85:1519–1524.
10. Nijholt W, Scafoglieri A, Jager-Wittenaar H, et al. The reliability and validity of ultrasound to quantify muscles in older adults: a systematic review. J Cachexia Sarcopenia Muscle 2017;8:702–712.
11. Freeman LM, Sutherland-Smith J, Prantil LR, et al. Quantitative assessment of muscle in dogs using a vertebral epaxial muscle score. Can J Vet Res 2017;81:255–260.
12. Freeman LM, Sutherland-Smith J, Cummings C, et al. Evaluation of a quantitatively derived value for assessment of muscle mass in clinically normal cats. Am J Vet Res 2018;79:1188–1192.
13. Laflamme D. Development and validation of a body condition score system for dogs. Canine Pract 1997;22(4):10–15.
14. Fluss R, Faraggi D, Reiser B. Estimation of the Youden index and its associated cutoff point. Biom J 2005;47:458–472.
15. 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.
16. Lang T, Streeper T, Cawthon P, et al. Sarcopenia: etiology, clinical consequences, intervention, and assessment. Osteoporos Int 2010;21:543–559.
