• View in gallery

    Scatterplot of urine osmolality versus USG for 60 dogs that were examined by the internal medicine service at a veterinary teaching hospital. The plot includes data points for individual dogs (black dots) and the univariate linear regression line (solid black line; regression equation, urine osmolality = −40,890 + [40,777 × USG]; R2 = 0.87) and 95% prediction bands (black dashed lines) for the association of USG with urine osmolality as well as regression lines for the association of USG with urine osmolality when severity of ketonuria was controlled (mild ketonuria = solid yellow line; moderate ketonuria = solid blue line; and severe ketonuria = solid red line). Notice that the regression line for the association between urine osmolality and USG moved slightly to the right as the severity of ketonuria increased.

  • 1. Wellman ML, DiBartolla SP. Applied physiology of body fluids in dogs and cats. In: DiBartolla SP, ed. Fluid, electrolyte, and acid-base disorders in small animal practice. 3rd ed. St Louis: Saunders Elsevier, 2006;325.

    • Search Google Scholar
    • Export Citation
  • 2. Ganong WF. The general and cellular basis of medical physiology. In: Ganong WF, ed. Review of medical physiology. 22nd ed. Boston: McGraw-Hill, 2005;150.

    • Search Google Scholar
    • Export Citation
  • 3. Thrall MA, Weiser G, Allison R, et al. Urine concentration. In: Thrall MA, ed. Veterinary hematology and clinical chemistry. 2nd ed. Ames, Iowa: Wiley-Blackwell, 2012;333335.

    • Search Google Scholar
    • Export Citation
  • 4. Chadha V, Garg U, Alon US. Measurement of urinary concentration: a critical appraisal of methodologies. Pediatr Nephrol 2001; 16:374382.

    • Search Google Scholar
    • Export Citation
  • 5. Stevens LA, Lafayette RA, Perrone RD, et al. Laboratory evaluation of kidney function. In: Schrier RW, ed. Diseases of the kidney and urinary tract. 8th ed. Philadelphia: Lippincott Williams & Wilkins, 2006;299335.

    • Search Google Scholar
    • Export Citation
  • 6. Paris JK, Bennett AD, Dodkin SJ. Comparison of a digital and an optical analogue hand-held refractometer for the measurement of canine urine specific gravity. Vet Rec 2012; 170:463.

    • Search Google Scholar
    • Export Citation
  • 7. Dossin O, Germain C, Braun JP. Comparison of the techniques of evaluation of urine dilution/concentration in the dog. J Vet Med A Physiol Pathol Clin Med 2003; 50:322325.

    • Search Google Scholar
    • Export Citation
  • 8. DiBartolla SP. Applied physiology of body fluids in dogs and cats. In: DiBartolla SP, ed. Fluid, electrolyte, and acid-base disorders in small animal practice. 3rd ed. St Louis: Saunders Elsevier, 2006;4779.

    • Search Google Scholar
    • Export Citation
  • 9. Leech S, Penney MD. Correlation of specific gravity and osmolality of urine in neonates and adults. Arch Dis Child 1987; 62:671673.

  • 10. Imran S, Eva G, Christopher S, et al. Is specific gravity a good estimate of urine osmolality? J Clin Lab Anal 2010; 24:426430.

  • 11. Voinescu GC, Shoemaker M, Moore H, et al. The relationship between urine osmolality and specific gravity. Am J Med Sci 2002; 323:3942.

    • Search Google Scholar
    • Export Citation
  • 12. Fry MM. Urinalysis. In: Bartges J, Polzin DJ, eds. Nephrology and urology of small animals. Chichester, West Sussex, England: Wiley-Blackwell, 2011;4661.

    • Search Google Scholar
    • Export Citation
  • 13. Stockham SL, Scott MA. Urinary system In: Stockham SL, Scott MA, eds. Fundamentals of veterinary clinical pathology. 2nd ed. Ames, Iowa: Blackwell, 2008;458467.

    • Search Google Scholar
    • Export Citation

Advertisement

Association between urine osmolality and specific gravity in dogs and the effect of commonly measured urine solutes on that association

Jennifer A. Ayoub DVM1, Hugues Beaufrere DrMedVet, PhD2, and Mark J. Acierno MBA3
View More View Less
  • 1 Department of Veterinary Clinical Sciences, School of Veterinary Medicine, Louisiana State University, Baton Rouge, LA 70803.
  • | 2 Department of Zoological Medicine, School of Veterinary Medicine, Louisiana State University, Baton Rouge, LA 70803.
  • | 3 Department of Veterinary Clinical Sciences, School of Veterinary Medicine, Louisiana State University, Baton Rouge, LA 70803.

Abstract

Objective—To determine the association between urine osmolality and specific gravity (USG) in dogs and to evaluate the effect of commonly measured urine solutes on that association.

Animals—60 dogs evaluated by an internal medicine service.

Procedures—From each dog, urine was obtained by cystocentesis and USG was determined with a refractometer. The sample was divided, and one aliquot was sent to a diagnostic laboratory for urinalysis and the other was frozen at −80°C until osmolality was determined. Urine samples were thawed and osmolality was measured in duplicate with a freezing-point depression osmometer. The correlation between mean urine osmolality and USG was determined; the effect of pH, proteinuria, glucosuria, ketonuria, bilirubinuria, and hemoglobinuria on this relationship was investigated with multiple regression analysis.

Results—The Pearson correlation coefficient between urine osmolality and USG was 0.87. The final multivariable regression model for urine osmolality included USG and the presence of ketones; ketonuria had a small negative association with urine osmolality.

Conclusions and Clinical Relevance—Results indicated a strong linear correlation between osmolality and USG in urine samples obtained from dogs with various pathological conditions, and ketonuria had a small negative effect on that correlation.

Abstract

Objective—To determine the association between urine osmolality and specific gravity (USG) in dogs and to evaluate the effect of commonly measured urine solutes on that association.

Animals—60 dogs evaluated by an internal medicine service.

Procedures—From each dog, urine was obtained by cystocentesis and USG was determined with a refractometer. The sample was divided, and one aliquot was sent to a diagnostic laboratory for urinalysis and the other was frozen at −80°C until osmolality was determined. Urine samples were thawed and osmolality was measured in duplicate with a freezing-point depression osmometer. The correlation between mean urine osmolality and USG was determined; the effect of pH, proteinuria, glucosuria, ketonuria, bilirubinuria, and hemoglobinuria on this relationship was investigated with multiple regression analysis.

Results—The Pearson correlation coefficient between urine osmolality and USG was 0.87. The final multivariable regression model for urine osmolality included USG and the presence of ketones; ketonuria had a small negative association with urine osmolality.

Conclusions and Clinical Relevance—Results indicated a strong linear correlation between osmolality and USG in urine samples obtained from dogs with various pathological conditions, and ketonuria had a small negative effect on that correlation.

The concentration of particles dissolved in urine, or urine osmolality, is an often-measured biological variable that allows for assessment of the kidneys' ability to concentrate or dilute urine relative to plasma. A concentration of urine particles that is increased from the reference range is often associated with dehydration and water conservation, whereas a concentration of urine particles that is decreased from the reference range is generally associated with conditions such as polydipsia, central diabetes insipidus, secondary peripheral diabetes insipidus, hypoadrenocorticism, primary renal glucosuria, Fanconi syndrome, and hyperglycemic glucosuria.

Actual determination of osmotically active urine particle (osmolyte) concentration requires measurement of osmolality. Osmolality provides a quantification of the solute concentration per unit of solvent. This relationship is unaffected by the physical characteristics of solutes such as molecular weight and is only affected by the number of osmolytes per unit of solvent.1 It is typically reported as osmoles per kiligram. Urine osmolality is routinely measured in laboratories by use of freezing-point osmometers, which rely on the principle that each mole of dissolved solute will decrease the freezing point of a liquid by 1.86°C.2

Because urine osmolality is difficult to measure in a clinical setting, USG, as measured by the index of light refraction, is commonly used as a surrogate.3 Unlike freezing-point depression, the index of light refraction can be affected by the physical properties of a solute and the temperature of the solution.4 Although modern refractometers can compensate for the temperature of a solution, the presence of relatively large molecules in the urine may have a disproportionate effect on USG.5 Other studies6,7 have focused on the degree of association between urine osmolality and USG in dogs but did not assess the effect of specific solutes on urine osmolality. Urine osmolality correction factors specific for some molecules such as glucose have been proposed.8

Results of studies9–11 that involved human patients suggest that large molecules such as ketones, bilirubin, and hemoglobin can affect the relationship between osmolality and USG. The correlation between urine osmolality and USG is high in the absence of pathological concentrations of those molecules and ranges from 0.83 to 0.999,10; however, that correlation decreases to 0.63 when those molecules are present.10 Investigators of 1 study11 suggest that USG should not be used as a surrogate for urine osmolality in patients with conditions that result in ketonuria, bilirubinuria, or hemoglobinuria.

The purpose of the study reported here was to determine the association between urine osmolality and USG in dogs and to evaluate the effect of commonly measured urine solutes on that association. The hypothesis was that the association between urine osmolality and USG would be affected by the presence of glucose, ketones, bilirubin, and hemoglobin in the urine. Therefore, the presence of those molecules might necessitate specific correction factors for urine osmolality.

Materials and Methods

Animals—Dogs evaluated by the internal medicine section at the Louisiana State University School of Veterinary Medicine Teaching Hospital and determined by a clinician to require a urinalysis as part of the diagnostic workup were considered for study inclusion. No attempt was made to screen dogs on the basis of breed, sex, age, or medical condition. The Louisiana State University Clinical Protocol and Institutional Animal Care and Use Committees approved the study, and for each dog enrolled in the study, consent was obtained from the owner prior to its enrollment.

Urine sample collection and analysis—From each dog, a 2.5-mL urine sample was obtained by cystocentesis that was performed by a veterinarian or veterinary technician. An optical refractometera that was calibrated in accordance with the manufacturer's instructions prior to each measurement was used by 1 investigator (JAA) to determine the specific gravity of each urine sample after it had equilibrated to room temperature (approx 22°C). Each urine sample was then divided; 0.5 mL was frozen at −80°C until osmolality was measured, and the remainder was submitted to a clinical pathology laboratoryb for urinalysis. On the day of sample collection, all urine samples were analyzed with a urine dipstick,c which was read with an automated analyzer.d All results were confirmed by a registered medical technologist in accordance with guidelinese established by the American Society for Clinical Pathology. When results determined by the technologist differed from those determined by the automated analyzer, analyses were repeated. For each sample, the presence of ketones was confirmed with a semiquantitative testf and the presence of bilirubin was confirmed with another semiquantitative test.g The results of the ketone and bilirubin semiquantitative tests were used instead of the corresponding results from the urine dipstick analysis for statistical analyses.

The frozen urine samples were thawed, and the osmolality of each sample was measured in duplicate by means of a commercial freezing-point depression osmometer,h which was operated by an investigator (MJA) who was trained in its use. The urine samples were assayed in batches, and prior to each batch, the osmometer was calibrated by means of a 2-point calibration technique in accordance with the manufacturer's instructions. For each urine sample, the duplicate osmolality measurements were used to calculate a mean value. If the difference between the duplicate measurements was > 5%, the values were discarded and that dog was excluded from the study.

Statistical analysis—The correlation between urine osmolality and USG was determined with the Pearson correlation coefficient (r). Kendall rank-correlation coefficients (τ) were used to determine the respective correlations between urine osmolality or USG and each of the following variables: urine pH and the presence of proteinuria, glucosuria, ketonuria, bilirubinuria, and hemoglobinuria. The association between urine osmolality and USG was further assessed by use of univariate linear regression. The effect of other commonly measured urine solutes on the association between urine osmolality and USG was assessed with multivariable linear regression, in which urine osmolality was the outcome of interest and USG was included as a fixed effect. For the multivariable regression analysis, the results for each urine solute evaluated (protein, glucose, ketones, bilirubin, and hemoglobin) were categorized on an ordinal scale of 0 to 4 in a manner similar to that described,10 in which 0 indicated that the solute was absent from the urine, 0.5 was indicative of trace amounts of a solute in the urine, and categories 1 to 4 were indicative of increasing amounts of a solute in the urine. Each solute was added as an explanatory variable to the multivariable regression model in a stepwise manner by use of forward and backward selection, and multicollinearity among the explanatory variables was assessed with variance inflation factors. When multicollinearity between 2 explanatory variables was identified (VIF > 10), only 1 of those variables was retained in the model and the explanatory variable chosen was the 1 that resulted in the lowest Aikaike information criterion for the model. The assumptions for linear regression (ie, normally distributed data and homoscedasticity of residuals) were assessed with residual diagnostics, and evaluations of the data for outliers and influential data points were assessed with standardized residuals and Cook's distance, respectively. All analyses were performed with statistical software,i and values of P < 0.05 were considered significant.

Results

Animals—Sixty dogs were enrolled in the study and included 25 spayed females, 4 sexually intact females, 26 castrated males, and 5 sexually intact males with a median age of 10.5 years (range, 2 to 17 years). Pathological conditions of the study dogs included diabetes (n = 16), acute kidney injury (3), chronic kidney disease (3), immune-mediated hemolytic anemia (3), hyperadrenocorticism (3), cystolithiasis (2), pancreatitis (2), gastroenteritis (2), heartworm disease (2), and acute hepatopathy, adrenal tumor, anal gland adenocarcinoma, chronic colitis, chronic hepatitis, chronic lymphocytic leukemia, endocarditis, fibrosarcoma, hemangiosarcoma, hepatocellular carcinoma, hyperparathyroidism, hypoadrenocorticism, inflammatory bowel disease, mast cell tumor, allergic skin disease, mitral valve disease, optic neuritis, primary hypertension, pyelonephritis, suppurative hepatitis, tracheal collapse, transitional cell carcinoma, ureterolithiasis, and urinary incontinence (1 each).

Urine sample analysis—Urine osmolality did not differ by > 5% for any of the duplicate measures, and the interpretation of the urine dipstick results did not differ between the medical technologist and the automated analyzer for any of the urine samples evaluated. The categorized urinalysis results for various urine solutes were summarized (Table 1). Among the urine solutes evaluated, only bilirubin had a significant (P = 0.001) correlation with urine osmolality and USG (Table 2). Urine osmolality and USG had an excellent correlation (r = 0.87; P < 0.001).

Table 1—

Number of dogs within each concentration category for the urine solutes evaluated during urinalysis for 60 dogs that were examined by the internal medicine service at a veterinary teaching hospital.

 Concentration category 
Urine solute00.51234Total
Protein12414194558
Glucose433212859
Ketones483225060
Bilirubin3813360060
Hemoglobin17191436059

For each dog, a urine sample was obtained by cystocentesis for diagnostic purposes, and urinalysis and determination of urine osmolality and USG were performed. For each solute, urinalysis results were categorized on a scale of 0 to 4, in which 0 indicated that the solute was absent, 0.5 indicated that trace amounts of the solute were present, and categories 1 to 4 indicated increasing amounts of the solute. The total number of dogs for some solutes was < 60 because of missing data.

Table 2—

Correlation between urine osmolality or USG and each of the following: urine pH, proteinuria, glucosuria, ketonuria, bilirubinuria, and hemoglobinuria as determined by calculation of the Kendall rank-correlation coefficient (τ) for the dogs of Table 1.

 Urine osmolalityUSG
VariableτP valueτP value
Urine pH0.090.320.140.16
Proteinuria0.130.180.170.17
Glucosuria−0.050.570.030.81
Ketonuria−0.130.220.000.96
Bilirubinuria0.360.0010.350.001
Hemoglobinuria−0.160.14−0.110.33

See Table 1 for key.

During multivariable linear regression analysis, 2 outlier values were removed from the analysis because they significantly decreased the fit of the model. The best multivariable model for prediction of urine osmolality included USG and ketonuria. Urine osmolality was negatively associated with severity of ketonuria (ie, when USG was held constant, urine osmolality decreased as severity of ketonuria increased; Figure 1). However, the effect of ketonuria on urine osmolality was small, and inclusion of ketonuria in the model did not improve the fit of the model significantly from that of the univariate linear regression model for USG.

Figure 1—
Figure 1—

Scatterplot of urine osmolality versus USG for 60 dogs that were examined by the internal medicine service at a veterinary teaching hospital. The plot includes data points for individual dogs (black dots) and the univariate linear regression line (solid black line; regression equation, urine osmolality = −40,890 + [40,777 × USG]; R2 = 0.87) and 95% prediction bands (black dashed lines) for the association of USG with urine osmolality as well as regression lines for the association of USG with urine osmolality when severity of ketonuria was controlled (mild ketonuria = solid yellow line; moderate ketonuria = solid blue line; and severe ketonuria = solid red line). Notice that the regression line for the association between urine osmolality and USG moved slightly to the right as the severity of ketonuria increased.

Citation: American Journal of Veterinary Research 74, 12; 10.2460/ajvr.74.12.1542

Discussion

Results of the present study indicated that there was a strong positive correlation between urine osmolality and USG for dogs with various medical conditions that were examined by the internal medicine service at a veterinary teaching hospital. Ketones had a small negative association on the relationship between urine osmolality and USG; however, because the fit of the model that included ketonuria was not significantly improved from that of the univariate linear regression model for USG, the effect of mild to moderate ketonuria on urine osmolality is not likely to be clinically relevant. In the present study, proteinuria, glucosuria, bilirubinuria, and hemoglobinuria did not significantly affect the relationship between urine osmolality and USG.

Results of a similar study10 in which urine samples from human patients were analyzed indicate that ketonuria, bilirubinuria, and hemoglobinuria significantly affect the association between urine osmolality and USG when USG is determined by a refractometer. In the present study, bilirubinuria and hemoglobinuria had no effect on the relationship between urine osmolality and USG in dogs, despite the fact that 22 of 60 (37%) and 42 of 59 (71%) study dogs had bilirubinuria and hemoglobinuria, respectively. The majority (16/22 [73%]) of dogs had only trace or moderate bilirubinuria, which is considered clinically normal in dogs, especially males.12 Similarly, the majority (33/42 [79%]) of dogs had only trace or moderate hemoglobinuria. Given that the urine samples were collected by cystocentesis, it is possible that some were contaminated with small amounts of blood, and the hemoglobinuria was an artifact of that contamination and might not have been present had the samples been obtained by another method. Regardless, the extent of bilirubinuria or hemoglobinuria in the dogs of this study had no effect on the relationship between urine osmolality and USG, and studies with a larger study population are warranted to further refine the effect of bilirubinuria and hemoglobinuria on urine osmolality.

Surprisingly, glucosuria had no effect on the relationship between urine osmolality and USG in this study, despite the fact that most of the dogs with glucosuria had greater than trace amounts of glucose in their urine. Glucose is a large molecule commonly found in abundance in the urine of diabetic patients (all 16 study dogs with diabetes had glucosuria), and correction factors for urine osmolality for dogs and cats with glucosuria have been proposed.8 In human patients, glucosuria had no effect on the relationship between urine osmolality and USG when USG was measured with a refractometer; however, when USG was measured with a urine dipstick, glucosuria had a significant positive association on the relationship between urine osmolality and USG.10

A limitation of the present study was that the laboratory reported the concentration of protein, glucose, ketones, bilirubin, and hemoglobin in the urine as ordinal variables; however, the statistical model used accounted for this. Other study limitations were associated with interpretation of urine dipstick results. Pigmenturia, high alkaline pH, high USG, and environmental contaminants (ie, quaternary ammonium salts or chlorhexidine) interfere with the detection of proteinuria, and the dipstick assay detects albumin with greater sensitivity than it does globulins or other cellular proteins.13 A few nonphysiologic factors such as bleach, hydrogen peroxide, and exogenously administered ascorbic acid can interfere with the detection of glucosuria.13 In regard to ketonuria, the urine dipstick is most sensitive for detection of acetoacetate, is slightly less sensitive for detection of acetone, and is insensitive for detection of β-hydroxybutyrate13; additionally, an acidic urine pH or high USG can interfere with interpretation of ketonuria,13 which is why another diagnostic test for ketones was used in the present study to confirm the dipstick results. Similarly, although few things interfere with detection of bilirubinuria by the urine dipstick,13 urine samples were analyzed for the presence of bilirubin with another diagnostic test to confirm the dipstick results. A high USG and nonphysiologic factors such as bleach, captopril, ascorbic acid, and formaldehyde interfere with the dipstick results for hemoglobin.13 Finally, the small number of study dogs with bilirubinuria and hemoglobinuria limited the power to detect whether either of those solutes had a significant effect on the association between urine osmolality and USG.

In this population of dogs with various pathological abnormalities, USG as determined by refractometry was strongly correlated with urine osmolality. This correlation was negatively associated with ketonuria but was unaffected by proteinuria, glucosuria, bilirubinuria, or hemoglobinuria. Additional studies with larger study populations are warranted to further refine the predictive model for determining urine osmolality in dogs with pathological abnormalities.

ABBREVIATIONS

USG

Urine specific gravity

a.

Heska-Vet, Heska Corp, Loveland, Colo.

b.

Veterinary Clinical Pathology Laboratory, Louisiana State University, Baton Rouge, La.

c.

Multistix 10 SG, Siemens AG, Erlangen, Germany.

d.

Clinitek Status +, Siemens AG, Erlangen, Germany.

e.

American Society for Clinical Pathology, Chicago, Ill.

f.

Acetest, Siemens AG, Erlangen, Germany.

g.

Ictotest, Siemens AG, Erlangen, Germany.

h.

Osmette, Precision Systems Inc, Natick, Mass.

i.

R, version 3.0.1, R Foundation for Statistical Computing, Vienna, Austria. Available at: www.r-project.org/. Accessed May 8, 2013.

References

  • 1. Wellman ML, DiBartolla SP. Applied physiology of body fluids in dogs and cats. In: DiBartolla SP, ed. Fluid, electrolyte, and acid-base disorders in small animal practice. 3rd ed. St Louis: Saunders Elsevier, 2006;325.

    • Search Google Scholar
    • Export Citation
  • 2. Ganong WF. The general and cellular basis of medical physiology. In: Ganong WF, ed. Review of medical physiology. 22nd ed. Boston: McGraw-Hill, 2005;150.

    • Search Google Scholar
    • Export Citation
  • 3. Thrall MA, Weiser G, Allison R, et al. Urine concentration. In: Thrall MA, ed. Veterinary hematology and clinical chemistry. 2nd ed. Ames, Iowa: Wiley-Blackwell, 2012;333335.

    • Search Google Scholar
    • Export Citation
  • 4. Chadha V, Garg U, Alon US. Measurement of urinary concentration: a critical appraisal of methodologies. Pediatr Nephrol 2001; 16:374382.

    • Search Google Scholar
    • Export Citation
  • 5. Stevens LA, Lafayette RA, Perrone RD, et al. Laboratory evaluation of kidney function. In: Schrier RW, ed. Diseases of the kidney and urinary tract. 8th ed. Philadelphia: Lippincott Williams & Wilkins, 2006;299335.

    • Search Google Scholar
    • Export Citation
  • 6. Paris JK, Bennett AD, Dodkin SJ. Comparison of a digital and an optical analogue hand-held refractometer for the measurement of canine urine specific gravity. Vet Rec 2012; 170:463.

    • Search Google Scholar
    • Export Citation
  • 7. Dossin O, Germain C, Braun JP. Comparison of the techniques of evaluation of urine dilution/concentration in the dog. J Vet Med A Physiol Pathol Clin Med 2003; 50:322325.

    • Search Google Scholar
    • Export Citation
  • 8. DiBartolla SP. Applied physiology of body fluids in dogs and cats. In: DiBartolla SP, ed. Fluid, electrolyte, and acid-base disorders in small animal practice. 3rd ed. St Louis: Saunders Elsevier, 2006;4779.

    • Search Google Scholar
    • Export Citation
  • 9. Leech S, Penney MD. Correlation of specific gravity and osmolality of urine in neonates and adults. Arch Dis Child 1987; 62:671673.

  • 10. Imran S, Eva G, Christopher S, et al. Is specific gravity a good estimate of urine osmolality? J Clin Lab Anal 2010; 24:426430.

  • 11. Voinescu GC, Shoemaker M, Moore H, et al. The relationship between urine osmolality and specific gravity. Am J Med Sci 2002; 323:3942.

    • Search Google Scholar
    • Export Citation
  • 12. Fry MM. Urinalysis. In: Bartges J, Polzin DJ, eds. Nephrology and urology of small animals. Chichester, West Sussex, England: Wiley-Blackwell, 2011;4661.

    • Search Google Scholar
    • Export Citation
  • 13. Stockham SL, Scott MA. Urinary system In: Stockham SL, Scott MA, eds. Fundamentals of veterinary clinical pathology. 2nd ed. Ames, Iowa: Blackwell, 2008;458467.

    • Search Google Scholar
    • Export Citation

Contributor Notes

Address correspondence to Dr. Ayoub (jenaayoub@yahoo.com).

Dr. Ayoub's present address is Palm Beach Veterinary Specialists, 3884 Forest Hill Blvd, West Palm Beach, FL 33406. Dr. Beaufrere's present address is Health Sciences Center, Ontario Veterinary College, University of Guelph, Guelph, ON N1G 2W1, Canada.

The study was performed in its entirety at the School of Veterinary Medicine, Louisiana State University, Baton Rouge, LA 70803.

The authors thank Dr. Stephen D. Gaunt for technical assistance.