Bacterial UTI is estimated to affect 14% of dogs in their lifetime.1 As in humans, Escherichia coli accounts for most (60% to 75%) bacterial UTIs in dogs; gram-positive cocci (Staphylococcus spp and Enterococcus spp) are the next most common causative agents and are responsible for 25% to 35% of bacterial UTIs.2 Most UTI-causing bacteria are susceptible to one or more of the commonly used antimicrobials: penicillin derivatives, potentiated β-lactam antimicrobials, sulfonamides, and fluoroquinolones.2 Detection of proteinuria, hematuria, pyuria, or bacteriuria on urine sediment examination suggests bacterial cystitis1–3; however, the specificity and sensitivity of urine sediment examination alone for the diagnosis of UTIs is low. Aerobic microbiological culture performed on a urine sample collected via cystocentesis remains the gold standard for diagnosis.2 In ideal circumstances, antimicrobial susceptibility testing would be used to guide treatment of UTIs1–4; however, the timeliness of diagnosis and treatment is also important.
Initial treatment of uncomplicated bacterial UTIs is often empirical because of the time required to identify the causative bacteria and antimicrobial susceptibility patterns with traditional methods. Multidrug-resistant bacteria are emerging as a major health concern, impacting choices for empirical treatment. For example, a high proportion of pathogenic E coli recovered from dogs are resistant to commonly used antimicrobials, such as amoxicillin and amoxicillin trihydrate–clavulanate potassium.4 A rapid patient-side point-of-care test would facilitate treatment decisions based on a confirmed diagnosis of bacterial UTI and identification of antimicrobial resistance patterns of the responsible organism. Additional benefits to in-clinic testing include minimizing sample deterioration between sample collection and laboratory testing, decreasing the number of false-negative test results, and, potentially, improving cost-effectiveness.3
A major disadvantage of performing bacteriologic culture of urine in private practice is that the culture of infectious, pathogenic organisms results in biomedical waste that must be properly decontaminated before disposal. When urine samples or isolates are handled improperly, environmental contamination can occur as well as colonization or infection of personnel and patients and promotion of antimicrobial resistance. Another potential disadvantage, depending on the frequency with which cultures are attempted, might be maintenance of proper storage conditions and adherence to product expiration dates.
In 2002, the Statens Serum Institute developed a benchtop point-of-care diagnostic test to identify bacterial UTI in humans as well as to test antimicrobial susceptibility of the responsible organism.5 A similar CCSP is now marketed for veterinary use. To the authors’ knowledge, the CCSP has not been validated for use in dogs. The purpose of the study reported here was to determine the diagnostic accuracy of the CCSP in dogs for the identification of uropathogenic bacteria and for testing antimicrobial susceptibility. We hypothesized that the CCSP would have > 90% sensitivity and specificity for identification of urinary bacteria and in testing antimicrobial susceptibility of those bacteria to 5 classes of antimicrobial agents: ampicillin, amoxicillin trihydrate–clavulanate potassium, cephalothin, enrofloxacin, and trimethoprim-sulfamethoxazole. The sensitivity and specificity cutoff was extrapolated from a validation study6 of the CCSP used in human medicine, in which the overall concordance with minimal inhibitory concentrations for antimicrobial susceptibility testing was 95%.
Materials and Methods
Specimens—The study had 2 phases. In phase 1 (preliminary validation of the CCSP), 62 frozen, previously characterized bacterial isolates from canine urine cultures were used. In 2011, all clinical urinary bacterial isolates from the University of Tennessee Veterinary Medical Center (n = 418) were saved and stored at −80°C. Patient histories for the stored isolates, including whether the animal was receiving antimicrobials at the time of urine sample collection, were unknown. The speciation and antimicrobial susceptibility test results were determined before freezing with standard benchtop biochemical analysis and standard antimicrobial susceptibility testing guidelines established by the Clinical and Laboratory Standards Institute.7 In our laboratory, the most common bacterial isolates among the archived isolates were E coli (50%), Enterococcus spp (23%), Staphylococcus pseudintermedius (14%), Proteus mirabilis (13%), Pseudomonas aeruginosa (6%), Klebsiella pneumoniae (5%), and Enterobacter cloacae (2%).
For phase 2 (diagnostic evaluation of the CCSP), urine samples from dogs evaluated at the University of Tennessee Veterinary Medical Center between January 2012 and October 2012 were eligible for inclusion. The urine samples were required to have been collected by cystocentesis for SAMC as part of a diagnostic evaluation, with at least 0.5 to 10 mL of that urine sample remaining after SAMC. Dogs believed to have a high probability of a UTI were identified, when possible, before cystocentesis to ensure adequate urine volume was available for the study. Dogs of particular interest included those with signs of lower urinary tract disease, such as stranguria, pollakiuria, dysuria, hematuria, malodorous urine, evidence of pyuria or bacteriuria on urine analysis, or underlying disease and predisposition for UTI, including diabetes mellitus, hyperadrenocorticism, urinary incontinence, polyuria and polydipsia, prostatic disease, urinary bladder neoplasia, cystoliths, and paralysis or paresis. Multiple samples from the same dog were eligible for inclusion when the dog had an ongoing predisposition for developing a UTI. Likewise, samples from dogs receiving antimicrobials were eligible when a strong clinical suspicion of UTI existed. Urinalysis performed concurrently with bacterial culture was preferred but not necessary for the study. Excluded urine samples included those not collected via cystocentesis and samples of inadequate volume. Urine samples that fulfilled the inclusion criteria were sequentially entered into the study. The study protocol was approved by the University of Tennessee Institutional Animal Care Use Committee. Because residual urine samples were used, no client consent was obtained.
CCSP—The CCSPa is an agar plate with a large compartment for quantitative analysis and a chromogenic substrate allowing for bacterial identification and 5 smaller antimicrobial compartments impregnated with ampicillin (predictive of susceptibility to amoxicillin), amoxicillin–clavulanate potassium, cephalothin (predictive of susceptibility to all first-generation cephalosporins, except cefazolin), enrofloxacin, and trimethoprim-sulfamethoxazole.1,5,6 The CCSP can be used to screen for the bacteria that most commonly cause UTIs in dogs, including E coli, Staphylococcus spp, Enterococcus spp, Klebsiella spp, Enterobacter spp, Proteus spp, and Pseudomonas aeruginosa. The chromogenic substrate within the agar causes the bacterial colonies or agar gel to assume different colors, depending on the bacterial species. For example, E coli colonies appear reddish brown, whereas Klebsiella spp colonies appear dark blue. Most bacterial uropathogens have a unique appearance on the CCSP, which facilitates identification (Appendix).
After receipt from the distributor, each CCSP was stored at 2° to 8°C until used. Urine samples (up to 10 mL) were plated onto a CCSP in accordance with the manufacturer's instructions.5 Briefly, each sample was poured onto the plate, and after a few seconds, the sample was decanted so that runoff was directed away from the large compartment. The plates were inverted, incubated in an aerobic incubator with ambient atmospheric conditions at 35°C, and examined 18 to 24 hours after plating. Bacterial identification was determined on the basis of manufacturer guidelines.5 Bacterial quantification was based on the amount of growth in the large compartment. Results used to estimate the concentration of CFUs were categorized by order of magnitude as follows: no colonies present on the CCSP, no growth; < 15 colonies, < 103 CFUs/mL; 15 to 20 colonies, 103 CFUs/mL; semiconfluent growth, 105 CFUs/mL; and confluent growth, 107 CFUs/mL.5 When > 1 bacterial species were present, each species was quantified separately. Contamination was assumed to be present when only 1 or 2 colonies were present.
Antimicrobial susceptibility was determined for similar-appearing isolates with ≥ 103 CFUs/mL in the quantitative compartment. The isolate was recorded as susceptible to a particular antimicrobial when there was no growth in an antimicrobial compartment. When growth in an antimicrobial compartment was less than that in the quantitative compartment, the isolate was recorded as having intermediate susceptibility to that antimicrobial. When growth in an antimicrobial well was equal to that in the quantitative compartment, the isolate was considered resistant to that antimicrobial.5
Preliminary assay validation (phase 1)—One person (RDJ) selectively removed archived bacterial isolates from the freezer in proportion to their relative contribution to the overall incidence of UTIs. For revitalization, isolates were incubated on Columbia agarb with 5% sheep blood in an atmosphere containing 7% CO2 at 35°C for 24 hours. Three to 5 colonies with typical morphologic characteristics for the species were mixed and cultured a second time on the same medium. After 24 hours, 3 to 5 colonies from the second culture were used to make a suspension in sterile saline (0.9% NaCl) solution. The optical density of the suspension was adjusted to that of a 0.5-McFarland opacity standard (corresponding to approx 108 CFUs/mL).
Disk diffusion antimicrobial susceptibility testing was performed and interpreted in accordance with Clinical and Laboratory Standards Institute guidelines.7 The following antimicrobials were tested: ampicillin, amoxicillin–clavulanate potassium, cephalothin, enrofloxacin, and trimethoprim-sulfamethoxazole. Inoculated Mueller-Hinton agar platesb were stacked no more than 2 plates high, incubated in an aerobic incubator with ambient atmospheric conditions at 35°C, and examined after 18 to 24 hours, depending on the particular antimicrobial and bacterial species combination. The person reading the Mueller-Hinton agar plates (RDJ) was unaware of the CCSP results.
A 25-μL sample from the standardized antimicrobial susceptibility testing inoculum was transferred via pipette into a sterile glass test tube containing 5 mL of sterile saline solution to yield approximately 5 × 105 CFUs/mL The inoculated test tubes were mixed for < 5 seconds with a vortex devicec and stored at 2° to 8°C for no more than 8 hours. A second person (SJO) applied the bacterial suspensions to the CCSP as described. The CCSP was inverted, incubated in an aerobic incubator with ambient atmospheric conditions at 35°C for 18 to 24 hours, and then examined. The person reading the CCSP (SJO) was unaware of the standard disk diffusion results. The Mueller-Hinton agar plates and CCSP cultures were photographed at 24 hours, and the CCSPs were photographed at 48 hours if results differed from the 24-hour reading.
Diagnostic validation (phase 2)—One hundred forty-seven urine samples underwent CCSP testing within 30 minutes after collection or were stored at 2° to 8°C for no more than 24 hours. All CCSPs with positive results were photographed at 24 hours and, if results differed from the 24-hour reading, at 48 hours. The SAMC was considered the criterion reference standard, and cultures were performed in accordance with established laboratory protocol. Briefly, a sterile loop calibrated to hold a volume of 0.001 mL was used to inoculate urine onto 3 plates: Columbia agar with 5% sheep blood, colistin–nalidixic acid agar with 5% sheep blood, and MacConkey II agar. Per laboratory protocol, if the dog from which the urine originated was receiving antimicrobials or had antimicrobial treatment discontinued within the previous 48 hours, then brain-heart infusion broth dilutions (1:100 and 1:1,000) were also set up. The inoculated blood-containing plates were incubated at 35°C in an atmosphere containing 5% to 7% CO2. The inoculated MacConkey II plates and brain-heart infusion broth were incubated aerobically at 35°C. Per Clinical and Laboratory Standards Institute recommendations, Enterococcus spp were not tested for susceptibility to cephalothin and trimethoprim-sulfamethoxazole. The person who examined the CCSP (SJO) was blinded to the results of the other test method. All clinical decisions were based on SAMC results at the discretion of the attending clinician.
Statistical analysis—Data analysis was performed with statistical software.d,e For phase 1, the χ2 test was used to evaluate whether bacterial species, presence or absence of bacterial growth, and time of reading (24 vs 48 hours) influenced correct bacterial identification when the CCSP was used. Sensitivity and specificity could not be analyzed, and an ROC curve could not be created for phase 1 data because no negative control specimens were evaluated.
For phase 2, descriptive statistics were calculated. The Pearson χ2 test was used to test the null hypothesis that the CCSP results were no better than those obtained by chance alone. The overall performance of the CCSP to identify growth versus no growth, regardless of correct identification of bacterial organism, was assessed through creation of an ROC curve; an AUC of 1.0 was considered perfect, and an AUC of 0.5 was equal to the null hypothesis. Sensitivity, specificity, positive predictive value, and negative predictive value were calculated. Overall accuracy was calculated as the number of true-positive results plus the number of true-negative results, divided by the total of tests performed. The κ statistic was used to measure the degree of agreement between CCSP and SAMC above that expected by chance alone. The κ statistic has a maximum of 1 with perfect agreement, and other values were interpreted as follows: < 0.2, poor agreement; 0.2 to < 0.4, fair agreement; 0.4 to < 0.6, moderate agreement; 0.6 to < 0.8, good agreement; and 0.8 to 1, very good agreement. For statistical analysis of antimicrobial susceptibility results, isolates classified as susceptible or having intermediate susceptibility were grouped. For pairwise comparisons of antimicrobial resistance versus susceptibility, the 2-tailed Student t test was used. The κ statistic was used to determine the agreement between antimicrobial susceptibility results as determined with the CCSP and with SAMC, and 95% CIs were used to quantify uncertainty. Values of P < 0.05 were considered significant for all tests.
Results
Phase 1—The 62 archived isolates from dogs with a UTI represented 7 bacterial species: E coli (n = 26), Enterococcus spp (12), S pseudintermedius (8), P mirabilis (8), P aeruginosa (4), K pneumoniae (3), and E cloacae (1). At 24 hours after test initiation, 46 of 62 (74%) samples had growth on the CCSP and all samples had growth on SAMC. For the isolates with growth, 45 of 46 (98%) bacteria were correctly identified with the CCSP. All gram-negative bacteria, except E cloacae, had growth with correct bacterial identification on the CCSP at 24 hours. Enterobacter cloacae had growth but was misidentified as E coli with the CCSP. At 48 hours, 53 of 62 (85%) isolates had growth on the CCSP and 49 of 53 (92%) bacteria were correctly identified. The bacterial species (P < 0.001) and whether there was growth or no growth (P < 0.001) were associated with whether bacterial identification was correct. There was no significant (P = 0.145) difference in agreement of correct bacterial identification when the CCSP was examined at 24 and 48 hours.
At 24 hours, 16 of 62 (26%) CCSPs had no growth. Bacteria with no growth on the CCSPs at 24 hours were the gram-positive cocci S pseudintermedius (n = 7) and Enterococcus spp (9). Two of the Enterococcus isolates were interpreted to be contaminants (< 3 white colonies; 1 mm) in the quantitative compartment of the CCSP at 24 hours and were considered to have no growth. At 48 hours, no S pseudintermedius was evident on the CCSP. Seven Enterococcus isolates with no growth on the CCSP at 24 hours had growth at 48 hours; 2 Enterococcus isolates had no growth. Three Enterococcus isolates initially grew as small white colonies instead of the characteristic blue color; 2 isolates had white colonies at both 24 and 48 hours, and 1 isolate had white colonies at 24 hours and blue colonies at 48 hours. There was excellent agreement between 2 observers for bacterial identification (κ = 0.91; P < 0.001).
Phase 2—One hundred seven urine samples collected from 96 dogs (64 spayed females, 27 castrated males, 2 sexually intact females, and 3 sexually intact males) yielded no growth when SAMC was used (ie, true negatives). The mean age of the dogs was 7.9 years (range, 0.3 to 15.0 years), and mean body weight was 19.0 kg (41.8 lb; range, 2.2 to 44.5 kg [4.8 to 97.9 lb]). Breeds represented included mixed (n = 19), Labrador Retriever (9), Standard Poodle (5), Beagle (5), Dachshund (4), Yorkshire Terrier (4), Maltese (3), Boxer (3), Chesapeake Retriever (3), Doberman Pinscher (3), Golden Retriever (3), Jack Russell Terrier (2), Weimaraner (2), Bichon Frise (2), Fox Terrier (2), Giant Schnauzer (2), and 1 each of various other breeds. Use of the CCSPs led to correct identification of all of these urine samples as yielding no bacterial growth, with no false-positive results. Thirteen of the dogs were receiving antimicrobials at the time of bacteriologic culture of urine. Nineteen dogs had been treated with antimicrobials within 2 weeks preceding urine sample collection.
Standard aerobic microbiological culture yielded bacterial growth for 40 urine samples collected from 34 dogs (24 spayed females, 9 castrated males, and 1 sexually intact female). The mean age of the dogs was 8.8 years (range, 2.0 to 15.0 years), and the mean body weight was 22.5 kg (49.5 lb; range, 4.4 to 53.7 kg [9.7 to 118.1 lb]). Breeds represented included mixed (6), Boxer (4), Golden Retriever (3), Labrador Retriever (3), Yorkshire Terrier (2), and 1 each of various other breeds. Two dogs were receiving antimicrobials when their urine sample was collected for SAMC. Three dogs had been treated with antimicrobials within 2 weeks preceding urine sample collection.
Thirty-four samples yielded 1 bacterial species and 6 samples yielded multiple bacterial species through SAMC. Urine samples with only 1 organism identified were found to contain E coli (n = 12), an Enterococcus sp (6), S pseudintermedius (5), P mirabilis (3), P aeruginosa (2), K pneumoniae (2), and Streptococcus spp (2). Two samples yielded 1 bacterium with the SAMC method (Streptococcus sp or E coli) and 2 organisms (E coli and an Enterococcus sp) with the CCSP approach. Excluding these 2 samples, the CCSP yielded growth at 24 hours for 25 of 32 (78%) samples. Regarding the CCSPs with growth, 23 of 25 (92%) samples had correct bacterial identification. Bacterial species misidentified on CCSP included 1 each of Streptococcus spp and E coli; both were identified as Staphylococcus spp.
Heavier microbial growth was the primary difference between CCSPs read at 24 and 48 hours. Of the urine samples in which only 1 organism was identified, 23 of 25 had the same microbial identification and antimicrobial susceptibility pattern at both time points. One E coli isolate was susceptible to amoxicillin–clavulanate potassium at 24 hours and resistant at 48 hours; it was identified as susceptible to this drug with SAMC. One sample had growth of a second organism, an Enterococcus sp, at 48 hours that was not identified on SAMC. Two CCSPs yielded minimal growth at 48 hours (one with 2 colonies of E coli and the other with 1 colony each of E coli and an Enterococcus sp); neither of these samples yielded growth when SAMC was performed.
Six urine samples yielded multiple bacterial species on SAMC; 5 SAMC tests identified 2 organisms, and 1 SAMC test yielded 3 organisms. Bacteria represented on the SAMCs with multiple organisms included Enterococcus spp (n = 3), Staphylococcus spp (2), E coli (4), K pneumoniae (2), and unidentified gram-negative rods (2). Use of the CCSP method led to 4 of 6 urine samples yielding growth of multiple organisms. No change in microbial identification or antimicrobial susceptibility was evident between readings obtained at 24 and 48 hours with the CCSP.
Seven urine samples interpreted as having no growth on CCSP had bacterial growth detected by SAMC: Staphylococcus spp (n = 2), Enterococcus spp (2), P mirabilis (1), E coli (1), and Streptococcus spp (1). Three of these samples yielded < 10,000 colonies on SAMC, including 1 each of E coli, Staphylococcus sp, and Enterococcus sp. One of the Enterococcus isolates grew at 48 hours on the CCSP.
Diagnostic accuracy—The Pearson χ2 statistic was significant (P < 0.001), indicating that the null hypothesis that the CCSP and SAMC had different proportions of urine samples yielding growth and no growth should be rejected. For the CCSP technique, the overall diagnostic accuracy was 94%, with a sensitivity of 81% and specificity of 99%. The positive predictive value was 98% and negative predictive value was 92%. The κ statistic was 0.84 (95% CI, 0.75 to 0.94), which suggested good to very good agreement between results obtained with the CCSP and SAMC. The AUC was 0.75 (95% CI, 0.70 to 0.79; P < 0.001; Figure 1)
Receiver operating characteristic curve of the agreement between results of testing canine urine samples (n = 147) with a CCSP and the reference standard, SAMC. The AUC for the ROC curve is 0.75. The straight line represents the null hypothesis (AUC = 0.5).
Citation: Journal of the American Veterinary Medical Association 243, 12; 10.2460/javma.243.12.1719
Receiver operating characteristic curve of the agreement between results of testing canine urine samples (n = 147) with a CCSP and the reference standard, SAMC. The AUC for the ROC curve is 0.75. The straight line represents the null hypothesis (AUC = 0.5).
Citation: Journal of the American Veterinary Medical Association 243, 12; 10.2460/javma.243.12.1719
Receiver operating characteristic curve of the agreement between results of testing canine urine samples (n = 147) with a CCSP and the reference standard, SAMC. The AUC for the ROC curve is 0.75. The straight line represents the null hypothesis (AUC = 0.5).
Citation: Journal of the American Veterinary Medical Association 243, 12; 10.2460/javma.243.12.1719
Antimicrobials—Comparing antimicrobial resistance versus susceptibility for all organisms and all antimicrobials, the null hypothesis that the urine samples evaluated with the CCSP and SAMC were different populations was rejected (P < 0.001); therefore, antimicrobial susceptibility results of both methods were deemed similar. The κ statistic was 0.41 (95% CI, 0.28 to 0.53), suggesting fair to moderate agreement.
Because of the small sample size, gram-positive cocci (Staphylococcus spp, Enterococcus spp, and Streptococcus spp) and the gram-negative rods (E coli, P mirabilis, P aeruginosa, and K pneumoniae) were grouped for statistical comparison of antimicrobial resistance versus antimicrobial susceptibility to all antimicrobials. For gram-positive cocci, there was significant (P = 0.001) fair to moderate agreement between diagnostic tests (κ = 0.50; 95% CI, 0.24 to 0.76). For gram-negative rods, significant (P < 0.001) agreement was also achieved, but the degree of that agreement was only poor to fair (κ = 0.38; 95% CI, 0.25 to 0.51). When only P mirabilis, P aeruginosa, and K pneumoniae were included in the gram-negative group (ie, without E coli), significant (P < 0.001) fair to moderate agreement was obtained (κ = 0.50; 95% CI, 0.3 to 0.89). When E coli was analyzed alone, agreement was again significant (P = 0.005) but the degree of that agreement was only poor to fair (κ = 0.22; 95% CI, 0.08 to 0.35).
Results of statistical analysis of individual antimicrobials revealed significant agreement between test methods for identification of isolates susceptible to amoxicillin–clavulanate potassium, cephalothin, enrofloxacin, and trimethoprim-sulfamethoxazole; however, results were not in significant agreement for ampicillin (Table 1). Fair agreement between methods was obtained for ampicillin, amoxicillin–clavulanate potassium, and cephalothin; moderate agreement was achieved for enrofloxacin, and very good agreement was achieved for trimethoprim-sulfamethoxazole. The sensitivity and specificity of the CCSP for identification of isolate susceptibility to each antimicrobial could not be determined because of the small number of isolates available for testing.
Comparison of results of antimicrobial susceptibility testing of bacterial isolates obtained from urine samples collected from 34 dogs with a UTI as determined with a CCSP and SAMC (reference standard).
| Antimicrobial | No. of samples | Accuracy (%) | P value (2-tailed t test) | κ statistic | 95% CI | Agreement |
|---|---|---|---|---|---|---|
| Ampicillin | 31 | 58 | 0.14 | 0.22 | −0.06 to 0.50 | Fair |
| Amoxicillin–clavulanate potassium | 30 | 63 | 0.035 | 0.32 | 0.05 to 0.59 | Fair |
| Cephalothin | 24 | 62 | 0.026 | 0.34 | 0.09 to 0.60 | Fair |
| Enrofloxacin | 31 | 71 | 0.004 | 0.43 | 0.17 to 0.69 | Moderate |
| Trimethoprim-sulfamethoxazole | 24 | 95 | < 0.001 | 0.90 | 0.72 to 1.09 | Very good |
The sample size (n = 31) includes all bacterial species that were cultured on both the CCSP and SAMC, regardless of whether the sample had single or multiple bacterial species. There were 40 positive urine samples on SAMC. The varied sample size primarily reflects Clinical and Laboratory Standards Institute recommendations that Enterococcus spp not be tested for susceptibility to cephalothin and trimethoprim-sulfamethoxazole. One Enterococcus sample was not tested for susceptibility to amoxicillin–clavulanate potassium. The κ statistic has a maximum of 1 with perfect agreement, and other values were interpreted as follows: < 0.2, poor agreement; 0.2 to < 0.4, fair agreement; 0.4 to < 0.6, moderate agreement; 0.6 to < 0.8, good agreement; and 0.8 to 1, very good agreement.
Values of P < 0.05 were used to indicate a significant difference between CCSP and SAMC results.
Discussion
The results of the present study suggested that use of a CCSP was specific and accurate for ruling out a diagnosis of UTI in dogs but less sensitive and, therefore, less reliable for diagnosis of UTIs. Additional research is warranted, particularly in regard to optimal growth conditions for gram-positive cocci on CCSPs. No overall significant difference was identified for detection of microbial growth on a CCSP at 24 versus 48 hours, but insufficient isolates were available to determine whether a difference existed between growth of gram-positive cocci and gram-negative rods at these time points.
Identification of an optimal incubation temperature and duration for culture of gram-positive cocci may improve the diagnostic sensitivity of the CCSP. In the first phase of the study, revitalized, previously frozen bacterial isolates were used and all isolates were viable as evidenced by 100% growth when the SAMC method was used. Suspended isolates were refrigerated up to 8 hours before CCSP inoculation, and we assumed that the bacteria remained viable. However, a limitation of that phase of the study was lack of control viability cultures to evaluate the effect of storage on the bacteria, particularly gram-positive isolates. A decrease in viability or nonviable isolates could explain the lack of growth of S pseudintermedius despite 48 hours of incubation, and some Enterococcus organisms required longer than 24 hours to grow on the CCSP. The CCSP is marketed to be used with fresh urine samples; the effect of the revitalization process on bacterial growth on the CCSP is difficult to ascertain. Interestingly, a similar phenomenon of slow growth of gram-positive cocci was also noted during the second phase of the study that involved fresh urine samples; most false-negative CCSP results involved gram-positive cocci. Inoculated CCSPs were incubated overnight at 35°C, which may have accelerated growth, compared with incubation at room temperature; the manufacturer suggests incubation of CCSPs at either temperature.5 Further investigation is required to determine the ideal temperature to facilitate gram-positive cocci growth and whether the sensitivity of the CCSP for detection of for gram-positive cocci could be improved by plate incubation for 48 hours.
In our experience, the CCSP test was easy to set up and interpret. The color of bacterial colonies and agar were the primary data used in bacterial identification. Colony size was less helpful for bacterial identification because in both study phases, there was typically confluent bacterial growth on the CCSPs. The bacterial dilution used for phase 1 was arbitrary, but most samples in phase 2 also had confluent growth, which suggested that natural infections can have a high bacterial load.
In comparisons of antimicrobial susceptibility results between the CCSP and SAMC, the agreements were fair for amoxicillin–clavulanate potassium and cephalothin, moderate for enrofloxacin, and very good for trimethoprim-sulfamethoxazole. Trimethoprim-sulfamethoxazole susceptibility had the greatest concordance. Trimethoprim-sulfamethoxazole is a good empirical choice for treatment of an uncomplicated UTI.4,7 A recent study4 found a relatively low prevalence of resistance to trimethoprim-sulfamethoxazole among E coli isolated from dogs with UTI, presumably because its use is limited because of adverse effects.
Ampicillin was the only antimicrobial for which no significant agreement was identified in susceptibility testing results between the CCSP and SAMC. Most urine samples contained high numbers of bacteria, as suggested by the confluent growth observed on the CCSPs and through quantification with SAMC. It is possible that the large numbers of bacteria overwhelmed the antimicrobials within the agar. Another possibility is that the antimicrobial concentration in the agar degraded faster than expected; instability of ampicillin in agar plates is well known.8 Additional research is required to determine whether better agreement exists for antimicrobial susceptibility results between test methods at different bacterial concentrations within urine samples or with different antimicrobial concentrations within the agar.
The major limitation of the study was the small sample size used for evaluation of diagnostic accuracy of the CCSP. Power analysis revealed that to achieve 80% power, the study would have needed to include 4,000 urine samples with 790 bacteria-positive samples and 790 samples of each bacterial species. A study of that size would have exceeded the financial and time limitations for our study. Because of the lack of power, the study was prone to type II error (failure to detect a difference that truly existed). Another limitation was that the bacterial isolates were not subcultured from the CCSP to verify the antimicrobial susceptibility pattern predicted by SAMC. We assumed that the bacteria did not develop additional resistance between CCSP and SAMC testing because the tests were performed approximately simultaneously. Isolates were subcultured from a limited number of CCSPs (approx 4), and the antimicrobial susceptibility matched the antimicrobial susceptibility predicted by SAMC; ideally all isolates obtained on the CCSPs would have had their antimicrobial susceptibility verified. Lastly, a tertiary referral hospital was used as the source of the study samples, and there may have been a bias toward including samples from dogs with a complicated UTI. Whether the UTIs were complicated was unlikely to influence the results, given that few urine samples yielded multiple bacterial species. The study had insufficient power to assess the accuracy of CCSP for a multifactorial UTI.
Overall, use of the CCSP was highly specific for excluding infection but less reliable for diagnosing infection, particularly infections involving gram-positive cocci. The poor concordance of most of the CCSP antimicrobial susceptibility results with the SAMC results suggested that if the CCSP is used, susceptibility results should be interpreted cautiously. Standard aerobic microbiological culture remains the gold standard for detection of UTI in dogs.
ABBREVIATIONS
| AUC | Area under the curve |
| CCSP | Compartmented bacteriologic culture and antimicrobial susceptibility plate |
| CFU | Colony-forming unit |
| CI | Confidence interval |
| ROC | Receiver operating characteristic |
| SAMC | Standard aerobic microbiological culture |
| UTI | Urinary tract infection |
Flexicult VET, Atlantic Diagnostics, Miami, Fla.
Becton Dickinson Laboratories, Sparks, Md.
Baxter Healthcare Corp, Deerfield, Ill.
SPSS, version 20 for Windows, SPSS Inc, Chicago, Ill.
Analyse-it, version 3.10, Analyse-it Software Ltd, Leeds, West Yorkshire, England.
References
1. Bartges JW. Diagnosis of urinary tract infections. Vet Clin North Am Small Anim Pract 2004; 34:923–933.
2. Littman MP. Diagnosis of infectious diseases of the urinary tract. In: Bartges J, Polzin DJ, eds. Nephrology and urology of small animals. Chichester, West Sussex, England: Wiley-Blackwell, 2011; 241–251.
3. Weese JS, Blondeau JM, Boothe D, et al. Antimicrobial use guidelines for treatment of urinary tract disease in dogs and cats: antimicrobial guidelines working group of the international society for companion animal infectious diseases [published online ahead of print Jun 27, 2011]. Vet Med Intern doi: 10.4061/2011/263768.
4. Boothe D, Smaha T, Carpenter M, et al. Antimicrobial resistance and pharmacodynamics of canine and feline pathogenic E coli in the United States. J Am Anim Hosp Assoc 2012; 48:379–389.
5. Statens Serum Institute. Flexicult Vet Urinary Test. Available at: www.atlanticdiagnostics.com/Flexicult%E2%84%A2Vet/. Accessed Nov 11, 2012.
6. Blom M, Sorensen TL, Espersen F, et al. Validation of FLEXICULT SSI-Urinary Kit for use in primary health care setting. Scand J Infect Dis 2002; 34:430–435.
7. Clinical and Laboratory Standards Institute. Performance standards for antimicrobial disk and dilution susceptibility tests for bacteria isolated from animals. Approved standard M32–A3. 3rd ed. Wayne, Pa: Clinical and Laboratory Standards Institute, 2008.
8. Ryan KJ, Needham GM, Dunsmoor CL, et al. Stability of antibiotics and chemotherapeutics in agar plates. Appl Microbiol 1970; 20:447–451.
Appendix
Characteristics of various bacteria identified with the aid of a CCSP.
| Bacteria | Colony size | Colony color | Agar color |
|---|---|---|---|
| Escherichia coli | 2 to 4 mm | Reddish brown | — |
| Klebsiella and Enterobacter spp | > 2 to 4 mm; fat | Dark blue to purple | — |
| Proteus spp | 2 to 4 mm; swarm | Light brown | Brown |
| Proteus vulgaris | 2 to 4 mm; swarm | Green to brown | Brown |
| Pseudomonas aeruginosa | Large; swarm | Gray-white to greenish | Green |
| Staphylococcus spp | 1 to 2 mm | White to red | — |
| Enterococcus spp | 0.5 to 1 mm | Greenish blue | — |
— = No change in agar color.
