Abstract
OBJECTIVE To examine the effect of 24 hours of refrigeration on urine samples collected from dogs with signs of urinary tract infection (UTI).
DESIGN Prospective cross-sectional study.
ANIMALS 104 dogs with signs consistent with UTI that had a urine sample collected via cystocentesis as part of their diagnostic workup.
PROCEDURES A 1-mL aliquot of each urine sample was refrigerated at 5°C for 24 hours in a plain glass tube, then processed for quantitative bacterial culture (QBC). A 0.5-mL aliquot was added to 3 mL of tryptic soy broth (TSB) and refrigerated at 5°C for 24 hours, then processed for QBC. The remaining portion was immediately processed for QBC, with results reported as numbers of bacterial colony–forming units (CFUs). Sensitivity of the QBC for detection of bacteria (and therefore UTI) was determined for sample refrigeration in the 2 conditions, compared with immediate processing (reference standard).
RESULTS Bacterial growth was identified in 35.6% (n = 37), 33.7% (35), and 31.7% (33) of the immediately processed, refrigerated, and refrigerated-in-TSB urine samples, respectively. Sample refrigeration without TSB resulted in no significant difference in CFU counts relative to immediate processing; however, the sensitivity of this method was 95% (35/37). Sample refrigeration with TSB resulted in significantly lower CFU counts, and sensitivity was only 89% (33/37).
CONCLUSIONS AND CLINICAL RELEVANCE Canine urine samples collected for bacterial culture should be immediately submitted for testing. Although CFU counts for refrigerated and immediately processed samples were statistically similar in this study, sample refrigeration in enrichment broth resulted in imperfect sensitivity for UTI detection and is not recommended.
Urinary tract infections are common in dogs. Studies1,2 have shown that between 14% and 26% of all dogs will develop at least 1 UTI during their life. Nevertheless, clinical signs and urinalysis findings are nonspecific, and most urine samples submitted for bacterial culture fail to yield bacterial growth.3 This makes it difficult to accurately predict which dogs have an actual infection and not another urinary condition such as inflammation. For this reason, as well as for appropriate antimicrobial stewardship, the diagnosis of a UTI requires that a QBC be performed on urine collected by cystocentesis.4
Studies in the human and veterinary literature have shown that the storage and handling of urine samples can influence the results of QBC. In a small study,5 canine urine samples (n = 24) that were inoculated with bacteria and stored at room temperature (21° to 25°C) for 2 hours prior to culture yielded results that were identical to those for samples in which culture was performed immediately following inoculation; however, following storage for 24 hours, the samples yielded both false-negative (4%) and positive (50%) results.5 The investigators suggested that false-negative results were possible for canine urine samples refrigerated for 24 hours in plain glass tubes. In a larger study6 involving human urine samples submitted to a hospital microbiology laboratory, investigators concluded that such samples could be stored only at room temperature for no more than 2 hours before QBC results would be affected.
Given the limited information available for dogs, it has been recommended that urine samples collected for QBC be inoculated onto culture media immediately after collection7; however, such practice would be impractical for many practitioners. In the authors' experience, urine samples in most veterinary clinics are collected and stored in refrigerated conditions in plain glass tubes until they are transported by courier to a remote commercial laboratory. This practice may affect the diagnostic accuracy of the QBC.
Previous veterinary studies of the effect of urine storage on QBC results have involved small numbers of experimentally inoculated urine samples. The purpose of the study reported here was to use urine samples collected from dogs with suspected UTI. Specifically we wanted to determine whether storage of urine samples in refrigerated conditions in plain glass tubes for 24 hours would result in CFU counts that were statistically different from those for samples immediately inoculated onto culture media. A secondary aim of the study was to evaluate the effect of adding urine to standard TSB prior to sample refrigeration on QBC results. We hypothesized that storage of urine samples in refrigerated conditions would result in false-negative results and decrease CFU counts and that use of an enrichment broth could improve results.
Materials and Methods
Dogs
Dogs brought to the Louisiana State University Small Animal Hospital from May through December of 2013 with clinical signs consistent with a UTI and without current antimicrobial treatment were considered for inclusion in the study. Such signs included pollakiuria, stranguria, dysuria, hematuria, and inappropriate urination. Only dogs that had a sample of urine collected aseptically by cystocentesis as part of their diagnostic workup were eligible to participate in this study. No dog or sample could be included more than once. The study protocol was approved by the Louisiana State University Clinical Protocol Committee and the university's Animal Care and Use Committee.
Sample collection
After owner consent was obtained, urine samples were collected for general diagnostic testing, and 2 mL of each sample was diverted to this study specifically. Urine samples for the study were placed in separate, sterile glass tubesa labeled with a patient identification number.
Immediately after collection, each urine sample was processed by 1 of 2 investigators (MP and KW). Samples that could not be processed at the time of collection were discarded. With strict aseptic technique used, each 2-mL urine sample was divided into aliquots. A 1-mL aliquot was placed in a plain glass tubea and refrigerated at 5°C for 24 hours, after which it was processed for QBC. A 0.5-mL aliquot was added to 3 mL of TSBb and refrigerated at 5°C for 24 hours, after which it was processed for QBC. The remaining 0.5-mL aliquot was immediately processed for QBC.
QBC
For each aliquot of each sample, 5 serial dilutions were made (1:10 to 1:100,000). Five blood-supplemented tryptic soy agarc plates and 5 MacConkey agard plates were each divided into 3 sectors. Each plate was inoculated such that 50 μL of the same dilution was applied to each sector by use of standard spread plate techniques. Thus, there was 1 blood-supplemented tryptic soy agar plate and 1 MacConkey agar plate for each serial dilution, and the dilution was applied 3 times, once in each sector. Plates were placed in a standard bench top incubator for 24 hours at 37°C. After 24 hours, all plates were examined, and the plate that had colony counts in the range of 30 to 300 was selected. Colonies were counted in each sector of that plate and averaged to provide a final CFU count for the particular sample. In situations in which the 1:100,000 plate had > 300 colonies, the value was recorded as too numerous to count. In all situations, except when Proteus spp were recovered, the blood agar plate was used for quantification.
No blinding was performed to prevent the individual performing the colony counts and microorganism identification from knowing the results for the other samples, so those individuals may have known the results for the immediately processed samples while reading the results for the refrigerated and refrigerated-in-TSB samples.
Bacterial isolate identification
In each situation in which bacterial growth was identified, the responsible microorganisms were further identified by standard microbiological techniques. The bacteria of interest were those most commonly implicated in UTIs in dogs: Escherichia coli, Enterococcus spp, Staphylococcus spp, Streptococcus spp, Proteus spp, Klebsiella spp, and Pseudomonas spp. In this process, samples yielding bacterial growth were inoculated onto a MacConkey agar plate by use of a loop calibrated to deliver 0.001 mL. Conventional streaking technique for colony isolation was used. Plates were incubated for 24 hours at 37°C. Gram staining was performed. For gram-negative lactose fermenters, an indole test was performed to differentiate E coli from Klebsiella spp, and a gelatinase test was performed to distinguish Proteus spp from Pseudomonas spp. For gram-positive bacteria, a catalase test was performed to differentiate Staphylococcus spp from Streptococcus spp and a bile-esculin test was performed to differentiate Enterococcus spp from Streptococcus spp.
Statistical analysis
Distribution of the continuous data (CFU counts) was evaluated by assessment of Shapiro-Wilk test results, skewness, kurtosis, and q-q plots. Because the data were not normally distributed and a high proportion of samples yielded no growth under all 3 conditions, nonparametric methods were used to analyze the data, which are reported as the median, 10th to 90th percentiles, and range. Percentages of samples with positive results are reported with 95% CIs.
The χ2 test was used to determine whether culture results were associated with sex. The Cochran Q test was used to determine whether a difference existed among the 3 sample conditions in categorical bacterial growth results. The Friedman test for repeated measures was used to determine whether differences existed between conditions in absolute CFU counts. The Wilcoxon paired rank test was used to compare pairs of conditions. The Kruskal-Wallis test was used to determine whether an association existed between CFU count and the isolated bacterial species.
Sensitivity and specificity of QBC for detection of bacteria in refrigerated or refrigerated-in-TSB samples were calculated by use of the results for immediately processed samples as the reference standard. Statistical softwarec was used to analyze the data. A value of P < 0.05 was used to indicate a significant difference.
Results
Urine samples from 104 dogs were included in the study and processed for QBC in accordance with the study protocol. Bacterial growth was identified for 37 (35.6%; 95% CI, 26.3% to 44.9%), 35 (33.7%; 95% CI, 23.9% to 42.0%), and 33 (31.7%; 95% CI, 22.1% to 39.8%) immediately processed, refrigerated, and refrigerated-in-TSB samples, respectively. There were no instances in which an immediately processed sample yielded negative culture results but the corresponding refrigerated or refrigerated-in-TSB sample yielded positive results. Bacterial growth was identified for immediately processed urine samples from 13 of 51 (25%; 95% CI, 15% to 38%) males and 24 of 53 (45%; 95% CI, 32% to 59%) females, and this difference between sexes was significant (P = 0.04).
With results for the immediately processed samples as the reference standard, sensitivity of QBC for the detection of bacteria in refrigerated and refrigerated-in-TSB samples was 95% (35/37) and 89% (33/37), respectively. Specificity was 100% (67/67) for both processing methods (ie, there were no false-positive results).
No significant (P = 0.14) difference in the likelihood of bacterial growth was identified among the 3 processing conditions; however, a significant (P < 0.001) difference was identified in CFU counts (Table 1). Specifically, the median CFU count for refrigerated-in-TSB samples (8.03 × 104) was significantly less than the median CFU counts for immediately processed samples (1.39 × 106; P = 0.002) and refrigerated samples (2.34 × 105; P = 0.02). No significant (P = 0.34) difference in CFU counts was identified between immediately processed and refrigerated samples.
Results of QBC (CFU counts) for urine samples obtained via cystocentesis from 37 dogs and immediately processed for QBC or refrigerated at 5°C for 24 hours in a plain glass tube with or without TSB and then processed for QBC.
| Condition | Median | 10th–90th percentile | Range |
|---|---|---|---|
| Immediate processing | 1.39 × 106 | 1.74 × 103–3.34 × 108 | 6.27 × 102–6.00 × 108 |
| Refrigeration | 2.34 × 106 | 7.63 × 102–2.53 × 108 | 0.0–3.70 × 108 |
| Refrigeration in TSB | 8.03 × 104 | 0.04 × 102–1.66 × 108 | 0.0–1.89 × 108 |
Escherichia coli (n = 14 [33%; 95% CI, 19% to 47%]), Staphylococcus spp (13 [31%; 95% CI, 17% to 45%]), Klebsiella spp (7 [17%; 95% CI, 5% to 28%]), Enterococcus spp (5 [12%; 95% CI, 0.1% to 22%]), Pseudomonas spp (1 [2%; 95% CI, 0.1% to 7%]), Proteus spp (1 [2%; 95% CI, 0.1% to 7%]), and Streptococcus spp (1 [2%; 95% CI, 0.1% to 7%]) were recovered from the immediately processed urine samples with positive results. On the basis of these 95% CIs, E coli and Staphylococcus spp were deemed more likely to be recovered than Pseudomonas spp, Proteus spp, or Streptococcus spp. No significant (P = 0.41) association was identified between CFU counts for immediately processed samples and the identified bacterial species.
Discussion
In research5 involving experimentally inoculated canine urine samples, storage time and temperature have been shown to influence the results of QBC. The purpose of the present study was to examine the influence of 24 hours of refrigeration on results of QBC for detection of bacteria (and, hence, diagnosis of UTI) in urine samples collected from dogs with clinical signs of UTI. In addition, we explored the possibility that use of an enrichment broth such as TSB could help mitigate the previously reported adverse effects of refrigeration.
Of the 104 canine urine samples evaluated, immediate processing for QBC yielded bacterial growth in 37 (35.6%) samples. Female dogs were overrepresented among those with positive results, which is consistent with previously reported findings.3 Although the difference in CFU counts for refrigerated and immediately processed samples was not significant, only 35 refrigerated samples yielded bacterial growth, resulting in a sensitivity of 95% and specificity of 100% (ie, no false-positive results, given there was no instance in which bacteria were recovered from refrigerated samples but not from immediately processed samples). Although we had anticipated that refrigeration of urine samples in TSB would improve the sensitivity of QBC involving refrigerated samples, bacteria were detected in 33 samples (sensitivity, 89%), suggesting that TSB did not have the intended effect.
In previous studies, refrigeration of experimentally inoculated canine urine samples affected both the sensitivity of QBC5 and bacterial colony counts.8 In the present study, although sensitivity was affected, CFU counts did not differ significantly between immediately processed and refrigerated (without TSB) urine samples. Important differences exist among these studies. The previous studies5,8 involved collected, pooled, and sterilized canine urine samples that were then inoculated with known qualities of E coli, whereas the present study involved actual clinical samples and no attempt was made to standardize the bacterial concentration of the urine, thereby providing a better representation of what would occur in clinical practice. In addition, the previous studies involved only 1 bacterial species and the present study involved 6. Some strains of bacteria may be less affected by refrigeration; however, the present study was not designed to explore this possibility.
One possible limitation of the study reported here was that the 2 individuals who performed the colony counts were not blinded to previous results. Although this situation could have introduced bias, it was consistent with procedures in similar studies.5,6 In addition, the counting of colonies could be considered fairly objective and less prone to bias than more subjective approaches. Nevertheless, blinding of QBC interpreters would increase the robustness of future studies.
Results of the study reported here suggested that canine urine samples should be immediately processed for QBC. Although no significant difference in CFU counts was identified between refrigerated and immediately processed samples, the sensitivity of QBC to detect bacteria in clinical urine samples (and, hence, to detect UTI) was affected by sample refrigeration. Given that refrigeration of urine samples in enrichment broth resulted in lower CFU counts and lower sensitivity than immediate processing for QBC, this approach is not recommended for the diagnosis of UTI in dogs.
Acknowledgments
Dr. Partyka, who was a second-year veterinary student at the time of the study was supported in part by the Morris Animal Foundation (D14CA-609).
ABBREVIATIONS
| CFU | Colony-forming unit |
| CI | Confidence interval |
| QBC | Quantitative bacterial culture |
| TSB | Tryptic soy broth |
| UTI | Urinary tract infection |
Footnotes
BD Vacutainer, Becton-Dickinson, Franklin Lakes, NJ.
BD Trypticase soy broth, BD Diagnostic Systems, Sparks. Md.
5% sheep blood in tryptic soy agar, Hardy Diagnostics, Santa Maria, Calif.
Hardy Diagnostics, Santa Maria, Calif
SPSS, version 22.0, IBM Corp, Armonk, NY.
References
1 Kivistö AK, Vasenius H, Sandholm M. Canine bacteruria. J Small Anim Pract 1977;18:707–712.
2 Ling GV, Rohrich PJ, Ruby AL, et al. Canine urinary tract infections: a comparison of in vitro antimicrobial susceptibility test results and response to oral therapy with ampicillin or with trimethoprim-sulfa. J Am Vet Med Assoc 1984;185:277–281.
3 Ling GV, Norris CR, Franti CE, et al. Interrelations of organism prevalence, specimen collection method, and host age. sex, and breed among 8,354 canine urinary tract infections (1969–1995). J Vet Intern Med 2001;15:341–347.
4 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. Vet Med Int 2011;2011:263768.
5 Padilla J, Osborne CA, Ward GE. Effects of storage time and temperature on quantitative culture of canine urine. J Am Vet Med Assoc 1981;178:1077–1081.
6 Hindman R, Tronic B, Bartlett R. Effect of delay on culture of urine. J Clin Microbiol 1976;4:102–103.
7 Bartges JW. Diagnosis of urinary tract infections. Vet Clin North Am Small Anim Pract 2004;34:923–933.
8 Patterson CA, Bishop MA, Pack JD, et al. Effects of processing delay temperature, and transport tube type on results of quantitative bacterial culture of canine urine. J Am Vet Med Assoc 2016;248:183–187.
