Comparison of immediate versus delayed streak plate inoculation on urine bacterial culture and susceptibility testing in dogs and cats

Abstract Background Quantitative bacterial culture and susceptibility testing is the gold standard diagnostic for determining bacterial urinary tract infection. Transport of samples to external reference laboratories is common practice in veterinary medicine. Objective To compare bacterial culture and susceptibility results from clinical urine samples when streak plate inoculation is performed immediately after sample collection versus after transport to a reference laboratory. To determine the clinical implications of discrepant culture results. Animals One hundred and ninety‐four canine and 45 feline urine samples that were submitted for urinalysis and urine culture and susceptibility testing. Methods This was a prospective, cross‐sectional study. Streak plate inoculations were performed on urine samples immediately after collection and also after transport to a reference laboratory. Samples were stored in plain sterile tubes and refrigerated up to 24 hours before transport. Culture results were compared, and discordant results were evaluated for clinical relevance. Signalment, comorbidities, lower urinary tract signs, and antimicrobial history were recorded. Results Kappa coefficient for agreement between plating methods was 0.884. Twenty‐two (71%) of 31 discrepant results were determined to have no clinical impact. Though 35% of clean midstream samples had discrepant culture results, only 8% of these had clinical impact. Conversely, 8.6% from cystocentesis were discrepant, but 41% of these had clinical impact. Conclusions and Clinical Importance Provided urine samples are stored and transported appropriately, the immediate preplating of urine for culture and susceptibility testing is unnecessary in the majority of cases. Despite more discrepancies in plating methods for midstream samples, the minority were of clinical importance.


| INTRODUCTION
The emergence of antimicrobial resistance has prompted considerable discussion in the veterinary community regarding responsible antimicrobial usage and culture-guided treatment. Genitourinary tract infections are a common indication for antimicrobial treatment in small animals, representing approximately 12% of all PO antibiotic prescriptions in dogs. 1,2 Unfortunately, antimicrobials are commonly prescribed despite lack of confirmed diagnosis. 3  Immediate inoculation of culture media with urine is recommended for optimal diagnostic accuracy of QBC. 5 However, transport to external laboratories can delay media inoculation by 24 hours or more. [5][6][7] Studies evaluating the impact of storage on canine urine samples have yielded variable results. Though storage at 4 C is recommended if immediate processing is unavailable, some studies have shown a decreased sensitivity of bacterial recovery and an increased (4%) false-negative rate with delayed inoculation despite refrigeration. [5][6][7] Currently, veterinary studies evaluating the impact of delay of urine culture processing are restricted to experimentally inoculated samples or to clinical samples collected via cystocentesis. Guidelines created by the International Society of Companion Animal Infectious Diseases (ISCAID) recommend cultures are preferentially performed on cystocentesis samples, though other studies have demonstrated that voided samples can yield accurate results. [8][9][10] The guidelines recommend urine samples with delayed processing are refrigerated and plated for culture within 24 hours of collection. 9,10 However, there are no studies to confirm that such recommendations, when applied to clinical samples of various collection techniques, preserve the accuracy of QBC testing following transport to an external laboratory.
There is also no investigation into how altered results secondary to delayed processing could impact clinical decision-making.
The primary objective of this study was to compare QBC and susceptibility results from clinical canine and feline urine samples obtained via variable collection techniques when a streak plate inoculation was performed immediately after sample collection or performed following transport to a reference laboratory. We hypothesized that QBC results differ between samples with delayed versus immediate inoculation, both in bacterial species and concentration in colony-forming units (CFU)/mL. A secondary objective was to determine whether such differences would impact decisions regarding antimicrobial treatment. Samples were submitted in sterile, preservative-free, plastic tubes (Greiner Bio-One, Vacuette Tube, Kremsmünster, Austria). Aliquots of urine samples were transferred to Blood Agar Plates followed by transfer to MacConkey (Remel, ThermoFisher Scientific, Waltham, Massachusetts) by using a conventional streaking technique for quantitative colony isolation. Briefly, the plates were labeled with the dog or cat's information, date, and loop size. The loop was dipped into thoroughly mixed urine, and a primary streak was made down the center of the blood agar. With the side of the loop, the plate was streaked by means of a back-and-forth motion. 11 Ten microliter aliquots were inoculated for samples collected via cystocentesis, feline clean midstream catch, or feline catheterization. One microliter aliquots of urine were inoculated for canine clean midstream catch or canine catheterization specimens in accordance with laboratory protocol. 11,12 Plates were incubated aerobically at 35 C before transport to an outside reference laboratory (Marshfield Labs, Marshfield, Wisconsin). The remainder of the sample was refrigerated at 2 C-8 C while awaiting transport to the same reference laboratory. Culture plates were transported at room temperature, and urine specimens awaiting plate inoculation were transported in a refrigerated pack within 24 hours of sample collection. Approximate travel time to the reference laboratory was 5-6 hours after retrieval from the primary institution. Specimens were plated upon arrival to the reference laboratory by using the same protocol. Culture media was evaluated for growth at 24 hours for all samples, and again at 48 and 72 hours for samples collected via cystocentesis. Minimum inhibitory concentrations (MICs) of isolates were determined by broth dilutions (Sensititre, ThermoFisher Scientific, Waltham, Massachusetts). Susceptibility breakpoint interpretations were based on Clinical Laboratory Standards Institute guidelines. 13,14 The corresponding medical records were reviewed. The dog or cat's sex, species, and age at the time of sample collection were recorded. The presence or absence of clinical signs localizing to the lower urinary tract (LUT) was recorded. Clinical signs were defined as stranguria, hematuria, dysuria, malodorous urine, pollakuria, periuria, and urinary incontinence. The administration of any antibiotic within 3 days of sample collection was noted, but these samples were not excluded in the event that antibiotic administration impacted the growth on 1 plating method differently than the other. The presence of endogenous or exogenous immunosuppression was recorded and categorized as diabetes mellitus, hyperadrenocorticism, or iatrogenic from immunosuppressive medications. Immunosuppressive medications included chemotherapeutic agents, corticosteroids, cyclosporine, azathioprine, mycophenolate mofetil, or leflunomide. For corticosteroids to be considered immunosuppressive, the dose must have been ≥1 mg/kg/day. Urinalyses were performed at the same time as urine culture collection. White blood cells (WBCs) in urine sediment were recorded as follows: No WBC/high-powered field (HPF) (score 0); occasional WBC/HPF (score 1); 0-5 WBC/HPF (score 2); 5-20 WBC/HPF (score 3); 20-50 WBC/HPF (score 4); >50 WBC/HPF (score 5). These scoring categories were in accordance with institutional laboratory protocol for WBC/HPF reporting. Bacteriuria was recorded as none, few, moderate, or many bacteria per HPF, according to standardized laboratory protocol.

| MATERIALS AND METHODS
The QBC results were categorized as negative, minimal growth, or marked growth according to previously recommended definitions. 9,15 If obtained via cystocentesis, bacterial growth ≥10 3 CFU/mL was considered to be marked growth. For samples collected via catheter, bacterial counts ≥10 4 CFU/mL in males and ≥10 5 in females were considered to be marked growth. For samples collected via clean midstream catch, ≥10 5 CFU/mL in dogs and ≥10 4 CFU/mL in cats were considered to be marked growth. Any sample with positive bacterial growth but with a bacterial concentration below the criteria for infection, as determined by these definitions, was labeled as minimal growth.
Discrepancies in QBC and susceptibility results were defined as any difference in organism(s) cultured, the bacterial concentration, or the results of the susceptibility panel. A discrepancy was determined to have a clinical impact if differences were present in antibiotic susceptibility profile or bacterial species cultured, as well as if bacterial growth was considered negative or minimal growth with 1 plating method but marked growth with the other. However, if differences in susceptibility profile were within a single MIC dilution, this was not considered to have a clinical impact.

| Statistical analysis
Percent disagreement and percent clinical impact (for those that disagreed) were compared by collection method. Weighted and unweighted kappa agreement was calculated between QBC growth categories (negative, minimal growth, marked growth) of immediate and delayed culture methods. Weighted kappa agreement treated negative-versus-marked growth discrepancies as a stronger disagreement than negative-versus-minimal growth or minimal-versus-marked growth discrepancies. Percent infection by WBC category (<3 and ≥3) as well as percent of LUT signs by WBC category (<3 and ≥3), both including and excluding those with WBC score of 0, were compared.
Percent agreement by whether antibiotics were used or if the dog or cat was immunocompromised was evaluated, and the percent of immediate growth by immunocompromised status, LUT signs, and antibiotic usage were compared. Percentages are reported with 95% confidence intervals (CIs), by using the Agresti-Coull method, 16 and differences are tested for by using the chi-squared test with the N-1 correction. 17 All analyses performed in R version 3.5.1 (July 2, 2018; R Core Team 2018). 18 3 | RESULTS Streptococcus spp.  respectively). However, given that the difference in MIC was only 1 dilution apart, this change was not deemed clinically relevant. Neither antibiotic usage nor immunosuppression significantly impacted agreement between plating methods (P = .94 and P = .35 respectively).

| Study population
Twenty-two (71%) of the 31 discrepancies were deemed to have no clinical impact, but 9 (29%) were deemed to have clinical relevance (

| Urinalysis results
There were significantly lower WBC scores (score of 0-2) in negative culture samples, and significantly higher scores (scores 3-5) in samples with marked bacterial growth (P < .001) (  (Table 3)