Abstract
Objective:
To determine the effect of urine specific gravity (USG) on using microscopic evaluation of bacteriuria to predict urine culture results in dogs and cats.
Methods:
Urine samples from 481 dogs and 291 cats with paired urinalysis and urine culture results were evaluated retrospectively. The sensitivity, specificity, positive predictive value, and negative predictive value of microscopic bacteriuria for predicting urine culture results were calculated, stratified by USG. Multivariable regression was performed to test the effect of USG, pyuria, hematuria, and species on agreement between microscopic bacteriuria and culture results.
Results:
Microscopic bacteriuria had moderate sensitivity (76% in dogs, 64% in cats) and high specificity (97% in dogs, 96% in cats) for predicting urine culture bacterial growth. Samples with rod bacteria were more likely to have bacterial growth than those with cocci (OR = Infinity, 95% CI 4.8 – Infinity, P <0.001). Agreement was lower in moderately concentrated (OR = 0.44, 95% CI 0.19 – 0.91, P = 0.037) and concentrated samples (OR = 0.47, 95% CI 0.19 – 1.02, P = 0.071) than isosthenuric+hyposthenuric samples (USG ≤1.012). Absence of bacteriuria, pyuria and hematuria had a high negative predictive value for no bacterial growth (96%).
Clinical Significance:
Microscopic bacteriuria has a high specificity in predicting urine culture results, regardless of USG. The finding that microscopic bacteriuria has better agreement with urine culture results in isosthenuric+hyposthenuric urine argues against reflex culture in these samples, especially if pyuria and hematuria are also absent. Urine microscopy can aid clinicians in determining the likelihood of urine culture growth.
Keywords: bacteriuria, hypersthenuric, hyposthenuric, isosthenuric, minimal concentration
Introduction:
Diagnosing a bacterial urinary tract infection (UTI) includes the evaluating the patient for urinary tract signs (stranguria, hematuria, etc.), identification of bacteriuria on urinalysis, and the gold standard finding of bacterial growth on a urine culture. Although the presence of bacteriuria or a positive urine culture is not alone sufficient for diagnosis of bacterial cystitis or pyelonephritis, urinalysis findings may aid the clinician in determining if infection is present. (Weese et al. 2019)
A complete urinalysis is composed of a urine reagents dipstick evaluation, urine specific gravity by refractometer, macroscopic evaluation, and a urine sediment exam by microscopy. Urine microscopy is the commonly used diagnostic to screen for bacteriuria and pyuria, as the urine dipstick method of detecting leukocytes and nitrites are unreliable tools for diagnosing UTI in animals (Klausner et al. 1976, Vail et al. 1986, Bauer et al. 2008). Clinically, the presence or absence of bacteriuria can help aid the clinician in determining if antimicrobial therapy is indicated for a patient demonstrating symptoms suggestive of bacterial cystitis or pyelonephritis while urine culture results are pending. If antimicrobial therapy is indicated, the morphology of bacteria noted may also help to determine which class of antibiotic may be a better empiric choice for that specific patient until culture susceptibility testing has been completed (Weese et al. 2019).
Several studies have compared various methods, namely Gram staining, Wright Giemsa, and wet mount analysis of urine sediment evaluation in cats and dogs to urine culture results (Swenson et al. 2004, Swenson et al. 2011, O’Neil et al 2013). These studies have described the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of their respective laboratories. These studies have shown unstained urine microscopy to have a 76-89% sensitivity, 57-91% specificity, 40-50% PPV, and 96-99% NPV in predicting positive culture results.
Urine specific gravity (USG) may affect urine sediment examination as hyposthenuric urine may dilute bacteria and cellular components. If urine is not prepared via centrifugation to create a concentrated aliquot for microscopic examination, hyposthenuric samples may have false negative microscopy but positive bacterial culture. One study identified the sensitivity of unconcentrated urine samples (USG ≤1.013) in dogs was 58.5% in predicting positive bacterial culture, however concentrated urine samples were not compared in this study (Tivapasi et al. 2009). Medical databases (Pubmed and Google Scholar) queried on 04/01/2021 with the keywords “urine microscopy,” “sediment,” “bacteriuria,” “specific gravity,” “urinary tract infection,” “dog,” and “cat” did not identify any studies evaluating the effect of USG on using urine microscopy findings to predict urine culture results.
The objective of this study was to evaluate the impact of USG on the performance of urine microscopic evaluation of bacteriuria for predicting the urine culture results. We hypothesized that agreement between urine microscopic bacteriuria and culture results is greater in concentrated samples than those with a USG ≤1.012. A retrospective study was designed to evaluate this study objective.
Materials and Methods:
Criteria for case inclusion and Study Design
A retrospective medical record review of canine and feline patients evaluated at Friendship Hospital for Animals between August 2015 and November 2017 was performed to identify patients that had a urinalysis and urine culture submitted simultaneously to a veterinary diagnostic laboratory (IDEXX Labs).
Samples were included in analysis if both the urinalysis and urine culture results were available for review and the collection method was cystocentesis. Samples were excluded from analysis if the urinalysis and urine culture were not performed on the same urine sample, or if only in-house urinalysis and sediment examination was performed. Patients were stratified according to their USG, which was categorized using the International Renal Interest Society (IRIS) guidelines as dilute (USG < 1.008), isosthenuria (USG 1.008 - 1.012), moderately concentrated (USG 1.013 - 1.029 for dogs and 1.013 – 1.034 for cats), and concentrated (USG ≥ 1.030 for dogs and ≥1.035 for cats) (Watson et al. 2015). Hydration status was not available for all patients and was therefore not accounted for.
At least 3mL of urine was collected by various methods including cystocentesis, urine catheterization, and voiding and aseptically transferred to a sterile tube containing no additive. The urine samples were stored (≤8 hours) with refrigeration at a temperature of 46°F (7.7°C) until transport from the hospital to the laboratory. Transport from the hospital to the laboratory was performed with ice packs through a commercial shipper, with transport time of 30-60 minutes. Upon arrival to the facility the urinalysis was performed at room temperature upon receipt. After the urinalysis was performed, the urine was plated for urine culture within 24 hours of receipt and stored in a refrigerator at 2-8°C for any additional testing.
All urinalyses were performed at a commercial veterinary diagnostic lab by trained medical technologists. The urinalysis was performed in four stages; physical examination for clarity and color, urine specific gravity by digital refractometer (Clinitek Atlas), colorimetric test reagent strip (Siemens Multistix) read by an automated urinalysis analyzer, and urine sediment exam. The colorimetric test reagent strip provided results for pH, glucose, ketones, blood/hemoglobin, bilirubin, urobilinogen. All results were validated in accordance with guidelines established by the American Society of Clinical Pathology. Urine sediment exam was performed by trained medical technologists as unstained, wet mount exam by well plating 60 uL of urine and inverted microscopy evaluating for red blood cells (RBCs), white blood cells (WBCs), crystals, bacteria, casts, and epithelial cells. The sediment is examined using the 40X objective. Observed elements are counted and reported in semi-quantitative ranges. WBCs, RBCs, bacteria, epithelial cells, and crystals are reported per high power field (HPF) at 40X. Casts and mucus are reported per low power field at 10X. When comparing results of the reagent strip to the results of microscopy, the leukocytes on the reagent strip was not utilized. Red blood cells were reported as separate independent results; on the colorimetric strip results as blood/hemoglobin, and as seen on sediment exam per HPF. If RBCs are absent on the sediment exam, but the blood/hemoglobin result is positive, this was interpreted as hemoglobinuria or myoglobinuria.
Data recorded from the urinalysis included collection method, urine specific gravity, pH, red and white blood cells per HPF, bacterial presence and morphology of bacteria if noted (rod versus cocci). If bacteria were suspected, an air-dried, stained smear of the sediment was examined for confirmation if needed. The morphology of bacteria present was only reported by the outside veterinary laboratory in samples obtained after April 2017. Hematuria was defined as >5 RBC per HPF and pyuria was defined as >5 WBC/HPF, regardless of urine concentration.
For quantitative urine culture, urine samples were inoculated onto biplates containing trypticase soy agar II with 5% sheep blood and MacConkey II agar as well as into thioglycolate broth. If the patient received recent antibiotic therapy or the urine concentration was equal to or below 1.015, it was considered a low colony count sample and 10 uL of urine were used to inoculate the plates with calibrated loops. For all other samples, 1 uL of urine was used to inoculate the plates. All culture media were incubated at 35°C ± 2° for 24 to 48 hours ± 2 hours at which time bacterial species were determined with matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (Bruker MALDI-TOF Biotyper system). The number and species of bacterial isolates were recorded. For standard urine culture, an approximate concentration (colony forming units (CFU)/mL) of each isolate was determined by individually counting the isolates and multiplying by 1000. For low colony count urine cultures, isolates were individually counted and multiplied by 100. Results were considered positive when ≥ 1,000 CFU/mL were observed. Lack of bacterial growth after 48 hours was considered a negative urine culture until July 2017. There after a lack of bacterial growth after 24 hours was considered a negative culture.
Statistical analysis
Patient age was tested for normal distribution using the D’Agostino-Pearson test and found to be non-normal; it is reported as median and range. The microscopic identification of bacteriuria, pyuria, and hematuria was evaluated in comparison to the urine culture results. For the subset of samples where microscopic bacterial shape was reported, a Fisher’s exact test was used to compare the proportion with positive culture results between shapes (rods versus cocci). Sensitivity, specificity, and positive predictive value (PPV) and negative predictive value (NPV) for these three findings (bacteriuria, pyuria, and hematuria) in all patients were calculated using a 2x2 contingency table. Bacteriuria was further evaluated by stratifying samples according to USG categories (dilute, isosthenuric, moderately concentrated, and concentrated) and calculating sensitivity, specificity, PPV, and NPV from a 2x2 contingency table. The corresponding 95% confidence intervals were calculated using exact binomial confidence limits. A multivariable logistic regression was performed to determine if urine concentration, pyuria, hematuria, or species affected the overall agreement between microscopic bacteriuria and urine culture results. For the logistic regression, dilute and isosthenuric samples urine were combined into an unconcentrated (isosthenuric+hyposthenuric) category (USG ≤1.012, based on low numbers of dilute samples. Overall agreement was defined as concordant results for microscopic bacteriuria and urine culture results (both positive, aka true positive, or both negative, aka true negative). Statistical analyses were performed using R software for statistical computing (R, version 4.1.1. www.r-project.org). P values < .05 were considered significant.
Results:
The medical record database identified 1038 records from August 2015 to November 2017 that were available for evaluation. Two hundred and sixty-six records were excluded from the study. The most common reasons for exclusion were missing urine collection technique (178 samples) or collection by a different method than cystocentesis (65 samples). Nine were excluded due to duplicate records (same urine sample). Three were excluded due to submission of a urinalysis without subsequent urine culture. Three were excluded due to submission of a urine culture without urinalysis. Three of the submissions were cancelled prior to completion of urine culture. Three submissions were not performed at the same laboratory. Two submissions grew probable contaminants defined by the veterinary diagnostic laboratory and were excluded. The remaining 772 samples were included into the study after exclusions; 481 samples from dogs and 291 samples from cats were included. Of the 481 samples from dogs, 352 were from spayed females, 7 were from intact females, 114 were from neutered males, and 8 were from intact males. The median age of the dogs at the time of urine sampling was 10 years (range 0.25-17 years). Of the 291 samples from cats, 174 were from spayed females, 1 was from an intact female, 115 were from neutered males, and 1 was from an intact male. The median age of the cats at the time of urine sampling was 13 years (range 0.33-20.5 years).
Among the 772 samples submitted for urine culture and susceptibility testing, 254 (33%) were positive for bacterial growth. The breakdown of reported CFU/mL were: 172 (68%) ≥100,000, 31 (12%) 50 – 100,000, 29 (11%) 10 – 50,000, and 22 (9%) 1 – 10,000. Among canine samples, 188/481 (39%) were positive for bacterial growth (174 single isolates cultured, 14 mixed isolates cultured). Among the feline samples, 66/291/ (23%) were positive for bacterial growth (59 single isolates cultured, 7 mixed isolates cultured). The bacterium isolated in culture is presented in Table 1. When available, the morphology of bacteria identified via urine microscopy was compared to the morphology of the organism that was cultured. Bacterial morphology was only reported on samples collected after April 2017. The majority of samples included in this study did not have morphology of the bacteria (when present) reported. Out of 772 samples included, bacteriuria was noted in 202 samples (26%). Cocci were reported in 22 samples, rods were reported in 47 samples, and bacteria was reported but morphology was not reported in 133 samples. Out of the 47 samples reporting rods, 44 (94%) cultured a rod organism only, and the 3 remaining samples (6%) cultured a mixed population of rods and cocci bacteria. Of the 22 samples reporting cocci, 10 (45%) cultured a cocci organism only, 4 (18%) cultured a rod organism only, and 8 (36%) did not have any bacterial growth. Overall, the proportion of positive urine culture results in samples with rod bacteriuria was significantly greater than those in samples with cocci bacteria (OR = Infinity, 95% CI 4.8 – Infinity, P < 0.001).
Table 1.
Bacteriologic data from 254 positive urine cultures (188 canine and 66 feline samples with ≥1,000 colony forming unit/mL) out of a study of 772 urine samples (481 canine and 291 feline) with paired urinalysis and urine culture results.
| Dog samples (n = 188) | ||
|---|---|---|
| Single isolate | Mixed growth | |
| E. coli | 99/174 (56.9%) | 12/14 (85.7%) |
| Proteus mirabilis | 34/174 (19.5%) | 3/14 (21.4%) |
| Staphylococcus species | 22/174 (12.6%) | 3/14 (21.4%) |
| Klebsiella pneumoniae | 8/174 (4.6%) | 0/14 (0%) |
| Enterococcus species | 4/174 (2.3%) | 5/14 (35.7%) |
| Enterobacter species | 4/174 (2.3%) | 2/14 (14.3%) |
| Streptococcus species | 3/174 (1.7%) | 2/14 (14.3%) |
| Lactobacillus | 0/174 (0%) | 1/14 (7.1%) |
| Cat samples (n = 66) | ||
| Single isolate | Mixed growth | |
| E. coli | 39/59 (66.1%) | 4/7 (57.1%) |
| Staphylococcus species | 10/59 (16.9%) | 1/7 (14.3%) |
| Enterococcus species | 3/59 (5.1%) | 5/7 (71.4%) |
| Pseudomonas | 3/59 (5.1%) | 0/7 (0%) |
| Streptococcus species | 2/59 (3.4%) | 0/7 (0%) |
| Klebsiella pneumoniae | 1/59 (1.7%) | 1/7 (14.3%) |
| Corynebacterium uralyticum | 1/59 (1.7%) | 0/7 (0%) |
| Alloicoccus | 0/59 (0%) | 1/7 (14.3%) |
Canine Results
Thirty-eight urine samples from dogs were dilute, of which 8 (21%) had a positive culture. There were 65 isosthenuric samples; 23 (35%) had positive cultures. Of 210 moderately concentrated samples, 87 (41%) had a positive culture. And of 168 concentrated samples, 70 (42%) had positive cultures. The sensitivity, specificity, PPV, and NPV of urinalysis identified microscopic bacteriuria for detecting positive culture results in urine samples from dogs is listed in Table 2. The presence of hematuria and pyuria (regardless of the presence of bacteriuria) and its predictive capabilities in urine culture results is also presented in Table 2. Hematuria had lower specificity and PPV than pyuria or microscopic bacteriuria for predicting a positive culture.
Table 2.
Performance of microscopic bacteriuria (stratified by urine specific gravity, USG), hematuria, and pyuria on predicting positive urine culture results (≥1,000 colony forming unit/mL) in 481 urine samples from dogs with paired urinalysis and urine culture results.
| Sensitivity (95% CI) | Specificity (95% CI) | PPV (95% CI) | NPV (95% CI) | |
|---|---|---|---|---|
| Bacteriuria | ||||
| All dogs (n = 481) | 76 (69 - 81) | 97 (94 – 98) | 93 (88 – 97) | 86 (82 – 90) |
| Dilute (USG < 1.008, n = 38) | 100 (63 – 100) | 100 (88 – 100) | 100 (63 – 100) | 100 (88 – 100) |
| Isosthenuric (USG 1.008 – 1.012, n = 65) | 78 (56 - 93) | 100 (92 - 100) | 100 (81 - 100) | 89 (77 - 96) |
| Moderately concentrated (USG 1.013 – 1.029, n = 210) | 77 (67 - 85) | 94 (89 - 98) | 91 (81 - 96) | 85 (78 - 91) |
| Concentrated (USG ≥1.030, n = 168) | 70 (58 - 80) | 97 (91 - 99) | 94 (84 - 99) | 82 (74 - 88) |
| Pyuria | ||||
| 77 (70 – 83) | 93 (89 – 96) | 87 (81 – 92) | 86 (82 – 90) | |
| Hematuria | ||||
| 72 (65 – 78) | 70 (64 – 75) | 60 (54 – 67) | 79 (74 – 84) | |
NPV, negative predictive value; PPV, positive predictive value; USG, urine specific gravity
Feline Results
Only one urine sample from a cat was hyposthenuric, and its urine culture was positive. There were 41isosthenuric samples; 9 (22%) had positive cultures. Of the 178 moderately concentrated urine samples, 50 (28%) had positive cultures. And of the 71 concentrated samples 6 (8%) had positive cultures. The sensitivity, specificity, PPV, and NPV of microscopic bacteriuria to predict positive urine culture results in cats is shown in Table 3. The presence of hematuria and pyuria (regardless of the presence of bacteriuria) and its predictive capabilities in urine culture results is also presented in Table 3. Pyuria had a lower specificity than microscopic bacteriuria for predicting a positive culture, and hematuria had a lower specificity and PPV than both pyuria and microscopic bacteriuria for predicting a positive culture.
Table 3.
Performance of microscopic bacteriuria (stratified by urine specific gravity), hematuria, and pyuria on predicting positive urine culture results (≥1,000 colony forming unit/mL) in 291 urine samples from cats with paired urinalysis and urine culture results.
| Sensitivity (95% CI) | Specificity (95% CI) | PPV (95% CI) | NPV (95% CI) | |
|---|---|---|---|---|
| Bacteriuria | ||||
| All cats (291) | 64 (51 - 75) | 96 (93 – 98) | 84 (71 – 93) | 90 (86 – 94) |
| Dilute (USG <1.008, n = 1) | 100 (2 - 100) | NA | 100 (0 - 100) | NA |
| Isosthenuric(USG 1.008 – 1.012, n = 41) | 89 (52 - 100) | 94 (79 – 99) | 80 (44 – 97) | 97 (83 – 100) |
| Moderately concentrated (USG 1.013 – 1.034, n = 178) | 64 (49 – 77) | 96 (91 – 99) | 86 (71 – 95) | 87 (81 – 92) |
| Concentrated (USG ≥1.034, n = 71) | 17 (0 – 64) | 98 (92 – 100) | 50 (1 – 99) | 93 (84 – 98) |
| Pyuria | ||||
| 71 (59 – 82) | 88 (83 – 92) | 64 (52 – 74) | 91 (87 – 95) | |
| Hematuria | ||||
| 80 (69 – 89) | 42 (35 – 49) | 29 (22 – 36) | 88 (80 – 93) | |
NPV, negative predictive value; PPV, positive predictive value; USG, urine specific gravity
Overall Agreement between Microscopic Bacteriuria and Urine Culture Results
The results of a multivariable regression model to determine the effects of USG, pyuria, hematuria, and species on the overall agreement between microscopic bacteriuria and urine culture results are presented in Table 4. Moderately concentrated and concentrated samples were less likely to have agreement than those with a USG ≤1.012, and samples without pyuria or hematuria were more likely to have agreement than those with these sediment findings. The latter findings were driven by a greater negative agreement; the combined absence of bacteriuria, pyuria, and hematuria had a NPV of 96% (95% CI: 93-98%) for predicting urine culture results. There was no significant effect of species on overall agreement.
Table 4 –
Multivariable logistic regression model to determine the effects of urine concentration, pyuria, hematuria, and species on overall agreement of microscopic bacteriuria with urine culture results in 772 urine samples (481 canine and 291 feline) with paired urinalysis and urine culture results.
| Variable | OR (95% CI) | P value |
|---|---|---|
| Urine concentration | ||
| USG ≤1.012 | Referent | - |
| Moderately concentrated* | 0.44 (0.19 – 0.91) | 0.037 |
| Concentrated* | 0.47 (0.19 – 1.02) | 0.071 |
| Pyuria absent | 2.1 (1.2 – 3.4) | 0.0038 |
| Hematuria absent | 1.8 (1.1 – 3.3) | 0.027 |
| Feline sample | 1.1 (0.69 – 1.8) | 0.68 |
USG, urine specific gravity
Moderately concentrated defined as a USG of 1.013 – 1.029 for dogs and 1.013 – 1.034 for cats. Concentrated defined as a USG ≥ 1.030 for dogs and ≥ 1.035 for cats.
Discussion:
This retrospective study evaluated the sensitivity, specificity, positive and negative predictive values of urine microscopy where bacteriuria was subgrouped by USG, on predicting urine culture results in dogs and cats. In contrast to our hypothesis, results of microscopic evaluation for bacteriuria in moderately concentrated and concentrated samples (USG >1.012) in both dogs and cats had a lower agreement with the urine culture results than unconcentrated samples (USG ≤1.012), though this effect only reached statistical significance for moderately concentrated samples. Pyuria had a similar diagnostic performance to microscopic bacteriuria for prediction of urine culture results in dogs but a lower specificity in cats. Hematuria had lower specificity and PPV than the presence of bacteriuria or pyuria in both dogs and cats. However, the absence of pyuria or hematuria improved the agreement between microscopic bacteriuria and urine culture due to a high NPV of the absence of bacteriuria, pyuria, and hematuria.
Unconcentrated urine, as well as the underlying systemic disorder, have been proposed to increase the risk of developing bacteriuria (Lees et al. 1979). In support of this, in vitro growth of E. coli is greater in canine urine diluted to a USG of 1.010 than in concentrated conditions (Thornton et al. 2018). Urine microscopy in dilute and isosthenuric samples has been suggested to be less reliable in the detection of a urinary tract infection in dogs (Ling et al. 2001) partly due to fewer leukocytes observed on microscopy due to dilution. The hypotonic nature of hyposthenuric urine may cause erythrocyte lysis and lead to false negative microscopic examination (Barsanti & Finco 1979). Although a study reported that microscopic bacteriuria in canine urine samples with a USG <1.013 had a high (98.3%) specificity for identifying urinary tract infection, the sensitivity was poor (58.5%) (Tivapasi et al. 2009). However, the same study suggested that performing a reflex urine culture on dilute urine samples (i.e., automatically culturing any dilute sample) was not cost effective, as only 3.5% of unconcentrated urine samples had false negative urinalysis examination (Tivapasi et al. 2009). A study in cats found no association between USG and risk of positive urine culture and suggested performing a urine culture solely on the basis of isosthenuria is unwarranted (Bailiff et al. 2008).
The results of the present study challenge the suggestion that relying on urine sediment examination alone for prediction of urine culture results is more likely to result in misdiagnosis in unconcentrated (isosthenuric+hyposthenuric) urine samples. Our results demonstrated a greater likelihood of agreement between microscopic bacteriuria and urine culture in isosthenuric and hyposthenuric urine samples, regardless of species. The performance of urine microscopy at different USG requires further evaluation, as many of the 95% CI were wide due to relatively low numbers of samples. The apparent cause of this improved performance compared to prior reports is unknown. It is possible that dilute urine contains fewer cells and less amorphous debris which could interfere with microscopic examination and afford more accuracy.
Several publications have evaluated urine microscopy as a method for identifying urinary tract infection. Studies have shown identifying higher sensitivity of stained air-dried preparations compared to unstained wet mount preparations (Swenson et al. 2004, Swenson et al. 2011, Way et al. 2013). Wright-Giemsa and Gram stain preparations have shown identical performance in identifying bacteriuria in dogs and cats, and stained preparations had a higher PPV (88%) than that of unstained preparations (50%) (O’Neil et al 2013). The present study used wet mount preparations and found a higher PPV in both dogs and cats than previous reports.
The prevalence of positive urine culture in the present study was 39% in dogs and 23% in cats. While medical records were not reviewed to determine the justification for urine culture testing, it is reasonable to assume many of these patients had clinical suspicion for bacterial UTI. Prior studies have reported the prevalence of positive urine cultures in dogs to range from 9.9-62% (Swenson et al. 2004, O’Neil et al 2013, Way et al. 2013, McMeekin et al. 2017) and 6.1-16.5% in cats (Bailiff et al. 2008, Tivapasi et al. 2009, Swenson et al. 2011, O’Neil et al 2013).
The most common bacterial isolate that was cultured in both species was E. coli (59% of dogs, 65% of cats), which is consistent with prior studies. Overall, the bacteria isolated are similar to prior reports in both dogs and cats (Mayer-Roenne et al. 2007, White et al. 2013, Olin & Bartges 2015, Wong et al. 2015).
The morphology of bacteria was reported by the diagnostic laboratory after April 2017. When available, the morphology noted on urine microscopy was compared to the results of the urine culture. When bacilli (rod) bacteria were identified on urine microscopy, 94% of cultures identified rod-shaped bacterial species and the remaining 6% grew mixed rod and cocci bacteria. In contrast, when cocci bacteria were observed, only 45% of cultures grew cocci shaped bacterial species, rod-shaped species were isolated in 18% of these samples, and there was no bacterial growth for the remaining 36%. Overall, cocci bacteria were more likely to be false positives than rod bacteria. The high percentage of negative urine cultures despite the notation of cocci bacteria may be caused by misinterpretation of lipid droplets, cytoplasmic organelles, amorphous crystal, debris with similar shape, or very rare bacteria that would not yield >1,000 CFU/mL when plated for culture. A prior study in dogs found a higher discordance rate between bacterial culture results and bacterial morphology identified on urinalysis with unstained preparation (60.8%) compared to Wright-stained preparations (21.6%) (Swenson et al. 2004). Of those unstained urinalysis microscopic examinations, bacterial morphology was correctly identified in 16.7%, 40.4%, and 53.3% of cultures that yielded growth of cocci, rods, or a mixed population, respectively. Based on the current study, a cocci organism noted on urine microscopy should be interpreted with caution as the corresponding culture is more likely to be negative or isolate a rod-shaped species.
Hematuria had lower specificity and PPV than microscopic bacteriuria in predicting positive urine cultures in both dogs and cats, regardless of urine specific gravity. Pyuria also had a lower specificity than microscopic bacteria in cats. This study did not determine the likelihood of a positive urine culture being associated with bacterial cystitis or pyelonephritis, where hematuria and pyuria would often be observed. Patients with subclinical bacteriuria may or may not have hematuria or pyuria, which may affect the performance of these findings in predicting positive urine culture growth. However, evaluation for pyuria and hematuria can improve interpretation of results for microscopic bacteriuria. Specifically, the absence of bacteriuria, pyuria and hematuria has a high likelihood of a negative urine culture result.
There are several limitations to the present study. Due to the number of patients that were enrolled in the study, clinical history was not able to be reviewed and the association between urine culture results and bacterial cystitis, pyelonephritis, and subclinical bacteriuria was not evaluated. Diagnosing one of these clinical syndromes involves more than urine culture results; patient symptoms, diagnostic imaging, and other biochemical data need to be evaluated to diagnose disease (Weese et al. 2019). Furthermore, due to the lack of clinical history previous antibiotic use was unknown. This poses a limitation as bacteria identified via microscopy may have been dead in these patients, yielding a negative urine culture thereby affecting accuracy and predictive values. Despite the exact number being unknown, submission of a urine culture after antibiotic therapy has been initiated is an uncommon practice, so it is not expected that this scenario would be a common occurrence.
This study used a cutoff of >1,000 CFU/ml to define positive growth, consistent with numerous other studies (Swenson et al. 2004, Swenson et al. 2011, Way et al. 2013). The categorization performed in the current study is only between positive and negative growth. Positive growth in our study should not be considered as diagnostic for UTI, as this is a clinical diagnosis. Clinicians should always interpret quantitative urine culture results in light of the method of collection, patient symptoms, and clinicopathologic data.
Because the study only included one hyposthenuric cat in the analysis, it is possible that the performance of urine microscopy may be different if a larger group of cats were to be evaluated. The presence of clinical signs suggestive of UTI was not evaluated, however this study was focused on performance of urine microscopy to identify positive culture, not to identify subclinical bacteriuria, bacterial cystitis, or pyelonephritis. The morphology of the bacteria seen was only reported after April 2017, and therefore the majority of the patients enrolled in this study were not included in that assessment. The duration of plate incubation was shortened in July 2017, in response to internal data at IDEXX Laboratories suggesting this limited growth of environmental contaminants. However, it is unclear if this change may have resulted in rare false-negative culture results. All the samples were sent to one veterinary diagnostic laboratory, and therefore the results of this study are specific to this laboratory and may not be the same for other populations.
In summary, bacteriuria observed on urinalysis has a high specificity in predicting urine culture results regardless of urine concentrating ability in both cats and dogs. Contrary to our hypothesis, moderately concentrated and concentrated urine samples had lower agreement between urine microscopy and culture results than unconcentrated samples. Predictive values of urine microscopy can aid clinicians and guide antimicrobial treatment decisions pending urine culture results in patients with clinical suspicion of UTI and should be interpreted alongside patient clinical signs, historical findings, and diagnostic imaging results. Culture should still be performed in patients with no microscopic bacteriuria where infection is a possibility. However, in this study, the probability of a negative culture result was very high if samples lacked bacteriuria, pyuria, and hematuria. Cautious interpretation of cocci bacteriuria is recommended.
Acknowledgments
The authors acknowledge and thank the editor and reviewers of JSAP for accepting this manuscript for publication.
Funding source:
Partial support for EF was provided by a National Institutes of Health ORIP K01 Mentored Research Scientist Development Award (K01-OD019912)
Footnotes
Conflict of interest: The authors declare no conflicts of interest.
References:
- Bailiff NL, Westropp JL, Nelson RW, et al. (2008) Evaluation of urine specific gravity and urine sediment as risk factors for urinary tract infections in cats. Veterinary Clinical Pathology 37(3), 317–322 [DOI] [PubMed] [Google Scholar]
- Barsanti JA & Finco DR (1979) Laboratory Findings in Urinary Tract Infections. Veterinary Clinics of North America Small Animal Practice 9(4), 729–748 [DOI] [PubMed] [Google Scholar]
- Bauer N, Rettig S & Moritz A (2008) Evaluation the Clinitek status automated dipstick analysis device for semiquantitative testing of canine urine. Research in Veterinary science 85(3), 467–472 [DOI] [PubMed] [Google Scholar]
- Carter JM, Klausner JS, Osborne CA, et al. (1978) Comparison of collection techniques for quantitative urine culture in dogs. Journal of the American Veterinary Medical Association 173(3), 296–298 [PubMed] [Google Scholar]
- Klausner JS, Osborne CA & Stevens JB (1976) Clinical evaluation of commercial reagent strips for detection of significant bacteriuria in dogs and cats. American Journal of Veterinary Research 37(6), 719–722 [PubMed] [Google Scholar]
- Lees GE, Osborne CA & Stevens JB (1979) Urine: A medium for bacterial growth. Veterinary Clinics of North America Small Animal Practice 9(4), 611–616 [DOI] [PubMed] [Google Scholar]
- Ling GV, Norris CR, Franti CE, et al. (2001) Interrelations of organism prevalence, specimen collection method, and host age, sex, and breed among 8,354 canine urinary tract infections (1969-1995). Journal of Veterinary Internal Medicine 15(4), 341–347 [PubMed] [Google Scholar]
- Mayer-Roenne B, Goldstein RE & Erb HN(2007) Urinary tract infections in cats with hyperthyroidism, diabetes mellitus and chronic kidney disease. Journal of Feline Medicine and Surgery 9(2), 124–132 [DOI] [PMC free article] [PubMed] [Google Scholar]
- McMeekin CH, Hill KE, Gibson IR, et al. (2017) Antimicrobial resistance patterns of bacteria isolated from canine urinary samples submitted to a New Zealand veterinary diagnostic laboratory between 2005-2012. New Zealand Veterinary Journal 65(2), 99–104 [DOI] [PubMed] [Google Scholar]
- Olin SJ & Bartges JW (2015) Urinary Tract Infections: Treatment/Comparative Therapeutics. Veterinary Clinics of North America Small Animal Practice 45(4), 721–746 [DOI] [PMC free article] [PubMed] [Google Scholar]
- O’Neil E, Horney B, Burton S, et al. , (2013) Comparison of wet-mount, Wright-Giemsa and Gram-stained urine sediment for predicting bacteriuria in dogs and cats. Canadian Veterinary Journal 54(11), 1061–1066 [PMC free article] [PubMed] [Google Scholar]
- Sørensen TM, Jensen AB, Damborg P, et al. (2016) Evaluation of different sampling methods and criteria for diagnosing canine urinary tract infection by quantitative bacterial culture. Veterinary Journal 216, 168–173 [DOI] [PubMed] [Google Scholar]
- Swenson CL, Boisvert AM, Gibbons-Burgener SN, et al. (2011) Evaluation of modified Wright-staining of dried urinary sediment as a method for accurate detection of bacteriuria in cats. Veterinary Clinical Pathology 40(2), 256–264 [DOI] [PubMed] [Google Scholar]
- Swenson CL, Boisvert AM, Kruger JM, et al. (2004) Evaluation of modified Wright-staining of urine sediment as a method for accurate detection of bacteriuria in dogs. Journal of the American Veterinary Medical Association 224(8), 1282–1289 [DOI] [PubMed] [Google Scholar]
- Thornton LA, Burchell RK, Burton SE, et al.(2018) The effect of urine concentration and pH on the growth of Escherichia coli in canine urine in vitro. Journal of Veterinary Internal Medicine 32(2), 752–756 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tivapasi MT, Hodges J, Byrne BA, et al. (2009) Diagnostic utility and cost-effectiveness of reflex bacterial culture for the detection of urinary tract infection in dogs with low urine specific gravity. Veterinary Clinical Pathology 38(3), 337–342 [DOI] [PubMed] [Google Scholar]
- Vail DM, Allen TA & Weiser G (1986) Applicability of leukocyte esterase test strip in detection of canine pyuria. Journal of the American Veterinary Medical Association 189(11), 1451–1453 [PubMed] [Google Scholar]
- Watson ADJ, Lefebvre HP & Elliott J (2015) Urine specific gravity. http://www.iris-kidney.com/education/urine_specific_gravity.html [accessed 24 August 2021]
- Way LI, Sullivan LA, Johnson V, et al. (2013) Comparison of routine urinalysis and urine Gram stain for detection of bacteriuria in dogs. Journal of Veterinary Emergency and Critical Care 23(1), 23–28 [DOI] [PubMed] [Google Scholar]
- Weese JS, Blondeau J, Boothe D, et al. (2019) International Society for Companion Animal Infectious Diseases (ISCAID) guidelines for the diagnosis and management of bacterial urinary tract infections in dogs and cats. The Veterinary Journal. 247, 8–25 [DOI] [PubMed] [Google Scholar]
- White JD, Stevenson M, Malik R, et al. (2013) Urinary tract infections in cats with chronic kidney disease. Journal of Feline Medicine and Surgery 15(6), 459–465 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wong C, Epstein SE & Westropp JL (2015) Antimicrobial Susceptibility Patterns in Urinary Tract Infections in Dogs (2010-2013). Journal of Veterinary Internal Medicine. 29(4), 1045–1052. [DOI] [PMC free article] [PubMed] [Google Scholar]