Clinical features and risk factors for development of urinary tract infections in cats

Abstract

The clinical and diagnostic features of 155 cats with urinary tract infection (UTI) and 186 controls with negative urine culture/s were characterized retrospectively (signalment, clinical signs, urinalysis, urine culture, concurrent diseases, lower urinary tract diagnostic/therapeutic procedures). Multivariable logistic regression was used to identify risk factors associated with UTI. Cats of all ages were affected by UTI with no sex/breed predisposition. Lower urinary tract signs were absent in 35.5% of cats with UTI. Pyuria and bacteriuria had sensitivities of 52.9% and 72.9%, and specificities of 85.5% and 67.7% for detection of UTI, respectively. Risk factors significantly associated with increased odds of UTI were urinary incontinence [odds ratio (OR) = 10.78, P = 0.0331], transurethral procedures (OR = 8.37, P <0.0001), urogenital surgery (OR = 6.03, P = 0.0385), gastrointestinal disease (OR = 2.62, P = 0.0331), decreased body weight (OR = 0.81, P = 0.0259) and decreased urine specific gravity (OR = 0.78, P = 0.0055). Whilst not independently significant, renal disease and lower urinary tract anatomic abnormalities improved statistical model performance and contributed to UTI.

Introduction

Infections of the urinary tract are uncommon in cats, with a reported frequency between 1% and 3% of all cases of feline lower urinary tract disorders seen at different referral institutions in North America.^1–3 However, some authors have observed an increased proportional morbidity of urinary tract infection (UTI) in older cats (>10 years of age).^1,4,5

Multiple factors are responsible for maintaining sterility of the urinary tract, including a normal anatomy, mucosal integrity and defense barriers, urethral sphincter function, normal micturition with unidirectional urine flow, specific antimicrobial properties of the urine (osmolality, pH, etc), and local and systemic immunocompetence. ⁶ Infections usually develop secondary to concurrent conditions or procedures that impair one or more of these properties.^6–9 Several studies have identified a relatively high frequency of UTI associated with common feline diseases, such as renal failure,^10,11 hyperthyroidism, ¹¹ diabetes mellitus,^11–13 urolithiasis ¹⁴ or other systemic disorders. ¹⁵ An increased risk for iatrogenic UTI following indwelling urethral catheterization ⁹ and in cats with perineal urethrostomies has been described.^7,16 However, most of these reports have evaluated the prevalence of UTI as a comorbid condition associated with a particular disease or procedure. There have been few studies investigating the clinical features and risk factors for bacterial UTI in the general feline population.^1,8 In one retrospective study of 259 cats with UTI, univariable analyses identified an increased risk for UTI in Siamese cats and a reduced risk in intact male and female cats. ⁸ In another retrospective case-controlled epidemiologic study of 784 cats with bacterial UTI, univariable analyses revealed that Abyssinian cats, spayed female cats and cats older than 10 years had an increased risk of UTI. ¹ To our knowledge, studies comprehensively evaluating potential risk factors for development of UTI in cats using multivariable analyses have not been reported. The objectives of this retrospective case-controlled study were: (i) to characterize clinical and diagnostic features of UTI (signalment, lifestyle conditions, clinical signs, urinalysis and microbiologic findings) in a large population of affected cats admitted to the Michigan State University Veterinary Teaching Hospital (MSU-VTH); and (ii) to identify potential risk factors associated with UTI (any differentiating signalment characteristic, environmental factor, anatomical feature, urine property, concurrent disease or therapy that may cause or contribute to UTI) by comparing such features with non-infected cats (controls) using multivariable logistic regression modeling.

Materials and methods

Criteria for selection of cases

Medical records of cats presented to MSU-VTH between January 1989 and December 2003 that had positive results of a quantitative aerobic urine culture were reviewed. Cases were included if the culture revealed any microbial growth in specimens collected via cystocentesis or ≥1000 colony forming units/ml by urethral catheterization, and a concurrent urinalysis was available.^8,17 For cats with multiple episodes of UTI, the first one was considered for the purposes of the study. Previous or concurrent antibiotic therapy was not an exclusion criterion. Affected cats were stratified by age groups to allow age-group matching of controls. Three different age categories were established: young (<5 years old), middle-aged (5–9 years old) and older (≥10 years old) cats. ¹ A total of 155 cats were included as cases.

Criteria for selection of controls

A total of 186 control cats were identified. These consisted of randomly selected, client-owned patients of the MSU-VTH evaluated during the same time period that had one or more urine cultures performed with no growth and for whom a urinalysis was available. Each control cat was not specifically matched to a case cat. Control cats were randomly accrued and distributed by age group until the proportion in each age group matched that of the affected cats. Previous antibiotic therapy was not an exclusion criterion but no antibiotics were given at the time of culture.

Information collected from affected and control cats included age, breed, sex, body weight, body condition score (thin vs normal vs overweight), diet (dry vs moist or mixed), lifestyle (indoor only vs outdoor exposure), type of household (single cat vs multi-cat), number of prior episodes of lower urinary tract disease, clinical signs, concurrent diseases, prior transurethral procedures, prior urogenital surgery and current therapy. In addition, results of complete urinalysis, urine culture, complete blood count, serum chemistry profile and diagnostic imaging procedures were reviewed.

Lower urinary tract signs were coded as present or absent, and then further specified as gross hematuria, stranguria, pollakiuria, dysuria, periuria, urethral obstruction or urinary incontinence. For the purpose of analysis, concurrent diseases were classified in various non-excluding categories and subcategories, including: lower urinary tract disorders, renal, endocrine, gastrointestinal, hepatobiliary, cardiorespiratory, neurologic, infectious or immune-mediated, neoplastic, or other non-urinary tract diseases. Lower urinary tract disorders refers to any diseases of the lower urinary tract that were present at the time of the study (eg, urolithiasis, masses, anatomic abnormalities, urethral obstruction, urinary incontinence, etc), which in the UTI group would be concurrent with the UTI. Idiopathic cystitis, which is a diagnosis of exclusion and requires a negative urine culture, was only evaluable for control cats. Likewise, control cats with clinical signs of lower urinary tract disease (and obviously a negative urine culture) but lack of further diagnostic evaluation were considered to have undiagnosed lower urinary tract disease. Transurethral procedures were defined as diagnostic or therapeutic interventions that may iatrogenically disrupt the normal anatomy and defense barriers of the urethra and potentially introduce bacteria into the lower urinary tract, and included urethral catheterization, retrograde urethrocystography or cystoscopy performed 1 week before obtaining the urine culture. Urogenital surgery included cats that had a perineal urethrostomy or other surgical procedure involving the lower urinary tract before obtaining urine for culture. Current therapy information was obtained for cats receiving antibiotics and/or glucocorticoids at the time of urine culture.

Urinalysis and urine culture

Urine samples for urinalysis and culture were obtained by cystocentesis or urethral catheterization. All specimens were immediately refrigerated at 4ºC (39.2ºF) until processed, within 12 h of collection. Routine urinalysis was performed using standard methods. ¹⁸ Briefly, urine specific gravity (USG) was determined by refractometry (TS meter hand-held refractometer; Leica Microsystems). Urine pH and concentrations of glucose, ketones, protein and occult blood were semi-quantitatively measured using commercial reagent test strips (Labstix; Bayer). Positive test strip reactions for protein were confirmed using a 3% sulfosalicylic acid turbidometric method. Bilirubin was semi-quantitatively determined with a commercial test tablet (Ictotest; Bayer). Urine (10 ml) was centrifuged for 5 mins at 2000 rpm (300g), supernatant was decanted, and the sediment was re-suspended and evaluated microscopically for white blood cells (WBCs), red blood cells (RBCs), crystals, casts and bacteria. For each sample, 10 high power fields (HPF) at 400× magnification were examined and the resulting counts were reported as a range. Pyuria and hematuria were defined as median cell counts of >5 WBCs/HPF and >5 RBCs/HPF, respectively. Observation of any bacteria in urine sediment was defined as bacteriuria. For statistical analyses, USG was analyzed as a continuous variable with 0.01 increments and as a categorical variable with four categories comprising values <1.015, 1.015–1.023, 1.024–1.040 and >1.040.

Quantitative aerobic urine cultures were performed using standard methods. ¹⁸ Briefly, 10 µl of urine were inoculated onto enriched Columbia and MacConkey agar plates, and incubated at 37 ºC (98.6 ºF) in 5% carbon dioxide for up to 5 days. Plates were examined daily for growth and the number of colonies was recorded. Microbial isolates were identified using routine biochemical and automated systems (VITEK automated system; BioMerieux). Antimicrobial susceptibility profiles were determined using Kirby-Bauer and broth dilution techniques (Sensititre; TREK Diagnostic Systems).

Statistical analysis

Descriptive statistics were generated for all risk factors using computer software (SAS, version 9.1.3, SAS Institute). Potential risk factors considered for statistical analysis included sex, weight, USG, transurethral procedures, urogenital surgery, renal disease, renal failure, other renal disease, urethral obstruction, urolithiasis, urinary incontinence, anatomic abnormalities of the lower urinary tract, other lower urinary tract disorder, endocrine disease, diabetes mellitus, gastrointestinal disease, neoplastic disease, infectious or immune-mediated disease, hepatobiliary disease, cardiorespiratory disease, neurologic disease, and other disease. Variables, particularly related to signalment, for which information was missing for many cases and controls, were described, but not included, in the multivariable analyses.

Univariable logistic regression models, using UTI (present/absent) as a dichotomous outcome, were generated for all risk factors to identify factors to be included in multivariable analysis, and any risk factors with P-values ≤0.20 were considered for inclusion in the models. Once the list of potential risk factors was developed, then Spearman rank correlation statistics were calculated for all risk factors to identify any significant interactions between risk factors.

Multivariable logistic regression models were developed to describe the associations of multiple risk factors with UTI simultaneously (SAS). ¹⁹ Initially, a full-rank model was developed from risk factors identified through descriptive analyses. In the case of risk factors with potential interactions identified through correlation statistics, interaction terms were included in the full model. Next, a hierarchical backwards model-building approach was used to create a final model. In this approach, the Akaike information criterion (AIC) statistic, a measure of model goodness-of-fit, ²⁰ was calculated, all risk factors and interaction terms in the model were examined, then the factor or interaction term with the highest P value was eliminated from the model (if an interaction term was non-significant, it was removed from the model; however, if a risk factor was eliminated from the model, any interaction term, including that factor was also eliminated from the model at that time). The reduced model was executed, and the AIC of the reduced model was compared with the previous version of the model. If the new AIC was less than the AIC of the previous model, the reduction step was accepted, and the next risk factor was selected for removal. The process was repeated until the AIC no longer decreased, which indicated that model performance was not improved with elimination of the risk factor, or if no risk factors in the model had P-values below the selected cut-off point for inclusion (0.05 in this analysis).

Results

One hundred and fifty-five cats met the inclusion criteria and 186 age-group matched controls were selected. Statistically significant differences in the year of diagnosis were observed between UTI cases and controls, with a median year of 1998 for cats with UTI versus 2000 for controls [odds ratio (OR) = 0.76, P <0.0001]; therefore, further logistic regression was conducted controlling for study year.

Signalment and lifestyle conditions are described in Table 1. The median age of cats with UTI was 8 years (range: 4 months to 20 years) compared with 7.7 years for controls (range: 2 months to 17 years). When cats with UTI were stratified by age, 29.7% were <5 years of age, 27.1% were 5–9 years of age and 43.2% were ≥10 years of age. A similar distribution was seen in controls as they were age-group-matched to UTI cases. Male cats were slightly over-represented in both the UTI (56.8%) and control (62.4%) groups. However, this difference was not statistically significant (P = 0.2639). Domestic shorthair or mixed breed cats predominated in both groups.

Table 1.

Signalment and lifestyle conditions of cats with urinary tract infection (UTI) and control cats with negative urine cultures

Variable	UTI (n = 155)		Control (n = 186)
	#	%	#	%
Age
<5 years	46	29.7	58	31.2
5–9 years	42	27.1	52	27.9
≥10 years	67	43.2	76	40.9
Sex
Male intact	5	3.2	4	2.2
Male neutered	83	53.6	112	60.2
Female intact	7	4.5	3	1.6
Female spayed	60	38.7	67	36
Breed
Domestic shorthair/mix	128	82.6	153	82.3
Domestic longhair	9	5.8	5	2.7
Siamese	3	1.9	6	3.2
Persian	3	1.9	5	2.7
Himalayan	3	1.9	2	1.1
Domestic medium hair	2	1.3	7	3.8
Maine Coon	2	1.3	6	3.2
Angora	1	0.6	0	0
Bengal	1	0.6	0	0
Birman	1	0.6	0	0
Russian Blue	1	0.6	0	0
Tonkinese	1	0.6	0	0
Abyssinian	0	0	1	0.5
Manx	0	0	1	0.5
Body condition score
Thin	49	31.6	43	23.1
Normal	5	3.2	16	8.6
Overweight	26	16.8	51	27.4
Unknown	75	48.4	76	40.9
Housing
Indoor	96	61.9	113	60.8
Outdoor	3	1.9	1	0.5
Both	24	15.5	39	21
Unknown	32	20.7	33	17.7
Cats per household
Single cat	16	10.3	38	20.4
Multiple cats	71	45.8	97	52.2
Unknown	68	43.9	51	27.4
Diet
Canned	15	9.7	15	8.1
Dry	45	29	76	40.8
Both	38	24.5	45	24.2
Unknown	57	36.8	50	26.9

Body weight was significantly lower for cats with UTI (mean 4.43 kg) than controls cats (mean 5.19 kg; OR = 0.81; P = 0.0259). There were no substantial differences in housing between cats with UTI and controls and review of dietary information did not reveal substantial differences among food consistency (Table 1). Unfortunately, body condition scores, housing information and dietary histories were not available for a large number of cats, so these risk factors were not included in the statistical analyses.

One or more lower urinary tract signs were observed in 100 (64.5%) cats with UTI and 114 (61.3%) control cats, whilst 55 (35.5%) cats with UTI and 72 (38.7%) controls had no signs (Table 2). The number of previous episodes of lower urinary tract signs was not substantially different between cats with UTI (n = 57, 36.8%) and controls (n = 68, 36.6%).

Table 2.

Prevalence of lower urinary tract signs in cats with urinary tract infection (UTI) and control cats with negative urine cultures

Clinical feature	UTI (n = 155)		Control (n = 186)
	Number	%	Number	%
Presence of signs of LUTD	100	64.5	114	61.3
Gross hematuria	39	25.2	48	25.8
Periuria	37	23.9	62	33.3
Stranguria	36	23.2	41	22
Pollakiuria	27	17.4	38	20.4
Urethral obstruction	25	16.1	24	12.9
Urinary incontinence	16	10.3	1	0.5
Dysuria	12	7.7	16	8.6
Absence of signs of LUTD	55	35.5	72	38.7

LUTD = lower urinary tract disease

The most commonly isolated species of bacteria was Escherichia coli, cultured from 50.3% cats with UTI (Table 3). There were 120 (77.4%) cultures with a single microbial isolate, 27 (17.4%) with two, and 8 (5.2%) with three. Among the 35 cultures with multiple isolates, E coli was identified in 20 (57.1%) and Enterococcus species in 19 (54.3%); concurrent isolation occurred in 11 (31.4%) cases. Two (1.3%) cultures revealed yeast infections (one Candida albicans and one undetermined).

Table 3.

Organisms isolated from aerobic culture of urine obtained from 155 cats with urinary tract infection

Organism isolated	Number of isolates	%
Escherichia coli	78	50.3
Enterococcus	33	21.3
Staphylococcus	27	17.4
Streptococcus	20	12.9
Pseudomonas	8	5.2
Klebsiella	6	3.9
Lactobacillus	5	3.2
Enterobacter	4	2.6
Pasteurella	3	1.9
Corynebacterium	3	1.9
Proteus	2	1.3
Moraxella	1	0.6
Alcaligenes	1	0.6
Bacillus	1	0.6
Plesiomonas	1	0.6
Salmonella	1	0.6
Serratia	1	0.6
Xanthomonas	1	0.6
Yeast	2	1.3

Average USG in the UTI group was 1.030 (± 0.017), compared with 1.041 (± 0.018) for controls. As a continuous variable, lower USG was significantly correlated with UTI in both univariable and multivariable analysis (OR = 0.78; P = 0.0055). There were 36 (23.2%) cats in the UTI group with a USG <1.015 compared with 16 (8.6%) controls; this was significantly associated with UTI in univariable analysis (OR = 5.16; P = 0.0059). Pyuria was present in samples from 82 (52.9%) cats with UTI and 27 (14.5%) controls. Sensitivity and specificity of pyuria for prediction of UTI were 52.9% and 85.5%, respectively. Bacteriuria was reported in 113 (72.9%) cats with UTI and 60 (32.3%) controls. Sensitivity and specificity of bacteriuria for prediction of UTI were 72.9% and 67.7%, respectively. Cocci and mixed bacteria (both rods and cocci), but not exclusively rods, were reported in urine sediments from control cats; therefore, the specificity of rod-shaped bacteriuria as the single visualized microorganism was 100% for UTI. There was no difference in mean urine pH of cats with UTI and controls.

Concurrent diseases were identified in 132 (85.2%) cats with UTI and 184 (98.9%) controls (Table 4). Fifty-five (35.5%) cats with UTI had a concomitant renal disease. Forty-five (29%) were in renal failure (defined as concurrent azotemia with inadequate urine concentrating ability). In the control group, renal disease affected 32 (17.2%) cats and there were 26 (14%) cats in renal failure. Lower urinary tract disorders affected 54 (34.8%) cats with UTI and 115 (61.8%) controls. Fifty-six control cats had lower urinary tract signs, but diagnostic information was insufficient to determine the specific cause. Additionally, idiopathic cystitis was diagnosed in 23 control cats and is a diagnosis of exclusion, so it was not possible to determine if any cats with UTI were affected by this condition. Gastrointestinal disease was present in 20 (12.9%) cats with UTI and 15 (8.1%) controls. Endocrine disorders were present in 19 (12.3%) cats with UTI and 30 (16.1%) controls. The numbers and percentages of the remainder concurrent disorders are represented in Table 4.

Table 4.

Concurrent diseases in 155 cats with urinary tract infection (UTI) and 186 controls

	UTI		Controls
Concurrent disease	Number	%	Number	%
Lower urinary tract disorders	54	34.8	115	61.8
Urethral obstruction	25	16.1	25	13.4
Incontinence	16	10.3	1	0.5
Anatomic abnormalities	13	8.4	2	1.1
Urolithiasis	10	6.4	15	8.1
Other	4	2.6	0	0
Idiopathic cystitis	-	-	23	12.4
Undiagnosed LUTD	-	-	56	30.1
Gastrointestinal disease	20	12.9	15	8.1
Constipation	6	3.9	5	2.7
Inflammatory bowel disease	5	3.2	1	0.5
Megacolon	3	1.9	0	0
Pancreatitis	3	1.9	4	2.1
Recto-cutaneous fistula	1	0.6	0	0
Perineal ulcer	1	0.6	0	0
Intestinal parasites	1	0.6	1	0.5
Intestinal lymphoma	1	0.6	3	1.6
Acute gastroenteritis	0	0	1	0.5
Renal disease	55	35.5	32	17.2
Renal failure	45	29	26	14
Nephrolithiasis	10	6.4	7	3.8
Pyelonephritis	8	5.2	0	0
Hydronephrosis	3	1.9	2	1.1
Renal tumors	3	1.9	0	0
Nephrocalcinosis	2	1.3	3	1.6
Polycystic kidney disease	2	1.3	2	1.1
Glomerulonephritis	1	0.6	0	0
Neurologic disease	13	8.4	1	0.5
Paraparesis/paraplegia	8	5.2	0	0
Urinary incontinence	7	4.5	0	0
Fecal incontinence	3	1.9	0	0
Lower motor neuron disease	2	1.3	0	0
Cerebral lesion	2	1.3	0	0
Urinary bladder atony	1	0.6	0	0
Vestibular disease	0	0	1	0.5
Endocrine disorders	19	12.3	30	16.1
Diabetes mellitus	9	5.8	21	11.3
Hyperthyroidism	11	7.1	9	4.8
Acromegaly	0	0	2	1.1
Cardiorespiratory disease	17	11	18	9.7
Neoplasia	13	8.4	11	5.9
Infectious/immune-mediated	7	4.5	5	2.7
Hepatobiliary disease	5	3.2	9	4.8
Other non-urinary tract disease	4	2.6	8	4.3

LUTD = lower urinary tract disease

Transurethral procedures were performed in a significantly higher number of cats with UTI (n = 34, 21.9%) than in control cats (n = 7, 3.8%). Urinary catheters had been placed in 34 cats with UTI. In addition, 5/34 cats had urethrocystograms and one had urethrocystoscopy. Seven control cats had urinary catheters placed and three of these also had urethrocystograms. Previous urogenital surgery was performed in 13 cats with UTI (8.4%; 11 perineal urethrostomies, one widening of a prepucial stenosis, one scrotal resection) and two control cats (1.1%; both perineal urethrostomies). The majority of cats undergoing transurethral procedures and/or urogenital surgery were male (93% of cats with UTI and 88.9% of controls). Gender distribution for cats with UTI that did not undergo transurethral procedures or surgery was 42.9% males and 57.1% females. The median age of cats with UTI that underwent these procedures was 6.3 years compared with 10.1 years for cats that did not.

Medications administered consisted of antibiotics in 36 (23.2%) cats with UTI and glucocorticoids in 14 (9%), whilst 18 (9.7%) control cats had received prior antibiotics and 10 (5.4%) controls were on glucocorticoids. Owing to the heterogeneity of drugs, dosages, timing of administration and the duration of therapies, no statistical comparison was performed for medication variables. The effect of pharmacotherapy on the relative risk of UTI in cats requires further investigation.

Univariable logistic regression models identified several potential risk factors meeting the inclusion criteria (P ≤0.20) for evaluation in multivariable analyses (Table 5a, b). Transurethral procedures, urogenital surgery, renal disease, renal failure, other renal disease, urinary incontinence, anatomical abnormalities of the lower urinary tract, gastrointestinal disease and neurological disease were associated with increasing risk for UTI, whereas increasing weight, USG and diabetes mellitus were risk factors associated with decreasing risk for UTI. Of these risk factors, potential interactions were identified between renal disease and renal failure, USG and renal disease, weight and renal disease, and neurologic disease and urinary incontinence. These interaction terms were included in the full model but were not retained during model development.

Table 5(a).

Univariable analysis of continuous potential risk factors for urinary tract infection in cats, in 155 cases and 186 controls

Variable	Units/level	155 cases		186 controls		P	OR	>95% CI
		Mean	SD	Mean	SD
Weight*	kg	4.43	1.52	5.19	1.67	<0.0001	0.74	0.64–0.84
USG*	(0.01 change)	1.030	0.017	1.041	0.018	<0.0001	0.72	0.63–0.81

^*Variables were considered for inclusion in multivariable analyses

USG = urine specific gravity

Table 5(b).

Univariable analysis of categorical potential risk factors for urinary tract infection in cats, in 155 cases and 186 controls

Variable	155 cases		186 controls		P	OR	95% CI
	Number	%	Number	%
Sex: male	88	56.8	116	62.4	0.2639	0.79	0.51–1.22
Transurethral procedures*	34	21.9	7	3.8	<0.0001	7.18	3.08–16.73
Urogenital surgery*	13	8.4	2	1.1	0.0073	7.53	1.72–32.85
Renal disease*	55	35.5	32	17.2	0.0002	2.65	1.60–4.38
Renal failure*	45	29.0	26	14.0	0.0008	2.52	1.47–4.32
Other renal disease*	22	14.2	12	6.4	0.0203	2.40	1.15–5.02
Urethral obstruction	25	16.1	25	13.4	0.4852	1.24	0.68–2.26
Urolithiasis	10	6.4	15	8.1	0.5701	0.79	0.34–1.80
Urinary incontinence*	16	10.3	1	0.5	0.0032	21.28	2.79–162.33
Anatomical abnormalities*	13	8.4	2	1.1	0.0055	8.42	1.87–37.88
Other LUTD	4	2.6	0	0	0.9878	>999	<0.001–>999
All endocrine disease	19	12.3	30	16.1	0.3116	0.73	0.39–1.35
Diabetes mellitus*	9	5.8	21	11.3	0.0803	0.48	0.22–1.09
Gastrointestinal disease*	20	12.9	15	8.1	0.1460	1.69	0.83–3.42
Neoplastic disease	13	8.4	11	5.9	0.3763	1.46	0.63–3.35
Infectious/immune disease	7	4.5	5	2.7	0.3663	1.71	0.53–5.51
Hepatobiliary disease	5	3.2	9	4.8	0.4578	0.66	0.21–2.0
Cardiorespiratory disease	17	11.0	18	9.7	0.6960	1.15	0.57–2.31
Neurological disease*	13	8.4	1	0.5	0.0067	16.93	2.19–130.96
Other disease	4	2.6	8	4.3	0.3957	0.59	0.17–2.0

^*Variables were considered for inclusion in multivariable analyses

LUTD = lower urinary tract disease

The final multivariable logistic regression model identified eight risk factors associated with UTI (Table 6). Urinary incontinence, transurethral procedures, urogenital surgery and gastrointestinal disease were significantly associated with increased risk for UTI, whereas increasing weight and USG were associated with decreased risk for UTI. Renal disease and anatomic abnormalities of the lower urinary tract were associated with increasing risk for UTI but were not statistically significant at P ≤0.05. However, their presence improved model performance and were left in the final multivariable model.

Table 6.

Final multivariable logistic regression model for associations between selected risk factors and urinary tract infection (UTI) in 155 cats with UTI and 186 control cats with negative urine cultures, controlling for study year

Variable	UTI		Control		P	OR	95% CI
	#	%	#	%
Weight (continuous, kg)	NA		NA		0.0259	0.81	0.68–0.97
USG (continuous, per 0.01)	NA		NA		0.0055	0.78	0.66–0.93
Transurethral procedures	34	21.9	7	3.8	<0.0001	8.37	3.21–21.82
Urogenital surgery	13	8.4	2	1.1	0.0385	6.03	1.10–33.07
Incontinence	16	10.3	1	0.5	0.0331	10.78	1.21–95.97
Gastrointestinal disease	20	12.9	15	8.1	0.0331	2.62	1.08–6.37
Renal disease*	55	35.5	32	17.2	0.0915	1.85	0.91–3.77
Anatomical abnormalities*	13	8.4	2	1.1	0.1503	4.01	0.60–26.57
Model AIC = 351.4, R2 = 33.4%, Likelihood ratio χ2 = 138.5, 9 df, P < 0.0001

^*Anatomical abnormalities and renal disease retained during model development

USG = urine specific gravity, NA = not applicable

Discussion

Our data showed that cats of all ages were similarly affected by UTI. The proportions of young and middle-aged cats with infection were similar (29.7% and 27.1%, respectively), and only slightly lower than that of older cats (43.2%). Our observations are in contrast with previous studies suggesting that UTIs are overrepresented in older cats.^1,4,5 It has been hypothesized that declining immune competence associated with aging may increase the risk of UTI in older cats.^8,12 In contrast, our data suggests that UTIs have a more uniform distribution among cats of all ages. One reason for this discrepancy is the inclusion of cats that had transurethral procedures and/or urogenital surgery. These cats were younger than cats without such procedures in our study and were excluded from other studies, thereby elevating the median age of cats in those reports.^4,5 Additionally, over-representation of other common lower urinary tract disorders in younger cats may artificially decrease the relative proportional morbidity of UTIs in these cats while increasing the proportion of UTIs in older animals. This finding is supported by data from a previous study that showed an increased risk for urocystolithiasis, urethral obstruction and idiopathic cystitis in cats between the age of 4 and 10 years, and for urethral plugs, neurogenic disorders, congenital defects and iatrogenic injuries in cats between the age of 2 and 7 years. ¹ Despite differences with other reports, our study was not designed to evaluate age as a risk factor for UTI, as we used age-group-matched controls.

Gender was not found to be a significant risk factor for UTI. Males were slightly over-represented in both the UTI (56.8%) and control (62.4%) groups. If cats with transurethral procedures or surgery were excluded, then females would be slightly over-represented in the UTI group (57.1%). Our observations differ from those of previous studies, where female cats appeared to be more susceptible to UTI.^{1,4,5,11–13}

Decreased body weight was a significant risk factor for UTI. The association of lower weight with UTI has been suggested previously; this may be explained, in part, by the advanced status of concurrent conditions that predispose to UTI via patient debilitation, weakened immune system and increased susceptibility to infection.^4,12

Our observations and those of others indicate that a substantial number of cats with UTI (35.5% in the present study) do not manifest signs of lower UTI.^5,10–12 This is comparable to canines, where a large proportion of dogs with UTI but no clinical signs (up to 54%) has been reported. ²¹ Consequently, clinical signs alone are a poor predictor of UTI in cats and urine culture should be submitted when lower urinary tract signs are observed or any of the procedures and/or conditions identified as risk factors for UTI are present.

Species of bacteria isolated from our population of cats were similar to prior reports, showing E coli as the predominant isolate.^1,8 Enterococcus species was the second most common isolate (21.3%), both as a single agent and as a concurrent pathogen, in particular, with other bacteria (54.3% of cultures with more than one species, particularly E coli). This increased prevalence of Enterococcus species has been reported in recent studies.^4,5,11,22 In addition, two cases of UTI were fungal: one due to Candida albicans and one due to an unidentified yeast. Fungal UTIs are uncommon in cats and often associated with a concurrent disorder or iatrogenically disrupted lower urinary tract.^15,23 The most common organism is Candida species; frequently there is concomitant bacterial growth. Bacteria were not found in either cat with fungal UTI in this study. Both cats had chronic kidney disease and one had a perineal urethrostomy, stricture and a urethral catheter in place.

One hundred and twenty (77.4%) cases of UTI in the present study had a single microbial isolate on culture, whilst 35 (22.6%) were polymicrobial, including 27 (17.4%) with two bacterial isolates and eight (5.2%) with three isolates. These numbers are similar to other studies that have reported frequencies of polymicrobial feline UTI in 14–16% of cases (including one study of asymptomatic cats).^5,8,22 These proportions are also comparable with those reported in canine UTI, where a large study revealed a single species isolated in 75–79% of cultures, two species in 17–20% and three species in 3–5% of cultures. ²⁴ In human medicine, polymicrobial cultures are commonly found in elderly patients (up to 33% of cultures) and particularly as a consequence of the use of indwelling urethral catheters, for example, in patients with spinal cord lesions.^25,26 Historically, urine cultures that contain more than one organism have been considered contaminated, but few studies have evaluated the clinical significance of polymicrobial growth from urine. In properly collected urine samples, and following quantitative criteria to determine significant bacteriuria based on method of collection, multiple growth often represents true mixed infection. ²⁷ If performed correctly, all bacteria cultured from cystocentesis samples, regardless of number, are an unequivocally abnormal finding indicative of true UTI.^17,24 In our study, of the 35 cats with polymicrobial UTI, 28 (80%) had urine collected by cystocentesis (including all cats with three isolates per culture), which is considered a gold standard technique; however, a limitation of the retrospective nature of this study is that the technique for cystocentesis could not be reviewed. Fifteen (42.3%) cats had a recent history of urethral catheterization and/or perineal urethrostomy, which we believe may have predisposed them to acquire the polymicrobial infection. Additionally, 17 (48.6%) cats had a history of past episodes of lower urinary tract disease (but no documented prior UTI), 16 (45.7%) cats had a concurrent lower urinary tract disease and 12 (34.3%) cats had concurrent renal disease. Interestingly, 5/8 (62.5%) cats with three bacterial isolates had renal disease, but the clinical significance of this finding is unknown.

Pyuria and bacteriuria were found to be neither sensitive (52.9% and 72.9%, respectively) nor specific (85.5% and 67.7%, respectively) indicators of UTI in cats. Our observations are in contrast with other feline studies, but in agreement with previous studies in dogs and one study in cats.^{4,5,11,12,18,28} Dilute urine and decreased neutrophil chemotaxis in conditions such as renal failure, diabetes mellitus or other immunosuppressive disorders may decrease the likelihood of observing bacteria and WBCs in urine sediment, accounting for their low sensitivities. Also, UTI cats receiving antibiotics and anti-inflammatories were allowed in our study, which may have negatively influenced the numbers of bacteria and WBCs present in urine, and may have falsely decreased sensitivity. This may account for differences from other studies that excluded cats on antibiotics and reported higher sensitivities of pyuria and bacteriuria.^4,11,12 Low specificity may be explained because pyuria is present in a variety of lower urinary tract diseases. In addition, small particles that resemble bacteria (usually confounded as cocci), such as lipid molecules, cytoplasmic organelles, amorphous crystals or debris in the urine sediment, may results in false-positive bacteriuria. ¹⁸ An exception in this study is identification of rods as a single bacterial population in urine sediment; our data confers this finding a specificity of 100% for UTI.

Urinary tract infections are seldom a primary disorder in cats. More commonly, they may develop following iatrogenic disruption of normal lower urinary tract anatomy or may arise secondary to underlying diseases that disturb normal host urinary tract defense mechanisms. Decreased USG was significantly associated with UTI upon multivariable analysis. This finding has been previously suggested to represent impaired intrinsic urine antimicrobial properties. ¹¹ Decreased osmolality may favor proliferation of bacteria that would be inhibited in concentrated urine. Our results are in contrast with a recent study evaluating USG as a risk factor for UTI, which did not find a correlation. ⁴ That study included only cats with concurrent chronic kidney disease, diabetes mellitus, uncontrolled hyperthyroidism or lower urinary tract disease, and used cats with the same diseases and a negative urine culture as controls. Some of these diseases decrease urine concentrating ability. Therefore, the control cat population may have had a lower USG than the general cat population not affected by the same diseases, making a correlation of USG with UTI more difficult to establish. We hypothesize that this may explain, in part, the differing results of our study, which used control cats regardless of concurrent conditions. In our study, renal disease was not a statistically significant risk factor for UTI by itself, but it improved the multivariable model performance owing to interaction of factors and was left in the final model, although its association with UTI is not as strong as other factors. Decreased USG as a result of impaired renal function is a common rational used to explain why UTIs frequently develop in cats with renal failure, and the observation of an interaction between USG and renal disease in our patients supports this hypothesis.^10,11 Other potential factors that may contribute to increased risk of UTI in cats with renal disease include changes in urine osmolality and pH, abnormal frequency of micturition and increased urine volume with possible residual urine retention, as well as potential impairment of the immune system depending on the type of renal disorder.

Urinary incontinence was the risk factor with the strongest association with UTI in this study, similar to previous reports in dogs. ²⁹ Mechanisms that may favor development of UTI in patients with incontinence include urine retention due to detrusor muscle areflexia or partial urethral obstruction, urethral incompetence due to urethral sphincter hyper- or hypoactivity, and development of perivulvar dermatitis or vestibulitis. These functional abnormalities may allow bacteria to ascend through the urethra and create a nidus of infection. ³⁰

Anatomic abnormalities of the lower urinary tract did not prove to be significant as a risk factor for UTI, but improved the statistical model performance. None of the remaining lower urinary tract disorders were shown to predispose to UTI. However, this is most likely influenced by intrinsic selection bias in the control population because of the need to have a urine culture. Because urine is not routinely cultured on every cat admitted to MSU-VTH, a bias was intrinsically introduced related to the clinician’s decision to order the test. We suspect that cats showing lower urinary tract signs were more likely to have urine cultured than cats without such signs in order to rule out a UTI. This would likely lead to an over-representation of cats with various lower urinary tract disorders in the pool of cats with negative urine cultures from where control cats were selected. This selection bias would diminish the impact of concurrent lower urinary tract disorders in the UTI group and increase the proportion of such disorders in the control population. Additionally, 30.1% of control cats showed lower urinary tract signs but did not have an established cause for these signs owing to lack of enough diagnostic evaluation to arrive at a final diagnosis. This is a limitation of the study because of its retrospective nature, because criteria for inclusion did not require additional diagnostic evaluation of the lower urinary tract, such as imaging studies, endoscopy or tissue pathology, and many cats did not have these procedures done. We suspect that a number of the control cats with undiagnosed lower urinary tract signs may have suffered from undiagnosed idiopathic cystitis; however, we cannot conclude this because idiopathic cystitis is a diagnosis of exclusion and further diagnostic evaluation was not performed. In any case, it is difficult to establish the importance of idiopathic cystitis as a potential risk factor for UTI because we cannot determine if any of the cats in the UTI group had been affected by this condition prior to developing the UTI.

Surprisingly, gastrointestinal disorders were a significant risk factor for UTI in our study. To our knowledge, this relationship has not been reported in cats or other species. The etiology of the gastrointestinal problems was variable, but most cats suffered from a spectrum of common clinical signs, including diarrhea, constipation and vomiting. Whilst we have no definitive explanation for this relationship, we hypothesize that diseases triggering diarrhea or constipation may result in increased perineal fecal bacterial contamination, decreased grooming efficacy and easier access of bacteria to the external urethral meatus, favoring an ascending infection. Other possible contributing factors may include imbalances in normal host intestinal flora or an altered immune system. A recent study showed that cats with lower urinary tract signs were more likely to manifest gastrointestinal signs than control cats without such signs. ³¹ In humans, certain lower urinary tract disorders, such as interstitial cystitis, appear to be identified with relative frequency concurrently with gastrointestinal disturbances. The authors of the above mentioned study compared the findings in cats with those in humans and hypothesized potential comorbidity of feline idiopathic cystitis with primary gastrointestinal disorders. If an association of feline idiopathic cystitis with clinical signs of gastrointestinal disease were present, this could favor the development of ascending bacterial infection into an already compromised lower urinary tract, increasing the incidence of UTIs amongst cats with idiopathic cystitis. However, as feline idiopathic cystitis is an exclusive diagnosis, it is impossible to determine whether cats with UTI were coincidentally affected by this condition.

Endocrine diseases, particularly diabetes mellitus and hyperthyroidism, are commonly associated with concurrent UTI in dogs and cats, favored by factors such as dilute urine, glucosuria, abnormal neutrophil function and impaired local immunity.^{11,12,32–34} In our study, neither diabetes nor hyperthyroidism were associated with UTI. However, selection bias may be responsible, in part, for this finding. Given the documented correlation of diabetes mellitus with UTI in dogs, diabetic cats may be more likely to have urine routinely cultured than non-diabetic cats. This would lead to an over-representation of diabetic cats in the control group, decreasing the potential for identifying an association between diabetes mellitus and UTI. Consequently, an association between endocrinopathies and UTI cannot be excluded. However, the low prevalence of UTI in cats with endocrine disorders (nine diabetic and 11 hyperthyroid cats) is in contrast with previous studies.^11,12

Urogenital procedures, especially placement of indwelling urethral catheters, are known risk factors for the development of UTIs in humans, dogs and cats.^9,29,35–37 Disruption of normal urothelial mucosa, inflammation and iatrogenic introduction of bacteria are etiopathologic factors believed to be responsible for this increased risk of UTI. Likewise, our study showed a high correlation between transurethral procedures and UTI. All cats undergoing these procedures had urinary catheters placed and these represented the vast majority of transurethral procedures. The number of cats with contrast cystograms were low in both groups and only one cat had cystoscopy. Therefore, our data may largely reflect the impact of urethral catheterization as a risk factor for UTI and conclusions on other procedures cannot be determined from this study. Similarly, urogenital surgery was associated with increased risk of UTI. Previous studies have identified UTI in up to 23% of cats with perineal urethrostomies. ¹⁶ However, a study evaluating the incidence of UTI after perineal urethrostomy in healthy cats versus cats with urethral obstruction revealed infection only in animals with prior obstruction. ⁷ Our results support that surgical alteration of lower urinary tract anatomy in cats with underlying uropathies increases the risk for infection. In addition, it is common practice amongst many surgeons to administer peri-operative and postoperative antibiotics. This could mask the presence of UTI if given prior to obtaining urine for culture and reduce the association of urogenital surgery with UTI. Despite this possibility, our study revealed an association between urogenital surgery and UTI.

Given the retrospective nature of this study, our observations are subject to certain limitations. It is likely that a certain degree of control cat selection bias may have been introduced based upon the perceived need for a urine culture. This bias is more likely to interfere with factors that historically are suspicious for association with UTI, such as concurrent endocrine or lower urinary tract disorders. It is common practice at MSU-VTH to submit urine cultures from cats presenting with these diseases. This would lead to an over-representation of cats with such conditions in the database of negative urine culture results from which control cats were randomly selected. The end result of this type of bias would actually be reinforcement of any significant correlation between the variables studied and UTI, whilst simultaneously diminishing the impact of variables not found to be significant. In other words, positive associations may be even stronger than they appear (ie, renal disease with UTI), whilst negative associations must be viewed carefully and a correlation cannot be ruled out based on results of this study (ie, diabetes mellitus with UTI). One potential strategy to diminish the impact of this bias could be selection of a much larger number of control cats than cats with UTI in future studies. We would propose that four controls be selected for each case cat in subsequent studies in order to provide additional strength to the statistical model and results. To our knowledge, this is the first reported study that compares the features of cats with UTI with those of controls with proof of absence of UTI, regardless of concurrent disorders and/or procedures. This type of comparison is useful to establish definitive associations between any variables that show significance. Studies that do not use controls, or where controls are selected from the general cat population without a urine culture, are subject to error due to potential introduction of cats with UTI lacking clinical signs as controls. As 35.5% of our cats with UTI did not have lower urinary tract signs, results of those studies may be unreliable.

In conclusion, we observed that cats of all ages were similarly affected by UTI and 1/3 lacked clinical signs of lower urinary tract disease. In addition, our results suggest that urine sediment abnormalities, such as bacteriuria (except for the presence of rods only) and pyuria, are poor predictors of the presence of UTI. Finally, this study identified several factors significantly associated with the presence of UTI in cats, including urinary incontinence, gastrointestinal disease, transurethral procedures, urogenital surgery, decreased USG and decreased body weight, with additional contributions of renal disease and anatomic abnormalities of the lower urinary tract. We recommend that urine should be cultured every time clinical signs or predisposing factors for UTI are present, regardless of urine sediment findings.