Introduction
Urinary tract infections (UTIs) are among the most common bacterial infections, typically managed with empiric antibiotics to provide rapid relief and prevent complications like sepsis. Reference Gupta, Hooton and Naber1 However, rising antimicrobial resistance and the overuse of broad-spectrum agents have led to frequent misalignment between empiric choices and actual pathogens. Reference Hooton2
The urinary dipstick is a cost-effective point-of-care tool; combining leukocyte-esterase and nitrite tests yields a sensitivity of 75% and specificity of 82% for UTI diagnosis. Reference Bellazreg, Abid, Ben Lasfar, Hattab, Hachfi and Letaief3 Nitrites specifically indicate the presence of gram-negative (gram−) Enterobacterales, such as E. coli. Reference Chaudhari, Monuteaux and Bachur4 Consequently, current guidelines suggest that a positive nitrite test combined with clinical symptoms justifies empiric therapy targeting E. coli without further confirmation. Reference Gupta, Hooton and Naber1
However, standard empiric regimens may overlook gram-positive (gram+) pathogens, which are increasingly prevalent. Enterococcus species, now accounting for 8%–35% of UTIs, are particularly challenging due to inherent resistance to cephalosporins and aminoglycosides Reference Gajdács, Ábrók, Lázár and Burián5 . Notably, patients with E. faecalis UTIs are significantly more likely to receive inappropriate empirical therapy. Reference Madrazo, Esparcia and Alberola6
Since gram+ bacteria do not produce nitrites, nitrite-negative pyuria may serve as a clinical indicator to shift empiric coverage. Reference Kline and Lewis7
Materials and methods
Ethics approval
The Tel Aviv Sourasky Medical Center institutional review board (IRB) approved this study (TLV-0205-24).
Inclusion criteria
Adult patients presented to the Department of Emergency Medicine (ER) of Tel-Aviv Sourasky Medical Center, Israel, between 1.7.2008–31.7.2023, in whom urinary dipstick test and urine culture were performed.
Microbiology
Positive urine cultures were defined as those who grew pathogenic bacteria. Normal skin flora and vaginal flora were regarded as contaminants.
Statistical analysis
Categorical variables were described using frequency and percentage. Continuous variables were evaluated for normal distribution using histograms and Q-Q plots and reported as mean with standard deviation. Scale variables were reported as median and interquartile range (IQR). The χ2 test compared categorical variables. Mann-Whitney U test was used to compare continuous, not normally distributed variables. All statistical tests were two-tailed. P ≤ .05 was considered statistically significant. Statistical analyses were performed using SPSS (IBM SPSS Statistics for Windows, version 27, IBM Corp., Armonk, NY, USA, 2020).
Results
During the study period, 4,724 patients had both a dipstick test and a positive urine culture with growth of pathogenic bacteria (Table 1). Majority of cases were caused by gram− bacteria, accounting for 4,429 (93.8%) of the positive cultures. The most common uropathogen was e. coli (71.4%), followed by klebsiella pneumonia (10%), and e. faecalis (3.8%). Of the gram+ bacteria, e. faecalis was the most common (61%) (Table 1).
Table 1. Organisms identified by urine cultures in the study cohort
Cohort characteristics
Out of 4,724 cases, 3,446 of the UTI cases occurred in females. The overall female predominance stems mainly from gram− pathogens. The F:M ratio in gram− UTIs is 2.9 while the F:M ratio is 1.1 for gram+ pathogens.
Patients with gram+ infections were older (62.9 ± 24.2 vs 56.2 ± 25.0 years, P < .001), had a higher CCI (4 [IQR 1,6] vs 3 [IQR 0,5], P = .049), higher creatinine (1.27 ± .94 vs 1.06 ± 0.77 mg/dL, P < .001), BUN (21.8 ± 19.3 vs 18 ± 16.8 mg/dL, P < .001), WBC (11.5 ± 4.9 vs 10 ± 5 109 cells/L, P < .001), Neutrophil count (8.8 ± 4.6 vs 7.2 ± 4.2 109 cells/L, P < .001). C-reactive protein (CRP) levels were similar in both groups (57.9 ± 71.5 vs 54.7 ± 69.3 mg/L, P = .594).
Positive leukocyte-esterase test had a high sensitivity of 85.8% (cut-off value ≥ 250 leukocytes/μL), compared with poor sensitivity of 41.4% for nitrites. The leukocyte esterase test had a higher sensitivity for gram− infections compared with gram+ infections (86.3% vs 76.9%, respectively, P < .001). Higher rates of nitrite positivity were observed for gram− infections, but nitrites were also detected in samples with the eventual growth of gram+ pathogens (44.4% vs 12.2%, P < .001).
Sensitivity and specificity
Sensitivity and specificity analyses are shown in Table 2.
Table 2. Sensitivity and specificity analyses of urinary dipstick test results for predicting infections and their types
Note. SN, sensitivity; SP, specificity.
While nitrites were not sensitive, their pooled specificity for UTI was high (89%). However, this was also true for gram+ infections (81%). A positive leukocyte esterase test was poorly specific overall (38.3%) regardless of the pathogen type (37.7% and 32.1% for gram− and gram+). Finally, we examined if any combination of the urinary dipstick test results could be used to discern between gram− and gram+ infections. Disappointingly, all the combinations lack sensitivity. The combination of positive leukocyte esterase and nitrite test has higher sensitivity for gram-negative infections (38.7% vs 9.8%), while positive leukocyte esterase and negative nitrite test has a higher specificity for gram-positive infections (67.1% vs 47.6%).
Discussion
In our large cross-sectional study among adult patients, we found an overall prevalence of gram+ UTI of 6.2%, most of which were caused by e. faecalis (Table 1). However, gram-positive infections accounted for only 10.8% of nitrite-negative urine dipstick tests.
Strikingly, the presence of urinary nitrites did not exclude the presence of a gram-positive uropathogen, yielding a positive result in 12.2% of gram-positive UTIs.
UTIs are common and are usually diagnosed based on history, physical examination, and urinary dipstick test. Reference Gupta, Hooton and Naber1 Because most UTI treatment is empirical this is of special importance in the outpatient setting where urine cultures are not routinely obtained, or at the emergency department where the choice of antibiotic regiment is done quickly, and time to appropriate antibiotic treatment is closely associated with critically ill patients’ outcomes. Reference Weinberger, Rhee and Klompas8
Notably, inappropriate empiric antibiotic therapy was recently shown to result in increased risk for UTI-related hospitalization and re-consultation. Reference Aryee, Rockenschaub and Robson9
Given the importance of appropriate empirical antibiotic therapy, previous studies addressed predictive factors for e. faecalis UTI. Reference Madrazo, Esparcia and Alberola6 Consistent with our findings, the e. faecalis group had a higher proportion of men, older age, and more comorbidities. Disappointingly, the presence or absence of urinary nitrites lacks specificity and sensitivity for guiding empiric therapy.
Not only that relying on the absence of nitrites would lead to adding coverage for e. faecalis in ∼50% of gram-negative UTI cases, but we also show that nitrites may be present in cases caused by gram-positive organisms. Thus, the clinical presentation and patient characteristics, rather than the urinary dipstick test, should guide the choice of empiric antibiotic coverage.
There are several limitations to our study. First, it is limited by its retrospective design. Second, due to the locality of the cohort patient characteristics may largely vary in other places. Third, the case definition is complex and represents and inherent limitation. Reference Bilsen, Jongeneel and Schneeberger10 Due to the nature of retrospective cohort formation, the inclusion criterion were patients who presented to the ER in whom both a urine dipstick and a urine culture were performed. Although usually it is performed after a clinical decision by an ER physician or a nurse suggesting UTI was suspected, sometimes it is done as a workup of fever or sepsis without specific urinary complaints or when a patient is unable to pinpoint specific etiologies for their condition, affecting the rate of UTIs and asymptomatic bacteriuria in this large cohort. However, urinary colonization with bacteria, either in asymptomatic bacteriuria or urinary tract infections, that has nitrate reductase should still produce positive nitrites. Meaning that the production of nitrates and leukocyte esterase presence should be independent of symptoms and thus does not affect the validity of our conclusions.
In conclusion, nitrites or the lack of them are not necessarily correlated with infections and the pathogens causing them, and dipstick testing alone is not suitable for guiding empirical antibiotic coverage for suspected UTI.
Open access
