Useful Predictive Factors for Bacteremia among Outpatients with Pyelonephritis.

Objective The aim of this study was to identify predictive factors for bacteremia conveniently and quickly among outpatients diagnosed with pyelonephritis. Patients All patients who were diagnosed with pyelonephritis at the outpatient clinic in the Department of General Medicine of Juntendo University Hospital from April 1, 2008, to June 30, 2015, were enrolled. Patients from whom blood cultures had not been taken were excluded. Methods Clinical information was extracted from medical charts. Factors potentially predictive of bacteremia were analyzed using a t-test and Fisher's exact test, followed by a multivariable logistic regression model analysis. Results Blood cultures were drawn from 116 patients, and 25 (22%) presented with bacteremia. A multivariate analysis with the age, chills, platelet count and urine nitrite test results revealed that older age, positive urinary nitrite test results and chills tended to be associated with bacteremia, respectively. [older age: unit odds ratio (OR) 1.02, p=0.052, 95% confidence interval (CI) 1.00-1.05, positive urinary nitrite test findings: OR 2.5, p=0.092, 95% CI 0.86-7.7, chills: OR 2.5, p=0.096, 95% CI 0.84-7.65]. The area under the receiver operating characteristic (ROC) curve of this model was 0.77. Regardless of age, positive urinary nitrite test findings were significantly associated with bacteremia (OR 3.1, p=0.033, 95% CI 1.1-9.2), and chills tended to be associated with bacteremia (OR 2.7, p=0.07, 95% CI 0.93-7.9) The area under the ROC curve of this model was 0.75. Conclusion Bacteremia should be considered in pyelonephritis patients with rapidly assessable factors in outpatient clinic. In particular, a model including a urinary nitrite test has the potential to aid in the prediction of bacteremia.

Introduction

Pyelonephritis is a common infectious disease. Approximately 250,000 cases of pyelonephritis occur each year in the US (1). The management guidelines for urinary tract infections in the US and Japan recommend that patients with mild, uncomplicated pyelonephritis be treated in an outpatient clinic (2,3). However, previous studies have reported that 15-32% of pyelonephritis cases were complicated with bacteremia (4,5). In addition, severe pyelonephritis accompanied by bacteremia has a mortality rate of 10% to 20% (6,7).

Bacteremia is one of the most severe complications of pyelonephritis, so physicians must have a high index of suspicion in patients with pyelonephritis. To enhance the likelihood of good outcomes, it is important to initiate adequate antimicrobial treatment before blood culture results return as positive (8). Some previous studies have revealed predictive factors for pyelonephritis with bacteremia. (4,5,9) However, these studies did not include outpatients.

The aim of this study was to identify predictive factors for bacteremia conveniently and quickly among patients diagnosed with pyelonephritis in an outpatient clinic.

Materials and Methods

In this study, we retrospectively investigated the medical records of all patients who were diagnosed with pyelonephritis at the outpatient clinic in the Department of General Medicine in Juntendo University Hospital from April 1, 2008, to June 30, 2015. We excluded patients from whom blood cultures had not been taken. Bacteremic pyelonephritis was defined as the detection of identical causative bacteria from blood and urine cultures.

We collected demographic data, vital signs, subjective symptoms, objective physical findings, laboratory findings, results of blood culture and urine culture, antimicrobial course, surgical interventions, and outcomes of the treatment as shown in Table 1. All male participants and participants with any underlying conditions listed in Table 1 were categorized as complicated pyelonephritis patients. Other participants were recognized as uncomplicated patients.

Table 1.

Patient Characteristics and Clinical Classification.

	Bacteremia n=25		Non-bacteremia n=91		p value
Age, years; mean (SD)	62.0	(21)	48.1	(22)	0.006*
Female, n (%)	22	(88)	81	(89)	1.00
Underlying disorders, n (%)
Diabetes mellitus	2	(8.0)	3	(3.3)	0.29
Anatomic abnormality of urinary tract	0	(0)	6	(6.6)	-
Indwelling urinary catheter	0	(0)	0	(0)	-
Neurogenic bladder	1	(4.0)	0	(0)	-
Immunosuppressive agents	2	(8.0)	3	(3.3)	0.29
Uncomplicated pyelonephritis, n (%)	17	(68)	71	(88)	0.30
History of pyelonephritis, n (%)	6	(24)	16	(18)	0.56

Uncomplicated pyelonephritis patients were those without any factors of complications, male gender or any underlying disorders listed above. SD: standard deviation

Because of the retrospective study design, the requirement for informed consent was waived. Study approval was obtained from the ethical committee of Juntendo University Hospital, with the approval number 15-123. Data analyses were performed using the JMP software program (version 11.0.0; SAS Institute, Cary, USA).

We used Fisher's exact test to compare the proportions of categorical variables between the groups. A t-test was used to compare continuous variables between the groups. A multivariate logistic regression analysis was then conducted based on the results of the univariate analysis (p<0.05) and previous studies to investigate the model for predicting bacteremia in the study population. We chose “chills” as the variable for the multivariate analysis, regardless of the univariate analysis results, because “chills” has been reported as a predictive factor by previous studies and is quickly assessable in outpatients (5,9,10).

Results

During the study period, 141 patients were diagnosed with pyelonephritis at outpatient clinic. Blood cultures were drawn from 116 pyelonephritis patients, 25 of whom (22%) presented with bacteremia. Eighty-eight cases (75.9%) were categorized as uncomplicated pyelonephritis. Demographic factors are shown in Table 1. Bacteremia was significantly associated with an older age (bacteremia: 62.0±21 years old, non-bacteremia: 48.1±22 years old, p=0.006). No association was found between bacteremia and complications. Table 2 shows the results of urine cultures and blood cultures. Escherichia coli was the most frequent causative microorganism. Table 3 shows the clinical symptoms and laboratory results. A low platelet count (bacteremia: 19.8±6.7×103/μL, non-bacteremia: 23.0±7.5×104/μL, p=0.037) and positive urinary nitrite test findings (bacteremia: 48%, non-bacteremia: 31%, p=0.043) were associated with bacteremia. In contrast, general inflammatory parameters, such as body temperature, white blood cell count, neutrophil count and C-reactive protein, were not associated with bacteremia.

Table 2.

Results of Urine and Blood Cultures.

	Urine culture results (n=116)		Blood culture results (n=116)
Escherichia coli, n (%)	65	(56)	23	(20)
Proteus mirabilis, n (%)	3	(2.6)	1	(0.9)
Citrobacter koseri, n (%)	3	(2.6)
Group B Streptococcus, n (%)	2	(1.7)
Klebsiella pneumoniae, n (%)	1	(0.9)
Enterococcus faecalis, n (%)	1	(0.9)	1	(0.9)
Lactobacillus, n (%)	1	(0.9)
Polymicrobial*, n (%)	7	(6.0)
Negative, n (%)	33	(28)	92	(79)

*Esc herichia coli+Entero coccus faecalis, Esc herichia coli+Klebsi ella pneumoniae, Esc herichia coli+Klebsi ella pneumoniae+Pseu domonas aeruginosa, Escherichia coli+Pro teus mirabilis, Esch erichia coli+Klebsie lla pneumoniae, Prot eus vulgaris+Myroides odoratus+Staphyloc occus aureus+Enterococcus faecalis

Table 3.

Vital Signs, Clinical Symptoms and Laboratory Results.

	Bacteremia n=25		Non-bacteremia n=91		p value
Vital signs
Body temperature, °C (SD)	38.2	(1.17)	38.1	(1.06)	0.84
Symptoms
Macrohematuria, n (%)	1	(4.0)	4	(4.4)	1.00
Pain in urination, n (%)	3	(12)	10	(11)	1.00
Back pain, n (%)	8	(32)	34	(37)	0.81
Chills, n (%)	11	(44)	24	(26)	0.14
Vomiting, n (%)	4	(16)	9	(9.9)	0.47
Nausea, n (%)	0	(0)	7	(7.7)	-
Diarrhea, n (%)	5	(20)	7	(7.7)	0.13
Clinical signs
CVA tenderness (+), n (%)	17	(68)	60	(66)	1.00
Laboratory results
White blood cells, ×109/L (SD)	11.6	(5.6)	12.3	(4.4)	0.56
Neutrophils, ×109/L (SD)	10.5	(4.1)	9.7	(5.7)	0.54
Platelet, ×104/μL (SD)	19.8	(6.7)	23.0	(7.5)	0.037*
BUN, mg/dL (SD)	17.1	(12.9)	12.7	(6.2)	0.11
Creatinine, mg/dL (SD)	0.81	(0.44)	0.70	(0.30)	0.26
CRP, mg/dL (SD)	10.8	(8.9)	9.9	(7.3)	0.65
Urinary nitrite test (+), n (%)	12	(48)	28	(31)	0.043*

*: p<0.05. SD: standard deviation, CVA: costophrenic angle, BUN: blood urea nitrogen, CRP: C-reactive protein

Table 4 shows the clinical course of all included patients. Patients with bacteremia were prone to require hospitalization for treatment [bacteremia: 22 patients (88%), non-bacteremia: 31 patients (34%), p<0.001], longer hospitalization (bacteremia: 12.5±9.2 days, non-bacteremia: 4.2±8.7 days, p<0.001) and a longer total duration of antimicrobial treatment than non-bacteremia patients (bacteremia:15.0±2.3 days, non-bacteremia:12.4±6.2 days). No patients died during the treatment course.

Table 4.

Clinical Courses of the Patients.

	Bacteremic n=25		Non-bacteremic n=91		p value
Hospitalization required, n (%)	22	(88)	31	(34)	<0.001*
Length of total antimicrobials, day (SD)	15.0	(2.3)	12.4	(6.2)	0.002*
Hospital stay, days (SD)	12.5	(9.2)	4.2	(8.7)	<0.001*
Death, n (%)	0	(0)	0	(0)	-

*: p<0.05. SD: standard deviation

The results of the multivariate analysis are shown in Tables 5 and 6. For the multivariate analysis, we chose the variables that showed p<0.05 in the univariate analysis and “chills”, based on the findings of previous studies of bacteremia (5,9,10). Table 5 shows the results of a multivariate analysis including four factors: older age, positive urinary nitrite test, chills and a low platelet count. Older age, positive urinary nitrite test and chills all tended to be associated with bacteremia [age: unit odds ratio (OR) 1.02, p=0.052, 95% confidence interval (CI) 1.00-1.05, positive urinary nitrite test: OR 2.5, p=0.092, 95% CI 0.86-7.7, chills: OR 2.5, p=0.096, 95% CI 0.84-7.65]. The area under the receiver operating characteristic (ROC) curve of this model was 0.77. Regardless of age, a positive urinary nitrite test was significantly associated with bacteremia (OR 3.1, p=0.033, 95% CI 1.1-9.2), and chills tended to be associated with bacteremia (OR 2.7, p=0.07, 95% CI 0.93-7.9). The area under the ROC curve of this model was 0.75.

Table 5.

Multivariate Analysis 1.

	OR	95% CI	p value
Urinary nitrite test (+)	2.5	0.86-7.8	0.094
Age	1.02*	1.0-1.1	0.052
Platelet	1.0	0.99-1.0	0.20
Chills	2.5	0.86-7.7	0.095

R2 was 0.15 (p<0.01). *: Unit odds ratio.

OR: odds ratio, CI: confidence intervals

Table 6.

Multivariate Analysis 2.

	OR	95% CI	p value
Urinary nitrite test (+)	3.1	1.1-9.2	0.033**
Chills	2.7*	0.93-7.9	0.068
Platelet	0.99	0.99-1.01	0.11

R2 was 0.11 (p=0.01). *: Unit odds ratio, **: p<0.05.

OR: odds ratio, CI: confidence interval

Discussion

In this study, we investigated the predictive factors for bacteremia among pyelonephritis cases. In the study population, three factors were significantly associated with bacteremia in a univariate analysis: a positive urinary nitrite test, an older age and a lower platelet count. The results of the multivariate analysis showed that older age, positive urinary nitrite test and chills tended to be associated with bacteremia. Regardless of age, a positive urinary nitrite test was associated with bacteremia, and chills tended to be associated with bacteremia.

Our study found that positive urinary test results were associated with bacteremia. Positive urinary nitrite test findings have not been mentioned as a predictive factor of bacteremia in pyelonephritis patients. Many previous studies have reported that urinary tract occlusion (5,9,11), diabetes mellitus (4,9) or the presence of an indwelling urinary catheter (4), chills (5,9,10) and neutrophilia (5,9,12) were significantly associated with bacteremia in pyelonephritis. However, these factors are all related to complicated pyelonephritis, except for neutrophilia and chills. Because the present study mainly involved uncomplicated pyelonephritis patients, no factors related to complicated pyelonephritis showed any significant association with bacteremia.

The urinary nitrite test is a rapid and convenient point-of-care test for clinics and emergency rooms. It is useful for predicting bacteriuria, and its sensitivity and specificity are 27-35% and 97.5-99%, respectively (13-15). The urinary nitrite test is often used in combination with the urinary leukocyte esterase test in practice. While previous studies have suggested that pyelonephritis may be present when either urinary leukocyte esterase or nitrite is positive, with a sensitivity of 75% and a specificity of 82% (14,16,17), no studies have shown that a nitrate test is useful for predicting bacteremia in these patients. The microbial spectrum of uncomplicated cystitis and pyelonephritis consists mainly of nitrite-producing Escherichia coli and other species of Enterobacteriaceae (18-20). The prevalent causative bacteria of pyelonephritis in this study was family Enterobacteriaceae, so the positive urinary nitrite test may reflect a long incubation time of nitrite-producing bacteria in urinary tracts, resulting in bacteremia (21). The sensitivity and specificity of the urinary nitrate test of bacteremia in this study were not sufficiently high (48% and 75%, respectively), but to our knowledge, there have been no studies suggesting a positive urinary nitrite test as an associated factor of bacteremia in uncomplicated pyelonephritis. In this retrospective study, physicians might have tended to hospitalize patients when the blood culture results turned positive. As such, the urinary nitrite test may be useful for assisting physicians in deciding on a treatment plan for pyelonephritis patients.

Several limitations associated with this study warrant mention. First, the overall study population was small, and the study was conducted at a single center. Second, a common diagnostic criterion of pyelonephritis was not used because of the retrospective study design. These factors might have created bias in the results and should be resolved in a future prospective study.

In conclusion, pyelonephritis is common and often complicated with bacteremia. It is therefore important for physicians working in outpatient clinics not to miss a diagnosis of bacteremia due to limited information and tests. A model including the urinary nitrite test may be useful for predicting bacteremia in the outpatient setting and facilitating the direct early management of pyelonephritis, thereby potentially reducing any delay in hospitalization.

The authors state that they have no Conflict of Interest (COI).

Financial support

This study was supported in part by a Grant-in-Aid (S1201013) from the Japanese Ministry of Education, Culture, Sports, Science and Technology (MEXT)'s Support Program for Strategic Research Foundations at Private Universities, 2012-2016.