Quantifying the Relationship between Staphylococcus aureus Bacteremia and S. aureus Bacteriuria: A Retrospective Analysis in a Tertiary Care Hospital

It is popularly believed that Staphylococcus aureus bacteremia causes S. aureus bacteriuria, especially in the case of endocarditis [1–3]. On the basis of this assumption, it is often advised to seek a distant site of infection as a source of (transient) bacteremia in the case of S. aureus bacteriuria. However, data that confirm this widely assumed association are scarce, and among 39 patients with S. aureus bacteriuria (but without demonstrated bacteremia), a distant site of S. aureus infection was demonstrated in only 1 patient [4]. Moreover, in the most frequently cited study in this regard [5], an association between S. aureus bacteremia and S. aureus bacteriuria was reported, but causality was not claimed. In fact, it was concluded that “staphylococcal bacteriuria appears to be a frequent and unexplained concomitant of S. aureus bacteremia” (p. 306).

If the theory that S. aureus bacteriuria is a reflection of bacteremia is not true, adherence to it may lead to extensive and unnecessary “source investigations” in patients with S. aureus bacteriuria and to disregarding the urinary tract as a focus of S. aureus bacteremia. S. aureus is, however, known to cause urinary tract infection (UTI), which may well be the source of S. aureus sepsis syndrome [6]. Furthermore, it has been demonstrated that urinary catheters predispose to urinary colonization with S. aureus [6–8] and that colonization with this microorganism is a risk factor for subsequent infection [6, 9].

In this study, we aimed to quantify the relationship between S. aureus bacteremia and S. aureus bacteriuria. We hypothesized that, if S. aureus bacteremia truly causes S. aureus bacteriuria, we should observe this phenomenon in patients without UTI and without risk factors for urinary colonization with S. aureus (specifically, the presence of urinary catheters), because only such could, in our opinion, be unambiguously attributed to S. aureus bacteremia.

Methods. Patients who received a diagnosis of documented S. aureus bacteremia during the period from 1 June 2001 to 1 June 2006 were identified in the database of the bacteriology department of the University Medical Center Utrecht (a tertiary care hospital in Utrecht, The Netherlands, with 1024 beds). From these patients, 2 groups were selected: patients who had a urine sample obtained for culture on the same day as a positive result of blood culture result (group 1), and patients who had S. aureus bacteriuria on any other day during this 5-year period (group 2), with an isolate with a similar susceptibility pattern (difference in susceptibility of ≤1 antibiotic between the urine and blood isolates). Patients with a clinical diagnosis of S. aureus UTI were excluded from groups 1 and 2 and analyzed separately, because in these cases, the bacteriuria was regarded as the cause of S. aureus bacteremia rather than its effect.

For each patient, we recorded the age, the number of positive results of blood culture bottles in a window period of 2 days before until 2 days after obtaining urine for culture, results of urine cultures, and presence of an intravesical catheter. The definite source of S. aureus bacteremia was established on the basis of patient charts. In patients with a clinical diagnosis of endocarditis, we evaluated whether the case complied with the Duke criteria [10].

The diagnosis “obvious UTI” was defined by the following criteria (all 5 criteria had to be met): (1) UTI diagnosed by the clinician, (2) presence of leukocyturia, (3) no other probable focus of bacteremia, (4) presence of S. aureus bacteriuria (difference in susceptibility of ≤1 antibiotic between urine and blood isolate), and (5) presence of either localizing symptoms, a recent surgical procedure or manipulation in the urinary tract, or urinary obstruction. The medical charts of these patients were examined for previous invasive urologic procedures and for urologic neoplasms.

Blood cultures were incubated for a minimum of 5 days at 35°C in an automated detection system. Urine specimens were processed with a dip slide (Uricult; Orion Diagnostics), and, when the results were positive (≥10⁴ cfu/mL), they were subcultured onto blood agar. Identification and susceptibility testing were performed using an automated system and software; the identification was confirmed by testing free and bound coagulase.

The associations between S. aureus bacteriuria and the presumed risk factors were tested for statistical significance by χ² analysis and χ² analysis for trend, using SPSS software, version 12.0 (SPSS).

Results. In the studied 5-year period, a total of 64,256 urine samples were processed at our laboratory. Of the 16,667 samples with positive urine culture results, 8928 were midstream urine samples, and 7739 were catheter-obtained samples. Six hundred eighty-five urine cultures (4.1% of all positive culture results) yielded S. aureus; 222 were midstream urine cultures (2.5% of all midstream urine cultures with positive results), and 463 were catheter cultures (6.0% of all catheter cultures with positive results); samples were obtained from 470 patients (167 patients for midstream urine samples and 303 patients for catheter-obtained samples). During this period, 42,101 sets of blood samples for culture were obtained, of which 5519 yielded positive results, and 950 yielded S. aureus, from 515 patients.

In the total group of patients with S. aureus bacteriuria (n = 470), 64% had urinary catheters in place, and 9.6% had S. aureus bacteremia. This confirms previous observations that intravesical catheters are an important risk factor for S. aureus bacteriuria (table 1).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 1

Relationship between Staphylococcus aureus bacteriuria, S. aureus bacteremia, and urinary tract infection in different studies.

In group 1 (table 2), 153 patients had urine culture samples obtained on the same day that S. aureus bacteremia was detected and did not have a diagnosis of UTI; of these culture samples, 79 were midstream urine samples, and 74 were catheter-obtained urine samples. Twelve patients (7.8%) had S. aureus bacteriuria; 10 of these patients had intravesical catheters. The likelihood of S. aureus bacteriuria in the case of S. aureus bacteremia and in the absence of a catheter was, therefore, 2.5% (2 of 79 patients). Of note, in 1 of these 2 patients (a patient undergoing hemodialysis who had oliguria and renal failure associated with glomerulonephritis), it was questionable whether S. aureus bacteriuria could be attributed to S. aureus bacteremia.

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 2

Clinically identified focus of bacteremia in groups 1 and 2.

The only independent risk factor for S. aureus bacteriuria among patients with S. aureus bacteremia was the presence of indwelling urinary catheters (OR, 6.0; 95% CI, 1.1–41.4). We found no association of S. aureus bacteriuria with age, endocarditis, or number of positive blood culture results.

In group 2 (table 2), 23 patients had S. aureus bacteremia and S. aureus bacteriuria on different days. Eighteen patients (78%) had urinary catheters in place. In 13 patients, documented S. aureus bacteriuria preceded S. aureus bacteremia (median interval, 5 days; range, 1–140 days), and in 10 patients, S. aureus bacteremia preceded S. aureus bacteriuria (median interval, 19 days; range, 1–162 days).

Five patients in group 2 did not have urinary catheters in place. Of these patients, 1 had severe kidney disease (acute rejection of a kidney transplant) and was oliguric; in 2 patients, the urine culture results became positive >3 months after S. aureus bacteremia; and in 1 patient, bacteriuria (which occurred 5 days before S. aureus bacteremia and sepsis syndrome) may have indicated UTI. This leaves only 1 patient without an identified focus for S. aureus bacteremia in which “unexplained” S. aureus bacteriuria might have been attributable to the bacteremia.

Sixteen (31%) of the 51 patients with S. aureus bacteremia and S. aureus bacteriuria fulfilled the criteria for diagnosis of S. aureus UTI. Twelve of these patients (75%) had urinary catheters in place, 12 (75%) had previously undergone invasive urologic procedures (such as changing of a nephrostomy catheter, placement of a suprapubic urine catheter, radical cystectomy, transurethral resection of the prostate, and renal transplantation), and 8 (50%) had been given a diagnosis of urologic neoplasms.

Seven of the 46 patients with S. aureus bacteremia with an “unknown origin” fulfilled criteria 2–5 for UTI but lacked the treating physician's diagnosis to be qualified as such. Pyuria and bacteriuria were demonstrated in the week previous to the bacteremia in 6 of these patients and on the same day in 1. Six of these patients had urinary catheters in place.

Discussion. Our study is, to our knowledge, the first that specifically tried to identify patients with S. aureus bacteremia and S. aureus bacteriuria who did not have obvious risk factors for S. aureus bacteriuria (specifically, no UTI and no indwelling catheter in place). Of the 79 patients with S. aureus bacteremia who lacked these 2 risk factors, only 2 had positive urine culture results on the day that blood samples were obtained for culture; 1 additional patient had S. aureus bacteremia and S. aureus bacteriuria identified in a midstream urine specimen on separate days. We identified 7 patients who fulfilled all of the criteria for UTI except for criterion 1 (i.e., diagnosis made by the treating clinician). One may speculate that this finding suggests that S. aureus UTI was underdiagnosed as a cause of S. aureus bacteremia.

Because of its retrospective nature, this study has several limitations. Urine samples were not obtained in a protocolized manner, and we could analyze the diagnostic samples only as ordered by the treating physicians; this probably biased our observations toward a higher percentage of patients with UTI. Furthermore, it cannot be excluded that blood samples were obtained for culture first and that urine samples were obtained after antibiotic therapy had been started, therefore yielding false-negative results. Yet all of the previous studies also had this limitation—and probably to a larger extent, because wider intervals (up to 48 h) were accepted. Finally, only the diagnoses of UTI and endocarditis were confirmed on the basis of predetermined criteria.

In short, on the basis of the results of this study—and the evidence in literature—we refute the widely held belief that S. aureus bacteriuria is a common effect of S. aureus bacteremia and that S. aureus bacteriuria in itself justifies extensive source investigations. A prospective study is, however, required to provide a definitive answer to this question.

• Received December 5, 2006.
• Accepted February 21, 2007.

• © 2007 by the Infectious Diseases Society of America