Infection is the second most common cause of death in patients dependent on hemodialysis (HD) [1–4]. Fever is a common presenting symptom of infectious disease, and differentiating HD patients with simple febrile illness from those with potentially life-threatening bacterial infections continues to be a diagnostic challenge [5]. HD patients visit the emergency department (ED) about six times more often than their age-matched counterparts, and the second most common reason for admission to the hospital is septicemia [6, 7].

Micronesians have particularly high rates of hypertension, diabetes, and renal disease compared to other Asian and Caucasian ethnic groups, and a disproportionate burden of kidney failure requiring hemodialysis [8, 9]. The ED in our hospital sees a relatively high number of HD patients presenting with infectious symptoms.

While the burden of initially identifying patients with septicemia often falls on the ED provider, the studies that have been done to evaluate risk factors for potentially life-threatening bacterial disease have been largely retrospective studies from already-confirmed cases of bacteremia [10–16].

The aim of this study was to determine the rate of bacteremia in HD the patient presenting to the ED with fever and to identify independent risk factors for bacteremia in this cohort.

Over the 3-year study period, 269 unique HD patients made a total of 3424 visits to the ED. This corresponded to 2.6 visits/patient or 0.87 visits/patient/year. Of these visits, 358 (10.5%) were for complaint of fever and 353 (10.3%) prompted a septic workup that were ultimately included in the study. Over the study interval, 51.3% of all known HD patients in our community had one or more visits to the ER for fever.

The 353 visits included in the study were made by 138 unique HD-dependent patients. Average patient age was 54.6 years (SD 12.7 years), and more than half of patient visits were made by women (60.6%). The great majority of patients were ethnically Chamorro (66.9%), and the remainder was other Pacific Islander (18.1%) and Filipino (9.6%). 43.3% were being dialyzed using permanent tunneled catheter, 39.1% native AVF, and 17.6% AVG. 67.4% of patients were admitted to the hospital, 31.2% were discharged from the ED, and 1.4% left against medical advice. Total 7-day mortality was 2.8%, and 28-day mortality was 5.4%. In patients with confirmed bacteremia, 7-day mortality was 3.6%, and 28-day mortality 9.8%.

Outcome

The rate of bacteremia in HD patients presenting to the ED was 31.7%. Gram-positive organisms accounted for 61% of positive blood cultures, with a predominance of MSSA (36.3%%) followed by MRSA (13.3%). In the Gram-negative bacteremia group, major pathogens included Enterobacter (11.5%), Pseudomonas (11.5%), Escherichia coli (3.5%), and Klebsiella (3.5%) species (Table 2).

	n	(%)
Gram-positive bacteria	69	61.1
S. aureus
nMRSA	41	36.3
MRSA	15	13.3
Other Staph spp. a	5	4.4
Enterococcus spp. b	4	3.5
Streptococcusc	3	2.7
Bacillis spp.	1	0.9
Gram-negative bacteria	44	38.9
P. Aeruginosa	13	11.5
Enterobacter spp. d	13	11.5
E. Coli	4	3.5
K. pneumoniae	4	3.5
Providencia stuartii	4	3.5
Other Gram-negativese	6	5.3
Total number of isolates	113

^aCapitis, auricularis, lungdensis, epidermitis, cohnii

^bFaecium, faecalis

^cGroup A Strep

^dCloacae, aerogenes, absuriae

^eStrenotrophamonas maltophilia, Aeromonas hydrophilia, P. Mirabilis, C. Koseri

By univariate analysis, the factors most strongly associated with bacteremia were hemodialysis catheter use (RR 3.57 (2.49–5.13)), history of bacteremia (RR 3.87 (2.82–5.33)), and bandemia ≥ 5% (RR 2.05 (1.53–2.75)) (Fig. 1).

Compared to the general population, hemodialysis patients have higher rates of bacteremia and associated mortality [17, 18]. Previous work studying this population has relied largely on confirmed cases of bacteremia, but the authors of this study are not aware of any ED-based studies concerned with the evaluation of the yet undifferentiated HD patient in whom bacteremia is only suspected. The primary goal of this study was to evaluate the rate of bacteremia in the HD patient presenting to the ED with fever, in the hopes of developing a more clinically useful framework for clinicians managing initial presentation in the ED and other outpatient settings.

The incidence of bacteremia in the HD patients in our population was found to be similar to that of previous studies [19, 20]. Compared to other hemodialysis patients in Asia, Europe, and the USA, our patients had a higher percentage of comorbidities (DM, HTN, CAD), a higher rate of dialysis catheter use, and were younger on average [21–25]. Mortality from sepsis in bacteremic patients in our study was similar to previous studies, with 7-day mortality of 3.6% and 28-day mortality of 9.8% [16]. Consistent with other studies, the predominant sources of bacteremia were found to be infections of the hemodialysis catheter, lungs, urinary tract, and skin and soft tissue. Gram-positive organisms accounted for the majority of organisms identified in blood culture (61%)—predominantly Staphylococcus aureus—but unlike prior studies, methicillin-sensitive S. aureus was more common than MRSA [26–30].

History of bacteremia and presence of a hemodialysis catheter were found to be the most significant risk factors for bacteremia in HD patients presenting with fever. In our study, more than half (53.6%) of the patients with a catheter presenting with a fever were ultimately found to be bacteremic, and the majority of the bacteremic cases were found to be caused by catheter infections. The risk of infection with hemodialysis catheter has been well documented, and this result is consistent with prior studies [31, 32]. For non-catheter HD patients, the rate of bacteremia was significantly lower (15.0%), and pulmonary and skin and soft tissue sources accounted for the majority of cases of bacteremia. While bandemia greater than 5% had some correlation with bacteremia (RR 2.05 (1.53–2.75)), ultimately, the other diagnostics generally available at first presentation (documented temperature, white blood cell count, neutrophilia) were of minimal significant clinical value as risk factors in predicting bacteremia.

This study has several limitations. This is a retrospective study, from a single center, with a study population that may be difficult to generalize to other clinical settings. Also, only aerobic blood cultures were drawn in this study, so bacteremia caused by obligate anaerobes would have been missed in our data set, though these infections are thought to be relatively infrequent [33]. Finally, there was some data not available for all patients, and it is possible that this introduced some bias into our analysis.

Hemodialysis patients carry a high burden of infectious disease, a high mortality with serious infections, and it remains challenging to predict which patients presenting to the ED with fever are bacteremic. In this study, we found that previous history of bacteremia and presence of a hemodialysis catheter were strong risk factors for the presence of bacteremia. We also found that leukocytosis, documented fever, and neutrophilia offered little predictive value in identifying patients with bacteremia.