Reappraisal of Community-Acquired Bacteremia: A Proposal of a New Classification for the Spectrum of Acquisition of Bacteremia
- Siegman-Igra Yardena1,
- Fourer Boaz1,
- Orni-Wasserlauf Ruth1,
- Golan Yoav1,
- Noy Aliza1,
- Schwartz David2, and
- Giladi Michael1,2
- 1Infectious Diseases Unit and Tel Aviv University, Israel
- 2Microbiology Laboratory, Tel Aviv Sourasky Medical Center and Sackler Faculty of Medicine, Tel Aviv University, Israel
- Correspondence: Dr. Yardena Siegman-Igra, Infectious Diseases Unit, Tel Aviv Sourasky Medical Center, 6 Weizman St., Tel Aviv 64239, Israel (ZIHUM{at}tasmc.health.gov.il).
Abstract
In recent years, dramatic changes in health care systems have shifted much of the care of sick individuals from hospitals to the community. Consequently, infections traditionally classified as community-acquired or hospital-acquired infections cannot now be readily classified into either category. We thus propose a new classification based on a wider spectrum of acquisition. A total of 1028 episodes of bloodstream infection (BSI) were divided into 5 categories: true community-acquired infections (370 episodes [36%]), infections in recently discharged patients (110 [11%]), infections associated with invasive procedures performed just before or at the time of admission (56 [5%]), infections in patients admitted from nursing homes (68 [7%]), and hospital-acquired infections (424 [41%]). Thus, 234 (39%) of the 604 bloodstream infections traditionally defined as community acquired were reclassified into 3 newly defined groups, each of which has distinct epidemiologic, clinical, and bacteriologic characteristics, as well as distinct antimicrobial susceptibility profiles. There is a conceptual and practical need for such a new classification.
Infection that is detected within the first 48 h of hospitalization is defined as community-acquired infection, whereas infection that occurs later during the course of hospitalization is defined as nosocomial, or hospital-acquired, infection. These simple and easy-to-use definitions are widely used in most clinical studies of bloodstream infections (BSIs). Conceptually, the difference between the 2 types of acquisition is based on the ecologic environment and the presence or absence of medical interventions or devices at the time of acquisition of the bacteremia. A nosocomial infection typically is acquired in an environment of resistant microorganisms or microorganisms typically found in a hospital setting, and it is often associated with a procedure or with instrumentation. A community-acquired infection presumably develops spontaneously, without an association with a medical intervention, and occurs in an environment with fewer resistance pressures.
In reality, some infections are acquired under circumstances that do not readily allow for the infection to be classified as belonging to either of these 2 categories. Such infections include infections that are acquired as a result of invasive procedures performed in the hospital on the day of admission; those that are acquired in nursing homes or long-term care facilities; those that occur in ambulatory patients who have long-term invasive devices and/or frequent contact with health care facilities; and those that occur in patients who were discharged from the hospital shortly before the current admission.
During an ongoing surveillance of positive blood culture results, we noticed that many BSIs could not be classified as community- or hospital-acquired infection because they had characteristics of both types of infection. The purpose of the present study was to characterize and estimate the extent of these difficult-to-classify BSIs and to propose a new classification that is based on a wider spectrum of acquisition of bacteremia.
Patients and Methods
The Tel Aviv Sourasky Medical Center, a 1150-bed university hospital located in central Tel Aviv, Israel, serves a mostly elderly population. All bacteremic episodes that occurred in adults during 1997 (a total of 44,423 admissions, excluding those for pediatric, obstetric, and gynecologic services) were prospectively evaluated. Upon being notified of a positive blood culture result, each patient was seen by a member of the Infectious Diseases Unit. Medical records were reviewed, and pertinent demographic, bacteriologic, and clinical data were collected. The presumptive focus (or source) of infection was determined by a senior infectious diseases specialist, who took into consideration each patient's medical history, the findings of physical and laboratory examinations and imaging, and the individual factors associated with acquisition of infection. When information was incomplete—in particular, for cases in which resistant organisms were found—outpatient medical records were reviewed, and nursing home physicians were contacted to obtain the missing information. Diphtheroids, Bacillus species, or coagulase-negative Staphylococcus species that were yielded by single blood cultures were excluded as contaminants, and a few other isolates were excluded as contaminants on an individual basis.
Bacteriologic methods. Blood culture specimens were inoculated into BacT/Alert bottles (Organon Teknika). Identification of the isolates and susceptibility testing were performed by use of standard bacteriologic methods and an automated system (MicroScan; Baxter Healthcare).
Definitions. Bacteremias detected within the first 48 h of admission were classified into 4 groups (groups A–D).
“Group A” included true community-acquired bacteremias found in patients admitted to the hospital from home without having been hospitalized within the past 30 days and without a history of undergoing an invasive procedure either just before or at the time of admission. Bacteremias that occurred in patients receiving long-term dialysis or in patients admitted with intravascular devices were excluded from this group.
“Group B” consisted of bacteremias found in patients who were recently discharged from the hospital (2–30 days before the current admission). Bacteremia that occurred in patients who were discharged from the hospital 1 day before the time of the current admission was classified as hospital-acquired infection.
“Group C” included bacteremias associated with invasive procedures. This group was further classified into 5 subgroups:
“Group C1” included bacteremias that occurred in patients who underwent an invasive procedure (endoscopy, urethral dilatation, etc.) shortly before admission to the hospital and who were admitted because of an infectious complication that resulted from this procedure.
“Group C2” consisted of bacteremias that occurred in patients with noninfectious problems who underwent an invasive procedure (Foley catheter insertion, intravascular line insertion, etc.) at the time of admission to the hospital and who had complicating bacteremia develop within the first 48 h of admission.
“Group C3” included bacteremias that occurred in patients who had long-term intravenous devices in place, who received chemotherapy or parenteral nutrition, and who were admitted to the hospital with BSIs that resulted from the presence of the intravenous devices.
“Group C4” consisted of bacteremias that occurred in patients who had central intravascular lines inserted for hemodialysis and who were admitted to the hospital with BSIs that resulted from the presence of such lines.
“Group C5” included bacteremias that occurred in patients receiving long-term dialysis therapy who developed BSIs that were caused by any source other than an intravascular line.
“Group D” consisted of nursing home—acquired bacteremias that occurred in patients who were admitted to the hospital from nursing homes.
A fifth group, “group E,” consisted of bacteremias that were detected after the first 48 h of admission (including bacteremias detected after admission to another hospital, if the patient was transferred) and were defined as "hospital-acquired bacteremias."
In addition, "infection-related mortality" was defined as death that occurred within 8 days of the date that bacteremia occurred.
Statistical analysis. Patients with bacteremias classified as belonging to each of the 5 classification groups (groups A–E) were compared for binary variables first by use of the χ2 test or Fisher's exact test, as applicable. Whenever a significant result was obtained, pairwise comparisons between the groups were subsequently performed. To control for experiment-related error, 2 different sequential procedures were used for multiple comparisons: (1) the step-down and the step-up Bonferroni method [1, 2], and (2) the more powerful Benjamini and Hochberg adjustment [3]. The Bonferroni method imposes a strict control on the probability of obtaining a type I error (i.e., a false-positive result) at the expense of obtaining a higher rate of type II error (i.e., the probability of not detecting a true difference). Because the classical Bonferroni method is often regarded as too conservative, the more liberal Benjamini and Hochberg adjustment, which has a lower rate of type II error, was also used [3]. Thus, significant comparisons identified by the Bonferroni method were also invariably detected by the Benjamini and Hochberg adjustment, which revealed additional significant differences that were also regarded as possibly meaningful.
Patients with bacteremias classified as belonging to each of the 5 groups were compared for age by one-way analysis of variance. The Ryan-Einot-Gabriel-Welsch multiple-range test was used for pairwise comparisons between groups, controlling for the experiment-related error [4].
Results
During 1997, a total of 1028 episodes of bacteremia occurred among 912 patients who were admitted to the hospital 964 times. Six hundred four BSIs (59%) were detected in 569 patients (604 admissions) within the first 48 h of hospitalization; included in this group of patients were 2 patients with community-acquired endocarditis who had bacteremia detected on days 3 and 4 of hospitalization, when blood samples were obtained for culture for the first time. The 604 bacteremic episodes, which presumably represented community-acquired infections, were classified into 4 groups according to the circumstances of acquisition, as previously defined in the Patients and Methods section (see also table 1). Table 2 shows the types of invasive procedures performed either before or at the time of admission for patients with group C1 or C2 bacteremia. Because all subgroups of group C (table 1) represented a similar concept of infection acquisition, and because each subgroup was relatively small, it was decided to combine all of these subgroups in the "group C" classification (procedure-related infection). Four distinct modes of non–hospital-acquired infection were thus defined: true community-acquired infection, recent hospitalization–related infection, procedure-related infection, and nursing home–acquired infections.
Figure 1
Graph comparing the patterns of susceptibility to 5 selected antimicrobial agents for 622 Enterobacteriaceae isolates, according to the 5 classifications of bacteremia defined in the "Definitions" subsection of the Patients and Methods section of the text. acq., acquired; gr, group.
Table 1
Distribution of 604 episodes of community-acquired bacteremia that occurred within 48 h of admission to the hospital, according to the proposed new classification of circumstances of acquisition of infection.
Table 2
Types of invasive procedures that were performed either before or at the time of admission to the hospital and that were associated with bacteremic infectious complication within 48 h of admission.
Of the 604 bacteremic episodes that occurred within 48 h of admission to the hospital, 234 (39%) did not meet the definition of true community-acquired infection and were reclassified, according to the new definitions, as belonging to group B, C, or D (table 1). The 1028 bacteremic episodes were analyzed according to demographic, clinical, outcome-related, and bacteriologic variables. The groups were compared in order to examine the unique features of each category.
Age, sex, outcome, and underlying conditions. The mean patient age differed significantly only between the older patients with nursing home–acquired BSIs and patients with bacteremias classified as belonging to any of the other groups (P < .001). The ratio of male to female patients was highest for individuals with procedure-related infections. The rate of infection-related mortality was significantly lower for patients with true community-acquired infection than for those with nursing home– or hospital-acquired infections (P < .0002) (table 3). There was no difference in the mortality rates among recently discharged patients, nursing home residents, and patients with hospital-acquired infections. Presence of malignant diseases and receipt of chemotherapy were more prevalent among recently discharged patients than among any other patient group (P < .02–.0001) (table 4).
Table 3
Comparison of demographic and outcome variables for 912 patients with 1028 episodes of bacteremia, according to bacteremia classification.
Table 4
Comparison of main underlying conditions (including the receipt of immune-suppressing therapy) associated with 1028 episodes of bacteremia, according to bacteremia classification.
Sources of infection. The most common source of bacteremia was urinary tract infection (associated with 39% of all bacteremias), but its prevalence differed substantially among patients with the various classifications of bacteremia. Urinary tract infection was the source of more than one-half of the true community- and nursing home–acquired BSIs, whereas it constituted the source of only a small proportion (16%) of the procedure-related group (P < .0001), where vascular device–associated infection was the most common source (46%). Skin and soft-tissue infection as the source of BSI had the highest representation among patients admitted to the hospital from nursing homes (18%; primarily as a result of pressure sores), but this and any other additional comparison of the distribution of sources of infection among the classification groups did not reach statistical significance (table 5).
Table 5
Comparison of sources of infection for 1028 episodes of bacteremia, according to bacteremia classification.
Types of microorganisms. A total of 1113 isolates were recovered from blood samples from the 1028 episodes of bacteremia; 79 (7.7%) of 1028 episodes were polymicrobial. Of the 1113 isolates, 754 (68%) were gram-negative organisms, 336 (30%) were gram-positive organisms, and 23 (2%) were fungi. The microbial blood isolates differed significantly among the groups. The microorganisms most commonly recovered from patients with true community-acquired infection were Escherichia coli and Streptococcus pneumoniae (associated with 49% and 9% of such infections, respectively), whereas Staphylococcus aureus was the most common isolate in procedure-related infections (associated with 22% of such infections). Klebsiella pneumoniae and S. aureus were the species most commonly found in patients with hospital-acquired BSIs, whereas individuals from nursing homes had a high representation of microorganisms from the Proteus-Morganella-Providencia group (table 6).
Table 6
Comparison of types of microorganisms among 1113 isolates recovered from 1028 bacteremic episodes, by bacteremia classification.
Antimicrobial susceptibility. Susceptibility test results of 622 isolates of the Enterobacteriaceae family to 5 commonly used antimicrobial agents (cefuroxime, ceftriaxone, ciprofloxacin, gentamicin, and ceftazidime) were clearly different among the 5 classification groups. Bacteria isolated from patients with true community-acquired infection were invariably more susceptible to antimicrobials, and the differences were highly significant when compared with nursing home–acquired or hospital-acquired isolates (figure 1). No differences in susceptibilities were noted in a comparison of nursing home–acquired and hospital-acquired infections. Forty-three (33%) of the 132 S. aureus isolates were methicillin resistant; the highest rate of methicillin-resistant S. aureus isolates was observed among nursing home residents (7 [64%] of 11 S. aureus isolates; table 6).
Discussion
The classical definition of nosocomial, or hospital-acquired, infection requires that the infection not be present or incubating at the time of admission to the hospital [5, 6]. The more practical definition—and the one that is used in almost all studies of nosocomial BSIs—is any significant positive blood culture result that is obtained ã48 h after hospitalization [7, 8]. Accordingly, any BSI that is detected within the first 48 h of admission is defined as community acquired (unless the patient was transferred from another hospital). This wellunderstood and easy-to-use definition has been in use for decades, since the earliest classical surveys of positive blood cultures results were performed, and only slight variations have been made in the wording or exact time cutoff used. For example, for nosocomial infection, time-cutoff definitions in the literature have varied: ã48 h after admission [9,10–11], on or after the third day of hospitalization [12, 13], ≥72 h after admission [14,15–16], and even after 5 days of hospitalization [17].
The dramatic changes in the health care system that have occurred in recent years have shifted a significant part of the care for sick individuals from the hospital to the community. This trend, as well as many other factors (partially listed below), has created a large population of ambulatory patients that need frequent contact with health care systems. The factors alluded to are such conditions, procedures, and treatments as organ and bone marrow transplantation, with its associated immune suppression; aggressive chemotherapy for malignant diseases; widespread use of intravenous lines, including long-term lines for hemodialysis, chemotherapy, or nutrition; the performance of more-complicated procedures in ambulatory settings; the trend to shorten postoperative hospitalization; and the aging of the general population. Consequently, the sharp distinction between hospital and community acquisition of infection has become less clear, with some infections having mixed characteristics of both types of acquisition. The definitions proposed in the present report were modified to cope with this evolving process.
Although some of the concepts in this classification have been mentioned in the past, they have not been translated into common practical use. For example, infections occurring among patients who were recently discharged from the hospital have been separated from community-acquired infections and have been defined as nosocomial infections in only a few studies [18,19,20–21]. However, our study showed that, in many aspects, this group of patients was different from patients with community-acquired infection, as well as from patients with hospitalacquired infection. Recently discharged patients who were admitted with bacteremia, as compared with those with true community-acquired infection, had (1) a higher prevalence of malignant diseases and chemotherapy use, (2) a higher rate of infection-associated mortality, (3) a lower rate of urinary tract infection as a source of bacteremia (probably in relation to the decreased likelihood of E. coli as a pathogen), and (4) a lower rate of susceptibility to ceftriaxone and cefuroxime among Enterobacteriaceae (possibly because of greater exposure to this type of drugs). Compared with patients with hospital-acquired infection, recently discharged patients had a higher prevalence of malignant diseases and chemotherapy use, a significantly higher rate of S. pneumoniae and E. coli as pathogens, and a lower crude mortality rate (although the rate of infectionassociated mortality was similar).
Bacteremia acquired in nursing homes usually is not separated from other community-acquired infections, although the exceptionally high antimicrobial resistance of organisms in long-term care facilities has been well documented [22,23–24]. Extensive antibiotic use, absence of antibiotic control, and lack of awareness of basic infection-control measures convert such institutions into breeding sites for resistant bacteria. However, we are aware of only a few studies that have classified nursing home–acquired BSIs either separately [25] or with the category of hospital-acquired infections [26]. Our study showed the rates of antimicrobial resistance of Enterobacteriaceae in episodes of nursing home–acquired bacteremia to be similar to those associated with hospital-acquired infections. Also, patients with nursing home–acquired bacteremia were significantly older and had different sources of infection, compared with those who had hospital-acquired infection, but no significant difference was observed with regard to the rate of infection-related mortality. Thus, nursing home–acquired infections clearly are different from community-acquired BSIs and deserve to be classified separately. However, for clinical decision-making purposes, future studies that adopt this approach may choose to combine the group of nursing home–acquired infections with the group of hospital-acquired infections because of the similarities between the 2 groups.
Another unique group of bacteremia consists of procedure-related bacteremias. Although this group has heterogeneous sources of infection, patients with procedure-related bacteremia have in common the presentation of BSI within the first 48 h of admission, although it is acquired in circumstances that are more characteristic of hospital setting. These mixed properties are reflected in the rates of antimicrobial resistance, which are midway between the susceptibility rates of hospital-acquired infections and those of community-acquired infections (figure 1). Not surprisingly, procedure-related bacteremia is associated with the highest rate of coagulase-negative staphylococcal infection. Although 2 studies have classified procedure-related bacteremia [17] or foreign body–associated bacteremia [27] as nosocomial infection, most studies have not distinguished these cases from community-acquired bacteremia.
The 4 proposed groups of non–hospital-acquired infection outlined in the present report represent distinct clinical situations. The differences in patient characteristics, sources of infection, types of microorganisms, and antimicrobial susceptibility patterns illustrate the uniqueness of each of the defined groups and support the need to reclassify bacteremias on the basis of the new groups rather than to unify them under the traditional definition of community-acquired infection.
Thus, we suggest that the classical dichotomy of community acquisition and hospital acquisition no longer holds true. In the present study, we defined 3 additional, distinct acquisition groups that constituted a significant proportion of all BSIs. A total of 234 (23%) of 1028 bacteremic episodes, or 39% of 604 community-acquired episodes, have been reclassified so that they are not considered pure community-acquired infections. In addition to presenting epidemiologic and research implications, this new classification also has practical applications as a tool to support decision-making regarding the choice of empiric therapy for patients admitted to the hospital with BSIs. To our knowledge, this report is the first attempt to systematically define, analyze, and quantify BSIs by use of expanded criteria.
Acknowledgments
We thank Esther L. Shabtai for assistance with the statistical analysis.
- Received May 11, 2001.
- Revision received December 14, 2001.
- © 2002 by the Infectious Diseases Society of America
References
This Article
- AbstractFree
- » Full Text (HTML)Free
- Full Text (PDF)Free
Classifications
Services
Citing Articles
Google Scholar
PubMed
Related Content
Share
- Email this article
-
CiteULike
Connotea
Delicious
Facebook
Google+
Mendeley
Twitter
Published on behalf of
IDSA Members: For your free access to this journal, log in via the IDSA members area.
The Journal
Impact Factor: 8.186
5-Yr impact factor: 7.898
For The Media
For Authors
- Services for authors
- Self-archiving Policy
- IDSA Conflict of Interest policy
- Instructions for Authors
- ICMJE Form
- Submit Now!
Open access options for authors visit Oxford Open
This journal enables compliance with the NIH Public Access Policy
Corporate Services
Alerting Services
Related IDSA Publications
