If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Divisions of Critical Care Medicine and Infectious Diseases, Faculty of Medicine and Dentistry, University of Alberta, 2–124 Clinical Sciences Building, 8440-112 Street, Edmonton, AB, T6G 2B7, Canada
The epidemiology and outcomes of intensive care unit-acquired Gram-negative bacteremia are described.
•
Resistance to ciprofloxacin and piperacillin/tazobactam was common.
•
Carbapenem resistance among pseudomonal isolates was surprisingly high.
•
Most patients received adequate empiric antimicrobial therapy.
•
Coronary disease, immunosuppression, and antimicrobial therapy predicted mortality.
Summary
Objectives
To describe the epidemiology, antimicrobial susceptibilities, treatment, and outcomes of intensive care unit (ICU)-acquired Gram-negative bacteremia.
Methods
Patients with ICU-acquired Gram-negative bacteremia from 2004 to 2012 were reviewed retrospectively. Independent predictors of mortality were examined using multivariable Cox regression.
Results
Seventy-eight cases of ICU-acquired Gram-negative bacteremia occurred in 74 patients. The infection rate was 0.97/1000 patient-days. Mean patient age was 55 years, 62% were male. The most common admission diagnoses were respiratory failure (34%) and sepsis/septic shock (45%). Mortality was 35% at 30 days. The most common source of bacteremia was pneumonia (33%). Of 83 Gram-negative isolates, Escherichia coli (20%) and Pseudomonas aeruginosa (18%) were most common. For aerobic isolates, susceptibilities to ciprofloxacin (61%) and piperacillin/tazobactam (68%) were low. For pseudomonal isolates, susceptibilities to ciprofloxacin (53%), piperacillin/tazobactam (67%), and imipenem (53%) were equally disappointing. Adequate empiric antimicrobial therapy was prescribed in 85% of bacteremia cases. On multivariable analysis, adequate empiric therapy (adjusted hazard ratio (aHR) 0.38, 95% confidence interval (CI) 0.16–0.89), immune suppression (aHR 3.4, 95% CI 1.4–8.3), and coronary artery disease (aHR 4.5, 95% CI 1.7–11.9) were independently associated with 30-day mortality.
Conclusions
ICU-acquired Gram-negative bacteremia is associated with high mortality. Resistance to ciprofloxacin, piperacillin/tazobactam, and carbapenems was common. Coronary artery disease, immune suppression, and inadequate empiric antimicrobial therapy were independently associated with increased mortality.
with approximately 30% due to Gram-negative bacilli. Mortality rates in Gram-negative sepsis are high, and can be up to 50% depending on patient factors, as well as timing and appropriateness of empiric antimicrobial therapy.
Although intensive care units (ICUs) account for less than 10% of the total number of beds in most hospitals, more than 20–30% of all nosocomial infections are acquired in the ICU,
National Nosocomial Infections Surveillance (NNIS) system report, data summary from January 1992–April 2000, issued June 2000. Am J Infect Control 2000;28:429–48.
NHSN annual update: antimicrobial-resistant pathogens associated with healthcare-associated infections: annual summary of data reported to the National Healthcare Safety Network at the Centers for Disease Control and Prevention, 2006-2007.
when compared with the general hospital population. Understanding the local epidemiology and antimicrobial susceptibility patterns in ICU-acquired Gram-negative bacteremia may facilitate the development of empiric therapy guidelines, formulary restrictions, and/or antimicrobial stewardship programs, resulting in improved patient outcomes.
The aim of this study was to provide up-to-date data on the epidemiology (including incidence, source, microbial etiology, antimicrobial susceptibilities, and outcomes) of nosocomial Gram-negative bacteremia in a population of critically ill patients. A comparison of rates, microbiology, antimicrobial susceptibilities, and appropriateness of empiric therapy over time was also performed, in particular a comparison to data published previously from the same patient population.
A retrospective observational cohort study was conducted at the University of Alberta Hospital, a 650-bed academic quaternary care referral hospital in Edmonton, Canada and the principal teaching hospital of the University of Alberta, Faculty of Medicine and Dentistry. It is a level-one trauma center, the largest solid organ transplantation center in Western Canada, and has a referral area of over two million.
The University of Alberta Hospital contains a 32-bed adult general systems ICU that admits general medical, surgical, trauma, and solid organ transplant patients. The ICU provides general invasive hemodynamic monitoring and support, mechanical ventilation, and renal replacement therapy, as well as extracorporeal liver support. Other specialized support therapies such as extracorporeal membrane oxygenation (ECMO) are provided in the cardiovascular ICU.
This study was conducted as part of a quality improvement study through Infection Prevention and Control and was therefore exempt from ethics submission as per the Research Ethics Board at the University of Alberta. Infection Prevention and Control (IPC) monitors all inpatient populations prospectively, including ICU patients, for the development of nosocomial infection(s). Positive blood cultures are identified by the clinical microbiology laboratory (BACTEC 9240 Blood Culture System; Becton Dickinson Biosciences) and reviewed by IPC daily. Charts are subsequently reviewed and blood culture isolates categorized as contaminants, community-acquired, or nosocomial using standardized US Centers for Disease Control and Prevention/National Healthcare Safety Network (CDC/NHSN) criteria.
For nosocomial bacteremia, the originating source is then identified and subcategorized.
All cases of ICU-acquired Gram-negative bacteremia from January 1, 2004 to December 31, 2012 were reviewed. Demographic data, the source of bacteremia, causative microorganisms, antimicrobial susceptibilities, choice of empiric antibiotic therapy, ICU and hospital lengths of stay (LOS), and 30-day mortality rates were collected. Data were subsequently compared to previously published data for the period 1999–2003 in the same patient population.
Immune compromise was defined as chronic steroid use (daily equivalent of 20 mg prednisone for ≥1 month per year), solid organ or hematopoietic stem cell transplantation, receipt of chemotherapy within the previous month, neutropenia (absolute neutrophil count <1.0 × 109 cells/L) at the time of admission, or HIV/AIDS (CD4 count <200 cells/μl). End-stage renal disease (ESRD) was defined as kidney failure necessitating either peritoneal or intermittent hemodialysis prior to admission. Empiric therapy was deemed effective if treatment was with an agent to which the microorganism was ultimately susceptible. Multi-drug resistance in Pseudomonas species was defined as resistance to three or more first-line antimicrobials in the following classes: β-lactams, carbapenems, aminoglycosides (AGs), and fluoroquinolones.
2.1 Statistical analysis
Epidemiological data were reported, including descriptive statistics such as proportions, percentages, means, and medians. Characteristics of subjects who died and survived were compared using the t-test or Mann–Whitney U-test (as appropriate based on their distribution) for continuous variables. The Chi-square test or Fisher's exact test was used for categorical variables.
Kaplan–Meier survival curves were then constructed and log-rank tests performed for variables associated with mortality on univariate analysis.
Multivariable Cox regression modeling was subsequently performed to identify variables independently associated with 30-day mortality. The APACHE II score (Acute Physiology and Chronic Health Evaluation) was included for clinical significance. Otherwise variables identified to be significantly associated with increased mortality (p < 0.1) on univariate analysis were entered into the model. There were no proportional hazards violations, which were tested using time covariate interaction terms. Results are presented as adjusted hazard ratios (aHR) and 95% confidence intervals (CI). All statistical tests were two-sided and p-values of <0.05 were considered statistically significant.
Statistical analyses were performed using IBM SPSS Statistics for Windows, version 22.0 (IBM Corp., Armonk, NY, USA).
3. Results
There were 11 602 admissions to the general systems ICU from January 1, 2004 to December 31, 2012. A total of 78 episodes of ICU-acquired Gram-negative bacteremia occurred in 74 patients, resulting in an overall infection rate of 6.7 per 1000 ICU admissions, or 9.7 per 10 000 patient-days. These rates are similar to those observed in the previous study (6.9 per 1000 ICU admissions and 11.3 per 10 000 patient-days). Annual infection rates fluctuated by year from 1.6 to 9.6 per 1000 admissions and 2.3 to 15.0 per 10 000 patient-days, consistent with expected year-to-year variation.
The mean patient age was 55 years (range 18–85, standard deviation (SD) 15.8 years) and 46/74 (62.2%) were male. A large number of patients (26/74, 35%) were immune compromised. Additional specific patient demographics can be found in Table 1.
The majority of patients were admitted with medical diagnoses (51/72, 69%). Admitting diagnoses included respiratory failure (25/74, 34%), sepsis or septic shock (33/74, 45%), liver transplantation (7/74, 10%), and trauma/major burns (5/75, 7%). The mean APACHE II score was 25 (SD 8.3) at the time of admission, and 73/78 (94%) patients required invasive mechanical ventilation.
The most common source of bacteremia was pneumonia, in 26/78 (33%) patients, as was observed in the previous study. Other common sources included gastrointestinal tract infections (not endoscopy-related) (17/78, 22%) and catheter-related bloodstream infections (10/78, 13%) (Figure 1). A decrease in catheter-related bloodstream infections was observed from 1999–2003 to 2004–2012 (22% vs. 13%, respectively).
Bacterial species identified can be found in Figure 2. In summary, the most common bacterial pathogen was Escherichia coli (17/83, 21%), followed by Pseudomonas aeruginosa (15/83, 18%) and Klebsiella species (13/83, 16%).
The majority of isolates were aerobic Gram-negative bacilli (71/83, 86%), of which 49/83 (59%) were Enterobacteriaceae. Glucose non-fermenting organisms (Pseudomonas and Stenotrophomonas) accounted for 21/83 (25%) cases of bacteremia.
Multidrug-resistant (MDR) pathogens were isolated in 14/83 (17%) cases, including extended-spectrum β-lactamase producers (ESBLs; 5/83, 5%), carbapenem-resistant Enterobacteriaceae (CRE; 2/83, 2%), and MDR Pseudomonas (7/83, 8%).
The majority of infections were monomicrobial (63/78, 81%). In 15 polymicrobial infections, three were with another Gram-negative organism, 10 with Gram-positive organisms, and there were two events in which additional Gram-negative and Gram-positive organisms were identified.
Overall aerobic Gram-negative and pseudomonal susceptibilities are shown in Table 2. Of all aerobic isolates tested, susceptibilities to ciprofloxacin (40/66, 61%) and piperacillin/tazobactam (44/65, 68%) were low. Similar numbers were observed for Enterobacteriaceae alone. For pseudomonal isolates, susceptibility to ciprofloxacin (8/15, 53%), piperacillin/tazobactam (10/15, 67%), and imipenem (8/15, 53%) were equally disappointing. Changes in antimicrobial susceptibilities over time are demonstrated in Figure 3. Comparison of the two study periods demonstrated the following: (1) sustained resistance to ciprofloxacin in 30–50% of isolates (in aerobic isolates as well as Pseudomonas alone), (2) increased resistance to piperacillin/tazobactam in aerobic isolates (from 11% to 32%) and in pseudomonal isolates alone (0% to 33%), and (3) substantially increased carbapenem resistance in Pseudomonas (9% to 47% for imipenem). Surprisingly, only 2/6 (33%) strains of Stenotrophomonas maltophilia were susceptible to trimethoprim/sulfamethoxazole (TMP/SMX).
Empiric therapy was appropriate in 66/78 (85%) cases of aerobic Gram-negative bacteremia and 11/15 (73%) cases of pseudomonal bacteremia. The most commonly prescribed empiric antibiotics were piperacillin/tazobactam (31/78, 40%) and anti-pseudomonal carbapenems (29/78, 37%). After excluding cases of S. maltophilia, for which empiric therapy is not common or standard of care, effective empiric coverage increased to 63/72 (88%). In the six cases of S. maltophilia bacteremia, three (50%) patients received adequate empiric therapy.
The crude 30-day mortality was 35% (26/74) and hospital mortality was 49% (36/74). Data from the previous study demonstrated higher mortality rates (53% and 60%, respectively). Comparing rates over time, there was a trend to decreased 30-day mortality only (p = 0.07) (p = 0.3 for in-hospital mortality). The median ICU LOS was 20.5 days (interquartile range (IQR) 7–33 days), while hospital LOS was substantially longer (49.5 days, IQR 25.3–109 days). Median ICU LOS following the first date of bacteremia was 6.5 days (IQR 2–20 days).
On univariate analysis (see Table 3), immune suppression, steroid use, and adequate empiric antimicrobial therapy were associated with mortality. A trend towards increased mortality with pseudomonal bacteremia and coronary artery disease was observed.
Table 3Univariate analysis of 30-day mortality (78 bacteremic events)
On multivariable analysis (see Table 4), adequate empiric antimicrobial therapy (aHR 0.38, 95% CI 0.16–0.89; p = 0.03), immune suppression (aHR 3.44, 95% CI 1.43–8.27; p = 0.006), and coronary artery disease (aHR 4.48, 95% CI 1.70–11.85; p = 0.003) were associated with 30-day mortality.
Table 4Multivariable regression analysis of 30-day mortality (78 bacteremic events)
The incidence of ICU-acquired Gram-negative bacteremia in our general systems ICU has been stable over the past 14 years. Catheter-related bloodstream infections, however, have decreased over time, likely due to major preventative efforts (zero tolerance), including a prevention bundle introduced in 2006, as well as increased education and awareness.
Similar to the previous study by this group, a large cohort of immune suppressed patients was identified, which is not surprising for a quaternary care ICU with expertise in solid organ transplantation. In fact 13/74 (18%) patients were postoperative liver transplant recipients. In this subgroup, a higher risk-adjusted mortality was demonstrated. Several studies have demonstrated similarly poor outcomes in immunocompromised critically ill patients who develop sepsis,
The most common source of ICU-acquired Gram-negative bacteremia was pneumonia in both study periods, as has been observed in a number of other studies.
Consistently high rates of pseudomonal infection (22% and 18% of all isolates for both study periods) were also observed. MDR organisms were common (17% of all isolates) and included ESBL producers, CREs, and MDR pseudomonal infections.
Perhaps of most interest, antimicrobial susceptibility patterns demonstrated persistent and increasing resistance over time. Ciprofloxacin resistance remained high among pseudomonal isolates, making this drug an unattractive choice for empiric therapy. Ciprofloxacin resistance in Pseudomonas has been increasing steadily over the past decade due to high fluoroquinolone use in both community and hospital settings,
Relationship between fluoroquinolone use and changes in susceptibility to fluoroquinolones of selected pathogens in 10 United States teaching hospitals, 1991-2000.
this was not observed in the present data. Ciprofloxacin resistance among all aerobic isolates (including the Enterobacteriaceae) was high, essentially ruling out ciprofloxacin as an option for empiric therapy in ICU-acquired Gram-negative bacteremia.
More concerning was the substantial increase in both piperacillin/tazobactam and carbapenem resistance over time. This resistance was demonstrated in all aerobic isolates as well as pseudomonal isolates alone. Other studies have also demonstrated high rates of carbapenem resistance in Enterobacteriaceae (predominantly Klebsiella pneumoniae), as well as P. aeruginosa, which relates mainly to prior patient antimicrobial exposure and also to unit antimicrobial pressure.
Although AGs demonstrated the highest activity overall, substantial resistance to these agents was also observed, with approximately 17% resistance among aerobic isolates and up to one quarter resistance among pseudomonal isolates.
It is postulated that the increasing resistance observed may be due to the widespread use, as well as longer than necessary, or even unnecessary, courses of antimicrobial therapy in critically ill patients in general. The use of piperacillin/tazobactam or anti-pseudomonal carbapenems as empiric therapy in 77% of the cases presented herein demonstrates that these are preferred agents in our institution. In addition, the increasing incidence of MDR pathogens such as ESBLs and CREs likely contributed to the observed increase in antimicrobial resistance, as these organisms may be inherently resistant to specific antimicrobials (e.g., ESBLs to piperacillin/tazobactam). The incidence of drug-resistant Gram-negatives has become a global problem for a number of reasons,
Population distribution of beta-lactamase conferring resistance to third-generation cephalosporins in human clinical Enterobacteriaceae in the Netherlands.
particularly in ICUs, and will pose more and more of a challenge in the future given the lack of investment in antimicrobial research and development.
Despite increasing antimicrobial resistance, most patients (85%) received appropriate empiric therapy. After excluding cases of Stenotrophomonas, for which empiric therapy would not be considered standard of care, even higher effective therapy rates (88%) were observed. To compare, a recent study by Shorr et al. observed lower (66%) rates of appropriate therapy in patients with Gram-negative bacteremia and severe sepsis or septic shock.
Nonetheless, 85% is still suboptimal, as 15% of patients received ineffective empiric therapy for a serious, life-threatening infection. The results of this study, however, serve to provide us with the knowledge required to improve care. Effectiveness of empiric therapy is indeed an important quality indicator; recent studies have shown that both inadequate empiric therapies, as well as delays in therapy, are associated with increased mortality.
Bloodstream infections caused by antibiotic-resistant Gram-negative bacilli: risk factors for mortality and impact of inappropriate initial antimicrobial therapy on outcome.
Based on the data presented here, empiric therapy with an AG in addition to a carbapenem or ureidopenicillin is recommended for ICU-acquired Gram-negative bacteremia in our centre. However, the local epidemiology must be considered in each unit and empiric therapy prescribed based on local susceptibility patterns and patient risk factors. Regarding the short-term safety profile of AGs, the risk of nephrotoxicity and ototoxicity is minimal if administered for 24–48 h as a bridge to definitive therapy until susceptibility results are available. Despite the low risk of toxicity with one to two doses, many intensivists are still very reluctant to prescribe AGs even short-term, particularly in patients presenting with acute kidney injury. We would strongly argue that the risks of ineffective empiric therapy in patients with ICU-acquired Gram-negative bacteremia far exceed the risks of short-course AG therapy, and we would urge intensivists to reconsider their prescribing practices. However, when prescribing longer courses of AGs, which can result in toxicity due to cumulative doses, a more careful evaluation of risks versus benefits is necessary.
Once susceptibilities are confirmed, it is also imperative to narrow and discontinue antimicrobial therapy at the earliest time possible to decrease adverse impacts on both the individual patient (alterations in microbiome) and the unit microbial ecology. Data from several studies suggest that restricting antimicrobial use, with respect to duration and breadth, can decrease resistance rates.
Clearly, with increasing resistance this needs to be a priority. The institution of antimicrobial stewardship programs can be an effective way to decrease unnecessary and inappropriate antibiotic use and should be instituted in all ICUs.
An antimicrobial stewardship program improves antimicrobial treatment by culture site and the quality of antimicrobial prescribing in critically ill patients.
The observed mortality rates remained high in this patient population, with a 30-day mortality of 35%. However, when compared to the previous data for the period 1999–2003 (30-day mortality 53%), a substantial 18% reduction was observed, with a trend to statistical significance. This hopefully reflects improvements in care, but it is recognized that this may also be due to any number of confounders.
Independent predictors of mortality, identified on multivariable analysis, included coronary artery disease, immune suppression, and effective empiric antimicrobial therapy. Other studies have demonstrated similar results,
Bloodstream infections caused by antibiotic-resistant Gram-negative bacilli: risk factors for mortality and impact of inappropriate initial antimicrobial therapy on outcome.
Coronary flow reserve is associated with tissue ischemia and is an additive predictor of intensive care unit mortality to traditional risk scores in septic shock.
Coronary flow reserve is associated with tissue ischemia and is an additive predictor of intensive care unit mortality to traditional risk scores in septic shock.
Despite an association on univariate analysis, pseudomonal infection was not predictive of mortality on multivariable analysis. This may be due to lack of power given the relatively small sample size of the study, as pseudomonal infection has clearly been associated with inferior outcomes in a number of studies.
Lastly, it may seem surprising that APACHE II scores did not correlate with mortality in this study; however it is important to recognize that APACHE II scores were collected within the first 24 h of ICU admission, which was temporally distinct from the time of ICU-acquired bacteremia (which by definition occurred >48 h after ICU admission). So although the APACHE II score was included in the model for exploratory purposes (as it captures chronic as well as acute disease), the lack of association was not surprising.
Despite some strengths of this study, several limitations are also recognized. First, all observational data are inherently at risk of confounding even with rigorous risk adjustment. In this study, the number of bacteremic events was small (78 total events) with only 26 deaths at 30 days, thus there were limitations in the ability to risk-adjust and the study may have been underpowered to detect predictors of mortality. The small size of the study also limited the ability to perform sensitivity analyses. For example, analyses in specific subgroups such as those with immune suppression, septic shock, or specific microbial etiologies, were not feasible. Although this is a weakness of the study, we are proud to report so few nosocomial bloodstream infections over the past decade, as this reflects high quality care provided in our institution.
Second, empiric therapy was deemed to be adequate if treatment was prescribed with an effective agent once positive blood culture results were reported (all results are called directly to the bedside nurse at our institution); however actual antimicrobial administration times may have varied. Delays in effective antimicrobial administration have been associated with poor outcomes in a number of studies,
Bloodstream infections caused by antibiotic-resistant Gram-negative bacilli: risk factors for mortality and impact of inappropriate initial antimicrobial therapy on outcome.
Empiric antibiotic treatment reduces mortality in severe sepsis and septic shock from the first hour: results from a guideline-based performance improvement program.
and clearly could have confounded the present results. In addition, specific doses of empiric antimicrobial therapies were not collected. Antimicrobial under-dosing (due to obesity or inaccurate estimation of creatinine clearance), for example, may have occurred and altered patient outcomes. Lastly, the single-center design also limits generalizability of the results, particularly given the high proportion of immune suppressed (predominantly solid organ transplant) patients in the population presented.
In conclusion, ICU-acquired Gram-negative bacteremia is associated with high mortality. The most common source of ICU-acquired Gram-negative bacteremia continues to be pneumonia. Resistance to ciprofloxacin remains high, while resistance to piperacillin/tazobactam has increased substantially over the past decade. A substantial increase in carbapenem resistance in Pseudomonas aeruginosa was also observed. Despite increasing antimicrobial resistance, most patients received adequate empiric antimicrobial therapy. It is suggested that empiric treatment regimens be based on unit-specific data. Coronary artery disease, immune suppression, and inadequate empiric antimicrobial therapy were independently associated with increased 30-day mortality.
Acknowledgements
We thank the infection control practitioners at the University of Alberta for their collection of data.
Conflict of interest/source of funding: None declared for all authors.
References
Colpan A.
Akinci E.
Erbay A.
Balaban N.
Bodur H.
Evaluation of risk factors for mortality in intensive care units: a prospective study from a referral hospital in Turkey.
National Nosocomial Infections Surveillance (NNIS) system report, data summary from January 1992–April 2000, issued June 2000. Am J Infect Control 2000;28:429–48.
NHSN annual update: antimicrobial-resistant pathogens associated with healthcare-associated infections: annual summary of data reported to the National Healthcare Safety Network at the Centers for Disease Control and Prevention, 2006-2007.
Relationship between fluoroquinolone use and changes in susceptibility to fluoroquinolones of selected pathogens in 10 United States teaching hospitals, 1991-2000.
Population distribution of beta-lactamase conferring resistance to third-generation cephalosporins in human clinical Enterobacteriaceae in the Netherlands.
Bloodstream infections caused by antibiotic-resistant Gram-negative bacilli: risk factors for mortality and impact of inappropriate initial antimicrobial therapy on outcome.
An antimicrobial stewardship program improves antimicrobial treatment by culture site and the quality of antimicrobial prescribing in critically ill patients.
Coronary flow reserve is associated with tissue ischemia and is an additive predictor of intensive care unit mortality to traditional risk scores in septic shock.
Empiric antibiotic treatment reduces mortality in severe sepsis and septic shock from the first hour: results from a guideline-based performance improvement program.
00162-9&id=gr1.jpg){kind=link}
00162-9&id=gr2.jpg){kind=link}
00162-9&id=gr3.jpg){kind=link}