Customize your JAMA Network experience by selecting one or more topics from the list below.
Kaufman D, Haas CE, Edinger R, Hollick G. Antibiotic Susceptibility in the Surgical Intensive Care Unit Compared With the Hospital-Wide Antibiogram. Arch Surg. 1998;133(10):1041–1045. doi:10.1001/archsurg.133.10.1041
Copyright 1998 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.1998
To compare the antibiotic susceptibility of bacterial isolates from patients in the sugical intensive care unit (SICU) with hospital-wide bacterial susceptibility.
Retrospective cohort analytic study.
Eight-bed SICU in a university-affiliated teaching hospital.
All hospitalized patients with culture results positive for microorganisms.
Main Outcome Measures
Antibiotic susceptibility data were collected retrospectively for all bacterial isolates from SICU patients during July 1, 1994, to June 30, 1995. All duplicate and surveillance cultures were eliminated from the data set. Susceptibility testing was conducted using our standard laboratory methods. Results were compared with the hospital-wide antibiogram (HWA) for the same time period. Comparisons were made using the χ2 test with Yates correction or the Fisher exact test, as appropriate. Staphylococcus aureus (HWA, n=494; SICU, n=71) was significantly less susceptible to oxacillin (51% vs 28%; P <.001), ciprofloxacin (50% vs 25%; P <.001), erythromycin (46% vs 23%; P<.001), and clindamycin (51% vs 27%; P<.001) in the SICU. Coagulase-negative staphylococci (HWA, n=339; SICU, n=37) were significantly less susceptible to oxacillin (33% vs 16%; P=.04) and clindamycin (57% vs 34%; P=.02). Pseudomonas aeruginosa (HWA, n=513; SICU, n=96) was less susceptible to imipenem (85% vs 74%, P =.01) and more susceptible to ticarcillin–clavulanic acid (88% vs 100%, P<.001) in the SICU. Escherichia coli (HWA, n=474; SICU, n=36) was more susceptible to most penicillin-derivative antibiotics in the SICU (ampicillin [68% vs 83%, P=.06], ticarcillin [65% vs 86%, P =.01], mezlocillin [76% vs 95%, P=.01], and ticarcillin–clavulanic acid [88% vs 100%, P=.02]).
The 2 most commonly isolated bacterial pathogens in the SICU (S aureus and P aeruginosa) had significantly different susceptibility patterns compared with the HWA. Surprisingly, E coli isolated in the SICU tended to be more susceptible to penicillin-derivative antibiotics. These data indicate that empiric antibiotic choices in the SICU may be better guided by unit-specific antibiograms.
ANTIBIOTIC resistance is a substantial and growing problem, making the choice of empiric antibiotic therapy increasingly difficult. The potential differences in susceptibility between hospital-acquired and community-acquired infections is widely appreciated, and many clinicians choose initial empiric therapy based on this distinction. Susceptibility of microorganisms causing community-acquired infections is known to vary between geographic locales, while susceptibility of bacterial isolates causing hospital-acquired infections may vary widely among hospitals within the same geographic locale. The application of hospital-wide antibiograms to guide clinicians in the initial choice of antibiotics is a rational and recommended approach, given the differences in susceptibility patterns between hospitals.1 However, susceptibility patterns may vary among individual hospital units. If more resistant organisms are isolated from patients in the intensive care unit (ICU) but not in other hospital units, then this important information could be masked by the use of a hospital-wide antibiogram.2
The emergence of resistance has paralleled the widespread use of antimicrobial agents. The classic example of staphylococci developing resistance within a few years of the introduction of penicillin has been repeated with many other antimicrobial agents throughout the antibiotic era.3 The clinician must be aware of these developments given the high prevalence of nosocomial infections in the ICU compared with other hospital units.4 More specifically, the critical care practitioner must be cognizant of the possibility that the appropriate choice of empiric antimicrobial agents hinges on the unique susceptibility patterns in a geographic locale. This locale may be as broad as a region of a country or as specific as a unit in a hospital.
This study was conducted to compare the susceptibility of bacterial isolates from patients in a surgical intensive care unit (SICU) with the hospital-wide antibiogram.
The SICU is an 8-bed unit in a 535-bed acute care hospital that cares for general, vascular, neurologic, orthopedic, plastic, otolaryngologic, obstetric-gynecologic, and urologic surgery patients. It is 1 of 4 intensive care units in the institution. Antibiotic susceptibility data were collected for all bacterial isolates from SICU patients from July 1, 1994, to June 30, 1995. Hospital-wide antibiotic susceptibility data were also collected for the same period.
Antibiotic susceptibility data were provided for both the SICU and the hospital by the Vitek software package (bioMerieux Inc, Hazelwood, Mo), used routinely in our laboratory for archiving susceptibility data. In an attempt to minimize the effect of duplicate isolates, the program counts isolates according to the following algorithm. Successive isolates from the same anatomical site from a single patient are counted as a single isolate, provided the antibiotic susceptibility pattern is identical. Successive isolates for the same patient that have different susceptibility patterns are counted as separate isolates. The same organism isolated from different sites in a single patient is counted as separate isolates regardless of the antibiotic susceptibility patterns.
Surveillance cultures are not routinely performed in the SICU or elsewhere in the institution, with the exception of rectal swab cultures performed on ICU patients for the detection of vancomycin-resistant enterococci. Because rectal swab cultures are not routinely performed outside of the ICU, and the rectal cultures could not be easily separated from other wound cultures in the microbiology database, all data for enterococcal isolates were eliminated from this analysis.
Antibiotic susceptibility testing of all isolates was performed using our standard laboratory methods (Vitek), using commercially available panels provided by the manufacturer. Interpretations for susceptible and resistant tests were performed according to the National Committe for Clinical Laboratory Standards guidelines.5 Test results interpreted as intermediate were considered resistant for the purposes of this study. All results are presented as percentage of isolates susceptible to the tested antibacterial agents.
Stastistical comparisons between the SICU and hospital-wide susceptibility results were completed using the χ2 test with Yates correction or the Fisher exact test as appropriate for the available data. The a priori level of significance for all statistical tests was P<.05. All statistical tests were performed using Systat software, version 5.02 (Systat Inc, Evanston, Ill).
Table 1 presents susceptibility data for the bacterial isolates from the SICU and hospital, respectively. The method of identifying study isolates resulted in 314 bacterial isolates from 81 SICU patients among a total of 435 SICU admissions, and 2313 isolates from 1202 patients among a total of 20855 hospital admissions during the study period. The most prevalent gram-negative and gram-positive organisms isolated from both the SICU and entire hospital were Pseudomonas aeruginosa and Staphylococcus aureus, respectively. However, P aeruginosa was a significantly more prevalent gram-negative organism in the SICU as compared with the hospital-wide data (46.6% vs 25.9%; P=.001).
For the gram-positive isolates, S aureus was significantly less susceptible to oxacillin (51% vs 28%; P<.001), ciprofloxacin (50% vs 25%; P<.001), erythromycin (46% vs 23%; P<.001), and clindamycin (51% vs 27%; P<.001) in the SICU. Coagulase-negative staphylococci were significantly less susceptible to oxacillin (33% vs 16%, P=.04) and clindamycin (57% vs 34%; P <.02) in the SICU.
For the gram-negative isolates, P aeruginosa was significantly less susceptible to imipenem-cilastatin (85% vs 74%; P=.01) and more susceptible to ticarcillin–clavulanic acid (88% vs 100%; P <.001) in the SICU. Escherichia coli was more susceptible to most penicillin-derivative antibiotics in the SICU: ampicillin (68% vs 83%; P<.06), ticarcillin (65% vs 86%; P =.01), mezlocillin (76% vs 95%; P =.01), and ticarcillin–clavulanic acid (88% vs 100%; P =.02).
In 1988, the Infectious Disease Society of America made a series of recommendations to hospitals to improve the use of antimicrobial agents. The main purpose was to provide the patient with the safest and most effective antimicrobial agents. They suggested that each hospital establish a team concerned with the optimal use of antibiotics.6 Despite the merits of this hospital-wide approach, it is necessary for clinical practitioners to have detailed knowledge of the specific antibiotic susceptibilty patterns in their individual ICU.7
SEVERAL INVESTIGATORS have compared hospital-wide antibiograms to unit-specific antibiograms, noting the trend toward increased resistance in ICUs.8-12 Stratton et al8,9 reported not only differences in ICU and hospital-wide susceptibility patterns, but also important differences in susceptibility patterns for S aureus and P aeruginosa among individual ICUs in the same hospital. Unfortunately, their publications provided no statistical analyses of the data, making it difficult to know the significance of the differences reported.
Bryce and Smith10 compared the hospital-wide antibiogram with the ICU susceptibility pattern for gram-negative organisms. Resistance of P aeruginosa was considerably higher in the ICU, and this was not accurately reflected in the hospital-wide antibiogram. The susceptibility of the Enterobacteriaceae did not seem to be appreciably different between the ICU and the hospital-wide data. This study involved isolates from a combined ICU that housed medical, surgical, and burn patients, and therefore the results of this trial may not easily extrapolate to a specific ICU type such as the SICU.8,9
Gubbins et al11 reported significantly greater antibiotic resistance for many gram-negative organisms isolated from ICU patients compared with isolates from non-ICU patients. However, this study combined data from multiple ICUs, which may mask important differences in susceptibility between ICUs. Also, this study did not compare ICU antibiotic susceptibility with the hospital-wide antiobiogram, but limited the comparison group to non-ICU patients. Because most institutions report antibiotic susceptibility profiles on a hospital-wide basis, we believed it was important to use the hospital-wide data as the comparator to illustrate the limitations of this profile in predicting susceptibility in a specific ICU environment. These authors also did not report the total number of isolates evaluated, making it difficult to know the clinical relevance of the reported statistically significant differences.
In our study, the 2 most commonly isolated bacterial pathogens in the SICU (S aureus and P aeruginosa ) had significantly different susceptibility patterns compared with the hospital-wide antibiogram. Staphylococcus aureus was significantly more resistant to essentially all antistaphylococcal agents with the exception of vancomycin. Although the reason for the variability in staphylococcal resistance is not completely understood, it may be due to selection pressure arising from the common use of antistaphylococcal agents in the SICU.13 The susceptibility of coagulase-negative staphylococci was similarly lower in the SICU; however, the difference did not reach statistical significance for several antibacterials, likely due to a lower level of susceptibility outside of the ICU combined with the smaller sample size.
The susceptibility of P aeruginosa was significantly less for imipenem-cilastatin and greater for ticarcillin–clavulanic acid. However, there were no differences in susceptibility for third-generation cephalosporins and antipseudomonal penicillins. This pattern is similar to the results reported by Stratton et al,8 which did not demonstrate major differences in the susceptibility of P aeruginosa isolates comparing hospital-wide isolates with isolates from the SICU. The reason for this may be related to the fact that P aeruginosa infections are essentially all nosocomial infections, with many patients outside of the ICU developing infections from organisms that may have been originally acquired during an ICU stay earlier in their hospitalization.
Surprisingly, E coli isolated in our SICU were more susceptible to most penicillin-derivative antibiotics, and were highly susceptible to most other tested antibiotics for both the SICU and hospital-wide isolates. A possible explanation for this increased susceptibility in the SICU is that many isolates may come from patients who present to the SICU early in their hospital stay with intra-abdominal or other severe, life-threatening infections caused by their own community-acquired flora. However, for the hospital-wide isolates, E coli may more commonly represent a nosocomially acquired organism with a more resistant profile. Bryce and Smith10 and Gubbins et al11 reported greater resistance among gram-negative species from the ICU compared with non-ICU isolates; however, they did not report increased resistance for E coli . Other published studies8,9 did not provide specific data for E coli isolates.
The major limitation of our study is the small number of isolates available from the SICU for the less prevalent gram-negative organisms. Because of this limited sample size we cannot draw any definitive conclusions concerning differences in susceptibility for these organisms. Also, the number of coagulase-negative staphylococci was relatively small, limiting the ability to detect statistically significant differences for some of the antibiotics. However, the overall susceptibility with most of these antibiotics was low enough that the results would not have affected empiric antibiotic choices.
An additional potential limitation of this study is that the method of data collection could not differentiate between isolates considered to be causing infection and those that represented colonization or contamination. It is possible that, if the data set had been limited to isolates causing infection and therefore requiring antimicrobial therapy, the resulting susceptibility patterns would be different. Once actual infections were determined, specific knowledge about the antibiotics chosen and the duration of therapy would likely explain many of the differences in antibiotic susceptibility. However, collection of this type of data would require extensive retrospective review of the medical records or prospective data collection, which would not be a practical or reasonable method to generate unit-specific antibiotic susceptibility data for routine use in the ICU. These limitations are significant in generating or supporting a theory of selection pressure on the development of antibiotic resistance. However, the practical reality is that the hospital-wide antibiogram is already generated and the unit-specific antibiogram can be easily queried from this database.
In conclusion, hospital-wide antibiograms may mask ICU susceptibility patterns. Unit-specific antibiograms are inexpensive tools that may help guide empiric therapy in the ICU. The results of this study are not intended to provide specific recommendations concerning empiric antibiotic selection in the SICU, because susceptibility will vary from hospital to hospital, but are intended to illustrate the importance of knowing unit-specific antibiotic susceptibility patterns.
Presented as an abstract at the 25th Annual Educational and Scientific Symposium, Society of Critical Care Medicine, New Orleans, La, February 7, 1996.
Reprints: David Kaufman, MD, Rochester General Hospital, 1425 Portland Ave, Rochester, NY 14621-3095.