Pseudomonas aeruginosa is an important cause of healthcare-associated infections. In the United States, it is the sixth most common cause of healthcare-associated infections, accounting for 7.1% of all hospital infections.Reference Magill, Edwards and Bamberg 1
The choice of empirical antibiotics in the intensive care unit (ICU) setting is difficult. There needs to be a balance between excessively broad coverage and too narrow coverage. Empirical antibiotic coverage that covers P. aeruginosa but is broader than necessary may lead to the emergence of P. aeruginosa and other intestinal bacteria that are resistant to those broad-spectrum antibiotics. In contrast, empirical therapy that does not cover P. aeruginosa may lead to poor outcomes for ICU patients eventually found to have P. aeruginosa infection. Improvements in our understanding of which patients require broad-spectrum empirical coverage versus situations in which narrower-spectrum agents may be appropriate would be valuable from an antimicrobial stewardship perspective.
Knowledge of whether a patient is colonized with P. aeruginosa can be helpful in guiding selection of empirical antibiotics for suspected sepsis in the ICU setting. Colonization with P. aeruginosa is associated with subsequent infection with the same strain of P. aeruginosa, Reference Thuong, Arvaniti and Ruimy 2 , Reference Nesher, Rolston and Shah 3 but few studies have assessed the prevalence and predictors of P. aeruginosa colonization at admission. The objectives of this cohort study were as follows: (a) to determine the prevalence of P. aeruginosa colonization on ICU admission, (b) to determine risk factors for P. aeruginosa colonization, and (c) to determine the incidence of subsequent clinical culture with P. aeruginosa among those colonized and not colonized.
METHODS
Study Population and Sample Collection
We conducted a cohort study of patients admitted to the medical or surgical ICUs at the University of Maryland Medical Center from January 1, 2013, through December 31, 2013. Patients in the medical and surgical ICUs had admission, weekly, and discharge perirectal cultures performed as part of an active surveillance program for vancomycin-resistant enterococci infection prevention. The hospital is an 816-bed tertiary care facility. The medical ICU is a 29-bed unit that provides care to adult patients who have acute or potentially life-threatening medical conditions, including hematologic and other malignant tumors. The surgical ICU is a 19-bed unit admitting adult patients after surgery and with surgical complications. The primary outcome was presence of P. aeruginosa on ICU admission swab. Patients who did not have admission swabs were excluded. Patients with multiple admissions to either of the ICUs during the study period were allowed to enter the cohort as at-risk patients multiple times, as long as they were not positive for P. aeruginosa on any prior ICU admissions. This study was approved by the institutional review board of the University of Maryland, Baltimore.
Microbiologic Methods
Swabs were placed in tryptic soy broth (BD) and 15% glycerol and frozen at −80°C. The freezing method that we used has been validated and published.Reference Green, Johnson and Furuno 4 , Reference Lautenbach, Santana and Lee 5 Frozen swabs were thawed and 100 μL of tryptic soy broth with 15% glycerol was placed in 5 mL tryptic soy broth and incubated overnight at 37°C. The next day, 50 μL of tryptic soy broth was plated onto cetrimide agar (Remel). After overnight incubation, colonies that were blue-green or yellow-green were identified by Vitek 2 Compact (bioMérieux).
Risk Factors Analyzed
Risk factors analyzed included comorbidities at the time of hospital admission, age, sex, antibiotics received during current hospitalization before ICU admission, and type of ICU. Antibiotic exposures were analyzed as binary variables. Comorbidities were classified using International Statistical Classification of Disease, Ninth Revision, codes and admission medications; underlying comorbid diseases were analyzed as individual components and as part of composite scores as determined using the Elixhauser comorbidity index and the Chronic Disease Score. We used Quan’s enhanced International Statistical Classification of Disease, Ninth Revision, Clinical Modification, code to calculate the Elixhauser index, an aggregate comorbidity measure, using discharge codes as indicators for comorbid conditions.Reference Quan, Sundararajan and Halfon 6 , 7 The Elixhauser index contains 31 comorbid conditions and assigns each patient a score between 0 and 31. To determine the Chronic Disease Score, pharmacy records of patient medications ordered during the first 24 hours of a hospital admission were used as indicators for preexisting comorbid conditions.Reference Von Korff, Wagner and Saunders 8 Data contained within the tables of this repository have been validated for this and other research studies and were found to have positive and negative predictive values greater than 99%.Reference Pepin, Thom and Sorkin 9 – Reference Harris, Fleming and Bromberg 11 In addition, a random sample of 2% of records had all data elements validated and the accuracy of the data was 100% for this data set.
Subsequent Clinical Culture Positivity
For the cohort, we assessed the proportion of clinical culture positivity with P. aeruginosa on the same ICU admission and on the same hospital admission. We compared these proportions between patients colonized and those not colonized with P. aeruginosa. We then determined what proportion of the patients with clinical culture–positive samples represented actual infection using National Healthcare Safety Network definitions. 12 To accomplish this, 2 infectious disease physicians (S.L. and A.D.H.) reviewed each medical record and classified each isolate detected from a clinical culture as being a true infection or a colonization.
Statistical Analysis
Initial bivariable statistical comparisons were conducted by using the χ2 test for categorical data and the t test or Wilcoxon test for continuous data. We calculated odds ratios and 95% CIs using multivariable logistic regression. Because patients were allowed to enter the study multiple times, we also assessed the need to control for the correlated error structure of the data. This correlated analysis did not yield different results. All variables that were associated with the outcome colonization in the bivariable analysis at the P<.1 level were included in the model-building stages of the multivariable analysis. Variables were retained in the final model if they were significant at a P<.05 level or if they were observed to have a confounding effect on the association between another predictor and P. aeruginosa colonization status. We calculated an incidence risk ratio with 95% CI of subsequent positive clinical culture given P. aeruginosa colonization at admission. Statistical analysis was performed with SAS, version 9.3 (SAS Institute).
RESULTS
During the study period, 1,840 admissions had admission perirectal cultures and were included in this study. Compliance with obtaining perianal surveillance culture samples at ICU admission was 93%. A total of 1,538 patients (84%) had only 1 ICU admission, and 135 patients had repeated admissions. Some of these 135 patients had more than 2 admissions. The cohort consisted of 1,461 admissions to the medical ICU (79.4%) and 379 admissions to the surgical ICU (20.6%). The mean age of the patients was 57.5 years. The mean (SD) comorbidity score was 5.4 (3) as measured by the Elixhauser and 7.8 (4) as measured by the Chronic Disease Score. Median length of stay in hospital prior to ICU admission was 4.5 hours. Because this period was so short, we did not include antibiotic exposure in this period in the analysis.
In this cohort, 213 patients (11.6%) had the primary outcome of being colonized with P. aeruginosa on ICU admission. The results of the bivariable analysis are shown in Table 1. The mean Elixhauser comorbidity index among those with P. aeruginosa colonization on admission was 5.8 whereas among those not colonized it was 5.3 (P=.005). The mean Chronic Disease Score among those colonized on admission with P. aeruginosa was 7.9 whereas among those not colonized it was 7.7 (P=.55). The results of the multivariable analysis are shown in Table 2. Significant risk factors in the multivariable analysis for P. aeruginosa colonization were age (odds ratio, 1.02 [95% CI, 1.01–1.03]), anemia (1.90 [1.05–3.42]), and neurologic disorder (1.80 [1.27–2.54]).
TABLE 1 Chronic Disease Score (CDS), CDS-ID Components, Elixhauser Score, and Elixhauser Components for Patients With or Without Pseudomonas aeruginosa
NOTE. Data are no. (%) of patients unless otherwise indicated. CDS and Elixhauser components were included in the table only if P<.20. ID, infectious disease; IQR, interquartile range.
a Time at risk: time in hospital prior to intensive care unit admission.
b CDS components not shown: anticoagulants, cardiac agents (including angiotensin-converting-enzyme [ACE] inhibitors), loop diuretics, isoproterenol, beta-adrenergic, xanthine products, bronchodilators and mucolytics, epinephrine, glucocorticoid, gold salts, antihypertensives and calcium channel blockers (excludes ACE inhibitors), beta-blockers and diuretics, cimetidine, ophthalmic miotics, antitubercular agents, calcitrol, calcium acetate, hematopoetic agents, opioid agonists, narcotic antagonists, and immunosuppressive agents.
c Elixhauser components not shown: peripheral vascular disorder, hypertension uncomplicated, chronic pulmonary disease, diabetes complicated, peptic ulcer disease excluding bleeding, human immunodeficiency virus/AIDS, lymphoma, rheumatoid arthritis/collagen, coagulopathy, fluid and electrolyte disorders, blood loss anemia, psychoses, and depression.
TABLE 2 Adjusted Predictors of Colonization with Pseudomonas aeruginosa in Study of 1,840 Patients at Intensive Care Unit Admission
a Adjusted for deficiency anemia, other neurologic disorders, and age.
Among the 213 patients colonized with P. aeruginosa on admission, 41 (19.2%) had a subsequent clinical culture positive for P. aeruginosa on ICU admission and 60 (28.2%) had a subsequent clinical culture positive for P. aeruginosa on the current hospital admission (ICU period and post-ICU period). Thus, 3.3% of the entire cohort (60 of 1,840) had positive clinical cultures for P. aeruginosa. These 60 patients had 170 clinical cultures positive on the current hospital admission. The sources for these 170 clinical cultures were 71 (42%) sputum, 48 (28%) bronchial culture, 15 (9%) urine culture, 10 (6%) wound culture, 7 (4%) blood, and 19 (11%) miscellaneous. Sixty-five of the 170 clinical cultures had susceptibility tests performed. Susceptibilities of these clinical cultures were as follows: 35% were resistant to piperacillin-tazobactam, 26% were resistant to cefepime, 43% were resistant to imipenem. The clinical cultures occurred with the following frequency after surveillance culture: 25% occurred in the first half-day, another 25% within 5.2 days, another 25% within 14 days, and the remaining 25% after 14 days.
Using the National Healthcare Safety Network definitions, we found that 49 (81.7%) of the 60 patients had clinical infections; 35 had pneumonia, 5 bloodstream infection, 4 intra-abdominal infection, 2 osteomyelitis, 2 surgical site infection, and 1 catheter-associated urinary tract infection.
In contrast, among the 1,627 patients not colonized, only 31 (1.9%) had a subsequent clinical culture positive for P. aeruginosa on ICU admission and 68 (4.2%) had a subsequent clinical culture positive for P. aeruginosa on the current hospital admission. Patients colonized with P. aeruginosa were thus more than 6 times as likely to have a subsequent positive clinical culture than patients not colonized (incidence rate ratio, 6.74 [95% CI, 4.91–9.25]).
DISCUSSION
In this study, we found that 11.6% of ICU patients were colonized with P. aeruginosa on admission. Among these patients, 28.2% had a clinical culture during the same hospital admission with P. aeruginosa. The Elixhauser comorbidity index was higher among patients colonized with P. aeruginosa, and independent risk factors for colonization included age, neurologic disorders, and anemia. Patients colonized with P. aeruginosa were more than 6 times as likely as patients not colonized to have a subsequent clinical culture (indicating likely infection) with P. aeruginosa. This latter percentage identifies the need for clinicians to have a rapid method of identifying which patients are colonized with P. aeruginosa to better guide empirical antibiotic therapy.
Appropriate empirical therapy for P. aeruginosa and other gram-negative bacteria improves patient outcomes.Reference Zilberberg, Shorr, Micek, Vazquez-Guillamet and Kollef 13 , Reference Tumbarello, De Pascale and Trecarichi 14 This is especially true in the era of increasing antibiotic-resistance in gram-negative bacteria. However, the evidence of adverse effects of antibiotics on antibiotic resistance in the human microbiome continues to increase.Reference Zaborin, Smith and Garfield 15 , Reference Modi, Collins and Relman 16 These competing risks create a difficult situation for the antibiotic-prescribing clinician, which in turn creates a need for better testing or prediction rules to help guide empirical antibiotic choice.
Other studies have analyzed risk factors for P. aeruginosa colonization on admission but to our knowledge, none have been done in the ICU setting of the United States. A study in hematology patients identified that 8.2% of patients were positive on admission for P. aeruginosa but less than 1% developed subsequent infection.Reference Sidler, Frei and Tschudin-Sutter 17 A small study in France among 121 ICU patients identified 1.7% as positive on admission.Reference Agodi, Barchitta, Cipresso, Giaquinta, Romeo and Denaro 18 Nesher et alReference Nesher, Rolston and Shah 3 studied 800 stem-cell transplant patients and showed that 7.3% were colonized with P. aeruginosa. They also showed that 32.8% of these patients had subsequent infection with P. aeruginosa.
Our identification of age as a risk factor is biologically plausible; increasing age places individuals at risk for certain bacteria and antibiotic-resistant bacteria.Reference Safdar and Maki 19 Other studies have identified age as a risk factor for antibiotic-resistant Pseudomonas.Reference Ghibu, Miftode, Teodor, Bejan and Dorobat 20 , Reference Tuon, Gortz and Rocha 21 We found 1 study that identified anemia as a risk factor for Pseudomonas bacteremia.Reference Enoch, Kuzhively, Sismey, Grynik and Karas 22 We found 1 study that identified neurologic disease as a risk factor for antibiotic-resistant Pseudomonas infections.Reference Zavascki, Barth and Gaspareto 23 Anemia has previously been identified as a risk factor for bacteremia due to Pseudomonas Reference Enoch, Kuzhively, Sismey, Grynik and Karas 22 and neurologic disease has been identified as a risk factor for antibiotic-resistant Pseudomonas infection.Reference Zavascki, Barth and Gaspareto 23 However, the biological mechanism for this is not clear.
The major limitation of our study is that it is a single site. Another significant limitation is that we did not have accurate data as to whether a patient was admitted to the hospital from a long-term care facility or another healthcare facility, or the number of hospital admissions in the prior year, which may affect the rate of admission positivity.
In conclusion, we envision a day in the near future where either prediction rules or rapid diagnostic testing will help clinicians more appropriately choose empirical antibiotic therapy for both susceptible and antibiotic-resistant bacteria. With this goal, our results significantly add to the literature in identifying a need to determine which ICU patients are colonized with P. aeruginosa on admission.
ACKNOWLEDGMENTS
Financial support. Cubist and Merck; and National Institutes for Allergy and Infectious Diseases (K24AI079040 to A.D.H. and K23AI082450 to K.A.T.).
Potential conflicts of interest. All authors report no conflicts of interest relevant to this article.
