- Email: [email protected]
The Impact of Nosocomial Infections on Patient Outcomes Following Cardiac Surgery
The Impact of Nosocomial Infections on Patient Outcomes Following Cardiac Surgery* Marin H. Kollef, MD, FCCP; Linda Sharpless, CRNA; Jon Vlasnik, PhannD; Christina Pasque, MD; Denise Murphy, MPH; and Victoria]. Fraser, MD
Study objective: To evaluate the relationship between nosocomial infections and clinical outcomes following cardiac surgery, and to identify risk factors for the development of nosocomial infections in this patient population. Design: Prospective cohort study. Setting: Barnes-Jewish Hospital, St. Louis, a university-affiliated teaching hospital. Patients: Six hundred five consecutive patients undergoing cardiac surgery. Interventions: Prospective patient surveillance and data collection. Main outcome measures: Occurrence of nosocomial infections, multiorgan dysfunction, hospital mortality, and risk factors for the acquisition of nosocomial infections. Results: One hundred thirty-one (21.7%) patients acquired at least one nosocomial infection following cardiac surgery. Four independent risk factors for the development of a nosocomial infection were identified: the duration of mechanical ventilation, postoperative empiric antibiotic administration, the duration of urinary tract catheterization, and female gender. Thirty (5.0%) patients died during their hospitalization. The mortality rate of patients acquiring a nosocomial infection (11.5%) was significantly greater than the mortality rate of patients without a nosocomial infection (3.2%) (odds ratio [OR]=4.0; 95% confidence interval [CI]=2.7 to 5.8; p<0.001). Multiorgan dysfunction was found to be the most important independent determinant of hospital mortality (adjusted OR=23.8; 95% CI=13.5 to 42.1; p<0.001) along with the aortic cross-clamp time (adjusted OR=2.3; 95% CI=l.7 to 3.0; p=0.002) and severity of illness as measured by APACHE II (acute physiology and chronic health evaluation) (adjusted OR=l.1; 95% CI=l.1 to 1.2; p=O.Ol9). Ventilator-associated pneumonia, clinical sepsis, female gender, the cardiopulmonary bypass time, and severity of illness were identified as independent risk factors for the development of multiorgan dysfunction. Among hospital survivors, patients acquiring a nosocomial infection had longer hospital lengths of stay compared to patients without a nosocomial infection (20.1±13.0 days vs 9.7±4.5 days; p<0.001). Conclusions: Nosocomial infections, which are common following cardiac surgery, are associated with prolonged lengths of hospitalization, the development of multiorgan dysfunction, and increased hospital mortality. These data suggest potential interventions for the prevention of nosocomial infections following cardiac surgery that could substantially improve patient outcomes and decrease (CHEST 1997; 112:666-75) medical care costs. Key words: bacteremia; cardiac surgery; intensive care; nosocomial infection; outcomes; pneumonia; minary tract infection; wound infection Abbreviations: AOR=adjusted odds ratio; APACHE=acute physiology and chronic health evaluation; CI=confidence interval; Fio2 =fraction of inspired oxygen; OR=odds ratio; VAP=ventilator-associated pneumonia
*From the D epartment of Internal Medicine, Pulmonary and Critical Care Division (Dr. Kolle£), the Division of Infectious Diseases (Dr. Fraser), the Department of Surgery, Division of Cardiothoracic Surgery (Dr. Pasque), Washington University School of Medicine; and the Department of Pharmacy (Dr. Vlasnik), Department of Infection Control (Ms. Murphy), and Department of Nursing (Ms . Sharpless), Barnes-Jewish Hospital, St. Louis. Supported in part by grants from the American Lung Association of Eastern Missouri and Merck & Co, Inc. Manuscript received December 9,1996; revision accepted March 5, 1997. Reprint requests: Marin H. Kollef, MD, FCCP, Pulmonary and Critical Care Division, Washington University School of Medicine, Box 8052, 660 S Euclid Ave, St. Louis, MO 63110; email: marin@wupulm3. wustl. edu 666
Nosocomial infections are recognized as an important cause of increased patient morbidity and mortality. 1 •2 The reported prevalence rate for nosocomial infections most commonly ranges from 5 to 20%, but can be significantly greater among patients requiring intensive care. 1·3 -7 The most common sites of hospital-acquired infection include the urinary tract, the lung, surgical wounds, and the bloodstream.8 Patients undergoing cardiac surgery appear to be at increased risk for the development of nosocomial infections due to the presence of multiple surgical wounds (chest and lower extremity incisions), frequent postoperative utilization of invaClinical Investigations
sive devices (eg, intra-aortic balloon counterpulsation, pulmonary artery catheter), and the common use of prophylactic or empiric antibiotics in the perioperative period_9~ll Additionally, current antibiotic administration practices may help to explain the emergence of nosocomial infections due to antibiotic-resistant pathogens in this and other groups of critically ill patients. 10·11 The development of such antibiotic-resistant infections has been associated with significantly greater hospital mortality rates compared to similar infections caused by antibioticsensitive pathogens. 12 · 13 This observation has led, in part, to a recent call for health-care providers to develop strategies aimed at reducing the emergence of antibiotic-resistant infections through more effective antimicrobial prescribing practices .14 ·15 As part of our hospital's ongoing quality improvement program, we performed a prospective cohort study examining patients undergoing cardiac surgery. The main goals of our study were as follows: (1) to determine the incidence of nosocomial infections in this specific group of hospitalized patients; (2) to evaluate the relationship between hospital-acquired infections and patient outcomes, including hospital mortality; and (3) to identify clinical strategies for future evaluation aimed at reducing the occurrence of nosocomial infections following cardiac surgery. In selecting these goals, we hoped to gain a better understanding of the increasingly complex problem of nosocomial infections among cardiac surgery patients to develop more effective infection control strategies.
MATERIALS AND METHODS
Study Location and Patients The study was conducted at a university~affiliated teaching hos~ pita!: Bames~Jewish Hospital (900 beds), St. Louis. During an 11-month period (August 1995 to June 1996), all patients undergo~ ing cardiac surgery were potentially eligible for this investigation. Patients were excluded if they were younger than 18 years and if they undenvent heart transplantation. All cardiac surgeries were performed by members of the attending physician staff of the Division of Cardiothoracic Surgery at Washington University. Pa~ tients were transported to the cardiothoracic ICU (17 beds) immediately following surgery and subsequently transferred to the postoperative recovery ward according to improvement in their medical condition. The study was approved by the Washington University School of Medicine Human Studies Committee.
Study Design and Data Collection A prospective cohort study design was employed segregating study patients according to th e presence or absence of an acquired nosocomial infection. Hospital mortality was the main outcome compared between the two study groups. \Ve also assessed secondary outcomes, including the durations of hospi-
talization, intensive care, and mechanical ventilation, the development of clinical sepsis, and the occurrence of multiorgan dysfunction. Our ability to assess these outcomes in this specific patient population has been established previously 10·r6 For all study patients, the following characteristics were prospectively recorded by one of the investigators: age, sex, race, premorbid lifestyle score,l7 serum albumin level (giL), the duration of surgery from the first skin incision until closure, the time spent receiving cardiopulmonary bypass, the aortic crossclamp time, the ratio of arterial blood oxygen tension to the concentration of inspired oxygen (Pa0;/Flo 2 ) at the time ofiCU admission, severity of illness based on APACHE II (acute physiology and chronic health evaluation) scores,rs the presence of presurgical conditions including congestive heart failure requiring medical therapy with diuretics and!or vasodilators, COPD requiring medical therapy with inhaled bronchodilators or corticosteroids, underlying malignancy, and immunosuppression, the specific type of cardiac surgery pe rformed (coronary artery bypass, valve replacement, maze procedure for chronic atrial fibrillation), and whether the surgery was performed on an emergent basis. Specific processes of medical care examined following cardiac surgery included the use of hemodialysis, intra-aortic balloon counterpulsation, extracorporeal membrane oxygenation, placement of a ventricular assist device, reintubation, use of a nasogastric tube, presence of a tracheostomy, administration of antacids, histamine-2-receptor antagonists, sucralfate, or aerosol the rapy (including bronchodilators, mucolytics, and antibiotics), fiberoptic bronchoscopy, positioning of the bead of the bed, use of chemical paralysis, postoperative administration of antibiotics, the clinical indications for postoperative antibiotics, duration of pulmonary artery catheter (or other central venous line) placement, and the duration of urinary tract catheterization. One of the investigators made daily rounds on all study patients recording relevant data from the medical records, bedside flow sheets, and the hospital's main frame computer for reports of microbiological studies (sputum Gram's stains and sputum, blood, pleural fluid, wound, and lower respiratOty tract cultures). All chest radiographs were prospectively reviewed by one of the investigators (M.H.K.) and the computerized radiographic reports were also reviewed (24 to 48 h l ater).The development of a nosocomial infection (wound, pneumonia, urinary tract, bloodstream) was also recorded prospectively. Patients were evaluated for nosocomial pneumonia during mechanical ventilation and for 48 h following extubation (ie, ventilator-associated pneumonia [VAP]). This was based on our previous experience which suggested that nosocomial pneumonia occurring vvithout mechanical ventilation was uncommon in this patient population.10·r6 Antibiotic administration, both perioperative prophylactic antibiotics and postsurgical antibiotics, was evaluated using patients ' medical records and verified with the pharmacy's central computer database.
Definitions All definitions were selected prospectively as part of the original study design. Nosocomial infections (urinary tract, bloodstream, wound infection) were defined according to criteria established by the Centers for Disease Control and Prevention.r 9 The diagnostic criteria for VAP were those established by the American College of Chest Physicians. 20 We routinely perform ed mini-BAL in patients \vith suspected VAP to obtain lower respiratory specimens for culture and Gram's stain. 21 The accuracy of mini-BAL for establishing a microbiological diagnosis of VAP has been established previously. 2 r·22 Microbiologically conCHEST I 112 I 3 I SEPTEMBER, 1997
667
firmed cases of VAP required the isolation of bacteiia in significant quantity fi·mn mini-BAL fluid (2:: 103 colony-forming units/ mL).2J.22 All patients were also prospectively screened for possible alternative causes for fever and radiographic c hest d ensities as suggested by other investigators. 23·24 The presence of atelectasis was defin ed b y the complete disappearance of the radiographic densities within 48 h of evaluation. Congestive heart failure was defin ed by a suggestive hemodynamic profile on pulmonary artery catheterization (i.e, increased pulmonaty artery occlusion pressure) and the resolution of the pulmonary infiltrates following treatment with diuretics. Alveolar hemorrhage was diagnosed if min i-BAL demonstrated agrossly bloody return along with the presence of hemosiderin-laden macrophages on microscopic examination. The presence of th e ARDS was defin ed as impaired oxygenation (ie , PaOj Fio 2 :S200, regardless of the level of positive end-expiratory press ure), th e presence of bilateral pulmonaty infiltrates, and a pulmonary artety occlusion press ure :S 18 mm Hg or no clinical evidence of elevated left atrial pressure on the basis of the c hest radiograph and other clinical data. 25 The pre morbid lifestyle score was used as previously defin ed. 17 We calculated APACHE II scores on the basis of clinical data available from the first 24-h peiiod of intensive care following cardiac surgety. Patients found to have th e head of th eir beds below 30°, other than for the performance of biief procedures (eg, hemodynamic assessments, repositioning), were classified as being in the supine position. Multiorgan dysfun ction was defin ed as derangemen ts of three or more organ systems using the ctiteria of Rubin and coworkers. 26 We have shown previously th at this definition of multiorgan dysfun ction is an excellent dete rminant of patient outcomes, including hospital mortality, for patients undergoing cardiac surge1yw·16 The definitions used for the systemic inflammatory respo nse syndrome, sepsis, severe sepsis, and septic shock, were those proposed by the Ameiic
All comparisons were unpaired and all tests of significance were two tailed. Continuous variables were compared using th e Student t test for normally disttibuted variables and th e vVilcoxon rank-sum tes t for nonnormally distiibuted variables. The x2 or Fisher's Exact Test were used to compare categori cal variables. The plimary data analysis compared patients who acquired a nosocomial infection with patients who did not develop such infections . To determine the relationships between hospital mortality and multiorgan dys function (dependent valiables) and the acquisition of nosocomial infections (independent valiables), multiple logistic regression models were used to control for the 668
effects of confounding valiables. 28 ·29 Each site of nosocomial infection (pulmonaty, urinmy tract, wound, and bloodstream) was entered into the multivariate models as separate dichotomous variables . Multiple logistic regression analysis was also used to identifY independent risk factors for acquiiing any nosocomial infection and for the development of the four individual nosocomial infections examined. Discrete variabl es (eg, dialysis, bronchoscopy) were ente red into th e multivariate models only if they were present prior to the occurrence of the outcome variable being examined. Similarly, continuous variables (durations of urina1y tract catheteiization, central vein and pulmonary artety catheterization, an d mechanical ventilation) were entered into the rnultivaliate models according to the durations of patient exposure to these variables occurring plior to the onset of the outcome variable of interest. A step\vise approach was used to enter new terms into the logistic regression models where 0.05 was set as the limit for the acceptance or removal of new terms. Model overfltting was examined by evaluating the ratio of outcome events to the total numbe r of independent variables in the final models and specific testing for inte ractions between the independent vaiiables was included in our analyses. 30 Results of th e logistic regression analyses are reported as adjusted odds ratios (AORs) with 95% confidence inte rvals (Cis). Relative Jisks and their 95% Cis were calculated using standard methodsa' Values are expressed as the mean::'::SD (con tinuous variables) or as a percentage of the group from which th ey were derived (categoiical variables). All p values were two tailed and p values of :S 0.05 were considered to indicate statistical significance.
RESULTS
Patients
A total of 605 consecutive patie nts undergoing cardiac surgery was prospectively evaluated (Tables 1 and 2). The mean age of the patients was 64.0±12.7 years (range, 18 to 90 years); 397 (65.6%) patie nts w e re men and 208 (34.4%) were wome n. The m e an APACHE II score of the entire study cohort was 12.6±4.1 (range, 1 to 38). The surgical procedures performed on these patients included 504 (83.3%) coronary artery bypass ope rations (28 with an accompanying valve r e place ment ), 69 (11.4%) valve operations, 25 (4.1%) maze procedures, and seven (1.2%) miscellane ous surgeries (one aortic arch repair, three atrial se ptal defect closures , two pericardiectomies , and one left ventricle stab wound repair). Sixty -nine (11.4%) of the cardiac surgeries were classified as e m e rgent. The mean duration of surge ry was 5.5± 1.7 h with a mean cardiopulmonary bypass time of 2.5±0.8 h and a mean aortic cross-clamp time of 1.4±0.6 h. Incidence of Nosocomial Infection One hundred thirty-one (21.7%) patients acquired 179 nosocomial infections during their hospital stay (1.4 nosocomial infections p e r infecte d patient). The remaining 474 (78.3%) patients did not acquire a nosocomial infection. Ninety-six (73.3%) patients Clinical Investigations
Table !-Baseline and Surgical Characteristics of the Study Cohort Nosocomial Infection Characteristic Age, y Sex, M/F Race, No. (%) White Black Other Premorbid lifestyle score, No. (%) 0 2 3 4 Congestive heart failure, No. (%) COPD, No. (%) Underlying malignancy, No. (%) Immunosuppressed, No. (%) Albumin, giL PaO!f'Io 2 APACHE II score Surgery type, No. (%) CABG* Valve Other Emergent surgery, No. (%) Duration of surgery, h Cardiopulmonary bypass tim e, h A01tic cross-clamp time, h
Present (n=l3 l )
Absent (n=474) p Value
65.7±13.5 63.5±12.5 326/148 71/60
0.087 0.002
110 (84. 0) 20 (153) 1 (0.7)
424 (89.4) 45 (9.5) 5 (l.l )
0.140
33 (25.2) 88 (67.2) 7 (53) 1 (0.8 ) 2 (1.5) 38 (29.0) 13 (9.9) 8 (6. 1) 4 (3.1) 38.5±4.2 274±110 14.0±4.4
0.035 164 (34.6) 278 (58.6) 27 (5.7) 5 (l.l ) 0 (0.0) 91 (19 2) 0.022 0.037 23 (4.9) 25 (5.3) 0.710 0.107 5 (l.l ) 39.8±5.8 < 0.001 0.265 285±104 12.2±4.0 < 0.001
112 (85.5) 12 (9.2) 7 (5.3) 11 (8.4) 6.1±2.1 2.7±1.1 1.5±0.7
392 (82.7) 0.658 57 (12.0) 25 (5.3) 58 (122) 0.430 5.3±1.5 <0.001 0.008 2.4±0.7 0.038 1.4±0.6
*CABG =coronary artery bypass grafting.
developed a single infection and 35 (26. 7%) patients had multiple nosocomial infections. The overall rate of acquiring a nosocomial infection was 1.8/100 patient hospital days. The distribution of the nosocomial infections and their associated hospital mortality rates are shown in Figure l. VAP was the most common infection (59 patients) followed by urinary tract (54 patients), wound (45 patients), and bloodstream infections (21 patients). The mean (median) delay from the time of surgery to the occurrence of each nosocomial infection was as follows: VAP, 6.5±4.3 days (5 days ); urinary tract, 9.0±8.2 days (6 days); wound, 11.6±7.8 days (9 days ); and bloodstream, 9.5± 11.8 days (5 days ). The distribution of the microorganisms associated with a nosocomial infection is shown in Table 3. Eleven patients with clinical evidence of VAP, and no other explanation for their fever and radiographic infiltrates, had negative mini-BAL culture results. All 11 of these patients had been started on a regimen of broad-spectrum antibiotics at least 12 h prior to obtaining the mini-BAL samples. The most common isolated pathogens associated with a nosocomial infection were Gram-negative bacteria (50.3%) fol-
lowed by Gram-positive bacteria (31.1%), yeast and molds (15.5%), and viruses (3. 1%). Among these isolated pathogens, 15 (25%) of the Gram-positive bacteria and 73 (75.3%) of the Gram-negative bacteria were classified as antibiotic resistant. Patients acquiring a nosocomial infection had statistically greater APACHE II scores, greater premorbid lifestyle scores, and longer durations of surgery, cardiopulmonary bypass, and aortic cross-clamping; and were more likely to be female, have a preoperative diagnosis of congestive heart failure or COPD , and have a lower serum albumin level compared to patients who did not acquire a nosocomial infection (Table 1). Differences in the process of medical care follovving cardiac surgery for patients with and without nosocomial infections are shown in Table 2. Patients developing a nosocomial infection more frequently underwent dialysis, reintubation, tracheostomy, intra-aortic balloon counterpulsation, administration of histamine type-2-receptor antagonists, aerosolized medications, bronchoscopy, and chemical paralysis. Similarly, the durations of pulmonary artery catheterization, other central vein catheterization, urinary tract catheterization, and mechanical ventilation were significantly longer among patients developing a nosocomial infection. Infected patients were also more likely to have a witnessed aspiration, to have received the combination of a first-generation cephalosporin plus vancomycin for surgical wound prophylaxis, and to have received empiric postoperative antibiotics compared to noninfected patients (Table 2). Risk Factors for Nosocomial Infections Multivariate analysis demonstrated that female gender (AOR=2.13; 95% CI=l.65 to 2.75; p=0.003), the duration of mechanical ventilation (AOR = 1.26; 95% CI=l.17 to 1.36; p=0.003), the administration of empiric postoperative antibiotics (AOR = 1.90; 95% CI=l.69 to 2.13; p<0.001), and the duration of urinary tract catheterization (AOR=l.16; 95% CI=l.ll to 1.21; p< 0.001 ) were independent risk factors for the development of a nosocomial infection. Independent risk factors associated with the acquisition of the four individual nosocomial infections examined are shown in Table 4. Hospital Mortality Thirty (5.0%) patients died during their hospitalization following cardiac surgery. The mortality rate of patients developing a nosocomial infection (11 .5%) was significantly greater than the mortality rate of patients without a nosocomial infection (3. 2% ) (odds ratio [OR]=4.0; 95% CI=2.7 to 5.8; p < 0.001). However, among patients developing a CHEST I 112 I 3 I SEPTEMBER, 1997
669
Table 2-Postoperative Process of Care Variables osocomial Infection Variable Antibiotic prophylaxis, No. (%) Class of antibiotic prophylaxis, No. (%) Vancomycin First-generation cephalosporin Vancomyci n plus first-generation cephalosporin Antibiotic prophylaxis start tim e, No. (%) Preoperative Intraoperative Postoperative Any postoperative antibiotic administration, No. (%) Empiric postoperative antib iotic administration, No. (%) Dialysis, No. (%) Reintubation , No. (%) Tracheostomy, No. (%) Intra-aortic balloon counterpulsation, No. (%) Extracorporeal membrane oxygenation, No. (%) Ventricular assist device, No. (%) Nasogastric tube, No. (%) Antacids, No. (%) Histamine type-2 antagonists, No. (%) Sucralfate, No. (%) Aerosol administration, No. (%) Witnessed aspiration , No. (%) Bronchoscopy, No. (%) Supine head positioning, No. (%) Che mi cal paralysis, No. (%) Duration of PA * catheterization, d Duration of other central vein catheterization, d Duration of urinary catheterization, d Duration of mechanical ventilation , d
Present (n= 131)
Absent (n=474 )
p Value
128 (97 7)
468 (987)
0.415
18 (14.0) 82 (64.1 ) 28 (219)
79 (16.9) 332 (70.9) 57 (12.2)
0.026
125 (97.7) 1 (0.8 ) 2(1 5) 131 (1000) 78 (59.5) 7 (5.3) 28 (214) 15 (11.5) 41 (3 1 3) 0 (0.0) 4 (3. 1) 129 (98.5) 12 (9.2) 30 (22.9) 123 (93.9) 32 (24.4) 5 (38) 17 (13.0) 7 (5.3) 29 (22 1) 3.3:!:: 2.7 5.5:!::7.7 9.1:!::9.9 6.0:!::8.0
460 (98 3) 4 (0 9) 4 (0 9) 150 (316) 150 (316) 6 (13) 7 (15) 2 (0.4) 70 (14.8) 1 (0.2) 6 (13) 461 (97.3) 26 (5 5) 63 (13.3) 438 (92.4) 49 (10.3) 2 (0.4) 13 (2.7) 23 (4.9) 60 (12 7) 17:!::13 1.0:!::2.2 2.9 :!:: 2.2 14:!::1.1
0.727
<0.001 <0.001 0.004 < 0.001 < 0.001 < 0.001 > 0.990 0.235 0.543 0.126 0.007 0.704 < 0.001 0.006 < 0.001 0.821 0.007 < 0.001 <0.001 < 0.001 < 0.001
*PA=pulmonary artery.
nosocomial infection (Fig 1), the mortality rates of patients with VAP (23. 7%) and bloodstream infections (23.8%) were significantly greater than tl1e
~
~
35
35
30
30 :I:
VI
c: 25 0 :;::; u ~ 20
25 0
VI
"tl
20
-= 0
0
15 :::1.
c: 15
!!!. ~
0 :;::; ::I
..c 10
:s c
~
!!:
10~
Multiorgan Dysfunction
VI
5
Lung
Urinary Tract
Wound
Blood
Source of Nosocomial Infection FIGURE l. Bar graph showing the distribution of the nosocomial infections (black bars ) and the hospital mortality rates associated with the individual infections (white bars ).
670
mortality rates associated with urinary tract (4.4%) and wound infections (9.3%) (p=O.Ol2) . Hospital nonsurvivors were statistically more likely to have developed a bloodstream infection, VAP, sepsis, severe sepsis, septic shock, and individual organ system derangements compared to hospital survivors (Table 5). Additionally, multiorgan dysfunction occurred in 20 (66.7%) of the hospital nonsurvivors compared to 32 (5.6%) of the hospital survivors (p < O.OOl ). Multivariate analysis identified the development of multiorgan dysfunction, the aortic cross-clamp time, and severity of illness as independent risk factors for hospital mortality (Table 6).
Patients acquiring a nosocomial infection were significantly more likely to develop multiorgan dysfunction compared to noninfected patients (OR=l0.8; 95% CI=5.8 to 20.3; p=
Table 3-Microorganisms Associated With Nosocomial Infections* VAP 1 (n=59 ) Pseudomonas aeruginosa Enterobacter species Serratia marcescens OSSA Haemophilus influenzae Klebsiella pneumoniae Aspergillus species Xanthomonas nwltophilia Herpes simplex yjrusl Cytomegalovirus 1 Proteus mirabilis Escherichia coli Citrobacter freundii Candida species ORSA Morganella morganii Streptococcus pneumoniae Alcaligenes xylosoxidans
Blood (n=21 )
Urinary Tract (n=54) 9 8 7 7 3 3 3 3 3 3 2 2 2 1 1 1 1
Candida species E coli P mirabilis Enterococcus species P aeruginosa Klebsiella pneumoniae Enterobacter species M morganii S marcescens C freundii Lactobacillus s pecies H afnia alvei Acinetobacter baumannii
15 7 6 5 4 4 3 3 2 2 1 1
Wound (n =45)
Candida species OSSA Enterococcus species§ Enterobacter species P aeruginosa S marcescens K pneumoniae A baunwnnii
8 6 5 3 3 2 1 1
OSSA CNS P aeruginosa Enterococcus species Enterobacter species P mirabilis ORSA Candida species S marcescens
18 lO 5 4 3 3 3 3 1
*n = the number of documented nosocomial infections within each categmy, some being polymicrobial; OSSA =oxacillin-sensitive Staphylococcus aureus; ORSA=oxacillin-resistant S aureus; C NS =coagulase-negative Staphylococcus. 1 Eieven patients with clinical e\~de n ce ofVAP had no gro,,th from mini-BAL fluid. trsolated in conjunction with an accompanying bacterial pathogen. §One vancomycin-resistant strain.
cardiopulmonary bypass time, and severity of illness were independent risk factors for the development of multiorgan dysfunction.
Antibiotic Administration Five hundred ninety-six (98.5%) patients received antibiotics for surgical wound prophylaxis in the perioperative period. Patients who did not receive antibiotic prophylaxis were significantly more likely to develop a surgical wound infection (three of nine patients) compared to patients who received prophylactic antibiotics (42 of 596 patients) (OR=6.6; 95% CI=l.6 to 27.0; p=0.024) . No other significant relationships were found between the use or timing of prophylactic antibiotics and the occurrence of other postoperative nosocomial infections. Patients developing a nosocomial infection were statistically more likely to have received empiric antibiotics in the postoperative period compared to patients who did not develop a nosocomial infection (OR=3.2; 95% CI=2.1 to 4.7; p=
Table 4-Independent Risk Factors for Specific Nosocomial Infections Nosomial Infection and Its Risk Factors Bacteremia Duration of central vein catheteri zation (1-d increments ) Duration of pulmonary artery catheterization (1-d increments) Duration of surgery (1-h increments ) YAP Duration of mechanical ventilation (1-d increments ) Administration of empiric antibiotics* Multiorgan dys function Urinary tract Duration of urinary catheterization ( 1-d increments) Female gender Wound Absence of antibiotic prophylaxis Congestive heart failure Female gender Duration of surgery (1-h increments)
AOR
95% CI
p Value
1.13
1.09- 1.17
< 0.001
1.27
1.16-1.39
< 0.001
1.25
1.12- 1.40
0.045
1.77
1.62- 1.92
< 0.001
1.86
1.54-2.23
< 0.001
5.07
2.85--8.99
0.004
1.18
1.15-1.21
< 0.001
2.39
1.73- 3.30
0.006
8.45
3.11-22.97
0.005
2.37 2.14 1.22
1.68-3.35 1.52-3.01 1.12-1.33
0.002 0.024 0.042
*In the postoperative setting. CHEST I 112 I 3 I SEPTEMBER, 1997
671
Table 5-Incidence of Postoperative Nosocomial Infection, Sepsis, and Acquired Organ System Derangements According to Hospital Mortality Nonsurvivors Survivors (n=30) (n=575 )
Variable
Nosocomial infections Bloodstream infection, No. (%) 5 (16.7) Urinary tract infection, No. (%) 5 16.7) ( 14 (46.7) VAP, No. (%) 2 (67) Wound infection,* No.(%) 2 (6.7) Sternum 0 (0.0) Vein harvest site Sepsis classification SIRS, 1 No. (%) 30 (100.0) Sepsis, No. (%) 9 (30.0) 6 (20.0 ) Severe sepsis, No. (%) 3 (10.0) Septic shock, No. (%) Acquired organ system derangements 3.3::!:: 1.4 No. of organ system derangementsl 21 (70.0) Lung, No. (% ) 23 (76.7) Heart, No. (%) Kidney, No. (%) 15 (50.0) Liver, No. (%) 9 (30.0) Brain, No. (%) 8 (26.7) Bone marrow, No. (%) 17 (56.7) GI, No. (%) 5 (16.7)
16 (2.8) 49 (8.5) 45 (7.8) 43 (75) 31 5.4) ( 15 (2.6)
p Value 0.002 0.18 <0.001 0.87 0.675 >0.99
547 (95.1) 0.39 58 (10.1) <0.001 17 (3.0) <0.001 3 0.5) ( <0.001
0.7::!::1.0
<0.001
39 (6.8) 182 (31. 7) 34 (5.9) 18 (3.1) 17 (3.0) 83 (14.4) 30 (52)
<0.001 <0.001 <0.001 <0.001 <0.001 <0.001 0.02
*Three surviving patients had both sternum and vein harvest site wound infections. 1 SIRS=systemic inflammatory response syndrome. IPer individual patient.
Lengths of Stay Patients developing a nosocomial infection had significantly longer lengths of stay in the ICU (7.8±8.0 days vs 2.3±1.8 days; p < 0.001 ) and in the hospital (21.0±13.7 days vs 9.7±4.7 days; p<0.001) compared to patients without a nosocomial infection (Fig 2). Similarly, the duration of mechanical ventilation was greater among infected patients (6.0±8.0 days vs 1.4±1.1 days; p < 0.001 ). Similar differences in lengths of stay and duration of mechanical ventilation were observed when only the hospital survivors were examined (p < 0.001 ).
Table 1-Variables Independently Associated With Multiorgan Dysfunction Variable
AOR
95% CI
p Value
Sepsis VAP Female gender Cardi opulmonary bypass time (1-h increments) APACHE II score (1-point increments)
7.6 5.2 3.4 2.0
4.3- 13.3 3.1-8.8 2.2-5.3 1.6- 2.4
<0.001 <0.001 0.005 0.003
1.3
1.2-1.4
<0.001
DISCUSSION
This study demonstrated that nosocomial infections are common among patients undergoing cardiac surgery. VAP was the most frequent hospitalacquired infection, followed by urinary tract, surgical wound, and bloodstream infections. These results confirm the findings of other recent investigations demonstrating that VAP is the most common nosocomial infection occurring among patients requiring intensive care. 7 ·32 More importantly, we showed that nosocomial infections, particularly nosocomial pneumonia and clinical sepsis, are associated with the subsequent development of multiorgan dysfunction, which we identified as the major determinant of hospital mortality. Similarly, Vincent and colleagues7 recently demonstrated that nosocomial pneumonia, clinical sepsis, and bloodstream infections were independent risk factors for hospital mortality among critically ill patients in Europe. Our study also
-en ,ns
-s
~ 80 0
-... ~
en 0
.c
0
60
§
40
C)
r::::
Cl)
...J 20
ns
Table 6-Variables Independently Associated With Hospital Mortality p
Variable
AOR
95% CI
Value
Multiorgan dysfunction Aortic cross-clamp time (1-h increments) APACHE II score (1point increments )
23.8 2.3
13.5-42.1 1.7-3.0
<0.001 0.002
l.l
l.l- 1.2
672
0.019
~
Q.
en
0
J:
o~--------~--------------------Absent Present
Nosocomial Infection
FIGURE 2. Box plots for hospital length of stay (days ) for patients classified as having nosocomial infection absent or present. Boxes represent 25th to 75th percentiles with 50th percentile (solid line ) and mean (broken line ) values shown within the boxes. The lOth and 90th pe rcentiles are shown as capped bars, and symbols mark all data outside the percentiles. Clinical Investigations
provides a perspective on the relationship between ICU-acquired infections and clinical outcomes for patients undergoing cardiac surgery. More than 300,000 cardiac operations requiring cardiopulmonary bypass are performed each year in the United States. 33 Extrapolating from our results, and those from other investigations,7 •32 approximately 60,000 of these patients would be expected to develop a nosocomial infection following cardiac surgery. The potential magnitude of this problem suggests that more effective infection control measures are needed to reduce these infection rates. Establishing a direct causal link between nosocomial infections and increased hospital m01tality has been problematic due to the frequent occurrence of hospital-acquired infections among critically ill patients. Nevertheless, several studies have attempted to estimate the attributable mortality associated \'lith specific nosocomial infections. 34 ·35 Although we cannot absolutely conclude from our study that nosocomial infections increased the mortality rate of patients following cardiac surgery, our results do show that certain nosocomial infections are associated with an increased risk of death. We also found that the development of a nosocomial infection increased the length of stay for patients surviving their hospitalization. Therefore, even if reductions in mortality are not possible, shorter hospital stays and decreased medical care costs should be achievable by reducing these infection rates. Additionally, hospital-acquired infections and other iatrogenic complications usually represent the major identifiable risk factors amenable to change that are associated with increased mortality and prolonged lengths of stay.36,37 We identified four risk factors associated \'lith the development of nosocomial infection (duration of mechanical ventilation, duration of urinary tract catheterization, postoperative empiric antibiotic administration, and female gender). The first three of these risk factors appear to be amenable to interventions aimed at reducing the occurrence of nosocomial infections. Mechanical ventilation and urinary tract catheterization are well-established risk factors for nosocomial infections. 7 Decreasing the durations of time these medical devices are in use should help to reduce the overall incidence of hospital-acquired infections. Several studies have already demonstrated that early extubation and removal of vascular lines, urinary catheters, and chest tubes can be done safely among selected patients undergoing cardiac surgery.38,39 Unfortunately, the role of postoperative empiric antibiotic administration in the pathogenesis of nosocomial infections has been more difficult to demonstrate. Therefore, the design and implemen-
tation of interventions aimed at reducing or eliminating unnecessary empiric postoperative antibiotics are problematic. Many clinical investigations have documented the misuse of antibiotic administration in both community and academic hospital settings. 9·14·15 One of the consequences of these antimicrobial prescribing practices has been the emergence of highly virulent, antibiotic-resistant bacterial infections among hospitalized patients. 13 -L5 Studies specifically examining hospital-acquired bacteremia and pneumonia have shown that infection due to antibiotic-resistant pathogens is usually associated vvith greater patient mortality compared to similar infections attributed to antibiotic-sensitive bacteria. 40·41 Therefore, the suggestion has been made that more conservative administration of antibiotics, particularly in the absence of a demonstrable site of infection (ie, empiric therapy), is needed to reverse this trend of emerging antimicrobial resistance.l4·15 A2 In an attempt to accomplish this goal, several strategies have been developed, including restricted hospital formularies, formalized guidelines for the prescription of antibiotics, and antibiotic rotation programs. 43 To date, widespread acceptance and utilization of such strategies has not occurred and the administration of antibiotics is still usually performed without considering the risks of superinfection. 14 Additionally, interventional tiials are required to prospectively determine whether the use of empiric antibiotic treatment (ie, ,'fithout a localized source of infection) can be safely reduced among this group of patients. While the use of empiric antibiotics is associated v.rith an increased risk of developing VAP, the absence of prophylactic antibiotic administration appears to increase the risk of acquiring a nosocomial wound infection (Table 4). Additionally, the combination of vancomycin plus a first-generation cephalosp01in was associated \'lith a greater risk of wound infection compared to the use of either antibiotic alone (Table 2). This latter association may simply represent a selection bias, whereby patients having the greatest risk of acquiring a wound infection received combination antibiotics for prophylaxis. However, the critical importance of appropriate administration of prophylactic antibiotics among surgical patients has recently been demonstrated by Classen and colleagues. 44 These authors found considerable variation in the timing of prophylactic antibiotic administration. More importantly, patients receiving their prophylactic antibiotics >2 h prior to surgery or in the postoperative period had a significantly greater risk for wound infection compared with patients receiving their antibiotics in the preopCHEST I 112 I 3 I SEPTEMBER, 1997
673
erative or perioperative periods (ie, within 2 h prior to beginning surgery and 3 h after the start of surgery). 44 We also identified female gender as an independent risk factor for nosocomial infections, particularly wound and urinary tract infections. Anatomic differences between the genders that may predispose to such infections offer one possible explanation for this observation. 45 Additionally, the durations of central vein and pulmonary artery catheterization, along with the duration of surgery, were identified as other important risk factors for the individual nosocomial infections examined (Table 4). Therefore, infection control efforts aimed at reducing the length of surgery, avoiding the use of unnecessary postoperative antibiotics, decreasing the use of central vein and urinary tract catheters, and ensming that preoperative prophylactic antibiotics are administered should yield the greatest decrease in nosocomial infection rates among cardiac surgery patients. In summary, we have demonstrated that nosocomial infections are common following cardiac surgery and their occurrence rates are comparable to those found in other groups of critically ill patients .7 We hope that these data will serve as a reference point for the further development and implementation of interventions aimed at reducing these infection rates. Accomplishing such reductions should improve patient outcomes and lower the costs associated with cardiac surgery. Our study suggests that decreasing patient exposures to invasive devices, including mechanical ventilation, and providing more effective antibiotic prescribing practices should have the greatest overall impact on hospital-acquired infections. Additional studies are needed to develop and rigorously evaluate infection control strategies employing such measures. Until such investigations are performed, clinicians should consider the prevention of nosocomial infections an important priority in the care of these patients .
REFERENCES 1 Haley RW, Culver DH, White JW, et al. The nationwide nosocomial infection rate: a new need for vital statistics. Am J Epidemiol 1985; 121:159-67 2 Martone WJ, Jarvis WR , Culver DH, et al. Incidence and nature of endemic and epide mic nosocomial infections. In: Bennett JV, Brachman PS, eds. Hospital infections. 3rd ed. Boston: Little Brown, 1992; 577-96 3 Wenzel RP, Thompson RL, Landry SM , et al. Hospitalacquired infections in intensive care unit patients: an overview with emphasis on epidemics. Infect Control 1983; 4:371-75 4 Craven DE, Kunches LM, Lichtenberg DA, e t al. Nosocomial infection and fatality in medical and surgical intensive care unit patients. Arch Intem M ed 1988; 149:1161-68 5 Trilla A. Epidemiology of nosocomial infections in adult 674
intensive care units. Intensive Care Med 1994; 20:S1-S4 6 Mertens R, Kegels G, Stroobant A, et la. The national prevalence of nosocomial infection in Belgium , 1984. J Hosp Infect 1987; 9:219-29 7 Vincent JL, Bihari DJ, Suter PM , et al. The prevalence of nosocomial infection in intensive care units in Europe: results of the European Prevalence of Infection in Intensive Care (EPIC) Study; EPIC International Advismy Committee. JAMA 1995; 274:639-44 8 Standfast SJ, Michelsen PB, Baltch AL, et al. A prevalence survey of infections in a combined acute and long-term care hospital. Infect Control 1984; 5:177-84 9 Dunagan WC, Woodward RS , Medoff G, et a!. Antibiotic misuse in two clinical situations: positive blood cultures and administration of aminoglycosides. Rev Infect Dis 1991; 13:405-12 10 Kollef MR. Ventilator-associated pneumonia: a multivariate analysis. JAMA 1993; 270:1965-70 11 Beck-Sague CM , Sinkowitz RL, Chinn RY, eta!. Risk factors for ventilator-associated pneumonia in surgical intensivecare-unit patients. Infect Control Hosp Epidemiol 1996; 17:374-76 12 de Vera ME , Simmons RL. Antibiotic-resistant enterococci and the changing face of surgical infections. Arch Surg 1996; 131:338-42 13 Meyer KS , Urban C, Eagan JA, e ta!. Nosocomial outbreak of Klebsiella infection resistant to late-generation cephalosporins. Ann Intern Med 1993; 119:353-58 14 Goldmann DA, Weinstein RA, Wenzel RP, eta!. Strategies to prevent and control the emergence and spread of antimicrobial-resistant microorganisms in hospitals. JAMA 1996; 275: 234-40 15 Gold HS , Moellering RC. Antimicrobial-drug resistance. N Engl J Med 1996; 33.5:1445-53 16 Kollef MH , Wragge T, Pasque C. Determinants of mortality and multiorgan dysfunction in cardiac surgery patients requiring prolonged mechanical ventilation. Chest 1995; 107: 1395-1401 17 Menzies R, Gibbons W, Goldberg P. Dete rminants of weaning and surviving among patients with COPD who require mechanical ventilation for acute respiratory failure. Chest 1989; 95:398-405 18 Knaus WA, Draper EA, Wagne r DP, et al. APACHE II: a severity of disease classification system. Crit Care Med 1985; 13:818-29 19 Garner JS, Jarvis WR, Emori TG, et al. CDC definitions for nosocomial infections. Am J Infect Con trol 1988; 16:128-40 20 Pingleton SK, Fagon JY, Leeper KY. Patient selection for clinical investigation of ventilator-associated pneumonia: crite tia for evaluating diagnostic techniques. Chest 1992; 102: 553S-56S 21 Kollef MH, Bock KR, Richards RD , e t al. The safety and diagnostic accuracy of minibronchoalveolar lavage in patients with suspected ventilator-associated pneumonia. Ann Intern Med 1995; 122:743-48 22 Papazian L, Thomas P, Garbe L, et l.a Bronchoscopic or blind sampling techniques for the diagnosis of ventilator-associated pneumonia. Am J Respir Crit Care Med 1995; 152:1982-91 23 Meduti GU , Mauldin GL, Wunderink RG , e t al. Causes of fever and pulmonary densities in patients with clinical manifestations of ventilator-associated pneumonia. Chest 1994; 106:221-35 24 Marquette CH, Georges I-1 , Wallet F, et a!. Diagnostic efficiency of endotracheal aspirates with quantitative bacterial cultures in intubated patients with suspected pneumonia: comparison vvith the protected specimen brush. Am J Respir Crit Care Med 1993; 148:138-44 Clinical Investigations
25 Bernard GR, Artigas A, Brigham KL, et al. The AmeJicanEuropean Consensus Conference on ARDS: definitions, mechanisms, relevant outcomes, and clinical trial coordination. Am J Respir Crit Care Med 1994; 149:818-24 26 Rubin DB, Wiener-Kronish JP, Murray JF, et al. Elevated von Willebrand factor antigen is an early phase predictor of acute lung injury in nonpulmonary sepsis syndrome. J Clin Invest 1990; 86:474-80 27 American College of Chest Physicians/Society of Critical Care Medicine Consensus Committee. Definitions for sepsis and organ failures and guidelines for the use of innovative therapies in sepsis. Chest 1992; 101:1644-55 28 Meinert CL, Tonascia S. Clinical trials: design, conduct, and analysis. New York: Oxford University Press, 1986; 194-95 29 SAS/STAT user's guide. Cary, NC: SAS Institute, 1990; 1071-1126 30 Concato J, Feinstein HR, Holford TR. The risk of determining risk with multivariable models. Ann Intern Med 1993; 118:201-10 31 Rothman KJ. Analysis of crude data. In: Rothman K, ed. Modern epidemiology. Boston: Little Brown, 1986; 153-74 32 Brun-Buisson C, Doyon F, Carlet J, et la. Incidence, risk factors , and outcome of severe sepsis and septic shock in adults: a multicenter prospective study in intensive care units. JAMA 1995; 274:968-74 33 Coyle JP, Higgins TL. Cardiac surgery. In: Hall JB, Schmidt GA, Wood LDH, eds. Principles of clitical care. New York: McGraw-Hill , 1992; 1012-29 34 Leu HS, Kaiser DL, Mori M, e t al. Hospital-acquired pneumonia: attributable mortality and morbidity. Am J Epidemiol 1989; 129:1258-67 35 Fagan JY, Chastre J, Hance AJ, et al. Nosocomial pneumonia
36 37 38 39 40 41 42 43 44 45
in ventilated patients: a cohort study evaluating attributable mortality and hospital stay. Am J M ed 1993; 94:281-88 Giraud T, Dhainaut JF, Vaxelaire JF, et a!. Iatrogenic complications in adult intensive care units: a prospective twocenter study. Crit Care Med 1993; 21:40-51 Bueno-Cavanillas A, Delgado-Rodriguez M, Lopez-Luque A, et a!. Influence of nosocomial infection on mortality rate in an intensive care unit. Crit Care Med 1994; 22:55-60 Hickey RF, Cason BA. Timing of tracheal extubation in adult cardiac surgery patients. J Card Surg 1995; 10:340-48 Engelman RM , Rousou JA, Flack JE III, et al. Fast-track recovery of the coronary bypass patient. Ann Thorac Surg 1994; 58:1742-46 Chow JW, Fine MJ, Shlaes DM, et al. Enterobacter bacteremia: clinical features and emergence of antibiotic resistance dUJing therapy. Ann Intern Med 1991; 115:585-90 Kollef MH , Silver P, Murphy DM, et a!. The effect of late-onset ventilator-associated pneumonia in determining patient mortality. Chest 1995; 108:1655-62 Wunderink RG. Mortality and ventilator-associated pneumonia: the best antibiotics may be the el ast antibiotics [editorial]. Chest 1993; 104:993-95 McGowan JE Jr. Do intensive hospital antibiotic control programs prevent the spread of antibiotic resistance? Infect Control Hosp Epidemiol 1994; 15:478-83 Classen DC, Evans RS, Pestotnik SL, et a!. The timing of prophylactic administration of antibiotics and the risk of surgical-wound infection. N Eng! J M ed 1992; 326:281-86 Copeland M, Senkowski C, Ergin MA, et al. Macromastia as a factor in sternal wound dehiscence following cardiac surgery: management combining chest wall reconstruction and reduction mammoplasty. J Card Surg 1992; 7:275-78
CHEST / 112 /3/ SEPTEMBER, 1997
675
Study objective: To evaluate the relationship between nosocomial infections and clinical outcomes following cardiac surgery, and to identify risk factors for the development of nosocomial infections in this patient population. Design: Prospective cohort study. Setting: Barnes-Jewish Hospital, St. Louis, a university-affiliated teaching hospital. Patients: Six hundred five consecutive patients undergoing cardiac surgery. Interventions: Prospective patient surveillance and data collection. Main outcome measures: Occurrence of nosocomial infections, multiorgan dysfunction, hospital mortality, and risk factors for the acquisition of nosocomial infections. Results: One hundred thirty-one (21.7%) patients acquired at least one nosocomial infection following cardiac surgery. Four independent risk factors for the development of a nosocomial infection were identified: the duration of mechanical ventilation, postoperative empiric antibiotic administration, the duration of urinary tract catheterization, and female gender. Thirty (5.0%) patients died during their hospitalization. The mortality rate of patients acquiring a nosocomial infection (11.5%) was significantly greater than the mortality rate of patients without a nosocomial infection (3.2%) (odds ratio [OR]=4.0; 95% confidence interval [CI]=2.7 to 5.8; p<0.001). Multiorgan dysfunction was found to be the most important independent determinant of hospital mortality (adjusted OR=23.8; 95% CI=13.5 to 42.1; p<0.001) along with the aortic cross-clamp time (adjusted OR=2.3; 95% CI=l.7 to 3.0; p=0.002) and severity of illness as measured by APACHE II (acute physiology and chronic health evaluation) (adjusted OR=l.1; 95% CI=l.1 to 1.2; p=O.Ol9). Ventilator-associated pneumonia, clinical sepsis, female gender, the cardiopulmonary bypass time, and severity of illness were identified as independent risk factors for the development of multiorgan dysfunction. Among hospital survivors, patients acquiring a nosocomial infection had longer hospital lengths of stay compared to patients without a nosocomial infection (20.1±13.0 days vs 9.7±4.5 days; p<0.001). Conclusions: Nosocomial infections, which are common following cardiac surgery, are associated with prolonged lengths of hospitalization, the development of multiorgan dysfunction, and increased hospital mortality. These data suggest potential interventions for the prevention of nosocomial infections following cardiac surgery that could substantially improve patient outcomes and decrease (CHEST 1997; 112:666-75) medical care costs. Key words: bacteremia; cardiac surgery; intensive care; nosocomial infection; outcomes; pneumonia; minary tract infection; wound infection Abbreviations: AOR=adjusted odds ratio; APACHE=acute physiology and chronic health evaluation; CI=confidence interval; Fio2 =fraction of inspired oxygen; OR=odds ratio; VAP=ventilator-associated pneumonia
*From the D epartment of Internal Medicine, Pulmonary and Critical Care Division (Dr. Kolle£), the Division of Infectious Diseases (Dr. Fraser), the Department of Surgery, Division of Cardiothoracic Surgery (Dr. Pasque), Washington University School of Medicine; and the Department of Pharmacy (Dr. Vlasnik), Department of Infection Control (Ms. Murphy), and Department of Nursing (Ms . Sharpless), Barnes-Jewish Hospital, St. Louis. Supported in part by grants from the American Lung Association of Eastern Missouri and Merck & Co, Inc. Manuscript received December 9,1996; revision accepted March 5, 1997. Reprint requests: Marin H. Kollef, MD, FCCP, Pulmonary and Critical Care Division, Washington University School of Medicine, Box 8052, 660 S Euclid Ave, St. Louis, MO 63110; email: marin@wupulm3. wustl. edu 666
Nosocomial infections are recognized as an important cause of increased patient morbidity and mortality. 1 •2 The reported prevalence rate for nosocomial infections most commonly ranges from 5 to 20%, but can be significantly greater among patients requiring intensive care. 1·3 -7 The most common sites of hospital-acquired infection include the urinary tract, the lung, surgical wounds, and the bloodstream.8 Patients undergoing cardiac surgery appear to be at increased risk for the development of nosocomial infections due to the presence of multiple surgical wounds (chest and lower extremity incisions), frequent postoperative utilization of invaClinical Investigations
sive devices (eg, intra-aortic balloon counterpulsation, pulmonary artery catheter), and the common use of prophylactic or empiric antibiotics in the perioperative period_9~ll Additionally, current antibiotic administration practices may help to explain the emergence of nosocomial infections due to antibiotic-resistant pathogens in this and other groups of critically ill patients. 10·11 The development of such antibiotic-resistant infections has been associated with significantly greater hospital mortality rates compared to similar infections caused by antibioticsensitive pathogens. 12 · 13 This observation has led, in part, to a recent call for health-care providers to develop strategies aimed at reducing the emergence of antibiotic-resistant infections through more effective antimicrobial prescribing practices .14 ·15 As part of our hospital's ongoing quality improvement program, we performed a prospective cohort study examining patients undergoing cardiac surgery. The main goals of our study were as follows: (1) to determine the incidence of nosocomial infections in this specific group of hospitalized patients; (2) to evaluate the relationship between hospital-acquired infections and patient outcomes, including hospital mortality; and (3) to identify clinical strategies for future evaluation aimed at reducing the occurrence of nosocomial infections following cardiac surgery. In selecting these goals, we hoped to gain a better understanding of the increasingly complex problem of nosocomial infections among cardiac surgery patients to develop more effective infection control strategies.
MATERIALS AND METHODS
Study Location and Patients The study was conducted at a university~affiliated teaching hos~ pita!: Bames~Jewish Hospital (900 beds), St. Louis. During an 11-month period (August 1995 to June 1996), all patients undergo~ ing cardiac surgery were potentially eligible for this investigation. Patients were excluded if they were younger than 18 years and if they undenvent heart transplantation. All cardiac surgeries were performed by members of the attending physician staff of the Division of Cardiothoracic Surgery at Washington University. Pa~ tients were transported to the cardiothoracic ICU (17 beds) immediately following surgery and subsequently transferred to the postoperative recovery ward according to improvement in their medical condition. The study was approved by the Washington University School of Medicine Human Studies Committee.
Study Design and Data Collection A prospective cohort study design was employed segregating study patients according to th e presence or absence of an acquired nosocomial infection. Hospital mortality was the main outcome compared between the two study groups. \Ve also assessed secondary outcomes, including the durations of hospi-
talization, intensive care, and mechanical ventilation, the development of clinical sepsis, and the occurrence of multiorgan dysfunction. Our ability to assess these outcomes in this specific patient population has been established previously 10·r6 For all study patients, the following characteristics were prospectively recorded by one of the investigators: age, sex, race, premorbid lifestyle score,l7 serum albumin level (giL), the duration of surgery from the first skin incision until closure, the time spent receiving cardiopulmonary bypass, the aortic crossclamp time, the ratio of arterial blood oxygen tension to the concentration of inspired oxygen (Pa0;/Flo 2 ) at the time ofiCU admission, severity of illness based on APACHE II (acute physiology and chronic health evaluation) scores,rs the presence of presurgical conditions including congestive heart failure requiring medical therapy with diuretics and!or vasodilators, COPD requiring medical therapy with inhaled bronchodilators or corticosteroids, underlying malignancy, and immunosuppression, the specific type of cardiac surgery pe rformed (coronary artery bypass, valve replacement, maze procedure for chronic atrial fibrillation), and whether the surgery was performed on an emergent basis. Specific processes of medical care examined following cardiac surgery included the use of hemodialysis, intra-aortic balloon counterpulsation, extracorporeal membrane oxygenation, placement of a ventricular assist device, reintubation, use of a nasogastric tube, presence of a tracheostomy, administration of antacids, histamine-2-receptor antagonists, sucralfate, or aerosol the rapy (including bronchodilators, mucolytics, and antibiotics), fiberoptic bronchoscopy, positioning of the bead of the bed, use of chemical paralysis, postoperative administration of antibiotics, the clinical indications for postoperative antibiotics, duration of pulmonary artery catheter (or other central venous line) placement, and the duration of urinary tract catheterization. One of the investigators made daily rounds on all study patients recording relevant data from the medical records, bedside flow sheets, and the hospital's main frame computer for reports of microbiological studies (sputum Gram's stains and sputum, blood, pleural fluid, wound, and lower respiratOty tract cultures). All chest radiographs were prospectively reviewed by one of the investigators (M.H.K.) and the computerized radiographic reports were also reviewed (24 to 48 h l ater).The development of a nosocomial infection (wound, pneumonia, urinary tract, bloodstream) was also recorded prospectively. Patients were evaluated for nosocomial pneumonia during mechanical ventilation and for 48 h following extubation (ie, ventilator-associated pneumonia [VAP]). This was based on our previous experience which suggested that nosocomial pneumonia occurring vvithout mechanical ventilation was uncommon in this patient population.10·r6 Antibiotic administration, both perioperative prophylactic antibiotics and postsurgical antibiotics, was evaluated using patients ' medical records and verified with the pharmacy's central computer database.
Definitions All definitions were selected prospectively as part of the original study design. Nosocomial infections (urinary tract, bloodstream, wound infection) were defined according to criteria established by the Centers for Disease Control and Prevention.r 9 The diagnostic criteria for VAP were those established by the American College of Chest Physicians. 20 We routinely perform ed mini-BAL in patients \vith suspected VAP to obtain lower respiratory specimens for culture and Gram's stain. 21 The accuracy of mini-BAL for establishing a microbiological diagnosis of VAP has been established previously. 2 r·22 Microbiologically conCHEST I 112 I 3 I SEPTEMBER, 1997
667
firmed cases of VAP required the isolation of bacteiia in significant quantity fi·mn mini-BAL fluid (2:: 103 colony-forming units/ mL).2J.22 All patients were also prospectively screened for possible alternative causes for fever and radiographic c hest d ensities as suggested by other investigators. 23·24 The presence of atelectasis was defin ed b y the complete disappearance of the radiographic densities within 48 h of evaluation. Congestive heart failure was defin ed by a suggestive hemodynamic profile on pulmonary artery catheterization (i.e, increased pulmonaty artery occlusion pressure) and the resolution of the pulmonary infiltrates following treatment with diuretics. Alveolar hemorrhage was diagnosed if min i-BAL demonstrated agrossly bloody return along with the presence of hemosiderin-laden macrophages on microscopic examination. The presence of th e ARDS was defin ed as impaired oxygenation (ie , PaOj Fio 2 :S200, regardless of the level of positive end-expiratory press ure), th e presence of bilateral pulmonaty infiltrates, and a pulmonary artety occlusion press ure :S 18 mm Hg or no clinical evidence of elevated left atrial pressure on the basis of the c hest radiograph and other clinical data. 25 The pre morbid lifestyle score was used as previously defin ed. 17 We calculated APACHE II scores on the basis of clinical data available from the first 24-h peiiod of intensive care following cardiac surgety. Patients found to have th e head of th eir beds below 30°, other than for the performance of biief procedures (eg, hemodynamic assessments, repositioning), were classified as being in the supine position. Multiorgan dysfun ction was defin ed as derangemen ts of three or more organ systems using the ctiteria of Rubin and coworkers. 26 We have shown previously th at this definition of multiorgan dysfun ction is an excellent dete rminant of patient outcomes, including hospital mortality, for patients undergoing cardiac surge1yw·16 The definitions used for the systemic inflammatory respo nse syndrome, sepsis, severe sepsis, and septic shock, were those proposed by the Ameiic
All comparisons were unpaired and all tests of significance were two tailed. Continuous variables were compared using th e Student t test for normally disttibuted variables and th e vVilcoxon rank-sum tes t for nonnormally distiibuted variables. The x2 or Fisher's Exact Test were used to compare categori cal variables. The plimary data analysis compared patients who acquired a nosocomial infection with patients who did not develop such infections . To determine the relationships between hospital mortality and multiorgan dys function (dependent valiables) and the acquisition of nosocomial infections (independent valiables), multiple logistic regression models were used to control for the 668
effects of confounding valiables. 28 ·29 Each site of nosocomial infection (pulmonaty, urinmy tract, wound, and bloodstream) was entered into the multivariate models as separate dichotomous variables . Multiple logistic regression analysis was also used to identifY independent risk factors for acquiiing any nosocomial infection and for the development of the four individual nosocomial infections examined. Discrete variabl es (eg, dialysis, bronchoscopy) were ente red into th e multivariate models only if they were present prior to the occurrence of the outcome variable being examined. Similarly, continuous variables (durations of urina1y tract catheteiization, central vein and pulmonary artety catheterization, an d mechanical ventilation) were entered into the rnultivaliate models according to the durations of patient exposure to these variables occurring plior to the onset of the outcome variable of interest. A step\vise approach was used to enter new terms into the logistic regression models where 0.05 was set as the limit for the acceptance or removal of new terms. Model overfltting was examined by evaluating the ratio of outcome events to the total numbe r of independent variables in the final models and specific testing for inte ractions between the independent vaiiables was included in our analyses. 30 Results of th e logistic regression analyses are reported as adjusted odds ratios (AORs) with 95% confidence inte rvals (Cis). Relative Jisks and their 95% Cis were calculated using standard methodsa' Values are expressed as the mean::'::SD (con tinuous variables) or as a percentage of the group from which th ey were derived (categoiical variables). All p values were two tailed and p values of :S 0.05 were considered to indicate statistical significance.
RESULTS
Patients
A total of 605 consecutive patie nts undergoing cardiac surgery was prospectively evaluated (Tables 1 and 2). The mean age of the patients was 64.0±12.7 years (range, 18 to 90 years); 397 (65.6%) patie nts w e re men and 208 (34.4%) were wome n. The m e an APACHE II score of the entire study cohort was 12.6±4.1 (range, 1 to 38). The surgical procedures performed on these patients included 504 (83.3%) coronary artery bypass ope rations (28 with an accompanying valve r e place ment ), 69 (11.4%) valve operations, 25 (4.1%) maze procedures, and seven (1.2%) miscellane ous surgeries (one aortic arch repair, three atrial se ptal defect closures , two pericardiectomies , and one left ventricle stab wound repair). Sixty -nine (11.4%) of the cardiac surgeries were classified as e m e rgent. The mean duration of surge ry was 5.5± 1.7 h with a mean cardiopulmonary bypass time of 2.5±0.8 h and a mean aortic cross-clamp time of 1.4±0.6 h. Incidence of Nosocomial Infection One hundred thirty-one (21.7%) patients acquired 179 nosocomial infections during their hospital stay (1.4 nosocomial infections p e r infecte d patient). The remaining 474 (78.3%) patients did not acquire a nosocomial infection. Ninety-six (73.3%) patients Clinical Investigations
Table !-Baseline and Surgical Characteristics of the Study Cohort Nosocomial Infection Characteristic Age, y Sex, M/F Race, No. (%) White Black Other Premorbid lifestyle score, No. (%) 0 2 3 4 Congestive heart failure, No. (%) COPD, No. (%) Underlying malignancy, No. (%) Immunosuppressed, No. (%) Albumin, giL PaO!f'Io 2 APACHE II score Surgery type, No. (%) CABG* Valve Other Emergent surgery, No. (%) Duration of surgery, h Cardiopulmonary bypass tim e, h A01tic cross-clamp time, h
Present (n=l3 l )
Absent (n=474) p Value
65.7±13.5 63.5±12.5 326/148 71/60
0.087 0.002
110 (84. 0) 20 (153) 1 (0.7)
424 (89.4) 45 (9.5) 5 (l.l )
0.140
33 (25.2) 88 (67.2) 7 (53) 1 (0.8 ) 2 (1.5) 38 (29.0) 13 (9.9) 8 (6. 1) 4 (3.1) 38.5±4.2 274±110 14.0±4.4
0.035 164 (34.6) 278 (58.6) 27 (5.7) 5 (l.l ) 0 (0.0) 91 (19 2) 0.022 0.037 23 (4.9) 25 (5.3) 0.710 0.107 5 (l.l ) 39.8±5.8 < 0.001 0.265 285±104 12.2±4.0 < 0.001
112 (85.5) 12 (9.2) 7 (5.3) 11 (8.4) 6.1±2.1 2.7±1.1 1.5±0.7
392 (82.7) 0.658 57 (12.0) 25 (5.3) 58 (122) 0.430 5.3±1.5 <0.001 0.008 2.4±0.7 0.038 1.4±0.6
*CABG =coronary artery bypass grafting.
developed a single infection and 35 (26. 7%) patients had multiple nosocomial infections. The overall rate of acquiring a nosocomial infection was 1.8/100 patient hospital days. The distribution of the nosocomial infections and their associated hospital mortality rates are shown in Figure l. VAP was the most common infection (59 patients) followed by urinary tract (54 patients), wound (45 patients), and bloodstream infections (21 patients). The mean (median) delay from the time of surgery to the occurrence of each nosocomial infection was as follows: VAP, 6.5±4.3 days (5 days ); urinary tract, 9.0±8.2 days (6 days); wound, 11.6±7.8 days (9 days ); and bloodstream, 9.5± 11.8 days (5 days ). The distribution of the microorganisms associated with a nosocomial infection is shown in Table 3. Eleven patients with clinical evidence of VAP, and no other explanation for their fever and radiographic infiltrates, had negative mini-BAL culture results. All 11 of these patients had been started on a regimen of broad-spectrum antibiotics at least 12 h prior to obtaining the mini-BAL samples. The most common isolated pathogens associated with a nosocomial infection were Gram-negative bacteria (50.3%) fol-
lowed by Gram-positive bacteria (31.1%), yeast and molds (15.5%), and viruses (3. 1%). Among these isolated pathogens, 15 (25%) of the Gram-positive bacteria and 73 (75.3%) of the Gram-negative bacteria were classified as antibiotic resistant. Patients acquiring a nosocomial infection had statistically greater APACHE II scores, greater premorbid lifestyle scores, and longer durations of surgery, cardiopulmonary bypass, and aortic cross-clamping; and were more likely to be female, have a preoperative diagnosis of congestive heart failure or COPD , and have a lower serum albumin level compared to patients who did not acquire a nosocomial infection (Table 1). Differences in the process of medical care follovving cardiac surgery for patients with and without nosocomial infections are shown in Table 2. Patients developing a nosocomial infection more frequently underwent dialysis, reintubation, tracheostomy, intra-aortic balloon counterpulsation, administration of histamine type-2-receptor antagonists, aerosolized medications, bronchoscopy, and chemical paralysis. Similarly, the durations of pulmonary artery catheterization, other central vein catheterization, urinary tract catheterization, and mechanical ventilation were significantly longer among patients developing a nosocomial infection. Infected patients were also more likely to have a witnessed aspiration, to have received the combination of a first-generation cephalosporin plus vancomycin for surgical wound prophylaxis, and to have received empiric postoperative antibiotics compared to noninfected patients (Table 2). Risk Factors for Nosocomial Infections Multivariate analysis demonstrated that female gender (AOR=2.13; 95% CI=l.65 to 2.75; p=0.003), the duration of mechanical ventilation (AOR = 1.26; 95% CI=l.17 to 1.36; p=0.003), the administration of empiric postoperative antibiotics (AOR = 1.90; 95% CI=l.69 to 2.13; p<0.001), and the duration of urinary tract catheterization (AOR=l.16; 95% CI=l.ll to 1.21; p< 0.001 ) were independent risk factors for the development of a nosocomial infection. Independent risk factors associated with the acquisition of the four individual nosocomial infections examined are shown in Table 4. Hospital Mortality Thirty (5.0%) patients died during their hospitalization following cardiac surgery. The mortality rate of patients developing a nosocomial infection (11 .5%) was significantly greater than the mortality rate of patients without a nosocomial infection (3. 2% ) (odds ratio [OR]=4.0; 95% CI=2.7 to 5.8; p < 0.001). However, among patients developing a CHEST I 112 I 3 I SEPTEMBER, 1997
669
Table 2-Postoperative Process of Care Variables osocomial Infection Variable Antibiotic prophylaxis, No. (%) Class of antibiotic prophylaxis, No. (%) Vancomycin First-generation cephalosporin Vancomyci n plus first-generation cephalosporin Antibiotic prophylaxis start tim e, No. (%) Preoperative Intraoperative Postoperative Any postoperative antibiotic administration, No. (%) Empiric postoperative antib iotic administration, No. (%) Dialysis, No. (%) Reintubation , No. (%) Tracheostomy, No. (%) Intra-aortic balloon counterpulsation, No. (%) Extracorporeal membrane oxygenation, No. (%) Ventricular assist device, No. (%) Nasogastric tube, No. (%) Antacids, No. (%) Histamine type-2 antagonists, No. (%) Sucralfate, No. (%) Aerosol administration, No. (%) Witnessed aspiration , No. (%) Bronchoscopy, No. (%) Supine head positioning, No. (%) Che mi cal paralysis, No. (%) Duration of PA * catheterization, d Duration of other central vein catheterization, d Duration of urinary catheterization, d Duration of mechanical ventilation , d
Present (n= 131)
Absent (n=474 )
p Value
128 (97 7)
468 (987)
0.415
18 (14.0) 82 (64.1 ) 28 (219)
79 (16.9) 332 (70.9) 57 (12.2)
0.026
125 (97.7) 1 (0.8 ) 2(1 5) 131 (1000) 78 (59.5) 7 (5.3) 28 (214) 15 (11.5) 41 (3 1 3) 0 (0.0) 4 (3. 1) 129 (98.5) 12 (9.2) 30 (22.9) 123 (93.9) 32 (24.4) 5 (38) 17 (13.0) 7 (5.3) 29 (22 1) 3.3:!:: 2.7 5.5:!::7.7 9.1:!::9.9 6.0:!::8.0
460 (98 3) 4 (0 9) 4 (0 9) 150 (316) 150 (316) 6 (13) 7 (15) 2 (0.4) 70 (14.8) 1 (0.2) 6 (13) 461 (97.3) 26 (5 5) 63 (13.3) 438 (92.4) 49 (10.3) 2 (0.4) 13 (2.7) 23 (4.9) 60 (12 7) 17:!::13 1.0:!::2.2 2.9 :!:: 2.2 14:!::1.1
0.727
<0.001 <0.001 0.004 < 0.001 < 0.001 < 0.001 > 0.990 0.235 0.543 0.126 0.007 0.704 < 0.001 0.006 < 0.001 0.821 0.007 < 0.001 <0.001 < 0.001 < 0.001
*PA=pulmonary artery.
nosocomial infection (Fig 1), the mortality rates of patients with VAP (23. 7%) and bloodstream infections (23.8%) were significantly greater than tl1e
~
~
35
35
30
30 :I:
VI
c: 25 0 :;::; u ~ 20
25 0
VI
"tl
20
-= 0
0
15 :::1.
c: 15
!!!. ~
0 :;::; ::I
..c 10
:s c
~
!!:
10~
Multiorgan Dysfunction
VI
5
Lung
Urinary Tract
Wound
Blood
Source of Nosocomial Infection FIGURE l. Bar graph showing the distribution of the nosocomial infections (black bars ) and the hospital mortality rates associated with the individual infections (white bars ).
670
mortality rates associated with urinary tract (4.4%) and wound infections (9.3%) (p=O.Ol2) . Hospital nonsurvivors were statistically more likely to have developed a bloodstream infection, VAP, sepsis, severe sepsis, septic shock, and individual organ system derangements compared to hospital survivors (Table 5). Additionally, multiorgan dysfunction occurred in 20 (66.7%) of the hospital nonsurvivors compared to 32 (5.6%) of the hospital survivors (p < O.OOl ). Multivariate analysis identified the development of multiorgan dysfunction, the aortic cross-clamp time, and severity of illness as independent risk factors for hospital mortality (Table 6).
Patients acquiring a nosocomial infection were significantly more likely to develop multiorgan dysfunction compared to noninfected patients (OR=l0.8; 95% CI=5.8 to 20.3; p=
Table 3-Microorganisms Associated With Nosocomial Infections* VAP 1 (n=59 ) Pseudomonas aeruginosa Enterobacter species Serratia marcescens OSSA Haemophilus influenzae Klebsiella pneumoniae Aspergillus species Xanthomonas nwltophilia Herpes simplex yjrusl Cytomegalovirus 1 Proteus mirabilis Escherichia coli Citrobacter freundii Candida species ORSA Morganella morganii Streptococcus pneumoniae Alcaligenes xylosoxidans
Blood (n=21 )
Urinary Tract (n=54) 9 8 7 7 3 3 3 3 3 3 2 2 2 1 1 1 1
Candida species E coli P mirabilis Enterococcus species P aeruginosa Klebsiella pneumoniae Enterobacter species M morganii S marcescens C freundii Lactobacillus s pecies H afnia alvei Acinetobacter baumannii
15 7 6 5 4 4 3 3 2 2 1 1
Wound (n =45)
Candida species OSSA Enterococcus species§ Enterobacter species P aeruginosa S marcescens K pneumoniae A baunwnnii
8 6 5 3 3 2 1 1
OSSA CNS P aeruginosa Enterococcus species Enterobacter species P mirabilis ORSA Candida species S marcescens
18 lO 5 4 3 3 3 3 1
*n = the number of documented nosocomial infections within each categmy, some being polymicrobial; OSSA =oxacillin-sensitive Staphylococcus aureus; ORSA=oxacillin-resistant S aureus; C NS =coagulase-negative Staphylococcus. 1 Eieven patients with clinical e\~de n ce ofVAP had no gro,,th from mini-BAL fluid. trsolated in conjunction with an accompanying bacterial pathogen. §One vancomycin-resistant strain.
cardiopulmonary bypass time, and severity of illness were independent risk factors for the development of multiorgan dysfunction.
Antibiotic Administration Five hundred ninety-six (98.5%) patients received antibiotics for surgical wound prophylaxis in the perioperative period. Patients who did not receive antibiotic prophylaxis were significantly more likely to develop a surgical wound infection (three of nine patients) compared to patients who received prophylactic antibiotics (42 of 596 patients) (OR=6.6; 95% CI=l.6 to 27.0; p=0.024) . No other significant relationships were found between the use or timing of prophylactic antibiotics and the occurrence of other postoperative nosocomial infections. Patients developing a nosocomial infection were statistically more likely to have received empiric antibiotics in the postoperative period compared to patients who did not develop a nosocomial infection (OR=3.2; 95% CI=2.1 to 4.7; p=
Table 4-Independent Risk Factors for Specific Nosocomial Infections Nosomial Infection and Its Risk Factors Bacteremia Duration of central vein catheteri zation (1-d increments ) Duration of pulmonary artery catheterization (1-d increments) Duration of surgery (1-h increments ) YAP Duration of mechanical ventilation (1-d increments ) Administration of empiric antibiotics* Multiorgan dys function Urinary tract Duration of urinary catheterization ( 1-d increments) Female gender Wound Absence of antibiotic prophylaxis Congestive heart failure Female gender Duration of surgery (1-h increments)
AOR
95% CI
p Value
1.13
1.09- 1.17
< 0.001
1.27
1.16-1.39
< 0.001
1.25
1.12- 1.40
0.045
1.77
1.62- 1.92
< 0.001
1.86
1.54-2.23
< 0.001
5.07
2.85--8.99
0.004
1.18
1.15-1.21
< 0.001
2.39
1.73- 3.30
0.006
8.45
3.11-22.97
0.005
2.37 2.14 1.22
1.68-3.35 1.52-3.01 1.12-1.33
0.002 0.024 0.042
*In the postoperative setting. CHEST I 112 I 3 I SEPTEMBER, 1997
671
Table 5-Incidence of Postoperative Nosocomial Infection, Sepsis, and Acquired Organ System Derangements According to Hospital Mortality Nonsurvivors Survivors (n=30) (n=575 )
Variable
Nosocomial infections Bloodstream infection, No. (%) 5 (16.7) Urinary tract infection, No. (%) 5 16.7) ( 14 (46.7) VAP, No. (%) 2 (67) Wound infection,* No.(%) 2 (6.7) Sternum 0 (0.0) Vein harvest site Sepsis classification SIRS, 1 No. (%) 30 (100.0) Sepsis, No. (%) 9 (30.0) 6 (20.0 ) Severe sepsis, No. (%) 3 (10.0) Septic shock, No. (%) Acquired organ system derangements 3.3::!:: 1.4 No. of organ system derangementsl 21 (70.0) Lung, No. (% ) 23 (76.7) Heart, No. (%) Kidney, No. (%) 15 (50.0) Liver, No. (%) 9 (30.0) Brain, No. (%) 8 (26.7) Bone marrow, No. (%) 17 (56.7) GI, No. (%) 5 (16.7)
16 (2.8) 49 (8.5) 45 (7.8) 43 (75) 31 5.4) ( 15 (2.6)
p Value 0.002 0.18 <0.001 0.87 0.675 >0.99
547 (95.1) 0.39 58 (10.1) <0.001 17 (3.0) <0.001 3 0.5) ( <0.001
0.7::!::1.0
<0.001
39 (6.8) 182 (31. 7) 34 (5.9) 18 (3.1) 17 (3.0) 83 (14.4) 30 (52)
<0.001 <0.001 <0.001 <0.001 <0.001 <0.001 0.02
*Three surviving patients had both sternum and vein harvest site wound infections. 1 SIRS=systemic inflammatory response syndrome. IPer individual patient.
Lengths of Stay Patients developing a nosocomial infection had significantly longer lengths of stay in the ICU (7.8±8.0 days vs 2.3±1.8 days; p < 0.001 ) and in the hospital (21.0±13.7 days vs 9.7±4.7 days; p<0.001) compared to patients without a nosocomial infection (Fig 2). Similarly, the duration of mechanical ventilation was greater among infected patients (6.0±8.0 days vs 1.4±1.1 days; p < 0.001 ). Similar differences in lengths of stay and duration of mechanical ventilation were observed when only the hospital survivors were examined (p < 0.001 ).
Table 1-Variables Independently Associated With Multiorgan Dysfunction Variable
AOR
95% CI
p Value
Sepsis VAP Female gender Cardi opulmonary bypass time (1-h increments) APACHE II score (1-point increments)
7.6 5.2 3.4 2.0
4.3- 13.3 3.1-8.8 2.2-5.3 1.6- 2.4
<0.001 <0.001 0.005 0.003
1.3
1.2-1.4
<0.001
DISCUSSION
This study demonstrated that nosocomial infections are common among patients undergoing cardiac surgery. VAP was the most frequent hospitalacquired infection, followed by urinary tract, surgical wound, and bloodstream infections. These results confirm the findings of other recent investigations demonstrating that VAP is the most common nosocomial infection occurring among patients requiring intensive care. 7 ·32 More importantly, we showed that nosocomial infections, particularly nosocomial pneumonia and clinical sepsis, are associated with the subsequent development of multiorgan dysfunction, which we identified as the major determinant of hospital mortality. Similarly, Vincent and colleagues7 recently demonstrated that nosocomial pneumonia, clinical sepsis, and bloodstream infections were independent risk factors for hospital mortality among critically ill patients in Europe. Our study also
-en ,ns
-s
~ 80 0
-... ~
en 0
.c
0
60
§
40
C)
r::::
Cl)
...J 20
ns
Table 6-Variables Independently Associated With Hospital Mortality p
Variable
AOR
95% CI
Value
Multiorgan dysfunction Aortic cross-clamp time (1-h increments) APACHE II score (1point increments )
23.8 2.3
13.5-42.1 1.7-3.0
<0.001 0.002
l.l
l.l- 1.2
672
0.019
~
Q.
en
0
J:
o~--------~--------------------Absent Present
Nosocomial Infection
FIGURE 2. Box plots for hospital length of stay (days ) for patients classified as having nosocomial infection absent or present. Boxes represent 25th to 75th percentiles with 50th percentile (solid line ) and mean (broken line ) values shown within the boxes. The lOth and 90th pe rcentiles are shown as capped bars, and symbols mark all data outside the percentiles. Clinical Investigations
provides a perspective on the relationship between ICU-acquired infections and clinical outcomes for patients undergoing cardiac surgery. More than 300,000 cardiac operations requiring cardiopulmonary bypass are performed each year in the United States. 33 Extrapolating from our results, and those from other investigations,7 •32 approximately 60,000 of these patients would be expected to develop a nosocomial infection following cardiac surgery. The potential magnitude of this problem suggests that more effective infection control measures are needed to reduce these infection rates. Establishing a direct causal link between nosocomial infections and increased hospital m01tality has been problematic due to the frequent occurrence of hospital-acquired infections among critically ill patients. Nevertheless, several studies have attempted to estimate the attributable mortality associated \'lith specific nosocomial infections. 34 ·35 Although we cannot absolutely conclude from our study that nosocomial infections increased the mortality rate of patients following cardiac surgery, our results do show that certain nosocomial infections are associated with an increased risk of death. We also found that the development of a nosocomial infection increased the length of stay for patients surviving their hospitalization. Therefore, even if reductions in mortality are not possible, shorter hospital stays and decreased medical care costs should be achievable by reducing these infection rates. Additionally, hospital-acquired infections and other iatrogenic complications usually represent the major identifiable risk factors amenable to change that are associated with increased mortality and prolonged lengths of stay.36,37 We identified four risk factors associated \'lith the development of nosocomial infection (duration of mechanical ventilation, duration of urinary tract catheterization, postoperative empiric antibiotic administration, and female gender). The first three of these risk factors appear to be amenable to interventions aimed at reducing the occurrence of nosocomial infections. Mechanical ventilation and urinary tract catheterization are well-established risk factors for nosocomial infections. 7 Decreasing the durations of time these medical devices are in use should help to reduce the overall incidence of hospital-acquired infections. Several studies have already demonstrated that early extubation and removal of vascular lines, urinary catheters, and chest tubes can be done safely among selected patients undergoing cardiac surgery.38,39 Unfortunately, the role of postoperative empiric antibiotic administration in the pathogenesis of nosocomial infections has been more difficult to demonstrate. Therefore, the design and implemen-
tation of interventions aimed at reducing or eliminating unnecessary empiric postoperative antibiotics are problematic. Many clinical investigations have documented the misuse of antibiotic administration in both community and academic hospital settings. 9·14·15 One of the consequences of these antimicrobial prescribing practices has been the emergence of highly virulent, antibiotic-resistant bacterial infections among hospitalized patients. 13 -L5 Studies specifically examining hospital-acquired bacteremia and pneumonia have shown that infection due to antibiotic-resistant pathogens is usually associated vvith greater patient mortality compared to similar infections attributed to antibiotic-sensitive bacteria. 40·41 Therefore, the suggestion has been made that more conservative administration of antibiotics, particularly in the absence of a demonstrable site of infection (ie, empiric therapy), is needed to reverse this trend of emerging antimicrobial resistance.l4·15 A2 In an attempt to accomplish this goal, several strategies have been developed, including restricted hospital formularies, formalized guidelines for the prescription of antibiotics, and antibiotic rotation programs. 43 To date, widespread acceptance and utilization of such strategies has not occurred and the administration of antibiotics is still usually performed without considering the risks of superinfection. 14 Additionally, interventional tiials are required to prospectively determine whether the use of empiric antibiotic treatment (ie, ,'fithout a localized source of infection) can be safely reduced among this group of patients. While the use of empiric antibiotics is associated v.rith an increased risk of developing VAP, the absence of prophylactic antibiotic administration appears to increase the risk of acquiring a nosocomial wound infection (Table 4). Additionally, the combination of vancomycin plus a first-generation cephalosp01in was associated \'lith a greater risk of wound infection compared to the use of either antibiotic alone (Table 2). This latter association may simply represent a selection bias, whereby patients having the greatest risk of acquiring a wound infection received combination antibiotics for prophylaxis. However, the critical importance of appropriate administration of prophylactic antibiotics among surgical patients has recently been demonstrated by Classen and colleagues. 44 These authors found considerable variation in the timing of prophylactic antibiotic administration. More importantly, patients receiving their prophylactic antibiotics >2 h prior to surgery or in the postoperative period had a significantly greater risk for wound infection compared with patients receiving their antibiotics in the preopCHEST I 112 I 3 I SEPTEMBER, 1997
673
erative or perioperative periods (ie, within 2 h prior to beginning surgery and 3 h after the start of surgery). 44 We also identified female gender as an independent risk factor for nosocomial infections, particularly wound and urinary tract infections. Anatomic differences between the genders that may predispose to such infections offer one possible explanation for this observation. 45 Additionally, the durations of central vein and pulmonary artery catheterization, along with the duration of surgery, were identified as other important risk factors for the individual nosocomial infections examined (Table 4). Therefore, infection control efforts aimed at reducing the length of surgery, avoiding the use of unnecessary postoperative antibiotics, decreasing the use of central vein and urinary tract catheters, and ensming that preoperative prophylactic antibiotics are administered should yield the greatest decrease in nosocomial infection rates among cardiac surgery patients. In summary, we have demonstrated that nosocomial infections are common following cardiac surgery and their occurrence rates are comparable to those found in other groups of critically ill patients .7 We hope that these data will serve as a reference point for the further development and implementation of interventions aimed at reducing these infection rates. Accomplishing such reductions should improve patient outcomes and lower the costs associated with cardiac surgery. Our study suggests that decreasing patient exposures to invasive devices, including mechanical ventilation, and providing more effective antibiotic prescribing practices should have the greatest overall impact on hospital-acquired infections. Additional studies are needed to develop and rigorously evaluate infection control strategies employing such measures. Until such investigations are performed, clinicians should consider the prevention of nosocomial infections an important priority in the care of these patients .
REFERENCES 1 Haley RW, Culver DH, White JW, et al. The nationwide nosocomial infection rate: a new need for vital statistics. Am J Epidemiol 1985; 121:159-67 2 Martone WJ, Jarvis WR , Culver DH, et al. Incidence and nature of endemic and epide mic nosocomial infections. In: Bennett JV, Brachman PS, eds. Hospital infections. 3rd ed. Boston: Little Brown, 1992; 577-96 3 Wenzel RP, Thompson RL, Landry SM , et al. Hospitalacquired infections in intensive care unit patients: an overview with emphasis on epidemics. Infect Control 1983; 4:371-75 4 Craven DE, Kunches LM, Lichtenberg DA, e t al. Nosocomial infection and fatality in medical and surgical intensive care unit patients. Arch Intem M ed 1988; 149:1161-68 5 Trilla A. Epidemiology of nosocomial infections in adult 674
intensive care units. Intensive Care Med 1994; 20:S1-S4 6 Mertens R, Kegels G, Stroobant A, et la. The national prevalence of nosocomial infection in Belgium , 1984. J Hosp Infect 1987; 9:219-29 7 Vincent JL, Bihari DJ, Suter PM , et al. The prevalence of nosocomial infection in intensive care units in Europe: results of the European Prevalence of Infection in Intensive Care (EPIC) Study; EPIC International Advismy Committee. JAMA 1995; 274:639-44 8 Standfast SJ, Michelsen PB, Baltch AL, et al. A prevalence survey of infections in a combined acute and long-term care hospital. Infect Control 1984; 5:177-84 9 Dunagan WC, Woodward RS , Medoff G, et a!. Antibiotic misuse in two clinical situations: positive blood cultures and administration of aminoglycosides. Rev Infect Dis 1991; 13:405-12 10 Kollef MR. Ventilator-associated pneumonia: a multivariate analysis. JAMA 1993; 270:1965-70 11 Beck-Sague CM , Sinkowitz RL, Chinn RY, eta!. Risk factors for ventilator-associated pneumonia in surgical intensivecare-unit patients. Infect Control Hosp Epidemiol 1996; 17:374-76 12 de Vera ME , Simmons RL. Antibiotic-resistant enterococci and the changing face of surgical infections. Arch Surg 1996; 131:338-42 13 Meyer KS , Urban C, Eagan JA, e ta!. Nosocomial outbreak of Klebsiella infection resistant to late-generation cephalosporins. Ann Intern Med 1993; 119:353-58 14 Goldmann DA, Weinstein RA, Wenzel RP, eta!. Strategies to prevent and control the emergence and spread of antimicrobial-resistant microorganisms in hospitals. JAMA 1996; 275: 234-40 15 Gold HS , Moellering RC. Antimicrobial-drug resistance. N Engl J Med 1996; 33.5:1445-53 16 Kollef MH , Wragge T, Pasque C. Determinants of mortality and multiorgan dysfunction in cardiac surgery patients requiring prolonged mechanical ventilation. Chest 1995; 107: 1395-1401 17 Menzies R, Gibbons W, Goldberg P. Dete rminants of weaning and surviving among patients with COPD who require mechanical ventilation for acute respiratory failure. Chest 1989; 95:398-405 18 Knaus WA, Draper EA, Wagne r DP, et al. APACHE II: a severity of disease classification system. Crit Care Med 1985; 13:818-29 19 Garner JS, Jarvis WR, Emori TG, et al. CDC definitions for nosocomial infections. Am J Infect Con trol 1988; 16:128-40 20 Pingleton SK, Fagon JY, Leeper KY. Patient selection for clinical investigation of ventilator-associated pneumonia: crite tia for evaluating diagnostic techniques. Chest 1992; 102: 553S-56S 21 Kollef MH, Bock KR, Richards RD , e t al. The safety and diagnostic accuracy of minibronchoalveolar lavage in patients with suspected ventilator-associated pneumonia. Ann Intern Med 1995; 122:743-48 22 Papazian L, Thomas P, Garbe L, et l.a Bronchoscopic or blind sampling techniques for the diagnosis of ventilator-associated pneumonia. Am J Respir Crit Care Med 1995; 152:1982-91 23 Meduti GU , Mauldin GL, Wunderink RG , e t al. Causes of fever and pulmonary densities in patients with clinical manifestations of ventilator-associated pneumonia. Chest 1994; 106:221-35 24 Marquette CH, Georges I-1 , Wallet F, et a!. Diagnostic efficiency of endotracheal aspirates with quantitative bacterial cultures in intubated patients with suspected pneumonia: comparison vvith the protected specimen brush. Am J Respir Crit Care Med 1993; 148:138-44 Clinical Investigations
25 Bernard GR, Artigas A, Brigham KL, et al. The AmeJicanEuropean Consensus Conference on ARDS: definitions, mechanisms, relevant outcomes, and clinical trial coordination. Am J Respir Crit Care Med 1994; 149:818-24 26 Rubin DB, Wiener-Kronish JP, Murray JF, et al. Elevated von Willebrand factor antigen is an early phase predictor of acute lung injury in nonpulmonary sepsis syndrome. J Clin Invest 1990; 86:474-80 27 American College of Chest Physicians/Society of Critical Care Medicine Consensus Committee. Definitions for sepsis and organ failures and guidelines for the use of innovative therapies in sepsis. Chest 1992; 101:1644-55 28 Meinert CL, Tonascia S. Clinical trials: design, conduct, and analysis. New York: Oxford University Press, 1986; 194-95 29 SAS/STAT user's guide. Cary, NC: SAS Institute, 1990; 1071-1126 30 Concato J, Feinstein HR, Holford TR. The risk of determining risk with multivariable models. Ann Intern Med 1993; 118:201-10 31 Rothman KJ. Analysis of crude data. In: Rothman K, ed. Modern epidemiology. Boston: Little Brown, 1986; 153-74 32 Brun-Buisson C, Doyon F, Carlet J, et la. Incidence, risk factors , and outcome of severe sepsis and septic shock in adults: a multicenter prospective study in intensive care units. JAMA 1995; 274:968-74 33 Coyle JP, Higgins TL. Cardiac surgery. In: Hall JB, Schmidt GA, Wood LDH, eds. Principles of clitical care. New York: McGraw-Hill , 1992; 1012-29 34 Leu HS, Kaiser DL, Mori M, e t al. Hospital-acquired pneumonia: attributable mortality and morbidity. Am J Epidemiol 1989; 129:1258-67 35 Fagan JY, Chastre J, Hance AJ, et al. Nosocomial pneumonia
36 37 38 39 40 41 42 43 44 45
in ventilated patients: a cohort study evaluating attributable mortality and hospital stay. Am J M ed 1993; 94:281-88 Giraud T, Dhainaut JF, Vaxelaire JF, et a!. Iatrogenic complications in adult intensive care units: a prospective twocenter study. Crit Care Med 1993; 21:40-51 Bueno-Cavanillas A, Delgado-Rodriguez M, Lopez-Luque A, et a!. Influence of nosocomial infection on mortality rate in an intensive care unit. Crit Care Med 1994; 22:55-60 Hickey RF, Cason BA. Timing of tracheal extubation in adult cardiac surgery patients. J Card Surg 1995; 10:340-48 Engelman RM , Rousou JA, Flack JE III, et al. Fast-track recovery of the coronary bypass patient. Ann Thorac Surg 1994; 58:1742-46 Chow JW, Fine MJ, Shlaes DM, et al. Enterobacter bacteremia: clinical features and emergence of antibiotic resistance dUJing therapy. Ann Intern Med 1991; 115:585-90 Kollef MH , Silver P, Murphy DM, et a!. The effect of late-onset ventilator-associated pneumonia in determining patient mortality. Chest 1995; 108:1655-62 Wunderink RG. Mortality and ventilator-associated pneumonia: the best antibiotics may be the el ast antibiotics [editorial]. Chest 1993; 104:993-95 McGowan JE Jr. Do intensive hospital antibiotic control programs prevent the spread of antibiotic resistance? Infect Control Hosp Epidemiol 1994; 15:478-83 Classen DC, Evans RS, Pestotnik SL, et a!. The timing of prophylactic administration of antibiotics and the risk of surgical-wound infection. N Eng! J M ed 1992; 326:281-86 Copeland M, Senkowski C, Ergin MA, et al. Macromastia as a factor in sternal wound dehiscence following cardiac surgery: management combining chest wall reconstruction and reduction mammoplasty. J Card Surg 1992; 7:275-78
CHEST / 112 /3/ SEPTEMBER, 1997
675