- Email: [email protected]
Severity-of-illness markers as predictors of nosocomial infection in adult intensive care unit patients
major articles
Severity-of-illness markers as predictors of nosocomial infection in adult intensive care unit patients Margaret E. McCusker, MD, MS André R.S. Périssé, MD, MS Mary-Claire Roghmann, MD, MS Baltimore, Maryland Background: Patients admitted to intensive care units (ICUs) are at high risk for acquiring nosocomial infections. We examined the association between markers of severity of illness at ICU admission and the development of ICU-attributable nosocomial infections. Methods: Retrospective cohort study of 851 patients admitted to the medical or surgical ICU in an urban teaching hospital from January 1997 to January 1998. Logistic regression analysis was used to identify predictors of nosocomial infection, including the Acute Physiology, Age, Chronic Health Evaluation III severity-of-illness scoring system. Results: Patients receiving mechanical ventilation on day 1 of ICU admission (OR, 1.99; 95% CI, 1.29-3.06) and patients transferred to the ICU from another unit within the same hospital (OR, 2.04; 95% CI, 1.24-3.34) were twice as likely to acquire an ICU-attributable nosocomial infection compared with patients admitted from other sources. The day-1 Acute Physiology, Age, Chronic Health Evaluation III score was not a significant predictor of nosocomial infection. Conclusion: The need for mechanical ventilation on ICU day 1 and transfer to the ICU from another unit are independent predictors of ICU-attributable nosocomial infections. Up to 50% of ICU patients who develop nosocomial infections could be easily identified at ICU admission, allowing for targeted use of preventive strategies to reduce the risk of nosocomial infections. (Am J Infect Control 2002;30:139-44.)
At least 2 million patients are diagnosed with nosocomial infections annually in the United States, at an estimated cost of more than $4.5 billion.1,2 In 1995, nosocomial infections contributed to approximately 88,000 deaths.2 The excess length of hospital stay due to nosocomial infection depends on the type of infection and has been estimated to add 1 to 4 days for urinary tract infections and 7 to 30 days for pneumonia. More than 20% of all hospital-acquired infections occur in intensive care unit (ICU) From the Department of Epidemiology and Preventive Medicine, University of Maryland School of Medicine. Reprint requests: Margaret E. McCusker, MD, MS, University of Maryland School of Medicine, Department of Epidemiology and Preventive Medicine, 660 W Redwood St, Baltimore, MD 21201. Copyright © 2002 by the Association for Professionals in Infection Control and Epidemiology, Inc. 0196-6553/2002/$35.00 + 0 17/46/121662 doi:10.1067/mic.2002.121662
patients.3 Approximately half of all nosocomial infections are related to the use of invasive medical devices.4 Devices such as antiseptic-coated urinary and vascular catheters decrease the risk of developing a nosocomial infection.5,6 The cost-effectiveness of these interventions is increased by their use in patients most likely to develop infections.7 Risk factors for nosocomial infection in ICU patients are often related to their severity of illness.4,8-14 Thus, many similar studies have investigated whether clinical prediction tools, developed to predict mortality and length of stay, are useful for determining which ICU patients have the highest risk of developing nosocomial infections. The Acute Physiology, Age, Chronic Health Evaluation (APACHE) severity-of-illness scoring system has been used in several of these studies, but no published study has examined the association between severity of illness as measured by the APACHE III score and the development of 139
McCusker, Périssé, and Roghmann
140 Vol. 30 No. 3
Table 1. Characteristics of patients admitted to the MICU or SICU for more than 2 days, January 1997January 1998 (n = 851) Characteristic
No. of patients (%)
Mean (SD)
Range
54 (17)
18-97
48 (24) 8 (8)
0-151 3-72
Demographics Age in years Male Race Black Other Type of ICU MICU SICU ICU admission diagnosis Respiratory Gastrointestinal Cardiovascular Transplant Other Location before ICU admission Operating/recovery room Hospital floor/special care unit Emergency room Another hospital Direct admission Other Mechanical ventilation, day 1 (n = 849)* APACHE III score, day 1 Length of stay, days Nosocomial infection, yes
450 (52.9) 359 (42.2) 492 (57.8) 326 (38.3) 525 (61.7) 222 178 147 139 165
(26.1) (20.9) (17.3) (16.3) (19.4)
302 (35.5) 248 (29.1) 152 (17.9) 124 (14.6) 10 (1.1) 15 (1.8) 373 (43.9)
109 (12.8)
MICU, Medical intensive care unit; SICU, surgical intensive care unit. *Data for 2 patients were missing for this variable.
overall nosocomial infections in a cohort of ICU patients. The purpose of this study was to determine whether markers of severity of illness, including the day-1 APACHE III score, were associated with the development of ICU-attributable nosocomial infections. Identification of predictors of nosocomial infections could help to reduce the impact of these infections on critically ill patients.
METHODS The University of Maryland Medical Center is a 747bed urban teaching hospital located in Baltimore, Maryland. The hospital has a 10-bed medical ICU (MICU), a 19-bed surgical ICU (SICU), and several specialty ICUs. A retrospective cohort study of patients admitted to the MICU or SICU between January 1997, and January 1998, was conducted. The study was approved by the Institutional Review Board of the University of Maryland. The exposure variable was severity of illness at ICU admission as measured by the day-1 APACHE III
score and other patient factors. The outcome was the development of a nosocomial infection during or up to 48 hours after discharge from the ICU. The University of Maryland Medical Center participates in the ICU component of the National Nosocomial Infection Surveillance System (NNIS; Centers for Disease Control and Prevention, Atlanta, Ga). All ICUattributable infections meeting NNIS criteria are recorded and reported to the Centers for Disease Control and Prevention on a quarterly basis. The first ICU admission of at least 2 days’ duration or the first stay during which an ICU-attributable infection occurred was analyzed for each patient. Patients who died or were transferred from the ICU less than 2 days after admission were excluded from the study because their shorter length of stay precluded them from being at risk for ICU-attributable infection. ICU nursing staff who had been trained to use the APACHE III system collected patient data and calculated APACHE III scores. The data were collected as part of routine patient care. A full-time nurse coordinator, trained by APACHE Medical Systems (McLean,
McCusker, Périssé, and Roghmann
May 2002 141
Table 2. Nosocomial infection status by patient characteristics (n = 851) NI (n = 109)
Characteristics Demographics Age in years, mean (SD) Sex, number (%) Male Female Race, number (%) Black Other Type of ICU, number (%) MICU SICU ICU admission diagnosis, number (%) Respiratory Gastrointestinal Cardiovascular ` Transplant Other Location before ICU admission, number (%) Operating/recovery room Hospital floor/special care unit Emergency room Another hospital Direct admission Other Mechanical ventilation, day 1, number (%) (n = 849)† APACHE III score, day 1, mean (SD) NI, Nosocomial infection; SD, standard deviation. *P value by Student t test for continuous variables and by †Data for 2 people were missing for this variable.
2
54 (16)
Non-NI (n = 742)
54 (17)
52 (47.7) 57 (52.3)
398 (53.6) 344 (46.4)
46 (42.2) 63 (57.8)
313 (42.2) 429 (57.8)
41 (37.6) 68 (62.4)
285 (38.4) 457 (61.6)
37 23 18 13 18
185 155 129 126 147
P value*
.89 .25
.99
.87
.29 (33.9) (21.2) (16.5) (11.9) (16.5)
(24.9) (20.9) (17.4) (17.0) (19.8) .05
34 (31.2) 46 (42.2) 13 (11.9) 13 (11.9) 1 (1.0) 2 (1.8) 62 (56.8) 51 (25)
268 (36.1) 202 (27.2) 139 (18.7) 111 (14.9) 9 (1.2) 13 (1.9) 311 (41.9) 47 (24)
.01 .08
for categorical variables.
Va), performed monthly quality assurance audits on 10% of the data entered for each ICU. Data collected for the APACHE III system included age; sex; race; dates of ICU admission and discharge; ICU admission diagnosis; source of ICU admission (another hospital, direct admission, emergency room, hospital floor, operating room, recovery room, or special care unit); and any major comorbidities present on admission, as defined by the APACHE III manual (AIDS, hepatic failure, lymphoma, metastatic cancer, leukemia/multiple myeloma, immunosuppression, or cirrhosis).15 The APACHE III score has a range of 0 to 299 points that are derived from 3 categories: the acute physiology score (APS; 0-252 points), age (0-24 points), and presence of 1 of 7 major comorbidities (0-23 points).15 Nosocomial infection was defined as an infection occurring 48 hours or more after ICU admission and up to 48 hours after discharge from the ICU. Each specific type of infection was defined according to the NNIS criteria for that infection.16 Although some patients had more than 1 nosocomial infection, only
the first infection diagnosed in each patient was included in this study. Infection control personnel collected data according to NNIS protocols. Mechanical ventilation was recorded as present or absent during the first day of ICU stay. This information was derived from the APACHE III data collected for the acute physiology score. Whether a patient was receiving mechanical ventilation at the time the respiratory rate was counted was recorded as a component of the vital signs, but it is not included as part of the score.15 Univariate analysis of patient characteristics included means and standard deviations for continuous variables and frequencies for categorical variables. Statistical significance was set at = .05. The incidence rate of nosocomial infections by source was calculated per 1000 patient-days of ICU stay for the combined study population and separately for the MICU and SICU. Bivariate analysis was performed with the Student t test for continuous variables and Pearson’s 2 test for categorical variables.
McCusker, Périssé, and Roghmann
142 Vol. 30 No. 3
Table 3. Logistic regression analysis of risk factors for nosocomial infection (n = 849) Characteristic APACHE III score 0-34 35-53 ≥54 Mechanical ventilation, day 1 Location before ICU admission Operating/recovery room Hospital floor/special care unit Emergency room Another hospital Direct admission
Logistic regression analysis was performed to build a model that measured the association between development of nosocomial infection and severity of illness, with use of a backward elimination process. Interaction terms between APACHE III score and type of ICU were also examined in the model. All analysis was performed with use of SAS version 8.0 (SAS, Cary, NC).
RESULTS During the 12-month study, 1261 patients were admitted to the medical and surgical ICUs. Of these patients, 851 (67.5%) had an ICU stay of more than 48 hours. Thirteen percent (n = 109) developed nosocomial infections. The baseline characteristics of the study population are shown in Table 1. APACHE III scores ranged from 0 to 151 out of 299 possible points. Overall incidence rate of nosocomial infections for both units was 17.1 infections per 1000 patient-days; there was no difference in overall incidence of nosocomial infections by unit. The 2 most common sites of nosocomial infection were the urinary tract (5.2 cases per 1000 patient-days) and the bloodstream (4.7 cases per 1000 patient-days). Incidence rate of bloodstream infections was significantly higher in MICU patients than in SICU patients (7.4 per 1000 in the MICU vs 3.0 per 1000 patient-days in the SICU; P = .01). No surgical wound infections were detected among patients in the MICU. Patients with and without a nosocomial infection are compared in Table 2. The most striking difference between the 2 groups is that a significantly higher proportion of patients who developed a nosocomial infection required mechanical ventilation on day 1 of their ICU stay. Patients who developed a nosocomial
Odds ratio
95% Confidence interval
1.00 1.22 1.16 1.99
Reference 0.72-2.07 0.69-1.97 1.29-3.06
1.00 2.04 0.81 0.97 1.17
Reference 1.24-3.34 0.41-1.60 0.49-1.93 0.14-9.73
infection also had a higher day-1 APACHE III score than patients who did not develop an infection and were significantly more likely to have been transferred to an ICU from another hospital floor. The mean APACHE III score for patients requiring mechanical ventilation on ICU day 1 was 55, significantly higher than the mean of 42 for nonventilated patients (P < .01). The highest mean APACHE III score for ICU admission source was 52, which was for patients transferred from wards within the hospital. The difference in scores for this variable was also statistically significant (P < .01). Logistic regression analysis of the association between the need for mechanical ventilation on ICU day 1 and the subtypes of nosocomial infection examined in this study indicated that the patients who required ventilatory support on ICU day 1 were more likely to develop both pneumonia (P = .04) and urinary tract infections (P = .01) than were patients who could breathe without assistance. Table 3 shows the final logistic regression model. Patients who required mechanical ventilation on day 1 of their ICU stay were twice as likely to develop a nosocomial infection as patients who could breathe without assistance. Location before ICU admission, specifically being on a hospital floor or in another special care unit before transfer to the ICU, was associated with a 2-fold increase in the odds of having a nosocomial infection. None of the other covariates was significantly associated with the acquisition of an ICU-attributable infection. A second logistic regression model that excluded patients with transplants identified the same associations between need for mechanical ventilation on ICU day 1 (OR, 2.05; 95% CI, 1.29-3.27) or transfer from another floor within the hospital (OR, 1.82; 95% CI, 1.053.15) and the development of nosocomial infection.
McCusker, Périssé, and Roghmann
The proportion of nosocomial infections attributable to the need for mechanical ventilation on ICU day 1 and/or transfer to the ICU from another floor within the same hospital was calculated to be 50%. When looking at these risk factors separately, 28% of nosocomial infections were attributable to the need for mechanical ventilation on ICU day 1, and 21% were attributable to transfer from another hospital unit.
DISCUSSION Two factors were significantly associated with a 2fold increase in the odds developing an ICU-attributable nosocomial infection: requiring mechanical ventilation on day 1 of ICU stay and transfer from another floor or special care unit within the hospital. The day-1 APACHE III score, which does not include these variables, was not a significant predictor of nosocomial infection in the bivariate analysis or in the logistic regression model. Our results indicate that patients requiring mechanical ventilation on day 1 of ICU stay have a higher risk of developing nosocomial infections. Mechanical ventilation is a well-described risk factor for nosocomial pneumonia. Hurr et al13 have also described an association similar to ours in trauma patients. In our study, mechanical ventilation may have been a surrogate marker of severity of illness and consequent device utilization. Since almost half of all nosocomial infections are device related,10 patients requiring mechanical ventilation may be more likely to develop nosocomial infections because of their increased exposure to other invasive devices. Nosocomial pneumonia only accounted for 16% of infections in our study population. The need for mechanical ventilation on ICU day 1 was associated with nosocomial infections in general and especially pneumonia and urinary tract infections. Mechanical ventilation indirectly predicts nosocomial infection, not simply pneumonia, by identifying patients with high-device utilization and prolonged length of stay. Our study is the only study to date that has identified location in another hospital unit before ICU admission as a risk factor for the development of ICU-attributable infections. A patient is usually transferred to the ICU from another unit because of deterioration of clinical condition and is therefore more likely to be critically ill. Thus, being admitted to the ICU from another unit could be another marker of severity of illness. In addition, transferred patients have a greater opportunity for exposure to invasive
May 2002 143
devices and of nosocomial pathogens such as methicillin-resistant Staphylococcus aureus. Other studies have also investigated the association between the APACHE score and nosocomial infection. Mylotte et al14 reported that patients with a higher APACHE III score were more likely to develop nosocomial infection. However, comparison to our data is difficult since their study was conducted in an acute rehabilitation unit and they used only the acute physiology score component of the APACHE III score. Higher day-1 APACHE II scores were significantly associated with an increased risk of nosocomial infection in a study of ICU patients in Spain4; however, 2 other studies that examined the connection between the APACHE II score and the development of nosocomial infections did not find an association.8,13 The changes in the APACHE scoring system between the second and third versions, especially the wider range of possible scores and increased weighting of physiologic variables, could affect the ability of the scoring system to predict nosocomial infection,17 but our study did not find that the APACHE III score predicted nosocomial infection. Several studies have reported an association between length of ICU stay and the development of nosocomial infections.8,10,11,13 Length of stay has been considered to be a confounder of the association between severity of illness scores and development of nosocomial infection.8,10,13 We suspect that length of stay is an intermediate variable in the causal pathway between severity of illness and nosocomial infection; severely ill patients tend to have longer hospitalizations and thus have a greater risk of developing a nosocomial infection. In addition, we wanted to identify predictors that could be used at ICU admission; therefore, we did not control for length of stay in our analysis. A number of issues should be considered when interpreting our results. The data used to conduct this study were collected for clinical purposes unrelated to our research objectives. This limited our ability to look at other potential predictors for nosocomial infection. Since our study was conducted in an urban teaching hospital, its generalizability is limited to ICUs in similar hospitals. The University of Maryland is a major center for organ transplantation. For this reason, 16% of our patients had undergone transplantation. Excluding these patients from the analysis did not affect the results of our study.
144 Vol. 30 No. 3 Two important strengths of this study were the use of the APACHE III score as it was designed and the use of the NNIS for detection of nosocomial infections. The use of these well-defined criteria enhances the reliability of our results. Another strength of this study was the focus on predictors for ICU-attributable infections that were present at the time of ICU admission. Our emphasis on characteristics predisposing a patient to nosocomial infection that can be identified at ICU admission provides a window of opportunity to implement preventive measures before complications arise. In this study, the need for mechanical ventilation on day 1 of ICU admission and transfer to the ICU from another unit within the same hospital were identified as independent risk factors for acquisition of nosocomial infection during ICU stay. Simple factors such as these may predict which patients will develop nosocomial infections. Up to 50% of patients who develop ICU-attributable infections could be recognized and targeted for preventive interventions with use of the risk factors identified in this study. Expensive devices such as antiseptic-coated urinary and vascular catheters could be used, in addition to standard infection control techniques, to reduce the risk of infection in these patients, ultimately reducing both the clinical and financial burden associated with these infections. The authors wish to thank the nurses and clinical staff of the medical and surgical ICUs and the infection control practitioners at the University of Maryland, all of whom worked to gather the data used in this study. We are grateful for the technical advice and critiques offered by Drs Patricia Langenberg and Mona Baumgarten. Drs Yingkai Cheng, Elaine Stanek, Paul Sedhev, Karyl Thomas, Ilene Zuckerman, and Priscilla Ryder, MPH, also provided guidance.
References 1. Public health focus: surveillance, prevention, and control of nosocomial infections. MMWR Morb Mortal Wkly Rep 1992;41:783-7.
McCusker, Périssé, and Roghmann
2. Weinstein RA. Nosocomial infection update. Emerg Infect Dis 1998;4:416-20. 3. Fridkin SK, Weibel SF, Weinstein RA. Magnitude and prevention of nosocomial infections in the intensive care unit. Infect Dis Clin North Am 1997; 11:479-96. 4. Fernandez-Crehuet R, Diaz-Molina C, de Irala J, Martinez-Concha D, Salcedo-Leal I, Masa-Calles J. Nosocomial infection in an intensive-care unit: identification of risk factors. Infect Control Hosp Epidemiol 1997;18:825-30. 5. Karchmer TB, Gianetta ET, Muto CA, Strain BA, Farr BM. A randomized crossover study of silver coated urinary catheters in hospitalized patients. Arch Intern Med 2000;160:3294-8. 6. Tcholakian RK, Raad II. Durability of anti-infective effect of long-term silicone sheath catheters impregnated with antimicrobial agents. Antimicrob Agents Chemother 2001;45:1990-3. 7. Saint S,Veenstra DL, Sullivan SD, Chenoweth C, Fendrick AM.The potential clinical and economic benefits of silver alloy urinary catheters in preventing urinary tract infection. Arch Intern Med 2000;160:2670-5. 8. Bueno-Cavanillas A, Rodriguez-Contreras R, Lopez-Luque A, DelgadoRodriguez M, Galves-Vargas R. Usefulness of severity indices in intensive care medicine as a predictor of nosocomial infection risk. Intensive Care Med 1991;17:336-9. 9. Chevret S, Hemmer M, Carlet J, Langer M. Incidence and risk factors of pneumonia acquired in intensive care units. Results from a multicenter prospective study on 996 patients. European Cooperative Group on Nosocomial Pneumonia. Intensive Care Med 1993;19:256-64. 10. Craven DE, Kunches LM, Lichtenberg DA, et al. Nosocomial infection and fatality in medical and surgical intensive care unit patients. Arch Intern Med 1988;148:1161-8. 11. Cunnion KM, Weber DJ, Broadhead WE, Hanson LC, Pieper CF, Rutala WA. Risk factors for nosocomial pneumonia: comparing adult critical care populations. Am J Respir Crit Care Med 1996;153:158-62. 12. Fagon JY, Chastre J,Vuagnat A,Trouillet JL, Novara A, Gibert C. Nosocomial pneumonia and mortality among patients in intensive care units. JAMA 1996;275:866-9. 13. Hurr H, Hawley HB, Czachor JS, Markert RJ, McCarthy MC. APACHE II and ISS scores as predictors of nosocomial infections in trauma patients. Am J Infect Control 1999;27:79-83. 14. Mylotte JM, Graham R, Kahler L, Young L, Goodnough S. Epidemiology of nosocomial infection and resistant organisms in patients admitted for the first time to an acute rehabilitation unit. Clin Infect Dis 200;30:425-32. 15. APACHE critical care—methodology and data collection manual. McLean (VA): APACHE Medical Systems, Inc; 1994. 16. NNIS manual. Atlanta: Centers for Disease Control and Prevention, US Department of Health and Human Services; 1994. 17. Barie PS, Hydo LJ, Fischer E. Comparison of APACHE II and III scoring systems for mortality prediction in critical surgical illness. Arch Surg 1995; 130:77-82.
Visit AJIC online! Your subscription to the print version of AJIC includes free access to the online version of AJIC. Activate the online portion of your subscription and get the most from AJIC, including “Online Exclusives”—content you’ll only find in the Web version! To activate your online subscription, visit www.mosby.com/ajic and follow the activation instructions. We invite you to visit today.
Severity-of-illness markers as predictors of nosocomial infection in adult intensive care unit patients Margaret E. McCusker, MD, MS André R.S. Périssé, MD, MS Mary-Claire Roghmann, MD, MS Baltimore, Maryland Background: Patients admitted to intensive care units (ICUs) are at high risk for acquiring nosocomial infections. We examined the association between markers of severity of illness at ICU admission and the development of ICU-attributable nosocomial infections. Methods: Retrospective cohort study of 851 patients admitted to the medical or surgical ICU in an urban teaching hospital from January 1997 to January 1998. Logistic regression analysis was used to identify predictors of nosocomial infection, including the Acute Physiology, Age, Chronic Health Evaluation III severity-of-illness scoring system. Results: Patients receiving mechanical ventilation on day 1 of ICU admission (OR, 1.99; 95% CI, 1.29-3.06) and patients transferred to the ICU from another unit within the same hospital (OR, 2.04; 95% CI, 1.24-3.34) were twice as likely to acquire an ICU-attributable nosocomial infection compared with patients admitted from other sources. The day-1 Acute Physiology, Age, Chronic Health Evaluation III score was not a significant predictor of nosocomial infection. Conclusion: The need for mechanical ventilation on ICU day 1 and transfer to the ICU from another unit are independent predictors of ICU-attributable nosocomial infections. Up to 50% of ICU patients who develop nosocomial infections could be easily identified at ICU admission, allowing for targeted use of preventive strategies to reduce the risk of nosocomial infections. (Am J Infect Control 2002;30:139-44.)
At least 2 million patients are diagnosed with nosocomial infections annually in the United States, at an estimated cost of more than $4.5 billion.1,2 In 1995, nosocomial infections contributed to approximately 88,000 deaths.2 The excess length of hospital stay due to nosocomial infection depends on the type of infection and has been estimated to add 1 to 4 days for urinary tract infections and 7 to 30 days for pneumonia. More than 20% of all hospital-acquired infections occur in intensive care unit (ICU) From the Department of Epidemiology and Preventive Medicine, University of Maryland School of Medicine. Reprint requests: Margaret E. McCusker, MD, MS, University of Maryland School of Medicine, Department of Epidemiology and Preventive Medicine, 660 W Redwood St, Baltimore, MD 21201. Copyright © 2002 by the Association for Professionals in Infection Control and Epidemiology, Inc. 0196-6553/2002/$35.00 + 0 17/46/121662 doi:10.1067/mic.2002.121662
patients.3 Approximately half of all nosocomial infections are related to the use of invasive medical devices.4 Devices such as antiseptic-coated urinary and vascular catheters decrease the risk of developing a nosocomial infection.5,6 The cost-effectiveness of these interventions is increased by their use in patients most likely to develop infections.7 Risk factors for nosocomial infection in ICU patients are often related to their severity of illness.4,8-14 Thus, many similar studies have investigated whether clinical prediction tools, developed to predict mortality and length of stay, are useful for determining which ICU patients have the highest risk of developing nosocomial infections. The Acute Physiology, Age, Chronic Health Evaluation (APACHE) severity-of-illness scoring system has been used in several of these studies, but no published study has examined the association between severity of illness as measured by the APACHE III score and the development of 139
McCusker, Périssé, and Roghmann
140 Vol. 30 No. 3
Table 1. Characteristics of patients admitted to the MICU or SICU for more than 2 days, January 1997January 1998 (n = 851) Characteristic
No. of patients (%)
Mean (SD)
Range
54 (17)
18-97
48 (24) 8 (8)
0-151 3-72
Demographics Age in years Male Race Black Other Type of ICU MICU SICU ICU admission diagnosis Respiratory Gastrointestinal Cardiovascular Transplant Other Location before ICU admission Operating/recovery room Hospital floor/special care unit Emergency room Another hospital Direct admission Other Mechanical ventilation, day 1 (n = 849)* APACHE III score, day 1 Length of stay, days Nosocomial infection, yes
450 (52.9) 359 (42.2) 492 (57.8) 326 (38.3) 525 (61.7) 222 178 147 139 165
(26.1) (20.9) (17.3) (16.3) (19.4)
302 (35.5) 248 (29.1) 152 (17.9) 124 (14.6) 10 (1.1) 15 (1.8) 373 (43.9)
109 (12.8)
MICU, Medical intensive care unit; SICU, surgical intensive care unit. *Data for 2 patients were missing for this variable.
overall nosocomial infections in a cohort of ICU patients. The purpose of this study was to determine whether markers of severity of illness, including the day-1 APACHE III score, were associated with the development of ICU-attributable nosocomial infections. Identification of predictors of nosocomial infections could help to reduce the impact of these infections on critically ill patients.
METHODS The University of Maryland Medical Center is a 747bed urban teaching hospital located in Baltimore, Maryland. The hospital has a 10-bed medical ICU (MICU), a 19-bed surgical ICU (SICU), and several specialty ICUs. A retrospective cohort study of patients admitted to the MICU or SICU between January 1997, and January 1998, was conducted. The study was approved by the Institutional Review Board of the University of Maryland. The exposure variable was severity of illness at ICU admission as measured by the day-1 APACHE III
score and other patient factors. The outcome was the development of a nosocomial infection during or up to 48 hours after discharge from the ICU. The University of Maryland Medical Center participates in the ICU component of the National Nosocomial Infection Surveillance System (NNIS; Centers for Disease Control and Prevention, Atlanta, Ga). All ICUattributable infections meeting NNIS criteria are recorded and reported to the Centers for Disease Control and Prevention on a quarterly basis. The first ICU admission of at least 2 days’ duration or the first stay during which an ICU-attributable infection occurred was analyzed for each patient. Patients who died or were transferred from the ICU less than 2 days after admission were excluded from the study because their shorter length of stay precluded them from being at risk for ICU-attributable infection. ICU nursing staff who had been trained to use the APACHE III system collected patient data and calculated APACHE III scores. The data were collected as part of routine patient care. A full-time nurse coordinator, trained by APACHE Medical Systems (McLean,
McCusker, Périssé, and Roghmann
May 2002 141
Table 2. Nosocomial infection status by patient characteristics (n = 851) NI (n = 109)
Characteristics Demographics Age in years, mean (SD) Sex, number (%) Male Female Race, number (%) Black Other Type of ICU, number (%) MICU SICU ICU admission diagnosis, number (%) Respiratory Gastrointestinal Cardiovascular ` Transplant Other Location before ICU admission, number (%) Operating/recovery room Hospital floor/special care unit Emergency room Another hospital Direct admission Other Mechanical ventilation, day 1, number (%) (n = 849)† APACHE III score, day 1, mean (SD) NI, Nosocomial infection; SD, standard deviation. *P value by Student t test for continuous variables and by †Data for 2 people were missing for this variable.
2
54 (16)
Non-NI (n = 742)
54 (17)
52 (47.7) 57 (52.3)
398 (53.6) 344 (46.4)
46 (42.2) 63 (57.8)
313 (42.2) 429 (57.8)
41 (37.6) 68 (62.4)
285 (38.4) 457 (61.6)
37 23 18 13 18
185 155 129 126 147
P value*
.89 .25
.99
.87
.29 (33.9) (21.2) (16.5) (11.9) (16.5)
(24.9) (20.9) (17.4) (17.0) (19.8) .05
34 (31.2) 46 (42.2) 13 (11.9) 13 (11.9) 1 (1.0) 2 (1.8) 62 (56.8) 51 (25)
268 (36.1) 202 (27.2) 139 (18.7) 111 (14.9) 9 (1.2) 13 (1.9) 311 (41.9) 47 (24)
.01 .08
for categorical variables.
Va), performed monthly quality assurance audits on 10% of the data entered for each ICU. Data collected for the APACHE III system included age; sex; race; dates of ICU admission and discharge; ICU admission diagnosis; source of ICU admission (another hospital, direct admission, emergency room, hospital floor, operating room, recovery room, or special care unit); and any major comorbidities present on admission, as defined by the APACHE III manual (AIDS, hepatic failure, lymphoma, metastatic cancer, leukemia/multiple myeloma, immunosuppression, or cirrhosis).15 The APACHE III score has a range of 0 to 299 points that are derived from 3 categories: the acute physiology score (APS; 0-252 points), age (0-24 points), and presence of 1 of 7 major comorbidities (0-23 points).15 Nosocomial infection was defined as an infection occurring 48 hours or more after ICU admission and up to 48 hours after discharge from the ICU. Each specific type of infection was defined according to the NNIS criteria for that infection.16 Although some patients had more than 1 nosocomial infection, only
the first infection diagnosed in each patient was included in this study. Infection control personnel collected data according to NNIS protocols. Mechanical ventilation was recorded as present or absent during the first day of ICU stay. This information was derived from the APACHE III data collected for the acute physiology score. Whether a patient was receiving mechanical ventilation at the time the respiratory rate was counted was recorded as a component of the vital signs, but it is not included as part of the score.15 Univariate analysis of patient characteristics included means and standard deviations for continuous variables and frequencies for categorical variables. Statistical significance was set at = .05. The incidence rate of nosocomial infections by source was calculated per 1000 patient-days of ICU stay for the combined study population and separately for the MICU and SICU. Bivariate analysis was performed with the Student t test for continuous variables and Pearson’s 2 test for categorical variables.
McCusker, Périssé, and Roghmann
142 Vol. 30 No. 3
Table 3. Logistic regression analysis of risk factors for nosocomial infection (n = 849) Characteristic APACHE III score 0-34 35-53 ≥54 Mechanical ventilation, day 1 Location before ICU admission Operating/recovery room Hospital floor/special care unit Emergency room Another hospital Direct admission
Logistic regression analysis was performed to build a model that measured the association between development of nosocomial infection and severity of illness, with use of a backward elimination process. Interaction terms between APACHE III score and type of ICU were also examined in the model. All analysis was performed with use of SAS version 8.0 (SAS, Cary, NC).
RESULTS During the 12-month study, 1261 patients were admitted to the medical and surgical ICUs. Of these patients, 851 (67.5%) had an ICU stay of more than 48 hours. Thirteen percent (n = 109) developed nosocomial infections. The baseline characteristics of the study population are shown in Table 1. APACHE III scores ranged from 0 to 151 out of 299 possible points. Overall incidence rate of nosocomial infections for both units was 17.1 infections per 1000 patient-days; there was no difference in overall incidence of nosocomial infections by unit. The 2 most common sites of nosocomial infection were the urinary tract (5.2 cases per 1000 patient-days) and the bloodstream (4.7 cases per 1000 patient-days). Incidence rate of bloodstream infections was significantly higher in MICU patients than in SICU patients (7.4 per 1000 in the MICU vs 3.0 per 1000 patient-days in the SICU; P = .01). No surgical wound infections were detected among patients in the MICU. Patients with and without a nosocomial infection are compared in Table 2. The most striking difference between the 2 groups is that a significantly higher proportion of patients who developed a nosocomial infection required mechanical ventilation on day 1 of their ICU stay. Patients who developed a nosocomial
Odds ratio
95% Confidence interval
1.00 1.22 1.16 1.99
Reference 0.72-2.07 0.69-1.97 1.29-3.06
1.00 2.04 0.81 0.97 1.17
Reference 1.24-3.34 0.41-1.60 0.49-1.93 0.14-9.73
infection also had a higher day-1 APACHE III score than patients who did not develop an infection and were significantly more likely to have been transferred to an ICU from another hospital floor. The mean APACHE III score for patients requiring mechanical ventilation on ICU day 1 was 55, significantly higher than the mean of 42 for nonventilated patients (P < .01). The highest mean APACHE III score for ICU admission source was 52, which was for patients transferred from wards within the hospital. The difference in scores for this variable was also statistically significant (P < .01). Logistic regression analysis of the association between the need for mechanical ventilation on ICU day 1 and the subtypes of nosocomial infection examined in this study indicated that the patients who required ventilatory support on ICU day 1 were more likely to develop both pneumonia (P = .04) and urinary tract infections (P = .01) than were patients who could breathe without assistance. Table 3 shows the final logistic regression model. Patients who required mechanical ventilation on day 1 of their ICU stay were twice as likely to develop a nosocomial infection as patients who could breathe without assistance. Location before ICU admission, specifically being on a hospital floor or in another special care unit before transfer to the ICU, was associated with a 2-fold increase in the odds of having a nosocomial infection. None of the other covariates was significantly associated with the acquisition of an ICU-attributable infection. A second logistic regression model that excluded patients with transplants identified the same associations between need for mechanical ventilation on ICU day 1 (OR, 2.05; 95% CI, 1.29-3.27) or transfer from another floor within the hospital (OR, 1.82; 95% CI, 1.053.15) and the development of nosocomial infection.
McCusker, Périssé, and Roghmann
The proportion of nosocomial infections attributable to the need for mechanical ventilation on ICU day 1 and/or transfer to the ICU from another floor within the same hospital was calculated to be 50%. When looking at these risk factors separately, 28% of nosocomial infections were attributable to the need for mechanical ventilation on ICU day 1, and 21% were attributable to transfer from another hospital unit.
DISCUSSION Two factors were significantly associated with a 2fold increase in the odds developing an ICU-attributable nosocomial infection: requiring mechanical ventilation on day 1 of ICU stay and transfer from another floor or special care unit within the hospital. The day-1 APACHE III score, which does not include these variables, was not a significant predictor of nosocomial infection in the bivariate analysis or in the logistic regression model. Our results indicate that patients requiring mechanical ventilation on day 1 of ICU stay have a higher risk of developing nosocomial infections. Mechanical ventilation is a well-described risk factor for nosocomial pneumonia. Hurr et al13 have also described an association similar to ours in trauma patients. In our study, mechanical ventilation may have been a surrogate marker of severity of illness and consequent device utilization. Since almost half of all nosocomial infections are device related,10 patients requiring mechanical ventilation may be more likely to develop nosocomial infections because of their increased exposure to other invasive devices. Nosocomial pneumonia only accounted for 16% of infections in our study population. The need for mechanical ventilation on ICU day 1 was associated with nosocomial infections in general and especially pneumonia and urinary tract infections. Mechanical ventilation indirectly predicts nosocomial infection, not simply pneumonia, by identifying patients with high-device utilization and prolonged length of stay. Our study is the only study to date that has identified location in another hospital unit before ICU admission as a risk factor for the development of ICU-attributable infections. A patient is usually transferred to the ICU from another unit because of deterioration of clinical condition and is therefore more likely to be critically ill. Thus, being admitted to the ICU from another unit could be another marker of severity of illness. In addition, transferred patients have a greater opportunity for exposure to invasive
May 2002 143
devices and of nosocomial pathogens such as methicillin-resistant Staphylococcus aureus. Other studies have also investigated the association between the APACHE score and nosocomial infection. Mylotte et al14 reported that patients with a higher APACHE III score were more likely to develop nosocomial infection. However, comparison to our data is difficult since their study was conducted in an acute rehabilitation unit and they used only the acute physiology score component of the APACHE III score. Higher day-1 APACHE II scores were significantly associated with an increased risk of nosocomial infection in a study of ICU patients in Spain4; however, 2 other studies that examined the connection between the APACHE II score and the development of nosocomial infections did not find an association.8,13 The changes in the APACHE scoring system between the second and third versions, especially the wider range of possible scores and increased weighting of physiologic variables, could affect the ability of the scoring system to predict nosocomial infection,17 but our study did not find that the APACHE III score predicted nosocomial infection. Several studies have reported an association between length of ICU stay and the development of nosocomial infections.8,10,11,13 Length of stay has been considered to be a confounder of the association between severity of illness scores and development of nosocomial infection.8,10,13 We suspect that length of stay is an intermediate variable in the causal pathway between severity of illness and nosocomial infection; severely ill patients tend to have longer hospitalizations and thus have a greater risk of developing a nosocomial infection. In addition, we wanted to identify predictors that could be used at ICU admission; therefore, we did not control for length of stay in our analysis. A number of issues should be considered when interpreting our results. The data used to conduct this study were collected for clinical purposes unrelated to our research objectives. This limited our ability to look at other potential predictors for nosocomial infection. Since our study was conducted in an urban teaching hospital, its generalizability is limited to ICUs in similar hospitals. The University of Maryland is a major center for organ transplantation. For this reason, 16% of our patients had undergone transplantation. Excluding these patients from the analysis did not affect the results of our study.
144 Vol. 30 No. 3 Two important strengths of this study were the use of the APACHE III score as it was designed and the use of the NNIS for detection of nosocomial infections. The use of these well-defined criteria enhances the reliability of our results. Another strength of this study was the focus on predictors for ICU-attributable infections that were present at the time of ICU admission. Our emphasis on characteristics predisposing a patient to nosocomial infection that can be identified at ICU admission provides a window of opportunity to implement preventive measures before complications arise. In this study, the need for mechanical ventilation on day 1 of ICU admission and transfer to the ICU from another unit within the same hospital were identified as independent risk factors for acquisition of nosocomial infection during ICU stay. Simple factors such as these may predict which patients will develop nosocomial infections. Up to 50% of patients who develop ICU-attributable infections could be recognized and targeted for preventive interventions with use of the risk factors identified in this study. Expensive devices such as antiseptic-coated urinary and vascular catheters could be used, in addition to standard infection control techniques, to reduce the risk of infection in these patients, ultimately reducing both the clinical and financial burden associated with these infections. The authors wish to thank the nurses and clinical staff of the medical and surgical ICUs and the infection control practitioners at the University of Maryland, all of whom worked to gather the data used in this study. We are grateful for the technical advice and critiques offered by Drs Patricia Langenberg and Mona Baumgarten. Drs Yingkai Cheng, Elaine Stanek, Paul Sedhev, Karyl Thomas, Ilene Zuckerman, and Priscilla Ryder, MPH, also provided guidance.
References 1. Public health focus: surveillance, prevention, and control of nosocomial infections. MMWR Morb Mortal Wkly Rep 1992;41:783-7.
McCusker, Périssé, and Roghmann
2. Weinstein RA. Nosocomial infection update. Emerg Infect Dis 1998;4:416-20. 3. Fridkin SK, Weibel SF, Weinstein RA. Magnitude and prevention of nosocomial infections in the intensive care unit. Infect Dis Clin North Am 1997; 11:479-96. 4. Fernandez-Crehuet R, Diaz-Molina C, de Irala J, Martinez-Concha D, Salcedo-Leal I, Masa-Calles J. Nosocomial infection in an intensive-care unit: identification of risk factors. Infect Control Hosp Epidemiol 1997;18:825-30. 5. Karchmer TB, Gianetta ET, Muto CA, Strain BA, Farr BM. A randomized crossover study of silver coated urinary catheters in hospitalized patients. Arch Intern Med 2000;160:3294-8. 6. Tcholakian RK, Raad II. Durability of anti-infective effect of long-term silicone sheath catheters impregnated with antimicrobial agents. Antimicrob Agents Chemother 2001;45:1990-3. 7. Saint S,Veenstra DL, Sullivan SD, Chenoweth C, Fendrick AM.The potential clinical and economic benefits of silver alloy urinary catheters in preventing urinary tract infection. Arch Intern Med 2000;160:2670-5. 8. Bueno-Cavanillas A, Rodriguez-Contreras R, Lopez-Luque A, DelgadoRodriguez M, Galves-Vargas R. Usefulness of severity indices in intensive care medicine as a predictor of nosocomial infection risk. Intensive Care Med 1991;17:336-9. 9. Chevret S, Hemmer M, Carlet J, Langer M. Incidence and risk factors of pneumonia acquired in intensive care units. Results from a multicenter prospective study on 996 patients. European Cooperative Group on Nosocomial Pneumonia. Intensive Care Med 1993;19:256-64. 10. Craven DE, Kunches LM, Lichtenberg DA, et al. Nosocomial infection and fatality in medical and surgical intensive care unit patients. Arch Intern Med 1988;148:1161-8. 11. Cunnion KM, Weber DJ, Broadhead WE, Hanson LC, Pieper CF, Rutala WA. Risk factors for nosocomial pneumonia: comparing adult critical care populations. Am J Respir Crit Care Med 1996;153:158-62. 12. Fagon JY, Chastre J,Vuagnat A,Trouillet JL, Novara A, Gibert C. Nosocomial pneumonia and mortality among patients in intensive care units. JAMA 1996;275:866-9. 13. Hurr H, Hawley HB, Czachor JS, Markert RJ, McCarthy MC. APACHE II and ISS scores as predictors of nosocomial infections in trauma patients. Am J Infect Control 1999;27:79-83. 14. Mylotte JM, Graham R, Kahler L, Young L, Goodnough S. Epidemiology of nosocomial infection and resistant organisms in patients admitted for the first time to an acute rehabilitation unit. Clin Infect Dis 200;30:425-32. 15. APACHE critical care—methodology and data collection manual. McLean (VA): APACHE Medical Systems, Inc; 1994. 16. NNIS manual. Atlanta: Centers for Disease Control and Prevention, US Department of Health and Human Services; 1994. 17. Barie PS, Hydo LJ, Fischer E. Comparison of APACHE II and III scoring systems for mortality prediction in critical surgical illness. Arch Surg 1995; 130:77-82.
Visit AJIC online! Your subscription to the print version of AJIC includes free access to the online version of AJIC. Activate the online portion of your subscription and get the most from AJIC, including “Online Exclusives”—content you’ll only find in the Web version! To activate your online subscription, visit www.mosby.com/ajic and follow the activation instructions. We invite you to visit today.