Journal of Hospital Infection 90 (2015) 15e21 Available online at www.sciencedirect.com
Journal of Hospital Infection journal homepage: www.elsevierhealth.com/journals/jhin
Differences in risk factors associated with surgical site infections following two types of cardiac surgery in Japanese patients K. Morikane a, *, H. Honda b, T. Yamagishi c, S. Suzuki d a
Division of Infection Control and Clinical Laboratory, Yamagata University Hospital, Yamagata, Japan Division of Infectious Diseases, Department of Medicine, Tokyo Metropolitan Tama Medical Centre, Tokyo, Japan c Infectious Disease Surveillance Centre, National Institute of Infectious Diseases, Tokyo, Japan d Department of Bacteriology II, National Institute of Infectious Diseases, Tokyo, Japan b
A R T I C L E
I N F O
Article history: Received 18 August 2014 Accepted 20 November 2014 Available online 7 January 2015 Keywords: Surgical site infection Risk factor Cardiac surgery
S U M M A R Y
Background: Differences in the risk factors for surgical site infection (SSI) following open heart surgery and coronary artery bypass graft surgery are not well described. Aim: To identify and compare risk factors for SSI following open heart surgery and coronary artery bypass graft surgery. Methods: SSI surveillance data on open heart surgery (CARD) and coronary artery bypass graft surgery (CBGB) submitted to the Japan Nosocomial Infection Surveillance (JANIS) system between 2008 and 2010 were analysed. Factors associated with SSI were analysed using univariate modelling analysis followed by multi-variate logistic regression analysis. Non-binary variables were analysed initially to determine the most appropriate category. Findings: The cumulative incidence rates of SSI for CARD and CBGB were 2.6% (151/5895) and 4.1% (160/3884), respectively. In both groups, the duration of the operation and a high American Society of Anesthesiologists’ (ASA) score were significant in predicting SSI risk in the model. Wound class was independently associated with SSI in CARD but not in CBGB. Implants, multiple procedures and emergency operations predicted SSI in CARD, but none of these factors predicted SSI in CBGB. Conclusions: There was a remarkable difference in the prediction of risk for SSI between the two types of cardiac surgery. Risk stratification in CARD could be improved by incorporating variables currently available in the existing surveillance systems. Risk index stratification in CBGB could be enhanced by collecting additional variables, because only two of the current variables were found to be significant for the prediction of SSI. ª 2015 The Healthcare Infection Society. Published by Elsevier Ltd. All rights reserved.
Introduction * Corresponding author. Address: Division of Infection Control and Clinical Laboratory, Yamagata University Hospital, 2-2-2 Iida-Nishi, Yamagata 990-9585, Japan. Tel.: þ81 23 628 5406; fax: þ81 23 628 5409. E-mail address: [email protected] (K. Morikane).
Surgical site infection (SSI) is among the most common healthcare-associated infections (HAIs), accounting for 17% of HAIs, and is estimated to occur at a rate of 240,000 cases per year in the USA.1 SSI also accounts for up to US$10 billion of additional costs per year.2,3
http://dx.doi.org/10.1016/j.jhin.2014.11.022 0195-6701/ª 2015 The Healthcare Infection Society. Published by Elsevier Ltd. All rights reserved.
16
K. Morikane et al. / Journal of Hospital Infection 90 (2015) 15e21
Surveillance is one of the key methods for preventing SSI.4 Surveillance data can be used as a quality indicator for SSI, and for implementing further strategies to prevent SSI. Caution should be exercised, however, when analysing SSI surveillance data because the simple incidence of SSI may be affected by ‘unmodifiable’ risk factors associated with patient characteristics. The Centers for Disease Control and Prevention (CDC) has been conducting the National Nosocomial Infections Surveillance (NNIS) system since 1970 in an effort to develop better risk stratification methods.5 The NNIS risk index, developed in 19916 and consisting of three variables [duration of operation, wound class and American Society of Anesthesiologists’ (ASA) score], is a good predictor of SSI in general, except in some operative procedures including coronary artery bypass graft (CGBG). The majority (95%) of operations have an index of 1 or 2 because CBGB is usually clean (Class I) and the ASA scores of the patients are normally 3. The NNIS risk index only dichotomizes patients based on the duration of the operation, which may be unmodifiable due to patient characteristics. A number of studies have indicated the inappropriateness of the NNIS risk index for stratifying patients who have undergone CBGB.7e9 Alternative risk factors specifically associated with CBGB have been studied extensively,10 and been proposed as candidates for better stratification, including diabetes,11e18 obesity and greater body mass index (BMI),11,13,14,16,17 female sex,12,13,19 admission-specific chronic disease score,20 and preoperative drug dispensing.21 In response to the results of these studies, the National Healthcare Safety Network (NHSN), successor to the NNIS system, developed a revised SSI risk model that includes some of the variables mentioned above. The revised model has been demonstrated to have better predictive power than the previous NNIS risk index.22,23 Open heart surgery (CARD) and CBGB both involve a sternotomy and are highly invasive surgical procedures, often complicated by mediastinitis, sternal osteomyelitis and postoperative pneumonia. However, these two procedures involve very different surgical techniques and manoeuvres, and may have different risk factors associated with SSI. In contrast to the extensive efforts made to identify risk factors associated with SSI in CBGB, relatively little has been done to identify the risk factors associated with SSI in CARD. In Japan, a national surveillance system for HAI e Japan Nosocomial Infection Surveillance (JANIS) e was established in 2000 by the Ministry of Health, Labour and Welfare of Japan. Surveillance for SSI using the NNIS surveillance definition with some modifications was commenced in 2002. JANIS currently collects SSI surveillance data from more than 400 institutions (as of March 2014), and is the largest SSI database in Japan. The aim of this study was to identify and compare the risk factors associated with SSI in patients who have undergone CARD and CABG in Japan, using the large Japanese national HAI database.
Methods Data on SSI in patients who underwent cardiac surgery between 2008 and 2010 were extracted from JANIS. This included CARD and CBGB. Instances of CBGB using only the internal mammary artery or gastroepiploic artery graft were not included in the CBGB code. In total, 5895 cases of CARD and
3884 cases of CBGB were included in the study. Approval for data extraction was granted by the Ministry of Health, Labour and Welfare of Japan. The Institutional Review Board at Yamagata University School of Medicine gave their approval for this project. The methodology of JANIS has been described elsewhere in detail.24 In summary, JANIS recruits hospitals on a voluntary basis each year, and participating hospitals are required to report SSI surveillance data for selected operative procedures electronically on a biannual basis. Variables collected for CARD and CBGB include those in the NNIS risk index (wound class, ASA score and duration of operation), age, sex, multiple procedures, emergency operation, implant, trauma surgery, general anaesthesia and same-day surgery. Due to the small number of cases available, the following variables were excluded from statistical analyses: same-day surgery [0.1% (3/5895) in CARD and 0.1% (4/3884) in CBGB] and trauma surgery [0.2% (13/5895) and 0.1% (4/3884), respectively]. General anaesthesia was also excluded from the analysis because it was performed in the vast majority of cardiac procedures [99.8% (5881/5895) and 99.9% (3879/3884), respectively]. Prior to performing statistical analysis for the factors associated with SSI in cardiac surgery, the association between the incidence of SSI and non-dichotomous variables was examined and modified to fit the statistical model. Continuous variables (duration of operation and age) were divided into four quartiles, and the incidence rates of SSI were compared in each quartile. If there was a definite cut-off point at which it was possible to rationalize dichotomization of the variable, it was dichotomized at the cut-off point. If there was no definite cutoff point, the value was divided by 10 and regarded as a continuous variable. Categorical variables (ASA and wound class) were examined in the same way as the continuous variables that were categorized into quartiles. Potential risk factors associated with SSI in each type of procedure were first assessed using univariate modelling analysis. Comparison between categorical variables was performed using Chi-squared test. Variables with P < 0.25 in the univariate modelling analysis were considered as potential candidate variables, and entered into the multi-variate logistic regression model which was developed using forward stepwise logistic regression. Variables were retained in the final model if two-tailed P was
Journal of Hospital Infection journal homepage: www.elsevierhealth.com/journals/jhin
Differences in risk factors associated with surgical site infections following two types of cardiac surgery in Japanese patients K. Morikane a, *, H. Honda b, T. Yamagishi c, S. Suzuki d a
Division of Infection Control and Clinical Laboratory, Yamagata University Hospital, Yamagata, Japan Division of Infectious Diseases, Department of Medicine, Tokyo Metropolitan Tama Medical Centre, Tokyo, Japan c Infectious Disease Surveillance Centre, National Institute of Infectious Diseases, Tokyo, Japan d Department of Bacteriology II, National Institute of Infectious Diseases, Tokyo, Japan b
A R T I C L E
I N F O
Article history: Received 18 August 2014 Accepted 20 November 2014 Available online 7 January 2015 Keywords: Surgical site infection Risk factor Cardiac surgery
S U M M A R Y
Background: Differences in the risk factors for surgical site infection (SSI) following open heart surgery and coronary artery bypass graft surgery are not well described. Aim: To identify and compare risk factors for SSI following open heart surgery and coronary artery bypass graft surgery. Methods: SSI surveillance data on open heart surgery (CARD) and coronary artery bypass graft surgery (CBGB) submitted to the Japan Nosocomial Infection Surveillance (JANIS) system between 2008 and 2010 were analysed. Factors associated with SSI were analysed using univariate modelling analysis followed by multi-variate logistic regression analysis. Non-binary variables were analysed initially to determine the most appropriate category. Findings: The cumulative incidence rates of SSI for CARD and CBGB were 2.6% (151/5895) and 4.1% (160/3884), respectively. In both groups, the duration of the operation and a high American Society of Anesthesiologists’ (ASA) score were significant in predicting SSI risk in the model. Wound class was independently associated with SSI in CARD but not in CBGB. Implants, multiple procedures and emergency operations predicted SSI in CARD, but none of these factors predicted SSI in CBGB. Conclusions: There was a remarkable difference in the prediction of risk for SSI between the two types of cardiac surgery. Risk stratification in CARD could be improved by incorporating variables currently available in the existing surveillance systems. Risk index stratification in CBGB could be enhanced by collecting additional variables, because only two of the current variables were found to be significant for the prediction of SSI. ª 2015 The Healthcare Infection Society. Published by Elsevier Ltd. All rights reserved.
Introduction * Corresponding author. Address: Division of Infection Control and Clinical Laboratory, Yamagata University Hospital, 2-2-2 Iida-Nishi, Yamagata 990-9585, Japan. Tel.: þ81 23 628 5406; fax: þ81 23 628 5409. E-mail address: [email protected] (K. Morikane).
Surgical site infection (SSI) is among the most common healthcare-associated infections (HAIs), accounting for 17% of HAIs, and is estimated to occur at a rate of 240,000 cases per year in the USA.1 SSI also accounts for up to US$10 billion of additional costs per year.2,3
http://dx.doi.org/10.1016/j.jhin.2014.11.022 0195-6701/ª 2015 The Healthcare Infection Society. Published by Elsevier Ltd. All rights reserved.
16
K. Morikane et al. / Journal of Hospital Infection 90 (2015) 15e21
Surveillance is one of the key methods for preventing SSI.4 Surveillance data can be used as a quality indicator for SSI, and for implementing further strategies to prevent SSI. Caution should be exercised, however, when analysing SSI surveillance data because the simple incidence of SSI may be affected by ‘unmodifiable’ risk factors associated with patient characteristics. The Centers for Disease Control and Prevention (CDC) has been conducting the National Nosocomial Infections Surveillance (NNIS) system since 1970 in an effort to develop better risk stratification methods.5 The NNIS risk index, developed in 19916 and consisting of three variables [duration of operation, wound class and American Society of Anesthesiologists’ (ASA) score], is a good predictor of SSI in general, except in some operative procedures including coronary artery bypass graft (CGBG). The majority (95%) of operations have an index of 1 or 2 because CBGB is usually clean (Class I) and the ASA scores of the patients are normally 3. The NNIS risk index only dichotomizes patients based on the duration of the operation, which may be unmodifiable due to patient characteristics. A number of studies have indicated the inappropriateness of the NNIS risk index for stratifying patients who have undergone CBGB.7e9 Alternative risk factors specifically associated with CBGB have been studied extensively,10 and been proposed as candidates for better stratification, including diabetes,11e18 obesity and greater body mass index (BMI),11,13,14,16,17 female sex,12,13,19 admission-specific chronic disease score,20 and preoperative drug dispensing.21 In response to the results of these studies, the National Healthcare Safety Network (NHSN), successor to the NNIS system, developed a revised SSI risk model that includes some of the variables mentioned above. The revised model has been demonstrated to have better predictive power than the previous NNIS risk index.22,23 Open heart surgery (CARD) and CBGB both involve a sternotomy and are highly invasive surgical procedures, often complicated by mediastinitis, sternal osteomyelitis and postoperative pneumonia. However, these two procedures involve very different surgical techniques and manoeuvres, and may have different risk factors associated with SSI. In contrast to the extensive efforts made to identify risk factors associated with SSI in CBGB, relatively little has been done to identify the risk factors associated with SSI in CARD. In Japan, a national surveillance system for HAI e Japan Nosocomial Infection Surveillance (JANIS) e was established in 2000 by the Ministry of Health, Labour and Welfare of Japan. Surveillance for SSI using the NNIS surveillance definition with some modifications was commenced in 2002. JANIS currently collects SSI surveillance data from more than 400 institutions (as of March 2014), and is the largest SSI database in Japan. The aim of this study was to identify and compare the risk factors associated with SSI in patients who have undergone CARD and CABG in Japan, using the large Japanese national HAI database.
Methods Data on SSI in patients who underwent cardiac surgery between 2008 and 2010 were extracted from JANIS. This included CARD and CBGB. Instances of CBGB using only the internal mammary artery or gastroepiploic artery graft were not included in the CBGB code. In total, 5895 cases of CARD and
3884 cases of CBGB were included in the study. Approval for data extraction was granted by the Ministry of Health, Labour and Welfare of Japan. The Institutional Review Board at Yamagata University School of Medicine gave their approval for this project. The methodology of JANIS has been described elsewhere in detail.24 In summary, JANIS recruits hospitals on a voluntary basis each year, and participating hospitals are required to report SSI surveillance data for selected operative procedures electronically on a biannual basis. Variables collected for CARD and CBGB include those in the NNIS risk index (wound class, ASA score and duration of operation), age, sex, multiple procedures, emergency operation, implant, trauma surgery, general anaesthesia and same-day surgery. Due to the small number of cases available, the following variables were excluded from statistical analyses: same-day surgery [0.1% (3/5895) in CARD and 0.1% (4/3884) in CBGB] and trauma surgery [0.2% (13/5895) and 0.1% (4/3884), respectively]. General anaesthesia was also excluded from the analysis because it was performed in the vast majority of cardiac procedures [99.8% (5881/5895) and 99.9% (3879/3884), respectively]. Prior to performing statistical analysis for the factors associated with SSI in cardiac surgery, the association between the incidence of SSI and non-dichotomous variables was examined and modified to fit the statistical model. Continuous variables (duration of operation and age) were divided into four quartiles, and the incidence rates of SSI were compared in each quartile. If there was a definite cut-off point at which it was possible to rationalize dichotomization of the variable, it was dichotomized at the cut-off point. If there was no definite cutoff point, the value was divided by 10 and regarded as a continuous variable. Categorical variables (ASA and wound class) were examined in the same way as the continuous variables that were categorized into quartiles. Potential risk factors associated with SSI in each type of procedure were first assessed using univariate modelling analysis. Comparison between categorical variables was performed using Chi-squared test. Variables with P < 0.25 in the univariate modelling analysis were considered as potential candidate variables, and entered into the multi-variate logistic regression model which was developed using forward stepwise logistic regression. Variables were retained in the final model if two-tailed P was
