Eur J Cardio-thorac
Surg (1992) 6:524-529
Stirjgery 0 Springer-Verlag
1992
Early prediction of septic complications after cardiac surgery by APACHE II score E. Kreuzer ‘, S. Kiiiib2, G. Pilz2, and K. Werdan2 Departments
of ’ Cardiac Surgery and ’ Medicine I, Grosshadern
University
Hospital,
Munich, FRG
Abstract. In 110 patients admitted to the intensive care unit after cardiac surgery, daily monitoring [clinical parameters, cardiac index (CI), left ventricular stroke work index (LVSWI) and APACHE II score] was performed in regard to its usefulness in the early prediction of septic complications, a major cause of postoperative mortality. Septic complications (defined as Elebute sepsis score of 112 on 2 2 days) occurred in 16 patients and were associated with a significantly worse prognosis (mortality 69% vs 1%, P < 0.0001) than was seen in patients without sepsis. While preoperative APACHE II score values did not differentiate between the patients with an uneventful postoperative course and those with septic complications, patients who ultimately developed septic complications had significantly (P < 0.001) higher scores as early as on the evening of the operation day (“day 0”). In addition, in contrast to patients without sepsis, whose scores dropped markedly (P < 0.001) between day 0 and day 1, patients with septic complications invariably had high scores. Compared to single parameters (fever, leucocyte count, CI, LVSWI), the APACHE II score proved to be superior in differentiating between patients who developed sepsis and those who did not. A score of 19 or more on the 1st postoperative day had a sensitivity of 75%, a specificity of 98%, a Youden index of 0.73, a positive predictive value of 86%, and a negative predictive value of 96% in regard to prediction of septic complications. The data from our subsequent prospective study (new group of 106 patients) confirmed an APACHE II score of 19 or more as the discriminating criterion (mortality 36% vs 0%, P < 0.0001). The practical APACHE II score seems to be useful in the early prediction of postoperative septic complications in patients who have undergone cardiac surgery. [Eur J Cardio-thorac Surg (1992) 6:524-5291 Key words: APACHE II score - Elebute score - Cardiac surgery - Risk estimation - Sepsis
Since sepsis and septic shock still constitute a major cause of postoperative intensive care unit (ICU) mortality [22,23], early identification of patients at risk of developing septic complications is a matter of considerable importance. However, the complexity and amount of pathophysiological data confronting ICU physicians makes it difficult to gain a picture of the overall status of the critically ill patient. Thus, there is a need for methods which can reliably and promptly distinguish these conditions and may ultimately be used prospectively in decision making regarding therapy. In view of the difficulties encountered in the initial diagnosis and estimation of severity of sepsis solely based on clinical judgement [9, 251, various diagnostic criteria have been proposed. Frequently, a combination of routinely available single parameters is used as the criterion of sepsis [3, 6, 14, 21, 241; another approach was the Presented at the 5th Annual Meeting of the European Association for Cardio-thoracic Surgery, London, UK, September 23-25,1991
introduction of scoring systems, such as the Elebute score [ll]. This score, established in surgical patients, divides the response to sepsis in four classes: local effects, temperature response, systemic effects, and laboratory data (for details, see [ll]). In the particular case of cardiac surgery patients, the intraoperative need for extracorporeal circulation, leading to unspecific sepsis-like changes in the early postoperative course, may further hamper prompt recognition of a developing septic state [l, 4, 171. In the present study, we therefore examined the applicability of several routinely used clinical variables, hemodynamic data, as well as the APACHE II score [18], in the early evaluation of postoperative septic complications in cardiac surgery patients. Briefly, the total APACHE II score is the sum of three components: an acute physiology score, derived from 12 physiologic measurements; an age-related score; and a chronic health evaluation score (for details, see [18]). With respect to practicability, it is of particular interest that this score can easily be calculated within
525
5- 10 min at the bedside using a computer program implemented in a conventional pocket computer [24].
Table 1. Prognosis study, n = 110)
in patients after cardiac surgery (observational
Operation
No septic complications
Septic complications (Elebute score 2 12, 2 2 days)
n
Died
n
Aortocoronary bypass grafting Valve surgery” Miscellaneousi’
68
1
2
1
1
22 4
-
7 7
5 5
4 (+I)”
Total
94
1*
16
II*
7 (+2)
Patients and methods All patients undergoing cardiac surgery during a 4-month period at the Department of Cardiac Surgery, Grosshadern Hospital, University of Munich, FRG, who fulfilled the following criteria were included in the study: (1) elective operation, excluding heart transplantation and pacemaker implantation; (2) presence of invasive perioperative hemodynamic monitoring by thermodilution pulmonary artery (Swan-Ganz) catheter [20]; (3) postoperative course longer than 24 h; (4) informed consent obtained. One hundred and ten patients fulfilled these criteria and were enrolled. Scores and laboratory data were assessed daily, beginning with preoperative baseline values (day - 1) and the immediate postoperative data, recorded on the evening of the operation day (day 0). This monitoring was continued in all patients for at least 3 postoperative days at 10 a.m. each day. In patients who stayed for longer, it was carried on until discharge from the KU. Patients’ hemodynamics were postoperatively monitored from day 0 until day 2. The preoperative value (day - 1) was recorded after induction of anesthesia and before the beginning of surgery. Cardiac output was invasively measured by the thermodilution method via a Swan-Ganz catheter. The cardiac index (CI) and left ventricular stroke work index (LVSWI) were calculated using standard formulas [27]. The scoring systems were assessed as originally described [I I, 181, using the most disturbed values recorded during the period of monitoring for score calculation. An exception was made with the Glasgow Coma Score [29] values in patients who were sedated during the period of observation: as suggested by Jacobs et al. [15], for this subgroup the best rather than the worst value was taken. Septic complications were defined, according to Grundmann et al. [13], as an Elebute sepsis score of at least 12, lasting for at least 2 days. In agreement with other investigators [12], negative blood cultures were not considered, incompatible with sepsis or septic shock. To consider sepsis as the cause of death (“fatal sepsis” in Table I), all the following criteria were required in addition to an Elebute score of 12 or more during the last 24 h: (1) strong clinical evidence of sepsis; (2) presence of at least six out of nine previously described sepsis criteria (for definition, see [24]); and (3) absence of any other potential cause of death. Only deaths in the ICU were counted in the mortality figures. Results are expressed as mean k SEM unless otherwise stated. Differences between groups were statistically analysed using the x2 test with Yates correction for categorical variables; for continuous variables, Student’s t test for unpaired data or the Mann-Whitney test were used, where appropriate. A P value of less than 0.05 was considered statistically significant. For the variables tested, the sensitivity, specificity, and predictive values for correct prediction of the development of septic complications (for definition, see above) were calculated and the cutoff point giving the best Youden index [30]) determined.
Results
Clinical results Details of the type of cardiac operations performed and of mortality are given in Table 1. Septic complications according to the Elebute score criterion (see Methods section) occurred in 16 patients and were associated with a significantly worse prognosis (Table 1). Deaths in this group were mainly due to fatal sepsis as defined above (Table 1). Pre-, intra-, and postoperative patient characteristics are shown in Table 2. While there was no significant
Died
Fatal sepsis
2 (+I)
* PtO.OOO1 a Aortic valve replacement (n= 14) mitral valve replacement (n = 12), combined aortic and mitral valve replacement (n = 3) b Combined aortocoronary bypass and valve surgery (n = 3), aortocoronary bypass and left ventricular aneurysmectomy (n = 3) myectomy for hypertrophic obstructive cardiomyopathy (n = I), partial pericardectomy for pericarditis constrictiva (n = I), closure of perforated aneurysm of sinus of Valsalva into right ventricle (n = I), left ventricular aneurysmectomy (n = I), surgery for WPW-syndrome @=I) ’ Numbers in parentheses refer to patients who fulfilled the criteria for sepsis but also had another coexisting condition which could potentially cause death Table 2. Patient preoperative, intraoperative, data (observational study, n = 110)
and postoperative
No septic complications (n = 94)
Septic complications (n=l6)
P
Surg (1992) 6:524-529
Stirjgery 0 Springer-Verlag
1992
Early prediction of septic complications after cardiac surgery by APACHE II score E. Kreuzer ‘, S. Kiiiib2, G. Pilz2, and K. Werdan2 Departments
of ’ Cardiac Surgery and ’ Medicine I, Grosshadern
University
Hospital,
Munich, FRG
Abstract. In 110 patients admitted to the intensive care unit after cardiac surgery, daily monitoring [clinical parameters, cardiac index (CI), left ventricular stroke work index (LVSWI) and APACHE II score] was performed in regard to its usefulness in the early prediction of septic complications, a major cause of postoperative mortality. Septic complications (defined as Elebute sepsis score of 112 on 2 2 days) occurred in 16 patients and were associated with a significantly worse prognosis (mortality 69% vs 1%, P < 0.0001) than was seen in patients without sepsis. While preoperative APACHE II score values did not differentiate between the patients with an uneventful postoperative course and those with septic complications, patients who ultimately developed septic complications had significantly (P < 0.001) higher scores as early as on the evening of the operation day (“day 0”). In addition, in contrast to patients without sepsis, whose scores dropped markedly (P < 0.001) between day 0 and day 1, patients with septic complications invariably had high scores. Compared to single parameters (fever, leucocyte count, CI, LVSWI), the APACHE II score proved to be superior in differentiating between patients who developed sepsis and those who did not. A score of 19 or more on the 1st postoperative day had a sensitivity of 75%, a specificity of 98%, a Youden index of 0.73, a positive predictive value of 86%, and a negative predictive value of 96% in regard to prediction of septic complications. The data from our subsequent prospective study (new group of 106 patients) confirmed an APACHE II score of 19 or more as the discriminating criterion (mortality 36% vs 0%, P < 0.0001). The practical APACHE II score seems to be useful in the early prediction of postoperative septic complications in patients who have undergone cardiac surgery. [Eur J Cardio-thorac Surg (1992) 6:524-5291 Key words: APACHE II score - Elebute score - Cardiac surgery - Risk estimation - Sepsis
Since sepsis and septic shock still constitute a major cause of postoperative intensive care unit (ICU) mortality [22,23], early identification of patients at risk of developing septic complications is a matter of considerable importance. However, the complexity and amount of pathophysiological data confronting ICU physicians makes it difficult to gain a picture of the overall status of the critically ill patient. Thus, there is a need for methods which can reliably and promptly distinguish these conditions and may ultimately be used prospectively in decision making regarding therapy. In view of the difficulties encountered in the initial diagnosis and estimation of severity of sepsis solely based on clinical judgement [9, 251, various diagnostic criteria have been proposed. Frequently, a combination of routinely available single parameters is used as the criterion of sepsis [3, 6, 14, 21, 241; another approach was the Presented at the 5th Annual Meeting of the European Association for Cardio-thoracic Surgery, London, UK, September 23-25,1991
introduction of scoring systems, such as the Elebute score [ll]. This score, established in surgical patients, divides the response to sepsis in four classes: local effects, temperature response, systemic effects, and laboratory data (for details, see [ll]). In the particular case of cardiac surgery patients, the intraoperative need for extracorporeal circulation, leading to unspecific sepsis-like changes in the early postoperative course, may further hamper prompt recognition of a developing septic state [l, 4, 171. In the present study, we therefore examined the applicability of several routinely used clinical variables, hemodynamic data, as well as the APACHE II score [18], in the early evaluation of postoperative septic complications in cardiac surgery patients. Briefly, the total APACHE II score is the sum of three components: an acute physiology score, derived from 12 physiologic measurements; an age-related score; and a chronic health evaluation score (for details, see [18]). With respect to practicability, it is of particular interest that this score can easily be calculated within
525
5- 10 min at the bedside using a computer program implemented in a conventional pocket computer [24].
Table 1. Prognosis study, n = 110)
in patients after cardiac surgery (observational
Operation
No septic complications
Septic complications (Elebute score 2 12, 2 2 days)
n
Died
n
Aortocoronary bypass grafting Valve surgery” Miscellaneousi’
68
1
2
1
1
22 4
-
7 7
5 5
4 (+I)”
Total
94
1*
16
II*
7 (+2)
Patients and methods All patients undergoing cardiac surgery during a 4-month period at the Department of Cardiac Surgery, Grosshadern Hospital, University of Munich, FRG, who fulfilled the following criteria were included in the study: (1) elective operation, excluding heart transplantation and pacemaker implantation; (2) presence of invasive perioperative hemodynamic monitoring by thermodilution pulmonary artery (Swan-Ganz) catheter [20]; (3) postoperative course longer than 24 h; (4) informed consent obtained. One hundred and ten patients fulfilled these criteria and were enrolled. Scores and laboratory data were assessed daily, beginning with preoperative baseline values (day - 1) and the immediate postoperative data, recorded on the evening of the operation day (day 0). This monitoring was continued in all patients for at least 3 postoperative days at 10 a.m. each day. In patients who stayed for longer, it was carried on until discharge from the KU. Patients’ hemodynamics were postoperatively monitored from day 0 until day 2. The preoperative value (day - 1) was recorded after induction of anesthesia and before the beginning of surgery. Cardiac output was invasively measured by the thermodilution method via a Swan-Ganz catheter. The cardiac index (CI) and left ventricular stroke work index (LVSWI) were calculated using standard formulas [27]. The scoring systems were assessed as originally described [I I, 181, using the most disturbed values recorded during the period of monitoring for score calculation. An exception was made with the Glasgow Coma Score [29] values in patients who were sedated during the period of observation: as suggested by Jacobs et al. [15], for this subgroup the best rather than the worst value was taken. Septic complications were defined, according to Grundmann et al. [13], as an Elebute sepsis score of at least 12, lasting for at least 2 days. In agreement with other investigators [12], negative blood cultures were not considered, incompatible with sepsis or septic shock. To consider sepsis as the cause of death (“fatal sepsis” in Table I), all the following criteria were required in addition to an Elebute score of 12 or more during the last 24 h: (1) strong clinical evidence of sepsis; (2) presence of at least six out of nine previously described sepsis criteria (for definition, see [24]); and (3) absence of any other potential cause of death. Only deaths in the ICU were counted in the mortality figures. Results are expressed as mean k SEM unless otherwise stated. Differences between groups were statistically analysed using the x2 test with Yates correction for categorical variables; for continuous variables, Student’s t test for unpaired data or the Mann-Whitney test were used, where appropriate. A P value of less than 0.05 was considered statistically significant. For the variables tested, the sensitivity, specificity, and predictive values for correct prediction of the development of septic complications (for definition, see above) were calculated and the cutoff point giving the best Youden index [30]) determined.
Results
Clinical results Details of the type of cardiac operations performed and of mortality are given in Table 1. Septic complications according to the Elebute score criterion (see Methods section) occurred in 16 patients and were associated with a significantly worse prognosis (Table 1). Deaths in this group were mainly due to fatal sepsis as defined above (Table 1). Pre-, intra-, and postoperative patient characteristics are shown in Table 2. While there was no significant
Died
Fatal sepsis
2 (+I)
* PtO.OOO1 a Aortic valve replacement (n= 14) mitral valve replacement (n = 12), combined aortic and mitral valve replacement (n = 3) b Combined aortocoronary bypass and valve surgery (n = 3), aortocoronary bypass and left ventricular aneurysmectomy (n = 3) myectomy for hypertrophic obstructive cardiomyopathy (n = I), partial pericardectomy for pericarditis constrictiva (n = I), closure of perforated aneurysm of sinus of Valsalva into right ventricle (n = I), left ventricular aneurysmectomy (n = I), surgery for WPW-syndrome @=I) ’ Numbers in parentheses refer to patients who fulfilled the criteria for sepsis but also had another coexisting condition which could potentially cause death Table 2. Patient preoperative, intraoperative, data (observational study, n = 110)
and postoperative
No septic complications (n = 94)
Septic complications (n=l6)
P
