Incidence and mortality by group.
Moore LJ, Moore FA, Todd SR, Jones SL, Turner KL, Bass BL. Sepsis in General SurgeryThe 2005-2007 National Surgical Quality Improvement Program Perspective. Arch Surg. 2010;145(7):695–700. doi:10.1001/archsurg.2010.107
To document the incidence, mortality rate, and risk factors for sepsis and septic shock compared with pulmonary embolism and myocardial infarction in the general-surgery population.
American College of Surgeons National Surgical Quality Improvement Program institutions.
General-surgery patients in the 2005-2007 National Surgical Quality Improvement Program data set.
Main Outcome Measures
Incidence, mortality rate, and risk factors for sepsis and septic shock.
Of 363 897 general-surgery patients, sepsis occurred in 8350 (2.3%), septic shock in 5977 (1.6%), pulmonary embolism in 1078 (0.3%), and myocardial infarction in 615 (0.2%). Thirty-day mortality rates for each of the groups were as follows: 5.4% for sepsis, 33.7% for septic shock, 9.1% for pulmonary embolism, and 32.0% for myocardial infarction. The septic-shock group had a greater percentage of patients older than 60 years (no sepsis, 40.2%; sepsis, 51.7%; and septic shock, 70.3%; P < .001). The need for emergency surgery resulted in more cases of sepsis (4.5%) and septic shock (4.9%) than did elective surgery (sepsis, 2.0%; septic shock, 1.2%) (P < .001). The presence of any comorbidity increased the risk of sepsis and septic shock 6-fold (odds ratio, 5.8; 95% confidence interval, 5.5-6.2) and increased the 30-day mortality rate 22-fold (odds ratio, 21.8; 95% confidence interval, 17.6-26.9).
The incidences of sepsis and septic shock exceed those of pulmonary embolism and myocardial infarction. The risk factors for mortality include age older than 60 years, the need for emergency surgery, and the presence of any comorbidity. This study emphasizes the need for early recognition of patients at risk via aggressive screening and the rapid implementation of evidence-based guidelines.
Prevention of perioperative complications is a major focus in the care of the general-surgery patient. In recent years, much attention has focused on the prevention of venous thromboembolism (postoperative deep vein thrombosis and pulmonary embolism [PE]), postoperative myocardial infarction (MI), and surgical site infections (SSIs). Through education and increased awareness, there has been a significant reduction in the incidence of postoperative venous thromboembolism.1 Likewise, preoperative cardiovascular evaluation and risk assessment of elective general-surgery patients have become standards of care. The issue of SSIs has been addressed through national guidelines.2 Minimizing the occurrence of these potentially preventable complications improves patient outcomes and reduces health care costs.
Within our institution, we have identified surgical sepsis to be a potentially preventable cause of morbidity and mortality in our general-surgery patients. Severe sepsis and septic shock are the leading causes of multiple organ failure and mortality in noncoronary intensive care units (ICUs).3 It is estimated that in the United States there are 751 000 cases per year of sepsis, with an annual cost of $17 billion.4By 2010, it is estimated that there will be 934 000 cases per year.3 Unfortunately, despite tremendous basic and clinical research efforts, mortality from septic shock remains unchanged at greater than 50%.5 Early intervention and implementation of evidence-based guidelines have been demonstrated to improve outcomes in patients with sepsis.6,7 However, effective intervention is contingent on the early identification of sepsis.
In an attempt to improve the early identification of sepsis, we have developed and instituted a sepsis screening tool for use in our surgical ICU (SICU).8 The use of this tool and implementation of early evidence-based care through computerized clinical decision support resulted in a substantial decrease in the rate of mortality from severe sepsis and septic shock.8 On the basis of this experience, we believe that sepsis screening could potentially prevent sepsis-associated morbidity and mortality in the general-surgery population. Before advocating mandatory sepsis screening programs in general-surgery patients, we need to further characterize and understand sepsis in these patients. Furthermore, we need to document the relative incidence and associated mortality of sepsis compared with the more commonly addressed preventable causes of postoperative mortality. The objective of this study was to document the incidence, mortality, and risk factors for sepsis and septic shock compared with PE and MI in the general-surgery population.
This study is an analysis of prospectively collected data from the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) data set. The 2005-2007 NSQIP data set contains prospectively gathered clinical data and outcomes on 363 897 patients collected from 121 academic and community-based hospitals. The NSQIP compiles data on 239 variables, including preoperative, intraoperative, and 30-day postoperative variables, for patients undergoing surgical procedures in inpatient and outpatient settings (a sample of general-surgery cases using an 8-day cycle). All data are collected by an institutional surgical clinical nurse reviewer, who receives extensive training in the NSQIP methods and data collection. All variables that are gathered in the NSQIP are predefined in the NSQIP data dictionary.
The 2005-2007 NSQIP Participant Use File was queried for demographics, comorbidities, elective vs emergency case, and 30-day mortality in general-surgery patients. Patients having severe sepsis and septic shock are classified in the NSQIP data set as the septic-shock stratum, whereas patients having only sepsis are classified as the sepsis stratum. Patients without either sepsis or septic shock are classified as the no-sepsis stratum. This strategy was adopted by NSQIP to avoid counting patients who had severe sepsis and septic shock twice. The NSQIP data dictionary defines sepsis as the systemic inflammatory response syndrome with a documented infection and defines septic shock as sepsis and documented organ and/or circulatory dysfunction. The NSQIP definition of septic shock captures those patients who are defined as having severe sepsis by the American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference Guidelines.9 Detailed descriptions of the NSQIP definitions are presented in Table 1. From the NSQIP data set, we were unable to determine the source of the sepsis.
A PE is defined by the NSQIP as “lodging of a blood clot in a pulmonary artery with subsequent obstruction of blood supply to the lung parenchyma.”10p31 The NSQIP data dictionary defines a postoperative MI as “a new transmural acute myocardial infarction occurring during surgery or within 30 days as manifested by new Q-waves on ECG [electrocardiogram].”10p28 It also states, “The blood clots usually originate from the deep leg veins or the pelvic venous system within 30 days of the operation. PE documented if the patient has a ventilation-perfusion scan interpreted as high probability of PE or a positive computed tomography spiral exam, pulmonary arteriogram, or computed tomography angiogram. Treatment usually consists of initiation of anticoagulation therapy or placement of mechanical interruption (eg, Greenfield filter), for patients whom anticoagulation is contraindicated or already instituted.”10p31
Patient body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. Patient records that were missing the weight or height were classified as unknown BMI. The patients were then stratified by their BMI following the 5 major categories of underweight, normal, overweight, obese, and severely obese (obese class III) as defined by the World Health Organization BMI classification guidelines.11 The classification breakdown for BMI is as follows: less than 18.50 is classified as underweight, between 18.50 and 24.99 as normal, between 25.00 and 29.99 as overweight, between 30.00 and 39.99 as obese, and 40.00 or higher as severely obese.
Patients were further categorized by ethnicity and age. Patients were classified based on their NSQIP-defined ethnicity. After initial data explorations, it was clear that patients older than 60 years formed a significant stratum of their own; thus, we stratified patients according to their age by whether or not they were older than 60 years at the time of the operation. Similarly, patients were categorized by whether the procedure was an emergency or elective. From the NSQIP data set, we were unable to determine whether the sepsis occurred before or after the emergency surgical procedures.
In comparing the study groups, a χ2 analysis was used for categorical data. Generalized linear models were used to calculate the adjusted relative risk between the presence of any of the NSQIP-documented comorbidities and risk of developing sepsis, septic shock, or 30-day mortality after adjustment for age, sex, ethnicity, BMI, and emergency case.12P < .05 was considered statistically significant. Stata statistical software, version 10 (StataCorp LP, College Station, Texas), was used for all statistical analyses. The review of data was approved by The Methodist Hospital Research Institute.
The 2005-2007 NSQIP data set contains information on 363 897 general-surgery patients. Of these, 349 570 (96.1%) had no sepsis, 8350 (2.3%) had sepsis, and 5977 (1.6%) had septic shock. Pulmonary embolism occurred in 1078 patients (0.3%) and MI in 615 (0.2%). The demographic breakdown for the entire population is listed in Table 2. The septic-shock group had a greater percentage of patients older than 60 years (no sepsis, 40.2%; sepsis, 51.7%; and septic shock, 70.3%; P < .001). The incidence of sepsis and septic shock in elective cases was 2.0% and 1.2%, respectively, compared with 4.5% and 4.9%, respectively, for sepsis and septic shock in emergency cases (P < .001). The sepsis and septic-shock groups had a higher incidence of patients with 1 or more NSQIP comorbidities (no sepsis, 69.0%; sepsis, 90.2%; and septic shock, 96.4%; P < .001). The presence of any of the NSQIP-documented comorbidities increased the odds of developing sepsis and septic shock by 6-fold (odds ratio [OR], 5.8; 95% confidence interval [CI], 5.5-6.2). In addition, the presence of any comorbidity when compared with patients without comorbidities increased the risk of 30-day mortality 22-fold (OR, 21.8; 95% CI, 17.6-26.9). Thirty-day mortality rates for each of the groups were as follows: no sepsis, 1.1%; sepsis, 5.4%; and septic shock, 33.7% (vs 9.1% for PE and 32.0% for MI). The incidence and mortality rates for the groups are depicted in the Figure. The 30-day mortality of sepsis and septic shock within the elective case groups (no sepsis, 0.6%; sepsis, 4.0%; and septic shock, 30.0%; P < .001) and the emergency-case groups (no sepsis, 4.9%; sepsis, 9.4%; and septic shock, 39.3%; P < .001) were different, but patients with septic shock in elective and emergency cases had a similar high mortality rate. The development of sepsis increased the risk of 30-day mortality 4-fold (OR, 3.9; 95% CI, 3.5-4.3). The development of septic shock increased the risk of 30-day mortality 33-fold (OR, 32.9; 95% CI, 30.9-35.1). The operative procedures most commonly associated with sepsis and septic shock are listed in Table 3.
Minimizing mortality by preventing postoperative complications is a key component of surgical care. In recent years there has been an increasing focus on minimizing the risk of perioperative complications, including venous thromboembolism, perioperative cardiac events, and SSIs. Multiple professional organizations have published guidelines addressing these issues.2,13,14 There is no question that the implementation of these guidelines has reduced the occurrence of these perioperative adverse events and subsequent mortality. However, the findings of this study demonstrate that sepsis continues to be a common and serious complication in general-surgery patients and occurs much more frequently than PE and MI. Of note, septic shock occurs 10 times more frequently than MI and has the same mortality rate; thus, it kills 10 times more people. These findings are consistent with other studies15- 17 demonstrating that sepsis continues to be a major cause of morbidity and mortality in surgical patients.
Surgical site infections are defined as infections occurring up to 30 days after surgery and affecting either the incision or deep tissue at the operative site.18 Surgical site infections are the second most common nosocomial infection in hospitalized patients, occurring in at least 2% of those undergoing surgical procedures.19 The Centers for Disease Control and Prevention have established guidelines to address the issue of SSI.2 These guidelines focus on a bundle approach, ie, optimization of patient factors, antisepsis in the operating room, administration of prophylactic antibiotics, and postoperative incision care.2 Although these initiatives can decrease the occurrence of SSI, they fail to address the systemic role of sepsis, of which SSIs are only a low-risk subset of potential causes. In particular, these guidelines fall short of addressing the issue of postoperative surveillance for the development of sepsis.
Within our institution, we have identified sepsis to be a major cause of morbidity and mortality in our general-surgery patients. As a result, our multidisciplinary sepsis research team has developed a sepsis-management protocol that uses computerized clinical decision support to ensure timely and consistent implementation of the evidence-based guidelines for the management of sepsis.20 Through our initial experience with implementation of our computerized clinical decision support sepsis-management protocol, we encountered the unanticipated problem of untimely and inaccurate recognition of sepsis by bedside physicians. The need for routine, accurate screening of all SICU patients for sepsis quickly became apparent. In an attempt to increase the early identification of sepsis, we developed a sepsis screening tool in our SICU.8 This tool is based on graded derangements in 4 variables that define the systemic inflammatory response syndrome: heart rate, white blood cell count, temperature, and respiratory rate. The ranges for these 4 variables are based on the Acute Physiology and Chronic Health Evaluation II scoring system.21 Our initial experience with this tool in our SICU showed promising results. The tool yielded a sensitivity of 96.5%, a specificity of 96.7%, a positive predictive value of 80.2%, and a negative predictive value of 99.5%. In addition, sepsis-related mortality decreased from 35.1% to 23.3%.8 On the basis of this experience, we have expanded use of the tool to our surgical ward and are currently evaluating its applicability in the non-ICU setting.
Although implementing mandatory sepsis screening on the inpatient surgical floor is likely to improve the early recognition of sepsis and allow for implementation of appropriate evidence-based care, implementing a systemwide screening protocol would require a significant amount of health care resources. By identifying risk factors for the development of sepsis and septic shock in general-surgery patients, we can better allocate the available resources and focus screening on those patients most likely to develop sepsis and/or septic shock. Previously identified risk factors for early death from sepsis delineated by a French ICU group include low blood pH, shock, preexisting liver or cardiac insufficiency, and hypothermia.22 In addition, a recent epidemiologic study23 of patients with septic shock from ventilator-associated pneumonia identified advanced age, lymphocytopenia, high blood glucose levels, and increased clinical pulmonary infection scores as independent predictors of the development of septic shock. The findings of advanced age and preexisting liver or cardiac insufficiency from these studies are consistent with our findings.
Our analysis of the NSQIP data set identified 3 major risk factors for the development of sepsis and septic shock and mortality from sepsis and septic shock: age older than 60 years, the need for emergency surgery, and the presence of any comorbidity. The association between age, the presence of comorbid conditions, and the need for emergency surgery and development of sepsis or septic shock has been demonstrated in larger epidemiologic studies.4,5,24 However, these studies were retrospective analyses of International Classification of Diseases, Ninth Revision (ICD-9)25 discharge data from state databases. These findings suggest that those patients with any of these 3 factors warrant a high index of suspicion for the development of sepsis and that this patient population would most likely benefit from mandatory sepsis screening.
A healthy respect for these risk factors, in addition to early sepsis identification by screening, has influenced our surgical management of this patient population. A distinct window of early intervention exists in which the septic source must be eliminated and physiologic derangements corrected. For the postoperative patient whose sepsis has a nonsurgical source, such as pneumonia or a urinary tract infection, this goal can be accomplished in an ICU setting in a straightforward manner. In the setting of abdominal surgical sepsis, however, this issue becomes more complicated. This patient often requires emergency exploration for source control but loses valuable resuscitation time on the operating room table with ongoing heat, intravenous volume, and blood loss. In this context, our practice has focused more on damage-control surgery in the setting of septic shock. We have used this concept for patients with de novo intra-abdominal sepsis and patients with postoperative abdominal complications requiring surgery. Of note, this surgery is performed for only those patients with significant physiologic derangements and predefined risk factors. We continue to further delineate these risk factors and physiologic cutoffs for truncated laparotomy with the goal of decreasing the morbidity and mortality of surgical sepsis even further. By incorporating damage-control surgery into our established sepsis resuscitation protocol, we are maximizing the paradigm of early goal-directed sepsis management for our surgical patients.
We identified some limitations involved with working with the NSQIP data set. The first limitation is the NSQIP definitions for sepsis and septic shock. The American College of Chest Physicians/Society of Critical Care Medicine consensus conference definitions of sepsis, severe sepsis, and septic shock9 were developed in an attempt to standardize patient classification. However, the NSQIP definitions deviate from these consensus conference standards. Within the NSQIP data dictionary there is no definition for severe sepsis. Instead, patients with severe sepsis are classified into the septic-shock category in the NSQIP. This misclassification of patients makes it difficult to compare NSQIP patients with those from other data sets. However, this phenomenon is consistent with our experience: it is sometimes difficult to differentiate severe sepsis and septic shock. The second limitation is related to lack of data regarding cause of death. Although the NSQIP reports mortality, there is not a defined category for cause of death. This makes it impossible to determine whether patients in the data set died as a result of sepsis or septic shock or from another cause. That is true of other large data sets: it is difficult to know the ultimate cause of mortality in patients with multiple organ failure.
This study demonstrates that the incidence of sepsis and septic shock far exceeds that of MI and PE in general-surgery patients. In addition, case mortality rates in patients with sepsis and septic shock exceed those of MI and PE combined by nearly 10-fold. Therefore, our level of vigilance in identifying sepsis and septic shock needs to mimic, if not surpass, our vigilance for identifying MI and PE. By identifying 3 major risk factors for the development of and death from sepsis and septic shock in general-surgery patients, we can heighten our awareness for sepsis and septic shock in these at-risk populations. The implementation of mandatory sepsis screening for these high-risk populations has resulted in decreased sepsis-related mortality within our institution. Further evaluation of the role of sepsis screening programs in other settings is critical and could significantly reduce sepsis-related mortality in general-surgery patients.
Correspondence: Laura J. Moore, MD, Department of Surgery, The Methodist Hospital, Weill Cornell Medical College, 6550 Fannin St, Smith Tower 1661, Houston, TX 77030 (firstname.lastname@example.org).
Accepted for Publication: March 11, 2010.
Author Contributions: Drs L. J. Moore, F. A. Moore, Todd, and Jones had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: L. J. Moore, F. A. Moore, Jones, and Turner. Acquisition of data: Jones, Turner, and Bass. Analysis and interpretation of data: L. J. Moore, F. A. Moore, Todd, Jones, and Turner. Drafting of the manuscript: L. J. Moore, F. A. Moore, and Jones. Critical revision of the manuscript for important intellectual content: L. J. Moore, F. A. Moore, Todd, Jones, Turner, and Bass. Statistical analysis: Jones. Administrative, technical, and material support: Todd, Jones, Turner, and Bass. Study supervision: L. J. Moore, F. A. Moore, and Turner.
Financial Disclosure: None reported.
Funding/Support: This study was supported by The Methodist Hospital Research Institute, Houston, Texas.
Disclaimer: The American College of Surgeons National Surgical Quality Improvement Program and the hospitals participating in the National Surgical Quality Improvement Program are the source of the data used herein; they have not verified and are not responsible for the statistical validity of the data analysis or the conclusions derived by the authors.
Previous Presentation: This paper was presented at the 117th Scientific Session of the Western Surgical Association; November 11, 2009; San Antonio, Texas; and is published after peer review and revision. The discussions that follow this article are based on the originally submitted manuscript and not the revised manuscript.
Gregory J. Jurkovich, MD, Seattle, Washington: Drs Moore and colleagues have shown us compelling data from the American College of Surgeon's NSQIP demonstrating that sepsis and septic shock are a significant problem in the surgical patient. This likely comes as no surprise to any surgeon in this audience. What may be more surprising is the severity of the problem. Overall, nearly 4% of the 363 000 patients analyzed over this 3-year time frame were diagnosed with sepsis or septic shock. This is 20-fold higher than the 0.2% incidence of pulmonary embolus and 15-fold higher than the 0.3% incidence of MI. While these data do not necessarily support the contention made in the title that sepsis and septic shock are preventable, this frequency alone supports the argument for better screening, intervention, and treatment tools. This is particularly so, given that the mortality from sepsis is 5% and for septic shock a remarkable 33%.
The authors have suggested they have the answer: a better screening tool for the early diagnosis and more rapid effective interventions in the treatment of sepsis. But what is this screening tool? Dr Todd, I fear you have left the audience hanging, like a chad, or like an anxious schoolgirl waiting for the text message, or well — the correct analogy clearly escapes me, but what I am trying to ask is, “Where's the beef?” Is a magical screening tool for the early recognition of impending sepsis really available? Is it some propriety formula you plan on marketing? Or is it an embarrassment that you have had to resort to something so simplistic in your ICU that you hesitate to discuss it publicly? Like paying attention to an elevated white blood cell count and fever? Less someone misconstrue my comments, I am simply teasing the authors, as they have teased us, and asking them to provide us more information.
I have the following 4 queries for the authors:
When did sepsis or septic shock occur in these patients? Was it always a postoperative problem, or could it have been a presenting symptom or diagnosis? This has significant implications for its recognition and interventions.
What was the cause of sepsis? At the very least, can you provide us the origin by body regions, such as lungs, gastrointestinal, renal, or soft tissue? Again, this carries significant implications for diagnosis and treatment.
You documented 3 risk factors for higher mortality: age older than 60, any comorbidity, and the need for emergency surgery. None of these seem to be of the nature or sort that we as clinicians can affect. And perhaps the emergency surgery was in fact done to treat sepsis or a closed-space infection. So what then should we do with this information of risk factors?
Since NSQIP is meant to be a quality improvement/quality assurance analysis, can you provide any information on the quality of care for those patients who developed sepsis or septic shock? Were they effectively treated? Did they receive the correct and timely antibiotics? Was there timely surgical management of closed-space infections?
I appreciate the effort gone into the analysis of the NSQIP database and the recognition of the limitations that can and do occur with any and all administrative databases. This is another example of the power of large numbers, as it gives us good insight into the magnitude of the problem of sepsis in the surgical patient.
Dr Todd: As for your first question, is there a magic bullet? Unfortunately, no, but we have developed a sepsis screening tool based on the systemic inflammatory response syndrome criteria and the Acute Physiology and Chronic Health Evaluation II cut points, which is quite effective in our SICU. Our SICU patients are screened every 12 hours using this tool. If they attain a score greater than or equal to 4, a secondary screen is performed by either a midlevel provider or a surgical resident. The purpose of this evaluation is to validate the presence of an infectious source or not. In evaluating this tool statistically, it has a sensitivity and specificity of more than 90%. We are currently evaluating this tool on our surgical wards and assessing its validity.
Your second question addressed the timing of the sepsis (preoperative or postoperative). In NSQIP, there is no preoperative variable for sepsis, so all of these counts of sepsis are postoperative.
What are the sources of sepsis? Unfortunately, that is not available in NSQIP.
Regarding the risk factors we identified as significant, you are correct, we cannot adjust these (their age, their need for an emergency procedure, or their comorbidities). Our objective here was to identify a high-risk group for screening in our hospital. We have a 1000-bed hospital and realistically cannot afford to screen every patient every 12 hours. These variables should provide us a narrower scope for focusing our limited resources.
Unfortunately, you are correct; the answer is no. One of the biggest struggles with the Surviving Sepsis Campaign Guidelines is that clinicians have a difficult time actually intervening and providing the needed therapies in a timely fashion.