Outcomes of Surgical Sepsis

Abstract

Background:

Discussion of outcomes of surgical sepsis is no longer straightforward. Definitions of sepsis have changed recently and updated data are scant. Surgical patient populations are often heterogeneous; the patient population being considered must be described with precision. Traditional 30-d operative mortality may not be the most relevant outcome to consider. What should change or be the emphasis going forward?

Methods:

Review and synthesis of pertinent English-language literature.

Results:

Epidemiologic data are abundant for short-term outcomes of sepsis in general, but despite the fact that approximately 30% of patients with sepsis are surgical patients, sepsis outcome data for surgical patients are scant, especially for durations longer than 30 d, and essentially non-existent for patients defined under the new Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3) criteria. Interpretability of extant data is hampered by non-standard and changing definitions.

Conclusions:

Sepsis and organ dysfunction may be decreasing in prevalence and magnitude among surgical patients, but terminology must be standardized to enhance the interpretability of data generated in the future. It behooves journal editors, reviewers, and authors to insist upon standardized definitions and rigorous study design and data interpretation. Longer term data (e.g., 90-d mortality as opposed to in-hospital or traditional 30-d mortality) will be needed to justify to payers the complex, expensive care that these patients require. There is an urgent need to redefine the research agenda for surgical infections.

Adiscussion of outcomes of surgical sepsis might seem straightforward at first glance, but far from it. Definitions of sepsis have changed: Which definition is being considered? Surgical patient populations can be heterogeneous: Which patient population is being discussed, exactly? Does “surgical sepsis” refer to infectious diseases that are treated primarily by surgery, or nosocomial infections that occur in the post-operative period, or both? If both, should they be considered separately or together? Are the most relevant outcomes being considered? Are there actually any data (there are few if any under the new Third International Consensus Definitions for Sepsis and Septic Shock [Sepsis-3] definition) [1]? What should change, or be the emphasis going forward?

What Is Sepsis?

The term sepsis has existed for two millennia [2] but the definition of such has changed repeatedly over time. For some time there has been a reasonable degree of consensus that sepsis is not infection, but rather the manifestation of the host's response to infection, defined primarily as the pro-inflammatory response. As a practical consideration, the concept of the systemic inflammatory response syndrome (SIRS) arose in the 1990s [3] as a simplistic but reasonably effective, quantifiable bedside tool to identify sepsis, which was differentiated by severity into sepsis (the pro-inflammatory response to infection), severe sepsis (sepsis plus organ dysfunction), or septic shock (sepsis with hypotension refractory to fluid administration). Quantitatively, SIRS (actually, even any one of the four criteria but certainly as defined by two or more criteria) portended the development and magnitude of organ dysfunction and the risk of death as early as the second day of critical illness [4], However, SIRS was derided conceptually as being too sensitive (e.g., patients with uncomplicated appendicitis who manifest fever and leukocytosis have the syndrome by definition), and SIRS performed only moderately well in predicting mortality (c statistic approximately 0.75 averaged across multiple reports).

The new definitions no longer consider sepsis to be present if organ dysfunction does not exist [1], thus describing a more serious disease entity that previously. The Sepsis-3 definition also incorporates a new criterion for septic shock (lactate concentration >2 mmol/L). Driessen et al. [5] compared retrospectively a prospectively collected cohort of 632 critically ill patients with sepsis. Mortality in the intensive care unit (ICU) was compared between patients fulfilling the Sepsis-2 definition of septic shock (n = 482, 76.3% of patients with sepsis) and those who met the new Sepsis-3 criteria (n = 300, 48.4% of patients with sepsis). Patients meeting the Sepsis-2 definition had lower mortality (34.0% vs. 38.9%). Notably, only serum lactate concentrations >6 mmol/L were associated with increased ICU mortality [5].

Also introduced along with the revised definitions was a new screening tool, reinforcing the idea that rapid detection of sepsis in its protean manifestations remains the most secure route to good outcomes. Perhaps in their zeal to move away completely from SIRS, the quick sepsis-associated organ failure assessment (qSOFA) tool [6,7] has been promulgated as superior for the identification of sepsis, but that may not be the case. Serafim et al. [8] performed a systematic review and meta-analysis of qSOFA and SIRS for the diagnosis of sepsis and prediction of mortality. Ten studies (229,480 patients) were included for analysis. The meta-analysis of sensitivity for the diagnosis of sepsis favored SIRS, although specificity was in favor of qSOFA in the single study that examined the same. By contrast, meta-analysis of the c statistic for mortality prediction (six studies) favored qSOFA as a predictor of in-hospital mortality [8]. As a prevalent feature of sepsis, SIRS should remain an important component of the diagnostic process [9].

Epidemiology of Sepsis

The most recently published data on sepsis epidemiology in the United States date from the period 2009–2014, well after the Barcelona Declaration (1992, see below) and the Surviving Sepsis Campaign began to exert their effects. Evidence from administrative claims data suggest that the incidence of sepsis (old definitions) is increasing and that mortality is decreasing. However, estimates based on administrative data may be inaccurate, and may reflect changing diagnosis and coding practices over time. Rhee et al. [10] estimated the United States national incidence of sepsis using detailed clinical data from the electronic health records of 173,690 sepsis cases extracted from the records of 409 academic, community, and federal hospitals. The authors “adapted” Sepsis-3 criteria for “objective and consistent” electronic health record-based surveillance. The incidence of sepsis was 6.0% with a mean age of 67 years; 42% were female. The mortality rate in-hospital was 15.0%, and an additional 6.2% were discharged to hospice. Claims data showed an increase in incidence and a decrease in mortality over time, whereas the clinical criteria showed no difference in either incidence or mortality over the time period.

Worldwide, it is estimated that more than 18 million people develop sepsis annually. Approximately 14 million (approximately 75%–80%) survive to hospital discharge. Among initial survivors, one-half recover fully, one-third die within one year, and one-sixth have severe, persistent impairments [11] that may be functional, cognitive, or behavioral (e.g., anxiety, depression, post-traumatic stress disorder). There is a 40% re-hospitalization rate within 90 d. Sepsis survivors are at increased risk of recurrent infection, acute kidney injury, and new cardiovascular events. Characteristics associated with post-discharge complications include poorer pre-morbid health status, higher severity of infection, greater impairment of host immunity, and low-quality hospital treatment (e.g., timeliness of initial sepsis care, avoidance of treatment-related morbidity). Referral to a rehabilitation facility has been associated with decreased risk of 10-y mortality, but may be a selected group that is well enough to participate in rehabilitation in a meaningful way.

Epidemiology of surgical sepsis

Data regarding outcomes of surgical sepsis are scant, and non-existent to date for sepsis as newly defined. Only individual reports of outcomes after emergency surgery are available, and most examine only short-term outcomes such as 30-d mortality. Despite the difficulty involved in tracking patients longer term, specifically that post-operative patients in large numbers tend to be lost to follow-up, it is imperative that reporting of longer term outcomes (e.g., 90-d mortality) become the norm, because it will become increasingly difficult to justify short-term event horizons to payers.

As many as two million operations are complicated by surgical site infections in the United States each year [12], and surgical patients account for 30% of patients with sepsis. Some epidemiologic data do exist examining outcomes of nosocomial infection after elective surgery. Vogel et al. [12] mined the Healthcare Cost and Utilization Project State Inpatient Databases for New Jersey to determine trends in sepsis incidence, severity, and mortality rate after surgical procedures for the period 1990–2006. A total of 1,276,451 surgery discharges (42.1% elective) were identified. The incidence of post-operative sepsis after elective surgery was 1.09% (severe sepsis, 0.52%). The incidence of sepsis after elective surgery increased from 0.67% to 1.74% over the time period, whereas the incidence of severe sepsis increased from 0.22% to 1.12%. The mortality rate of sepsis after elective surgery did not change. By contrast, the rates of post-operative sepsis (4.24%) and severe sepsis (2.28%) were higher after non-elective surgery. The incidences of sepsis (3.74% to 4.51%) and severe sepsis (1.79% to 3.15%) after non-elective surgery increased over time. After non-elective surgery, the in-hospital mortality rate from sepsis decreased from 37.9% to 29.8% over the time period. The differences in mortality trends between elective and non-elective surgery complicated by sepsis might be caused by lack of statistical power resulting from the much lower incidence of sepsis after elective surgery. Alternatively, the non-elective patients were more likely to receive post-operative care in a monitored setting (e.g., an intensive care unit), thus facilitating early recognition and intervention.

Vogel et al. mined the National Inpatient Sample for data between 2002–2006 to identify patients who developed sepsis after undergoing elective surgery [13], utilizing the Patient Safety Indicator “Postoperative Sepsis” (PSI-13). Case-mix-adjusted rates were calculated by using a multivariable logistic regression model for sepsis risk, and an indirect standardization method. Among 6,512,921 weighted elective surgical cases, 78,669 developed post-operative sepsis (1.21%). Esophageal, pancreatic, and gastric procedures represented the greatest risk, whereas in-hospital mortality was highest after thoracic, adrenal, and hepatic operations if sepsis did develop. Older males and patients of black or Latino ethnicity had a particular predilection. Lower socioeconomic status and larger hospital size were also associated with increased risk of sepsis.

Vogel et al. [14] also examined the long-term ramifications of hospital-acquired post-operative infections among U.S. Medicare patients after open abdominal vascular surgery, examining the re-admission rate and both 30- and 90-d mortality rates for the period 2005–2007. Among the 29,459 procedures identified, 4,016 were complicated by infection (13.6%). The hospital mortality rate was 13.7% (compared with 4.0% for uninfected patients). Infection complicated by sepsis carried a 50.9% rate of in-hospital mortality (versus 13.7% for post-operative infection without sepsis). The mortality rate for infected patients doubled between 30 and 90 d (4.4% vs. 1.2% at 30 d as opposed to 8.6% vs. 2.6% at 90 d).

Multiple organ dysfunction syndrome in critical surgical illness

Approximately three-quarters of cases of multiple organ dysfunction syndrome (MODS) are associated with sepsis (now by definition, in Sepsis-3), the remainder being associated with massive inflammation (e.g,, severe acute pancreatitis in its early phase) or injury (e.g., multiple trauma) in surgical patients [15]. The syndrome is variable in its manifestations and severity, but as many as 45% of critically ill surgical patients develop some component of MODS. Those who do are at more than 20-fold increased risk of death. Theories of pathogenesis abound, including ischemia-reperfusion syndrome and dysregulation of the host inflammasome and innate immunity. After four decades of work, contributed to in large part by surgeon–scientists, the incidence and magnitude of MODS appear to be decreasing [16].

Barie et al. [16] conducted a longitudinal 17-year prospective study of 11,314 critically ill or critically injured surgical patients, 5,157 (45.5%) of whom developed organ dysfunction to some degree. The mortality rate among those afflicted was 22%, versus 1% for those unaffected. Fully manifest MODS is nearly non-survivable except for anecdotal reports [17]. By multivariable analysis, only 6% of ICU mortality in surgical illness was not explained by MODS. From 1990–2006, the incidence and magnitude of MODS decreased when adjusted for severity of illness [16]. Improved resuscitation, fewer transfusions causing fewer nosocomial infections, glycemic control (which reduces mortality among surgical patients, although not among medical patients [18,19]), and understanding the importance of timely, definitive surgical source control of infection are all likely contributors to the diminution of this scourge.

The Research Agenda in Septic Shock

Sepsis and septic shock are less lethal than they were two decades ago. In October 2002, during the European Society of Intensive Care Medicine annual congress, the Surviving Sepsis Campaign was launched through the Barcelona Declaration, a document calling critical care providers, governments, health agencies, and lay people to join the fight against sepsis. The aim of the campaign was to reduce the sepsis mortality rate by 25% within five years. Although not achieved temporally, the overall goal has been achieved and improvement is ongoing. Improvements in outcomes came initially as much from early recognition and reaction and a standardized approach, outlined through guidelines promulgated by consensus, because at the time, there simply was not much evidence with scientific rigor. Now in their fourth iteration [20], the guidelines incorporate a large body of scientific evidence that has accumulated in the intervening years. Paradoxically, improved clinical outcomes may have hampered the research enterprise, because control group mortality in clinical trials of sepsis therapeutics became a moving (decreasing) target, thus under-powering several major trials [21]. However, there may be value in terms of new knowledge to emerge from even negative trials [22,23].

Sepsis care is expensive, as are the trials that have informed clinical practice. Recovery from sepsis is protracted, and patients may never be restored to their pre-morbid health status. Competition for research funding is intense. To maintain awareness and maximize funding opportunities, a group of intensivists (including surgical representation) has defined a research agenda for septic shock [24]. The top ten topics recommended to undergo clinical testing in the next decade are shown in Table 1. Note that the emphasis remains on therapeutics, although new diagnostics do hold promise and are especially worthy of study. Missing are recommendations for refining the definitions of outcomes. Note also that the list is not especially “surgical,” except perhaps for the proposed fluid resuscitation trial. It has been more than a decade since a research agenda has been outlined in surgical infection [25]; the surgical infection research agenda is in urgent need of redefinition.

Table 1.

Top Ten Research Topics in Adult Critical Care Medicine Recommended to Undergo Testing in the Next Ten Years

1. Restrictive vs. liberal fluid administration

2. Rapid microbiology diagnostics and antibiotic measuring devices to guide therapy

3. Reducing catecholamine use in septic shock

4. Counteracting endocrine/metabolic/bioenergetic failure

5. Stem cell therapies

6. Biomarker-guided trials

7. Novel anti-inflammatory therapies

8. Biomarker-guided immune stimulation

9. Use of machine learning algorithms/computerized decision support

10. Multi-arm, multi-stage trials of common sepsis therapies

Adapted from Perner et al. [24].

Selected 2017 Literature

Perforated sigmoid diverticulitis

Recent randomized trials show that laparoscopic lavage for Hinchey III perforated diverticulitis is associated with similar mortality and less prevalent stoma formation, but a higher rate of early re-intervention, i.e., re-operation when laparoscopic lavage fails. The Scandinavian Diverticulitis (SCANDIV) trial randomly assigned 1,999 patients at 21 hospitals in Norway and Sweden treated between 2010–2014 to compare laparoscopic lavage to primary sigmoid colectomy. Patients with perforated diverticulitis confirmed at the time of surgery numbered 89 and 83 patients, respectively [26]. At one year, neither severe complications (34% vs. 27%, p = 0.33) nor disease-related mortality (12% vs. 11%) differed among such confirmed patients. Nor did the complication rate or mortality differ among the 144 patients with purulent peritonitis. Laparoscopic lavage was associated with a higher rate of deep surgical site infections (SSIs) (32% vs. 13%, p = 0.006),” but fewer superficial surgical site infections (1% vs. 17%, p = 0.001). Un-planned re-operation occurred more frequently after lavage (27% vs. 10%, p = 0.01), but the overall re-operation rate did not differ after accounting for stoma reversals (28% vs. 29%), indicating that un-planned operation was usually not undertaken with stoma creation. Despite a substantially lower stoma creation rate in the lavage group (14% vs. 42%, p = 001), a quality of life assessment did not differ between groups.

Three aspects of this study are presented for the reader's consideration. First, the use of non-standard terminology to describe surgical site infections makes it difficult to parse what is meant by deep surgical site infections. How many were deep incisional SSIs and how many were organ/space SSIs? All authors and journal editors are encouraged to use standard terminology to describe findings so as to optimize interpretability. Second, a fundamental tenet of surgical infections management is to achieve definitive surgical source control at the index procedure, so as to minimize the risk of organ dysfunction and death (about the former, there is no description provided). From the data, laparoscopic lavage clearly does not achieve source control at the index procedure in more than one-quarter of patients. Although the overall re-operation rate did not differ, clearly the lavage patients underwent more emergency salvage procedures to achieve source control whereas the resection patients mostly underwent elective operation when healthy (presumably free of sepsis) for stoma reversal. Which is less dangerous for the patient? Third, the description of the statistical analysis plan makes no mention of when the final analysis would occur, or whether the one-year (interim?) analysis was planned a priori. It is well known that repetitive “peeks” at data via interim analysis degrades the statistical power of a study. Journal editors and reviewers must be sticklers on this point, lest peer-reviewed reports of studies undertaken at considerable cost and effort be tainted by scientific evangelism or worse, “salami-slicing” to identify the “minimum publishable data set.”

Omega-3 supplementation for immunomodulation

Nutritional supplementation of omega-3 fatty acids has been proposed to modulate the balance of pro- and anti-inflammatory mediators in sepsis. Most fatty acid supplementation occurs for patients receiving parenteral nutrition, but standard lipid formulas are rich in omega-6 fatty acids, which provide substrate for eicosanoid synthesis, perhaps the proverbial “pouring gasoline on a fire.” If omega-3 supplementation is proved to be effective, it would be inexpensive to implement, if only we knew how. Unfortunately, it may take several weeks to displace omega-6 with omega-3 as substrate for lipid biosynthesis, which may extend beyond the time of benefit, presumably when the patient has active or recently treated sepsis. Lu et al. [27] searched online databases for randomized trials on enteral or parenteral omega-3 supplementation for patients with sepsis or septic shock using GRADE methodology. Seventeen trials that enrolled 1,239 patients were selected for analysis, indicating a mean sample size in each arm of each trial of only 36 patients. Unfortunately, the included trials only compared omega-3 to no supplementation or placebo; none compared omega-3 with omega-6. Omega-3 supplementation had no effect upon mortality (relative risk [RR] 0.85, 95% confidence interval (CI) 0.71–1.03, p = 0.10 based on evidence of moderate quality). Duration of ICU stay (mean difference [MD] −3.79 d, 95% CI −5.49–2.09, p < 0.0001) and mechanical ventilation (MD −2.27 days, 95% CI −4.27–0.27, p = 0.030) were both reduced, although both analyses were based on very low-quality evidence. Sensitivity analysis called in to question the robustness of both of the latter observations. The authors opined that the very low quality of evidence overall was insufficient to justify routine use of omega-3 supplementation).

Two aspects of this study are presented for the reader's consideration. First, outcomes that rely on an assessment of the duration of something are dubious as they can be manipulated absent pre-specified protocols for management (in this case ICU discharge criteria and protocols for weaning of and liberation from mechanical ventilation). Such is impossible under the rubric of meta-analysis. Second, meta-analysis may suffer from the phenomenon of “garbage in, garbage out.” A well-done analysis of low-quality data simply cannot constitute evidence of benefit [28]. The secondary analyses reported by Lu et al. [27] unfortunately lack construct validity.

Bowel preparation for inflammatory bowel disease surgery

Mechanical and antibiotic bowel preparation prior to elective colon surgery was recognized to be efficacious based on validated work extending back more than 40 y [29,30]. The preparation protocol, emphasizing administration of non-absorbable oral antibiotics the day before surgery, was timed to reduce bacterial counts in the colon so as to nadir at the time of skin incision the next morning. Beginning in the 1980s and continuing through the 1990s, the pendulum swung away from this standard of care. Now un-supervised medically because bowel preparation occurred at home in the new era of same-day admission after major surgery, there was a view that the diarrhea engendered by mechanical bowel preparation made it unlikely that the oral antibiotics could exert their effect. Out went the oral antibiotic prophylaxis. The next step in the progression was to abandon mechanical preparation as well, particularly for right hemicolectomy after several studies questioned the practice. Unfortunately, it was the prior omission of oral antibiotic prophylaxis that rendered mechanical bowel preparation ineffective. Predictably, infection rates after elective colon surgery, as low as 6% with a combination of oral and mechanical bowel preparation and intravenous broad-spectrum antibiotic prophylaxis, soared to more than 20% in many reports after the new millennium [31]. In response, the pendulum has now swung back to combined mechanical and oral antibiotic bowel preparation.

Less attention has been paid to surgery for inflammatory bowel disease, which includes small intestinal surgery as well as colon surgery. In patients with inflammatory bowel disease, factors such as nutritional status, use of immunosuppressive agents, and intra-operative findings can influence surgical outcomes. Shwaartz et al. [32] queried the American College of Surgeons National Surgical Quality Improvement Program, Procedure-Targeted Colectomy database from 2012–2014 to identify 3,679 patients with inflammatory bowel disease who underwent elective colorectal resection with or without bowel preparation [32]. Among the patients, 42.5% had no bowel preparation whatsoever, 21.5% had mechanical bowel preparation only, 8.8% had antibiotic bowel preparation only, and only 27.2% underwent combined mechanical and antibiotic bowel preparation. Combined mechanical and antibiotic bowel preparation was associated with significantly lower risk of anastomotic leak, ileus, incisional surgical site infection, organ-space surgical site infection, wound dehiscence, and sepsis or septic shock (Table 2).

Table 2.

Multivariable Analysis of Outcomes Influenced by Mechanical and Antibiotic Bowel Preparation after Elective Colon Surgery for Inflammatory Bowel Disease

Post-operative complication	p	Odds ratio	95% confidence interval
Anastomotic leak	0.02	0.50	0.28–0.89
Ileus	0.001	0.65	0.51–0.83
Incisional surgical site infection	0.001	0.55	0.39–0.79
Organ/space surgical site infection	0.001	0.53	0.36–0.77
Wound dehiscence	0.03	0.19	0.04–0.83
Sepsis or septic shock	0.001	0.52	0.36–0.76
Blood transfusion	0.01	0.68	0.50–0.92

Adapted from Harhay et al. [32].

Sepsis following renal transplantation

Allareddy et al. [33] performed a retrospective analysis of the Nationwide Inpatient Sample for the years 2004–2010 to ascertain the impact of “septicemia” (another example of non-standard terminology) on outcomes of renal transplantation among adult patients [33]. During the period, 113,058 patients underwent kidney transplantation, of whom 2,459 developed septicemia. (It should go without saying that use of non-standard terminology is to be discouraged strongly, because it becomes difficult to ascertain exactly what they endeavor to describe). Patients who developed sepsis (?) were predominantly male (60% and white (54%) but that might reflect disparities in access to care. The mean age was 50 y. In addition to increased cost (mean, $528,980 vs. $182,165) and length of stay (mean increase of 18.7 d), the risk of in-hospital mortality was increased by more than 30-fold (odds ratio 31.33, 95% CI 20.25–48.48, p < 0.0001). Increased age (2% per year) and comorbidity (57% for each additional comorbidity) were associated with higher risk.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

References

Singer

, Deutschman

, Seymour

, et al. The third international consensus definitions for sepsis and septic shock (Sepsis-3). JAMA, 2016; 315:801–810.

Marshall

. Sepsis definitions: A work in progress. Crit Care Clin, 2018; 34:1–14.

American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference: Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Crit Care Med, 1992; 20:864–874.

Talmor

, Hydo

, Barie

. Relationship of systemic inflammatory response syndrome to organ dysfunction, length of stay, and mortality in critical surgical illness: Effect of intensive care unit resuscitation. Arch Surg, 1999; 134:81–87.

Driessen

RGH

, van de Poll

MCG

, Mol

, et al. The influence of a change in septic shock definitions on intensive care epidemiology and outcome: Comparison of Sepsis-2 and Sepsis-3 definitions. Infect Dis (Lond), 2017; Sep 26:1–7.

Vincent

, Moreno

, Takala

, et al. The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-Related Problems of the European Society of Intensive Care Medicine. Intensive Care Med, 1996, 22:707–710.

Seymour

, Liu

, Iwashyna

, et al. Assessment of Clinical Criteria for Sepsis: For the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA, 2016; 315:762–774.

Serafim

, Gomes

, Salluh

, Povos

. A comparison of the quick-SOFA (qSOFA) and SIRS criteria for the diagnosis of sepsis and prediction of mortality: A systematic review and meta-analysis. Chest (in press).

Simpson

. SIRS in the time of Sepsis-3. Chest, 2018; 153:34–38.

10.

Rhee

, Dantes

, Epstein

, et al. Incidence and trends of sepsis in US hospitals using clinical vs. claims data, 2009–2014. JAMA, 2017; 318:1241–1249.

11.

Prescott

, Angus

. Enhancing recovery from sepsis: A review. JAMA, 2018; 319:62–75.

12.

Vogel

, Dombrovskiy

, Lowry

. Trends in postoperative sepsis: Are we improving outcomes?. Surg Infect, 2009; 10:71–78.

13.

Vogel

, Dombrovskiy

, Carson

, et al. Postoperative sepsis in the United States. Ann Surg, 2010; 252:1065–1071.

14.

Vogel

, Dombrovskiy

, Lowry

. Impact of infectious complications after elective surgery on hospital readmission and late deaths in the U.S. Medicare population. Surg Infect, 2012; 13:307–311.

15.

Barie

, Hydo

, Pieracci

, et al. Multiple organ dysfunction syndrome in critical surgical illness. Surg Infect, 2008; 10:369–377.

16.

Barie

, Hydo

, Shou

, Eachempati

. Decreasing magnitude of multiple organ dysfunction syndrome despite increasingly severe critical surgical illness: A 17-year longitudinal study. J Trauma, 2008; 65:1227–1235.

17.

Kleiman

, Barie

. Survival in fully manifest multiple organ dysfunction syndrome. Surg Infect, 2014; 15:445–449.

18.

Eachempati

, Hydo

, Shou

, Barie

. Implementation of tight glucose control for critically ill surgical patients: A process improvement analysis. Surg Infect, 2009, 10:523–531.

19.

Griesdale

, de Souza

, van Dam

, et al. Intensive insulin therapy and mortality among critically ill patients: A meta-analysis including NICE-SUGAR study data. CMAJ, 2009; 180:821–827.

20.

Rhodes

, Evans

, Alhazzani

, et al. Surviving Sepsis Campaign: International guidelines for management of sepsis and septic shock: 2016. Intensive Care Med, 2017; 43:304–377.

21.

Ranieri

, Thompson

, Barie

, et al. Drotrecogin alfa (activated) in adults with septic shock. N Engl J Med, 2012; 366:2055–2064.

22.

Pocock

, Stone

. The primary outcome fails-what next?. N Engl J Med, 2016; 375:861–870.

23.

Harhay

, Wagner

, Ratcliffe

, et al. Outcomes and statistical power in adult critical care randomized trials. Am J Respir Crit Care Med, 2014; 189:1469–1478.

24.

Perner

, Gordon

, Angus

, et al. The intensive care medicine research agenda on septic shock. Intensive Care Med, 2017; 43:1294–1305.

25.

Nathens

, Cook

, Machiedo

, et al. Defining the research agenda for surgical infection: A consensus of experts using the Delphi approach. Surg Infect, 2006; 7:101–110.

26.

Schultz

, Walton

, Blecic

, et al. One-year results of the SCANDIV randomized clinical trial of laparoscopic lavage versus primary resection for perforated diverticulitis. Br J Surg, 2017; 104:1382–1392.

27.

, Sharma

, McIntyre

, et al. Omega-3 supplementation in patients with sepsis: A systematic review and meta-analysis of randomized trials. Ann Intensive Care, 2017; 7:58.

28.

Barie

. Does a well-done analysis of poor-quality data constitute evidence of benefit?. Ann Surg, 2012; 255:1031–1031.

29.

Nichols

, Broido

, Condon

, et al. Effect of preoperative neomycin-erythromycin intestinal preparation on the incidence of infectious complications following colon surgery. Ann Surg, 1973; 178:453–462.

30.

Lewis

, Goodall

, Marien

, et al. Is neomycin necessary for bowel preparation in surgery of the colon? Oral neomycin plus erythromycin versus erythromycin-metronidazole. Can J Surg, 1989; 32:265–270.

31.

, Barie

, Stein

, et al. Antibiotic regimen and the timing of prophylaxis are important for reducing surgical site infection after elective abdominal colorectal surgery. Surg Infect, 2011; 12:255–260.

32.

Shwaartz

, Fields

, Sobrero

, Divino

. Does bowel preparation for inflammatory bowel disease surgery matter?. Colorectal Dis, 2017, 19:832–839.

33.

Allareddy

, Rampa

, Rotta

, Allareddy

. The impact of septicemia occurring during hospitalization for renal transplantation procedures on outcomes of adults in United States. PLoS One, 2017; 12:e0179466.