Acute kidney injury is surprisingly common and a powerful predictor of mortality in surgical sepsis.

BACKGROUND: Acute kidney injury (AKI) is a common and often catastrophic complication in hospitalized patients; however, the impact of AKI in surgical sepsis remains unknown. We used Risk, Injury, Failure, Loss, End stage (RIFLE) consensus criteria to define the incidence of AKI in surgical sepsis and characterize the impact of AKI on patient morbidity and mortality. METHODS: Our prospective, institutional review board-approved sepsis research database was retrospectively queried for the incidence of AKI by RIFLE criteria, excluding those with chronic kidney disease. Patients were grouped into sepsis, severe sepsis, and septic shock by refined consensus criteria. Data including demographics, baseline biomarkers of organ dysfunction, and outcomes were compared by Student's t test and χ test. Multivariable regression analysis was performed for the effect of AKI on mortality adjusting for age, sex, African-American race, elective surgery, Acute Physiology and Chronic Health Evaluation II score, septic shock versus severe sepsis, and sepsis source. RESULTS: During the 36-month study period ending on December 2010, 246 patients treated for surgical sepsis were evaluated. AKI occurred in 67% of all patients, and 59%, 60%, and 88% of patients had sepsis, surgical sepsis, and septic shock, respectively. AKI was associated with Hispanic ethnicity, several baseline biomarkers of organ dysfunction, and a greater severity of illness. Patients with AKI had fewer ventilator-free and intensive care unit-free days and a decreased likelihood of discharge to home. Morbidity and mortality increased with severity of AKI, and AKI of any severity was found to be a strong predictor of hospital mortality (odds ratio, 10.59; 95% confidence interval, 1.28-87.35; p = 0.03) in surgical sepsis. CONCLUSION: AKI frequently complicates surgical sepsis, and serves as a powerful predictor of hospital mortality in severe sepsis and septic shock. LEVEL OF EVIDENCE: Prognostic and epidemiologic study, level III.

The epidemiology of sepsis in general surgery patients.

BACKGROUND: Sepsis is increasing in hospitalized patients. Our purpose is to describe its current epidemiology in a general surgery (GS) intensive care unit (ICU) where patients are routinely screened and aggressively treated for sepsis by an established protocol. METHODS: Our prospective, Institutional Review Board-approved sepsis research database was queried for demographics, biomarkers reflecting organ dysfunction, and mortality. Patients were grouped as sepsis, severe sepsis, or septic shock using refined consensus criteria. Data are compared by analysis of variance, Student's t test, and χ test (p<0.05 significant). RESULTS: During 24 months ending September 2009, 231 patients (aged 59 years ± 3 years; 43% men) were treated for sepsis. The abdomen was the source of infection in 69% of patients. Several baseline biomarkers of organ dysfunction (BOD) correlated with sepsis severity including lactate, creatinine, international normalized ratio, platelet count, and d-dimer. Direct correlation with mortality was noted with particular baseline BODs including beta natriuretic peptide, international normalized ratio, platelet count, aspartate transaminase, alanine aminotransferase, and total bilirubin. Most patients present with severe sepsis (56%) or septic shock (26%) each with increasing multiple BODs. Septic shock has prohibitive mortality rate (36%), and those who survive septic shock have prolonged ICU stays. CONCLUSION: In general surgery ICU patients, sepsis is predominantly caused by intra-abdominal infection. Multiple BODs are present in severe sepsis and septic shock but are notably advanced in septic shock. Despite aggressive sepsis screening and treatment, septic shock remains a morbid condition.

Computer protocol facilitates evidence-based care of sepsis in the surgical intensive care unit.

BACKGROUND: Care of sepsis has been the focus of intense research and guideline development for more than two decades. With ongoing success of computer protocol (CP) technology and with publication of Surviving Sepsis Campaign (SSC) guidelines, we undertook protocol development for management of sepsis of surgical intensive care unit patients in mid-2006. METHODS: A sepsis protocol was developed and implemented in The Methodist Hospital (TMH) (Houston, TX) surgical intensive care unit (27 beds) together with a sepsis research database. We compare paper-protocol (PP) (2008) and CP (2009) performance and results of the SSC guideline performance improvement initiative (2005-2008). TMH surgical intensive care unit sepsis protocol was developed to implement best evidence and to standardize decision making among surgical intensivists, nurse practitioners, and resident physicians. RESULTS: The 2008 and 2009 sepsis protocol cohorts had very similar number of patients, age, % male gender, Acute Physiology and Chronic Health Evaluation scoring system II, and Sequential Organ Failure Assessment scores. The 2008 PP patients had greater baseline lactate concentration consistent with greater mortality rate. Antibiotic agents were administered to 2009 CP cohort patients sooner than 2008 PP cohort patients. Both cohorts received similar volume of intravenous fluid boluses. Comparing 6-hour resuscitation bundle compliance, the 2009 CP cohort was substantially greater than SSC eighth quarter and 2008 PP cohorts (79% vs. 31% vs. 29%), and mortality rate was much less when using the CP (14% vs. 31% vs. 24%). CONCLUSIONS: Our comprehensive sepsis protocol has enabled rapid and consistent implementation of evidence-based care, and, implemented as a bedside CP, contributed to decreased mortality rate for management of surgical sepsis.

Identification of cardiac dysfunction in sepsis with B-type natriuretic peptide.

BACKGROUND: B-type natriuretic peptide (BNP) is secreted in response to myocardial stretch and has been used clinically to assess volume overload and predict death in congestive heart failure. More recently, BNP elevation has been demonstrated with septic shock and is predictive of death. How BNP levels relate to cardiac function in sepsis remains to be established. STUDY DESIGN: Retrospective review of prospectively gathered sepsis database from a surgical ICU in a tertiary academic hospital. Initial BNP levels, patient demographics, baseline central venous pressure levels, and in-hospital mortality were obtained. Transthoracic echocardiography was performed during initial resuscitation per protocol. RESULTS: During 24 months ending in September 2009, two hundred and thirty-one patients (59 ± 3 years of age, 43% male) were treated for sepsis. Baseline BNP increased with initial sepsis severity (ie, sepsis vs severe sepsis vs septic shock, by ANOVA; p < 0.05) and was higher in those who died vs those who lived (by Fisher's exact test; p < 0.05). Of these patients, 153 (66%) had early echocardiography. Low ejection fraction (<50%) was associated with higher BNP (by Fisher's exact test; p < 0.05) and patients with low ejection fraction had a higher mortality (39% vs 20%; odds ratio = 3.03). We found no correlation between baseline central venous pressure (12.7 ± 6.10 mmHg) and BNP (526.5 ± 82.10 pg/mL) (by Spearman's ρ, R(s) = .001) for the entire sepsis population. CONCLUSIONS: In surgical sepsis patients, BNP increases with sepsis severity and is associated with early systolic dysfunction, which in turn is associated with death. Monitoring BNP in early sepsis to identify occult systolic dysfunction might prompt earlier use of inotropic agents.

Validation of a screening tool for the early identification of sepsis.

BACKGROUND: Sepsis is the leading cause of mortality in noncoronary intensive care units. Recent evidence based guidelines outline strategies for the management of sepsis and studies have shown that early implementation of these guidelines improves survival. We developed an extensive logic-based sepsis management protocol; however, we found that early recognition of sepsis was a major obstacle to protocol implementation. To improve this, we developed a three-step sepsis screening tool with escalating levels of decision making. We hypothesized that aggressive screening for sepsis would improve early recognition of sepsis and decrease sepsis-related mortality by insuring early appropriate interventions. METHODS: Patients admitted to the surgical intensive care unit were screened twice daily by our nursing staff. The initial screen assesses the systemic inflammatory response syndrome parameters (heart rate, temperature, white blood cell count, and respiratory rate) and assigns a numeric score (0-4) for each. Patients with a score of > or = 4 screened positive proceed to the second step of the tool in which a midlevel provider attempts to identify the source of infection. If the patients screens positive for both systemic inflammatory response syndrome and an infection, the intensivist was notified to determine whether to implement our sepsis protocol. RESULTS: Over 5 months, 4,991 screens were completed on 920 patients. The prevalence of sepsis was 12.2%. The screening tool yielded a sensitivity of 96.5%, 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%. CONCLUSIONS: The three step sepsis screening tool is a valid tool for the early identification of sepsis. Implementation of this tool and our logic-based sepsis protocol has decreased sepsis-related mortality in our SICU by one third.

Decreasing ventilator-associated pneumonia in a trauma ICU.

BACKGROUND: The incidence of ventilator-associated pneumonia ranges from 10 to 25%, with mortality of 10 to 40%. It prolongs hospital stay and drives up hospital costs. Our Intensive Care Unit (ICU) ventilator-associated pneumonia (VAP) rates were hovering at the National Nosocomial Infection Surveillance (NNIS) 90th percentile (22.3-32.7 infections per 1,000 ventilator days from January 2002 through October 2002) necessitating a performance improvement initiative designed to decrease the incidence of VAP. METHODS: A ventilator bundle that incorporates the Center for Disease Control (CDC) Guidelines for Prevention of Nosocomial Pneumonia was instituted in June of 2002. In October 2002, an intervention that audited compliance with the ventilator bundle and provided real-time feedback to ICU staff was started. VAP rates were followed using NNIS criteria. Costs were evaluated using hospital TSI data. RESULTS: VAP did not decrease with institution of the ventilator bundle alone. However, VAP did significantly decrease when the compliance with the ventilator bundle was audited daily and weekly feedback was provided to the caregivers. From November 2002 through June 2003 VAP stayed between 0 and 12.8 per 1,000 ventilator days. The average cost of a VAP was 50,000 dollars. CONCLUSIONS: Prevention of VAP requires a concerted effort on the part of hospital administration, physicians, and ICU personnel. The program must be evidence-based, maintained, and accepted by ICU personnel. Continued education and feedback are crucial to maintaining a low VAP rate.

Preload optimization using "starling curve" generation during shock resuscitation: can it be done?

Preload-directed resuscitation is the standard of care in U.S. trauma centers. As part of our standardized protocol for traumatic shock resuscitation, patients who do not respond to initial interventions of hemoglobin replacement and fluid volume loading have optimal preload determined using a standardized algorithm to generate a "Starling curve." We retrospectively analyzed data from 147 consecutive resuscitation protocol patients during the 24 months ending August 2002. Fifty (34%) of these patients required preload optimization, of which the optimization algorithm was completed in 36 (72%). The average age of those who required preload optimization was 44 +/- 3 years vs. 34 +/- 1 years for patients who did not. Execution of the algorithm caused PCWP to increase from 18 +/- 1 mmHg to a maximum of 25 +/- 2 mmHg and CI to increase from 3.2 +/- 0.1 L/min m(-2) to 4.5 +/- 0.4 L/min m(-2). Algorithm logic determined PCWP = 24 +/- 2 to be optimal preload at the maximum CI = 4.8 +/- 0.4, and as the volume loading threshold for the remaining time of the resuscitation process. Starling curve preload optimization was begun 6.5 +/- 0.8 h after start of the resuscitation protocol and required 36 +/- 5 min and 4 +/- 0.4 fluid boluses (1.6 +/- 0.2 L). Comparison of early response of those patients who required preload optimization and those who did not indicated hemodynamic compromise apparent in the 1st 4 h of standardized resuscitation. We conclude that preload optimization using sequential fluid bolus and PCWP-CI measurement to generate a Starling curve is feasible during ICU shock resuscitation, but that there is the disadvantage that increasing and maintaining high PCWP may contribute to problematic tissue edema.

Goal-oriented shock resuscitation for major torso trauma: what are we learning?

Shock resuscitation is an obligatory intervention for severely injured patients who present in shock. During the past 15 years, with widespread acceptance of "damage control" surgery and early triage to the intensive care unit (ICU) to optimize resuscitation, the lives of many major trauma victims have been saved, and much has been learned about shock resuscitation. Due largely to the work of Shoemaker et al., a resuscitation strategy based on a standardized process using O(2) delivery index (DO(2)I) as an endpoint and physiologic performance goal for interventions has been developed, studied, and refined for resuscitation of shock caused by major trauma. DO(2)I >or=600 mL O(2)/min-m2 is the only resuscitation endpoint variable that has been tested in prospective randomized trials (PRTs) of trauma patient outcome. These PRTs are limited, and their results are not conclusive. Results from other investigators, including our group, using similar process and endpoints, are indicating similar performance and outcomes. We believe that DO(2)I is a useful endpoint because it integrates three important variables, ie, hemoglobin concentration [Hb], arterial hemoglobin O(2) saturation, and cardiac output. We have found DO(2)I >or=500 mL O(2)/min-m2 to be an endpoint with more general applicability, but we believe that the standardized process is more important than the specific endpoint. To standardize our process, we have developed a computerized decision support tool for shock resuscitation. This technology has provided novel data collection and has permitted refinement of the bedside process. Our data analysis indicates that the next challenge will be to develop a similar pre ICU resuscitation process that will use less invasive monitors and different endpoints. Identification of the high-risk resuscitation nonresponders early in the resuscitation process will be needed to redirect their clinical trajectories. As an endpoint for interventions for goal-directed resuscitation in the critically injured trauma patient, systemic O(2) delivery is the current state of the art and the basis for near future development of clinical processes for resuscitation of shock due to major trauma.

Both primary and secondary abdominal compartment syndrome can be predicted early and are harbingers of multiple organ failure.

BACKGROUND: Primary abdominal compartment syndrome (ACS) is a known complication of damage control. Recently secondary ACS has been reported in patients without abdominal injury who require aggressive resuscitation. The purpose of this study was to compare the epidemiology of primary and secondary ACS and develop early prediction models in a high-risk cohort who were treated in a similar fashion. METHODS: Major torso trauma patients underwent standardized resuscitation and had prospective data collected including occurrence of ACS, demographics, ISS, urinary bladder pressure, gastric tonometry (GAP(CO2) = gastric regional CO(2) minus end tidal CO(2)), laboratory, respiratory, and hemodynamic data. With primary and secondary ACS as endpoints, variables were tested by uni- and multivariate logistic analysis (MLA). RESULTS: From 188 study patients during the 44-month period, 26 (14%) developed ACS-11 (6%) were primary ACS and 15 (8%) secondary ACS. Primary and secondary ACS had similar demographics, shock, and injury severity. Significant univariate differences included: time to decompression from ICU admit (600 +/- 112 vs. 360 +/- 48 min), Emergency Department (ED) crystalloid (4 +/- 1 vs. 7 +/- 1 L), preICU crystalloid (8 +/- 1 vs. 12 +/- 1L), ED blood administration (2 +/- 1 vs. 6 +/- 1 U), GAP(CO2) (24 +/- 3 vs. 36 +/- 3 mmHg), requiring pelvic embolization (9 vs. 47%), and emergency operation (82% vs. 40%). Early predictors identified by MLA of primary ACS included hemoglobin concentration, GAP(CO2), temperature, and base deficit; and for secondary ACS they included crystalloid, urinary output, and GAP(CO2). The areas under the receiver-operator characteristic curves calculated upon ICU admission are primary= 0.977 and secondary= 0.983. Primary and secondary ACS patients had similar poor outcomes compared with nonACS patients including ventilator days (primary= 13 +/- 3 vs. secondary= 14 +/- 3 vs. nonACS = 8 +/- 2), multiple organ failure (55% vs. 53% vs. 12%), and mortality (64% vs. 53% vs. 17%). CONCLUSION: Primary and secondary ACS have similar demographics, injury severity, time to decompression from hospital admit, and bad outcome. 2 degrees ACS is an earlier ICU event preceded by more crystalloid administration. With appropriate monitoring both could be accurately predicted upon ICU admission.

Supranormal trauma resuscitation causes more cases of abdominal compartment syndrome.

HYPOTHESIS: Normal resuscitation (oxygen delivery index [DO2I] >/=500 mL/min per square meter), compared with supranormal trauma resuscitation (DO2I >/=600 mL/min per square meter), requires less crystalloid volume, thus decreasing the incidence of intra-abdominal hypertension (IAH) and abdominal compartment syndrome (ACS). DESIGN: Retrospective analysis of a prospective database. SETTING: Twenty-bed intensive care unit (ICU) in a regional level I trauma center. PATIENTS: Patients with major trauma (injury severity score >15, initial base deficit >/=6 mEq/L, or need for >/=6 units of packed red blood cells in the first 12 hours) or age 65 years or older with any 2 of the previous criteria. INTERVENTIONS: Shock/trauma resuscitation protocol: pulmonary artery catheter, gastric tonometry, urinary bladder pressure measurements, lactated Ringer infusion, packed red blood cell transfusion, and moderate inotrope support, as needed, in that sequence, to attain and maintain a DO2I greater than or equal to 600 mL/min per m2 (16 months, ending January 1, 2001, n = 85) or a DO2I greater than or equal to 500 mL/min per square meter (16 months, starting January 1, 2001, n = 71) for the first 24 hours in the ICU. MAIN OUTCOME MEASURES: Lactated Ringer infusion volume (liters) at ICU admission, gastric partial carbon dioxide minus end-tidal carbon dioxide(GAPCO2), IAH (urinary bladder pressure measurements >20 mm Hg), ACS (urinary bladder pressure measurements >25 mm Hg with organ dysfunction), multiple organ failure, and mortality. RESULTS: Demographics, injury severity, and shock severity parameters were similar in both groups. The supranormal resuscitation group required more lactated Ringer infusion volume in the first 24 hours in the ICU (mean +/- SD, 13 +/- 2 vs 7 +/- 1 L; P<.05) and had higher GAPCO2 (16 +/- 2 vs 7 +/- 1 mm Hg; P<.05). In the supranormal group, IAH (42% vs 20%; P<.05) and ACS (16% vs 8%; P<.05) were more frequent. The conventional trauma outcomes, such as multiple organ failure (22% vs 9%; P<.05) and mortality (27% vs 11%; P<.05) were less favorable in the supranormal resuscitation group. CONCLUSION: Supranormal resuscitation, compared with normal resuscitation, was associated with more lactated Ringer infusion, decreased intestinal perfusion (higher GAPCO2), and an increased incidence of IAH, ACS, multiple organ failure, and death.

Normal versus supranormal oxygen delivery goals in shock resuscitation: the response is the same.

BACKGROUND: Shock resuscitation is integral to early management of severely injured patients. Our standardized shock resuscitation protocol, developed in 1997 and implemented as a computerized intensive care unit (ICU) bedside decision support tool in 2000, used oxygen delivery index (Do I) > or = 600 mL/min/m as the intervention endpoint. In a recent publication, Shoemaker et al. refuted positive outcome effect of early supranormal Do (i.e., Do I > or = 600) resuscitation. In response to and because of ongoing concern for excessive volume loading, we decreased our Do I endpoint from 600 to 500. Our hypothesis was that by decreasing the Do I endpoint, less crystalloid would be administered. We compare resuscitation responses to the protocol with goals of Do I > or = 600 versus 500 in two patient cohorts. METHODS: A standardized protocol was used to direct bedside decisions for resuscitation of patients with major injury (Injury Severity Score > 15), blood loss (> or = 6 units of packed red blood cells), metabolic stress (base deficit > or = 6 mEq/L), and no severe brain injury. The protocol logic is to attain and maintain Do I > or = a specified goal for the first 24 ICU hours using primarily blood and volume loading. Two cohorts were compared: Do I > or = 500 (18 patients admitted February-August 2001) versus Do I > or = 600 (18 patients admitted during 2000 age and gender matched with the Do I > or = 500 group). Data were analyzed using analysis of variance, chi, and t tests (p < 0.05). RESULTS: Both groups had similar demographics (age 30 +/- 3 years; 78% men; Injury Severity Score 27 +/- 3), hemodynamics, and severity of shock at start of resuscitation in the ICU. Resuscitation response was Do I increase to > or = 600 for both cohorts within approximately 12 hours. Throughout the 24-hour ICU process, the Do I > or = 500 cohort received less lactated Ringer's volume than the Do I > or = 600 cohort (total of 8 +/- 1 vs. 12 +/- 2 L; p < 0.05) and tended to receive less blood transfusion (total of 3 +/- 1 vs. 5 +/- 1 units of packed red blood cells). CONCLUSION: Shock resuscitation using Do I > or = 500 was indistinguishable from Do I > or = 600 mL/min/m. Less volume loading was required to attain and maintain Do I > or = 500 than 600 using computerized protocol technology to standardize resuscitation during the first 24 ICU hours.

Standardized trauma resuscitation: female hearts respond better.

HYPOTHESIS: Women respond better to standardized shock resuscitation compared with similarly severely injured men. DESIGN: Severely injured patients who met specific criteria were resuscitated using a standardized protocol with no adjustment for gender. The resuscitation protocol was used to attain and to maintain an oxygen delivery index of 600 mL/min. m(2) or greater (DO(2)I > or = 600) for the first 24 hours in the intensive care unit (ICU). Interventions, responses, and outcomes for the 2 cohorts were compared. Data were analyzed using analysis of variance, chi(2), and t tests; P<.05 was considered significant. SETTING: A 20-bed regional level I trauma center ICU. PATIENTS: Patients at high risk of postinjury multiple organ failure (major organ or vascular injury and/or skeletal fractures, initial arterial base deficit of 6 mEq/L or greater, requirement for 6 units or more of packed red blood cells in the first 12 hours after hospital admission, or age > or = 65 years with any 2 previous criteria). INTERVENTIONS: Pulmonary artery catheter, packed red blood cell transfusion, crystalloid fluid infusion, inotrope, and vasopressor support, as needed, in that sequence, to maintain DO(2)I > or = 600. MAIN OUTCOME MEASURES: Hemodynamic response to resuscitation, fluid, and packed red blood cell volume. RESULTS: During 2000, 58 patients (38 men, 20 women) met criteria and were resuscitated using our standardized protocol. Demographics and outcomes were similar for both cohorts. Requirements for and responses to standardized resuscitation were also similar, except for volume loading. The female cohort required less lactated Ringer solution volume (12 +/- 1 vs 8 +/- 2 L, P<.05), required less Starling curve intervention (42% vs 15%, P<.05), and maintained the DO(2)I goal with average pulmonary capillary wedge pressure that was less than that of the male cohort. CONCLUSION: Review of prospective data from standardized shock resuscitation for female and male cohorts demonstrates that women respond better to standardized resuscitation compared with similarly severely injured men.