Validation of a Screening Tool for the Early Identification of Sepsis

Validation of a Screening Tool for the Early Identification of Sepsis
of 9
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  Validation of a Screening Tool for the Early Identification ofSepsis  Laura J. Moore, MD, Stephen L. Jones, MD, Laura A. Kreiner, MD, Bruce McKinley, PhD, Joseph F. Sucher, MD,S. Rob Todd, MD, Krista L. Turner, MD, Alicia Valdivia, RN, and Frederick A. Moore, MD Background:   Sepsis is the leadingcause of mortality in noncoronary intensivecare units. Recent evidence based guidelinesoutline strategies for the management of sepsis and studies have shown that earlyimplementation of these guidelines im-proves survival. We developed an extensivelogic-based sepsis management protocol;however, we found that early recognition of sepsis was a major obstacle to protocol im-plementation. To improve this, we devel-oped a three-step sepsis screening tool withescalating levels of decision making. We hy-pothesized that aggressive screening for sep-siswouldimproveearlyrecognitionofsepsisand decrease sepsis-related mortality by in-suring early appropriate interventions. Methods:   Patients admitted to the sur-gical intensive care unit were screened twicedaily by our nursing staff. The initial screenassessesthesystemicinflammatoryresponsesyndrome parameters (heart rate, tempera-ture, white blood cell count, and respiratoryrate) and assigns a numeric score (0–4) foreach. Patients with a score of   > 4 screenedpositive proceed to the second step of the toolinwhichamidlevelproviderattemptstoiden-tify the source of infection. If the patientsscreens positive for both systemic inflamma-tory response syndrome and an infection, theintensivist was notified to determine whetherto implement our sepsis protocol. Results:   Over 5 months, 4,991 screenswere completed on 920 patients. The prev-alence of sepsis was 12.2%. The screeningtoolyieldedasensitivityof96.5%,specificityof 96.7%, a positive predictive value of 80.2%, and a negative predictive value of 99.5%. In addition, sepsis-related mortalitydecreased from 35.1% to 23.3%. Conclusions:   The three step sepsisscreening tool is a valid tool for the earlyidentification of sepsis. Implementation of this tool and our logic-based sepsis proto-col has decreased sepsis-related mortalityin our SICU by one third. Key Words:   Sepsis, Sepsis screening,Evidence-based care, Sepsis protocol, Sur-gical sepsis.  J Trauma.  2009;66:1539–1547. S evere sepsis and septic shock are the leading causes of multiple organ failure and mortality in noncoronaryintensive care units (ICUs). 1 Additionally, sepsis islisted as the 10th leading cause of death by the Centers forDisease Control. It is estimated that in the United States, thereare 751,000 cases per year of sepsis with an annual cost of $17 billion. 2 By 2010, it is estimated that there will be934,000 cases per year. 2 Unfortunately, despite tremendousbasic and clinical research efforts, mortality from septicshock remains unchanged at  50%. 3 In an effort to improvesepsis-related mortality, several organizations have outlinedevidence-based guidelines (EBGs) for the management of severe sepsis and septic shock. 4–6 These EBGs provide acomprehensive list of recommended therapies for bedsideclinicians that if effectively implemented improve patientoutcome. 7,8 These EBGs include several time sensitive inter-ventions, emphasizing the importance of early sepsis recog-nition. The early identification of sepsis and implementationof early goal directed therapy has been shown to improveoutcomes and decrease sepsis-related mortality. 7 Administra-tion of empiric, broad spectrum antibiotic therapy is recom-mended within the first hour of recognition of severe sepsis orseptic shock. 4 Each hour of delay in the administration of antibiotic therapy is associated with an increased mortalityrate. 9 In addition, a recent study from the United Kingdomdemonstrated that failure to comply with a 6-hour resuscita-tion bundle resulted in a twofold increase in mortality whencompared with patients who did receive the 6-hour bundle. 10 Within our institution, we have identified sepsis to be amajor cause of morbidity and mortality in our surgical ICU(SICU) patients. In response to this, our multidisciplinarySICU team developed a computerized clinical decision sup-port (CCDS) tool to facilitate implementation of these EBGs.This group has previous experience with the use of CCDS forventilatory management of acute respiratory distress syndromeand the resuscitation of hemorrhagic shock patients. 11–14 Theuse of CCDS ensures implementation of the EBGs, decreasesvariability in management among clinicians, provides a stableplatform for clinical and translational research, and monitorsquality of care. 15 We have developed an evidence based,patient specific protocol for the management of patients withsepsis, severe sepsis, and septic shock. Submitted for publication October 10, 2008.Accepted for publication March 3, 2009.Copyright © 2009 by Lippincott Williams & WilkinsFrom the Division of Acute Care Surgery (L.J.M., S.L.J., B.K., J.F.S.,S.R.T., K.L.T., A.V., F.A.M.), Department of Surgery, The Methodist Hos-pital, Houston, Texas; Weill Cornell Medical College (L.J.M., K.L.T.,S.R.T., F.A.M.), New York, New York; and Department of Surgery(L.A.K.), University of Texas Health Science Center, Houston, Texas.Supported by The Methodist Hospital Research Institute, Houston, Texas.Presented at the 67th Annual Meeting of the American Association forthe Surgery of Trauma, September 24–27, 2008, Maui, Hawaii.Address for reprints: Laura J. Moore, MD, Department of Surgery, TheMethodist Hospital, Weill Cornell Medical College, 6550 Fannin Street,Smith Tower 1661, Houston, TX 77030; email: DOI: 10.1097/TA.0b013e3181a3ac4b The Journal of   TRAUMA    Injury, Infection, and Critical Care Volume 66   •  Number 6   1539  Implementation of our protocol for sepsis managementin mid-2007 encountered the unanticipated problem of un-timely and inaccurate recognition of sepsis by bedside clini-cians. Although the protocol included specific criteria toconfirm diagnosis and to discern early sepsis and severesepsis or septic shock, no mechanism had been provided tomonitor each ICU patient for onset or ongoing sepsis. Asbedside nurses and other team members focus on multiplepriorities and tasks, early signs of sepsis are often missed andinterventions are delayed. Need for routine, accurate screen-ing of all ICU patients for sepsis quickly became apparent. Inan attempt to increase the early identification of sepsis, wedeveloped a sepsis screening tool in our SICU. If the patientis screened positive for sepsis, the SICU intensivist is notifiedand decides whether to implement our sepsis protocol. Wehypothesized that aggressive screening for sepsis would im-prove early recognition of sepsis and decrease sepsis-relatedmortality by insuring early appropriate interventions. MATERIALS AND METHODS Study Site This retrospective observational study was performed inthe SICU at The Methodist Hospital which is an academictertiary referral hospital in The Texas Medical Center inHouston, TX. The SICU is a 31-bed unit that serves a diversegroup of patients including critically ill general, vascular,oncologic, transplant, thoracic, orthopedic, plastic, urologic,and head and neck surgical patients. All patients admitted tothe SICU were candidates for sepsis screening using ourthree-step sepsis screening tool. We started a multidisci-plinary ICU team in August 2006, and during the studyperiod, the SICU was an open unit where our team wasconsulted at the discretion of the admitting physician. SICU Sepsis Protocol After starting our multidisciplinary SICU team, wequickly recognized that sepsis was a major problem in ourSICU and focused our efforts on developing an evidencebased protocol to manage these patients. This protocol wasdeveloped by a sepsis working group that met on a weeklybasis. After several months of work, this group produced a“paper protocol” that was ready for implementation at thebedside. This protocol is divided into two phases. Phase 1 isfocused on patients who present with early signs of sepsis andinvolves basic interventions including obtaining blood cul-tures, administration of antibiotics within 1 hour, giving afluid challenge, basic laboratory testing, and monitoring forsigns of worsening sepsis. Phase 2 is focused on patients whopresent with severe sepsis or septic shock and involves esca-lating interventions. This treatment plan requires placementof a central line capable of measuring central venous hemo-globin oxygen saturation (ScvO 2 ) and an arterial line formonitoring, repeat labs every 4 hour for 24 hours, and screen-ing for activated protein C. In addition, there is detailed logicthat directs fluid/blood therapy, vasopressors/inotrope ther-apy, diagnosis of adrenal insufficiency/steroid therapy, andpulmonary artery catheter directed therapy if needed. We usethe following three criteria: (a) hypotension (mean arterialpressure   65), (b) lactate   4 mmol/L, and (c) urine output  0.5 mL/kg over past 1 hour to triage septic patients. Thesecriteria were derived fromthe literatureand localexpert opinion.Patients without one of these criteria are triaged into phase 1 of the protocol, whereas patients (new or monitored phase 1) withone or more triage criteria are triaged into phase 2.In January 2007, we implemented the protocol at thebedside under direct supervision by a physician member of the multidisciplinary team. Over the course of the next 4months, we revised the protocol based upon a case by caseanalysis of the process and ongoing discussion with the sepsisworking group. By April 2007, we had a functioning protocolwhich was instituted as our standard of care. With implemen-tation of the sepsis management protocol, we recognized thatthe early detection of sepsis was problematic. In an attempt toincrease the early recognition of sepsis in our SICU, wedeveloped a sepsis screening tool. SICU Sepsis Screening Tool The SICU sepsis screening tool was developed fromliterature evidence and expert consensus. The design strategywas to involve the bedside clinician team (i.e., bedside nurse,nurse practitioner or resident physician, and surgical inten-sivist) in a routine screen process for all SICU patients, but tominimize data gathering tasks. This was done by using se-lected current physiologic and clinical laboratory measure-ments routinely obtained by the bedside nurse and, whenthose measurements exceeded agreed thresholds, by request-ing information specifying possible infection and source de-rived from focused clinical observation and judgment of thenurse practitioner or resident physician. With this informa-tion, the surgical intensivist physician was required to estab-lish the diagnosis of sepsis, severe sepsis or septic shock, andto order start of our protocol directed sepsis management, ordiagnosis of a disease other than sepsis.The screen system developed is therefore based on a ruleset with numeric thresholds and escalation of decision mak-ing for diagnosis of sepsis, severe sepsis, septic shock, or adisease other than sepsis. Systemic inflammatory responsesyndrome (SIRS; defined by temperature [T], heart rate [HR],respiratory rate [RR], and white blood cell count [WBC]) andclinical suspicion of infection are the basis of the screensystem. SIRS severity is assessed by the bedside nurse onceper 12 hour shift by adding scores for current T, HR, RR, andWBC (from measurement ranging 0 [normal] to 4 [extremelyderanged]). The numeric ranges for these variables wereadapted from the Acute Physiology and Chronic Health Eval-uation score system. 16 SIRS score   4 was agreed as thethreshold indicating severe SIRS. A simple format was de-vised to record this data and to determine SIRS severity (seeFig. 1). This part of the sepsis screen form was determined torequire 2 minutes to 3 minutes of bedside nurse time to The Journal of   TRAUMA    Injury, Infection, and Critical Care 1540  June 2009  complete and, as part of data reconciliation and confirmation,was determined to be completed most efficiently at the end of each 12 hour nursing shift. Twice daily screening of recog-nized systemic indicators of stress caused by infection istherefore obtained, and all SICU bedside nurses activelyparticipate in decision making for recognition of sepsis.Escalation to the second step in decision making occurswhen the bedside nurse notifies the midlevel provider (i.e.,nurse practitioner or resident physician) that a patient has aSIRS score is   4 (severe SIRS). The midlevel providerassesses for possible source(s) of infection (invasive line,pneumonia, abdomen, soft tissue, urinary tract, or other). Asimple format was devised to record these data (see Fig. 2).This part of the sepsis screen form was determined to require5 minutes to 10 minutes of nurse practitioner or residentphysician time to complete, and, if a source was suspected, toprovide a quick reference for the surgical intensivist.If a patient screens positive for severe SIRS and possibleinfection, then the mid levels provider notifies the SICUintensivist to convey the current screen form with currentrecords of SIRS data and of possible infection source infor-mation to further escalation of decision making. In this thirdstep of the screening process, the surgical intensivist is re-quired to review the current form, assess severe sepsis/septicshock criteria, record a diagnosis, and, if criteria are met,initiate sepsis management protocol directed treatment. Forpatients who are not followed by our SICU team, the attend-ing physician was notified of the positive SIRS screen anddirected care as they thought appropriate.The screen system is implemented using a one page paperform (front and back sides) designed to be used easily androutinely by the clinician team to grade severity of physiologicderangement possibly caused by sepsis, to identify a possiblesource(s) of infection, to alert possible onset, and to establishdiagnosis of sepsis, severe sepsis, or septic shock.We started the screening process in May 2007 and overthe next 5 months; the bedside nurses identified 163 patientswho had a SIRS score  4. The nurse practitioner or residentphysician screen for possible infection was positive in 64 of these patients. The intensivists confirmed a sepsis diagnosisin 45 of these patients of which 30 were triaged and managedby phase 2 of our protocol. The mortality rate for this group of patients was 24%. 17 This initial experience with the screeningtooldemonstratedthatitwasfeasibletocompletethetwicedailyscreen and that the screening tool appeared to identify patients atrisk for sepsis. However, further validation of the tool was stillnecessary and the purpose of this study. Data Collection In September 2007, we began prospectively collectingdata from our sepsis management protocol patients for entryinto our Institutional Review Board approved sepsis databaseand collecting all of the sepsis screens performed in the SICUfor validation of the screening tool. We collected and ana- Fig. 1.  SICU bedside nurse SIRS score. Validation of a Screening Tool for Sepsis Volume 66   •  Number 6   1541  lyzed all of the screens completed from September 1, 2007,through January 31, 2008, for this study. Demographic datawere obtained for all screened patients for the current hospitalICU encounter. Patient chart review and demographic dataacquisition were approved by The Methodist Hospital Re-search Institute Institutional Review Board. Patients were cat-egorized in four groups: (1) true positive (positive screen withsepsis-related diagnosis recorded in hospital billing data), (2)true negative (negative screen without sepsis-related diagnosisrecorded in hospital billing data), (3) false positive (positivescreen without sepsis-related diagnosis recorded in hospitalbilling data), and (4) false negative (negative screen withsepsis-related diagnosis recorded in hospital billing data).The medical records were reviewed to determine the presenceof sepsis in all patients who had a positive SIRS screen andthose with a negative SIRS screen but a sepsis-related diag-nosis in hospital billing data (International Statistical Classi-fication of Diseases and Related Health Problems [ICD-9]codes). Table 1 depicts the diagnoses used to identify sepsisby ICD-9 codes. As mentioned previously, this screeningprocess was performed on all patients in an open SICU andnot all patients were under the care of our multidisciplinary Fig. 2.  SICU nurse practitioner/resident physician sepsis screening. The Journal of   TRAUMA    Injury, Infection, and Critical Care 1542  June 2009  SICU team. As a result, there were 24 patients that had apositive SIRS screen that did not have the second and thirdsteps of the screen completed. Upon review of the hospitalbilling data for these particular patients, none had a sepsisdiagnosis. Therefore, for the purposes of this analysis, thesepatients were categorized as false positives. Mortality Assessment As part of The Methodist Hospital’s Care Managementperformance improvement (PI) programs, sepsis-related out-comes (including mortality) for all of the hospital’s ICUswere independently monitored using a tool devised by theInstitute for Healthcare Improvement starting in January2006. It is important to note the PI mortality data were forentire years of 2006 and 2007, whereas our current screeningdata covers a 5-month period. Additionally, our screeningtool was devised to identify sepsis, severe sepsis, and septicshock, whereas the PI data were for patients meeting criteriafor severe sepsis and septic shock. Statistics Sensitivity and specificity of the three-step sepsis screen-ing tool were calculated using a standard two by two table.Sensitivity was calculated by dividing the number of truepositives by the sum of the true positives and false negatives.Specificity was calculated by dividing the number of truenegatives by the sum of the true negatives and false positives. RESULTS During 5 consecutive months starting September 1,2007, 4,991 sepsis screens were completed on 920 patientsrepresenting 927 ICU admissions. The distribution of theSIRS scores from all of screens is shown in Figure 3. Thedemographic data for all of the screened patients and the sub-group of patients who developed sepsis are shown in Table 2.Among the patients who screened positive for sepsis, therewas a slight predominance of men (55% vs. 45%) and olderadults (54% vs. 49% older age 60 years). There was a slightlyhigher ratio of African Americans among the sepsis-positivepatients (25%) versus the population as a whole (22%). The Table 1  International Statistical Classification of Diseasesand Related Health Problems (ICD-9) Codes Used toIdentify Sepsis ICD-9Code Diagnosis 38 Septicemia38.4 Septicemia caused by other gram-negative organisms38.49 Other septicemia caused by gram-negative organism38.9 Unspecified septicemia670 Major puerperal infection670 Major puerperal infection771.81 Septicemia (sepsis) of newborn785.5 Unspecified shock785.52 Septic shock (e.g., endotoxic, gram negative)785.59 Other shock without mention of trauma (hypovolemic,septic)995.9 Systemic inflammatory response syndrome (SIRS)995.91 Systemic inflammatory response syndrome caused byinfectious process without organ dysfunction995.92 Systemic inflammatory response syndrome caused byinfectious process with organ dysfunction SCORE STRATA 05001000150020002500 SCORE      S     C     R     E     E     N     S Screens 2093 1083 921 533 183 120 33 15 4 2 30 1 2 3 4 5 6 7 8 9 10 SIRS, systemic inflammatory response syndrome; for all abbreviations in the table. . . . Fig. 3.  Distribution of SIRS scores. Validation of a Screening Tool for Sepsis Volume 66   •  Number 6   1543
Similar documents
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks