Study of an Electronic Health Record-embedded Severe Sepsis Early Warning Alert

• ClinicalTrials.gov Identifier: NCT02376842
• Recruitment Status: Completed
• First Posted: March 3, 2015
• Last Update Posted: November 17, 2015
• Sponsor: Stanford University
• Information provided by (Responsible Party): N. Lance Downing, MD, Stanford University

Tracking Information

• First Submitted Date: October 3, 2014
• First Posted Date: March 3, 2015
• Last Update Posted Date: November 17, 2015
• Study Start Date: November 2014
• Actual Primary Completion Date: March 2015 (Final data collection date for primary outcome measure)

Current Primary Outcome Measures (submitted: February 25, 2015)

Percentage of patients with an antibiotic order within 3 hours of the alert [ Time Frame: 3 hours ]

Time from when the alert fires until appropriate antibiotics are ordered will be measured via the electronic health record and a sample of cases will be verified by manual chart review.

• Original Primary Outcome Measures: Same as current
• Current Secondary Outcome Measures: Not Provided
• Original Secondary Outcome Measures: Not Provided
• Current Other Pre-specified Outcome Measures: Not Provided
• Original Other Pre-specified Outcome Measures: Not Provided

Descriptive Information

• Brief Title: Study of an Electronic Health Record-embedded Severe Sepsis Early Warning Alert
• Official Title: Single-blind Randomized Trial of a Commercially Sold Electronic Health Record Based Severe Sepsis Early Warning Best Practice Alert.
• Brief Summary: The investigators hypothesize that implementing an electronic health record-based early warning system for severe infections (severe sepsis) will decrease the time to antibiotic order. The study will consist of an algorithm which will monitor lab values, vital signs, and nursing documentation for signs of severe sepsis. When these criteria are met, an alert will be delivered via the electronic health record to a nurse and doctor and simultaneously an alert via pager to another nurse. The investigators plan to randomize which patients will generate these alerts and analyze the data after collecting information for approximately 6 months which will be sufficient to detect a 10% difference in the two patient groups.

Detailed Description

Sepsis is the leading cause of mortality at Stanford Hospital and ranks only 54th out of 119 hospitals according to UHC data with approximately 60 episodes of documented sepsis per quarter. Based on some preliminary data, there is concern that sepsis is both being recognized late and not treated in a timely enough fashion. In fact, there are evidence and expert guidelines that suggestion-delaying antibiotics in a patient with septic shock can increase mortality by 6.7% per hour (1C recommendation in severe sepsis by the Surviving Sepsis Campaign authors). As part of a hospital wide initiative to improve our treatment of sepsis and ultimately reduce sepsis-related mortality, an EHR-based clinical decision support (aka BPA) will be implemented. This BPA will be an algorithm that will alert practitioners and trigger clinical workflow after criteria are met. Criteria include lab values, vital signs and nursing flow sheet descriptions of perfusion (Table 1). The algorithm will alert when, in a 24 hour period, three criteria from the manifestation group, one criteria from the suspected infection group and one criteria from the organ dysfunction group. Note that one variable (eg creatinine > 2) can fulfill criteria in more than one group. Figure 1 contains details of proposed EHR workflow. After criteria are met, whomever is next in the chart with RN or MD user-type, will receive an interruptive alert via the EHR; simultaneously a page will automatically be sent by the EHR to a crisis nurse who will assess the patient and notify the primary MD and RN.

Electronic early warning systems and predictive analytic tools lack rigorous evaluation and standardization. There are literature demonstrating unintended consequences and even harms from the implementation of electronic health records and clinical decision support tools. As such, this is a situation of clinical equipoise in which it is unclear whether this quality improvement initiative will benefit patient care or not. To evaluate this question, the severe sepsis BPA will be initiated in a randomized fashion with each patient randomly assigned to either potentially generate this alert as described above or to generate this alert silently such that only quality improvement staff will be aware that criteria have been met via the EHR.

This is a randomized, single-blind prospective quality improvement study. Patients will be randomized by encounter to have the BPA visible or invisible during hospital admission. If visible, the alert will display to the primary nurse and physician and send a page to a crisis nurse when BPA criteria are met. If invisible, the alert will be triggered but will be invisible to the care team (only visible to quality improvement staff via the EHR)

• Inclusion criteria

  o Admitted to Stanford Hospital (inpatient or observation status) to any medical or surgical service for at least 24 hours during the period of the study

• Exclusion criteria

  o Admitted to an intensive-care level service (MICU, SICU, CVICU, CCU)

• While ideally we will make this available to ICU patients, the alert is likely to be far less specific in ICU patients and less clinically useful given the high level of care (eg 1:1 nursing and hourly vital signs).

  o Patient code status is DNR/C (comfort care only)

• These patients would not be appropriate to treat with antibiotics and aggressive care generally give the comfort care goals of care.

  o Emergency Department patients (may be included in the near future)

• Primary endpoints

  o Percentage of patients receiving antibiotics within three hours

• Secondary endpoints

  • Percentage of patients with hypotension or lactate >= 4 who received at least 30 ml/kg fluid
  • Rate of 3 hour sepsis bundle completion (serum lactate, blood cultures, antibiotics)
  • Length of hospital stay
  • Cost of hospital stay per day and overall cost
  • Mortality at hospital discharge

• Sample size and duration

  • Based on prior studies, we predict a 10-15% difference in one or more primary endpoints (see Sawyer et. Al, Crit Care Med. 2011 March, 39:469)
  • We therefore anticipate enrolling 1500 patients, an estimated 150 of whom will have true sepsis, over approximately six months, to achieve 80% power.
• Analysis will compare endpoints among all patients in the treatment and control groups
• Early stopping will be decided by an oversight committee which will review data at 3 months. If there is statistically significant different of more than 10% between groups, the study will be terminated early and in discussion with hospital quality and safety committee, use whichever strategy is superior for all patients in the hospital.

• Study Type: Interventional
• Study Phase: Not Applicable
• Study Design: Allocation: Randomized; Intervention Model: Parallel Assignment; Masking: Single (Care Provider)
• Condition: Severe Sepsis

Intervention

• Behavioral: Severe sepsis early warning best practice alert
• Behavioral: Standard care

Study Arms

• Active Comparator: Severe sepsis early warning best practice alert
  Patients in this arm will actively generate the alert.
  Intervention: Behavioral: Severe sepsis early warning best practice alert
• Placebo Comparator: Standard care
  This arm will be the current standard of care and will not generate the alert.
  Intervention: Behavioral: Standard care

Publications *

• Westphal GA, Koenig A, Caldeira Filho M, Feijo J, de Oliveira LT, Nunes F, Fujiwara K, Martins SF, Roman Goncalves AR. Reduced mortality after the implementation of a protocol for the early detection of severe sepsis. J Crit Care. 2011 Feb;26(1):76-81. doi: 10.1016/j.jcrc.2010.08.001. Epub 2010 Oct 30.
• Nelson JL, Smith BL, Jared JD, Younger JG. Prospective trial of real-time electronic surveillance to expedite early care of severe sepsis. Ann Emerg Med. 2011 May;57(5):500-4. doi: 10.1016/j.annemergmed.2010.12.008. Epub 2011 Jan 12.
• Brandt BN, Gartner AB, Moncure M, Cannon CM, Carlton E, Cleek C, Wittkopp C, Simpson SQ. Identifying severe sepsis via electronic surveillance. Am J Med Qual. 2015 Nov-Dec;30(6):559-65. doi: 10.1177/1062860614541291. Epub 2014 Jun 26.

Recruitment Information

• Recruitment Status: Completed
• Actual Enrollment (submitted: November 16, 2015): 1149
• Original Estimated Enrollment (submitted: February 25, 2015): 1500
• Actual Study Completion Date: March 2015
• Actual Primary Completion Date: March 2015 (Final data collection date for primary outcome measure)

Eligibility Criteria

Inclusion Criteria:

• Admitted to Stanford Hospital (inpatient or observation status) to any medical or surgical service for at least 24 hours during the period of the study

Exclusion Criteria:

• Admitted to an intensive-care level service (MICU, SICU, CVICU, CCU)
• Patient code status is DNR/C (comfort care only)
• Patients less than 18 years of age at time of admission.
• Emergency Department patients (may be included in the near future)

Sex/Gender

• Sexes Eligible for Study: All

• Ages: 18 Years and older (Adult, Older Adult)
• Accepts Healthy Volunteers: No
• Contacts: Contact information is only displayed when the study is recruiting subjects
• Listed Location Countries: United States

Administrative Information

• NCT Number: NCT02376842
• Other Study ID Numbers: 3675309
• Has Data Monitoring Committee: No
• U.S. FDA-regulated Product: Not Provided
• IPD Sharing Statement: Not Provided
• Current Responsible Party: N. Lance Downing, MD, Stanford University
• Original Responsible Party: Same as current
• Current Study Sponsor: Stanford University
• Original Study Sponsor: Same as current
• Collaborators: Not Provided
• Investigators: Not Provided
• PRS Account: Stanford University
• Verification Date: November 2015