1 Creating High Reliability to Reduce Patient Harm Florida State University College of Medicine Center for Medicine and Public Health Tallahassee, FL Grand Rounds March 1, 2012 William Riley, Ph.D. Associate Dean School of Public Health University of Minnesota 1
2 Objectives 1. Define High Reliability for health care and describe the components. 2. Discuss the etiology of an organizational accident (sentinel event) in health care 3. Explain the nature of health care teams and impact on patient safety 4. Appreciate the importance of process design for patient safety 2
3 Acknowledgements I would like to express gratitude to the Robert Wood Johnson Foundation for funding The Evidence Base for Quality Improvement and the Public Health Practice Base Research Networks for supporting the Science of Improvement.
4 Overview of High Reliability Patient injury is at epidemic levels. IHI estimates 15 million instances of medical harm occur each year (IHI, 2007). Processes are not deliberately designed for defenses in depth. Professional training is not synonymous with team training. Current models focus on preventing error by individuals when the bulk of care is given by teams. (van der Schaaf, 2002). 4
Phase I 5 Patient Safety Safety is freedom from unintended harm Injury caused by: Unsafe Acts System Complexity 5
6 Unsafe Acts Errors The failure of a planned action to achieve desired goal Violations A deliberate deviation from safe practices and rules Negligence Failure to use degree of care required by law to protect others from harm 6
7 Medical Error The failure of a planned action to be completed as intended (error of execution) or the use of a wrong plan to achieve an aim (error of planning). Can be error of commission (prescribe wrong drug), or error of omission (failing to provide drug from which patient would benefit). Adverse Event Unintended harm to the patient by an act of commission or omission rather than the underlying disease of condition. Results in at least 1 additional day LOS. Source: IOM, Patient Safety, 2004 7
8 Relationship between Error and Adverse Event Medical Error Near Miss Adverse Reaction Adverse Event 8
9 High Reliability Organizational design principles and management approaches that prevent patient injury and improve quality(riley, 2009) Characteristics of a high-reliability health organization (HRO) include an organizational commitment to safety, back-up steps built into processes, safety measures, and a culture of continuous Quality Improvement(Riley et al 2010)
Accident Investigation Shows 10 Best people can make worst mistakes Greatest calamities can happen to the most conscientious and well run organizations Latent conditions are always present. Goal is to identify recurring patterns in the way defenses fail. 10
11 Components of High Reliability High Reliability = Technical Skills + Non-technical skills + Designed Processes Culture of Safety Source: Riley, Davis and Miller, 2010
Overview of Health Care Teams 12 Teams are stable a myth. Leadership is constant unsupported by evidence. Leadership transfer is performed poorly. 12
13 Teams and High Reliabilty Fundamental feature of HRO s: defensive barriers, protect, and prevent. Current health services research approaches may be inadequate to meet the needs and problems associated with patient safety (Battles and Lilford, 2003). The present culture of naming, blaming and shaming is counterproductive and anachronistic. 13
14 Defenses in Depth Model Defenses Hazards Losses 14
Phase II 15 Defensive Weaknesses Some defensive weaknesses will always be present from the very beginning. Local factors combine with human condition to produce errors and violations. Large numbers of unsafe acts will be made. 15
16 Swiss Cheese Model Defenses Hazards Losses
17 Active Failure (Sharp End) An unsafe act committed by those at the patient/provider interface. Damaging consequences are immediately apparent Latent Condition (Blunt End) Created as a result of decisions at higher echelons of organizations. Damaging consequences may lie dormant for long time until triggered by local conditions. 17
18 "Breaches" That May Affect Safety Comm 23% Policy 24% Active End 50.2% Shared MM 15% S. A. 12% Equip 10% System 16% Latent Conditions 49.8% TOTAL BREACHES: 35 Simulations in 6 Hospitals AVE: 24.2 BREACHES/ SIMULATION
19 Defenses in Depth Swiss Cheese Model Defenses Hazards Losses Accident Trajectory 19
20 Health Care Team Communication failures responsible for preponderance of sentinel events. A team consists of two or more individuals, Who have specific roles, Perform interdependent tasks,are adaptable, and Share a common goal Team Stages: form, storm, norm, perform Salas, E., T. L. Dickinson, and S. A. Converse. 1992. "Toward an Understanding of Team Performance and Training." In Teams: Their Training and Performance, edited by R. W. Swezey and E. Salas, pp. 3 29. : Ablex. 20
21 The Nature of Critical Event Teams, Quality Control, and Variability: Obstetricians 93 L&D Nurses 50 Anesthesiologists 16 NNPs 12 Scrub Techs 14 CRNAs 35 Reliability How many C/S teams are possible with these staff numbers? 437.5 Million 21
22 Health Care Teams Conventional View a team has established hierarchy and known leadership with stable composition with extensive training together, with many team performances over long periods of time New View: (For Certain types of teams) Teams rarely train together Team building processes, such as establishing team performance norm, building personal trust, socialization, are abbreviated or absent. The team comes together for a specific purpose, extremely short duration, likelihood of same team again is non-existent. Team members come from separate discipline with diverse educational programs 22
23 Team Leadership That person who is physically present and performs three specific tasks: 1)prioritizes decisions, 2) co-ordinates activity of other team members, and 3) communicates a shared mental model for the other team members. 23
24 Leadership Transfer The handoff of leadership from one member of the team to another member. The leadership transfer occurs (usually) when a new person physically joins the team and assumes the three specific leadership tasks. The leadership transfer occurs when the existing leader explicitly concedes leadership to the new leader, and the new leader explicitly assumes the leadership roles. The leadership role during a critical event is dependent on the phase of the group formation during the critical incident and the specific role of the team member rather than the authoritarian hierarchy. 24
25 Team Formation Team Formation: a group of persons with special expertise assemble to execute a specific task. Can be a critical event or more routine event. During a critical event participants continually flow in and out of the team. The team is not stable and leadership is not constant. 25
26 Team Reformation Team reformation is defined as instances when the team membership changes in a significant way by the addition or deletion of team members directly involved in the critical incident. Team formation and reformation is characterized by recurrent stages that occur in predictable patterns during a critical event. Most of the leadership transfer occurs concurrently with the stages of team formation and reformation. The nature of the tasks is different in each of the phases. 26
27 Joint Commission Article TeamSTEPPS intervention at three community hospitals, totaling 1800 deliveries per year over a four year period(2005-2008). Measured Weighted Adverse Outcomes Score (WAOS) for all deliveries Interdisciplinary team training and simulations were utilized in delivery units to improve communication and teamwork 37% decrease in perinatal harm
28 A TEAM OF EXPERTS IS NOT AN EXPERT TEAM 28
29 PHASE III Premier Perinatal Safety Initiative January 2008-December 2010 16 hospitals in 12 states 18 comparison group hospitals Introduced three clinical care bundles Interventions: Bundle Compliance Harm Measures: AOI, WAOS, SI, Association between Interventions and Harm Measures 29
30 Participating Hospitals in the Initiative Illinois -Methodist Medical Center of Illinois (Peoria, IL) Kentucky -Baptist Hospital East, Baptist Healthcare System (Louisville, KY) Massachusetts -Baystate Medical Center, Baystate Health (Springfield, MA) Minnesota -Fairview Ridges Hospital, Fairview Health System (Burnsville, MN) -Univ. of Minnesota Medical Center, Fairview, Fairview Health System (Minneapolis, MN) New Mexico -Presbyterian Hospital, Presbyterian Healthcare Services (Albuquerque, NM) Ohio -Bethesda North Hospital, TriHealth (Cincinnati, OH) Ohio (continued) -Good Samaritan Hospital, TriHealth (Cincinnati, OH) -Summa Health System, Akron City Hospital (Akron, OH) Tennessee -Indian Path Medical Center, Mountain States Health Alliance (Kingsport, TN) Texas -Texas Health Harris Methodist Fort Worth Hospital (Forth Worth, TX) -Texas Health Presbyterian Hospital of Dallas (Dallas, TX) Washington -St. Joseph Hospital, PeaceHealth (Bellingham, WA) Wisconsin -West Allis Memorial Hospital, Aurora Health Care (West Allis, WI) 30
31 (1) Interventions: Bundle Compliance 31
32 Definition of Bundles Elective Induction Augmentation Vacuum Gestational age 39 weeks Normal Fetal status (per NICHD tiers) prior to onset of Oxytocin Pelvic exam prior to the onset of Oxytocin Recognition and management of tachysystole Documentation of estimated fetal weight Normal fetal status (per NICHD tiers) Pelvic exam prior to the onset of Oxytocin Recognition and management of tachysystole Alternative labor strategies considered Patient prepared High probability of success Maximum application time and # of pop-offs predetermined and documented Cesarean and resuscitation teams available at delivery 32
33 Induction Bundle Compliance Conformity with the elective induction bundle increased from 40 percent to a peak of 91 percent in October of 2010. Phase I compliance increased 52 percent. 33
Induction Bundle Individual Hospitals (n=6) with Criterion Compliance 34 34
35 (2) Harm Measures: AOI, WAOS, SI, 35
Adverse Outcome Index: 36 10 Indicators of Harm Complication Maternal death 750 Intrapartum neonatal death of a neonate > 2500 grams (excluding 400 cases with a congenital anomaly or fetal hydrops) Uterine rupture 100 Unexpected internal or external maternal transfer to an ICU for a 65 postpartum complication Birth trauma 60 Return to OR or labor and delivery 40 Admission of neonate > 2500 grams and > 37 weeks to NICU within 35 one day of birth for > 24 hours (excluding cases with a congenital anomaly or fetal hydrops) APGAR 5 < 7 (excluding cases with a congenital anomaly or fetal 25 hydrops) Maternal blood transfusion 20 3rd or 4th degree perineal laceration. 5 Weight 36
Incremental Change: 37 Intervention vs Comparison Baseline (Q1 2006 Q4 2007) Follow-up (Q3 2008 Q2 2010) Incremental Change Collaborative Comparison Collaborative Comparison Collaborative Comparison AOI 53.6 55.0 50.5 55.0-3.1* 0 WAOS 1.18 1.30 1.18 1.39-0.0 +0.08 SI 22.10 22.95 23.27 24.19 +1.17* +1.24 *Significant change AOI is presented per 1,000 deliveries 1) 3.1 fewer adverse events for every 1,000 births 2) The Collaborative outpeformed the Comparison on all three indicies
Trending in Obstetrical Harm 38 Measures: Adverse Outcome Index AOI Rate per 1,000 Deliveries 60.00 58.00 56.00 54.00 52.00 50.00 48.00 46.00 44.00 Q1 2006 Q2 2006 Q3 2006 Aggregate Trending of AOI through Phase I: Run Chart Q4 2006 Baseline Q1 2007 Q2 2007 Aug/Ind bundles initiated Q3 2007 Q4 2007 Q1 2008 Q2 2008 Q3 2008 Q4 2008 Median = 51.7 Q1 2009 Q2 2009 Follow-up Vacuum bundle initiated Q3 2009 Q4 2009 Q1 2010 Q2 2010 Q3 2010 Q4 2010 38
39 Item Analysis of AOI Scale: Collaborative Baseline vs. Follow-up 4.7 fewer lacerations for every 1,000 births 1.39 fewer NICU admissions for every 1,000 births 1 fewer birth trauma events for every 3,226 births 1 more APGAR 5<7 for every 1,408 births 39
40 (3) Association between Interventions and Harm Measures 40
Augmentation, Induction 41 Bundles: Multivariate Regression Analysis Increased Augmentation bundle compliance appears to be associated with: a decrease in AOI, WAOS, and SI scores a decrease in adverse events fewer complications of anesthesia, maternal hypertension, hypoxia and birth asphyxia. Increased Induction bundle compliance appears to be associated with fewer respiratory problems 41
Phase IV 42 AHRQ Phase IV AHRQ patient safety and medical liability grant awarded in 2010 to: University of Minnesota School of Public Health Fairview Health System Premier healthcare alliance/pims