Healthcare Failure Mode and Effect AnalysisSM Edward J. Dunn, MD, MPH VA National Center for Patient Safety edward.dunn@med.va.gov www.patientsafety.gov Location in our VA NCPS Curriculum Toolkit Content - Patient Safety Introduction - Human Factors Engineering Instructor Preparation -Swift and Long Term Trust - “Selling the Curriculum” - Etc… -HFMEA ppt & exercise Alternative Education Formats - Pt Safety Case Conference (M&M) - Pt Safety on Rounds (Modulettes) - HFMEA participation - Etc… Why use prospective analysis? Aimed at prevention of adverse events Doesn’t require previous bad experience (patient harm) Makes system more robust JCAHO requirement JCAHO Standard LD.5.2 Effective July 2001 Leaders ensure that an ongoing, proactive program for identifying risks to patient safety and reducing medical/health care errors is defined and implemented. Identify and prioritize high-risk processes Annually, select at least one high-risk process Identify potential “failure modes” For each “failure mode,” identify the possible effects For the most critical effects, conduct a root cause analysis Who uses failure mode effect analysis? Engineers worldwide in: Aviation Nuclear power Aerospace Chemical process industries Automotive industries Has been around for over 40 years Goal has been, and remains, to prevent accidents from occurring Healthcare Version - HFMEASM Combines: – Traditional Failure Mode Effect Analysis – Hazard Analysis and Critical Control Point – VA Root Cause Analysis Adapted and Tested in Healthcare Settings – 163 VA hospitals (with some success) – Still a complex process/time commitment (see NIH) The Healthcare Failure Mode Effect Analysis Process Step 1- Define the Topic Step 2 - Assemble the Team Step 3 - Graphically Describe the Process Step 4 - Conduct the Analysis Step 5 - Identify Actions and Outcome Measures HFMEATM Hazard Scoring Matrix Severity Probability Catastrophic Major Moderate Minor Frequent 16 12 8 4 Occasional 12 9 6 3 Uncommon 8 6 4 2 Remote 4 3 2 1 Does this hazard involve a sufficient likelihood of occurrence and severity to warrant that it be controlled? (e.g. Hazard Score of 8 or higher) HFMEATM Decision Tree NO YES Is this a single point weakness in the process? (e.g. failure will result in system failure) (Criticality) NO YES Does an Effective Control Measure exist for the identified hazard? YES STOP NO Is the hazard so obvious and readily apparent that a control measure is not warranted? (Detectability) NO PROCEED TO HFMEA STEP 5 YES ICU Alarm Example Monitoring Patient Alarms in ICU Isolation Room 1 2 3 4 Patient is being Transferred to ICU Isolation Room Connect to necessary physiological monitor and equipment Provide care and monitor Alarms Intervene as appropriate Sub Process Steps A. Periodically check monitor status B. Respond to alarms Sub Process Steps A. Verify validity of alarm B. Reconnect equipment (if necessary) C. Medically intervene (if necessary) D. Silence alarm E. Readjust alarm parameters (if necessary) Sub Process Steps A. Apply transfer acceptance checklist B. Determine type of isolation and post C. Determine parameters to be monitored D. Gather and calibrate monitor and accessories (e.g. transducers) Sub Process Steps A. Don Personal Protective Equipment B. Connect to ventilator if appropriate C. Connect monitoring devices to patient D. Set Alarm parameters as appropriate E. Test Alarm Broadcast ICU Alarm Example 3A 3B Periodically check monitor status Respond to alarms Failure Modes 3A1 Did not check status 3A2 Misread or misinterpret 3A3 Partially check Failure Modes 3B1 Did not respond 3B2 Respond slowly or late ICU Alarm Example HFMEA Subprocess Step: 3B1 - Respond to Alarms HFMEA Step 4 - Hazard Analysis 3B1e Didn't hear alarm; remote location (doors closed to isolation room) Caregiver busy; alarm does not broadcast to backup Y C N Y E 16 N N Y C N Person Responsible Management Concurrence Action Type (Control, Accept, Eliminate) C N N Unw anted alarms on floor are reduced by 75% w ithin 30 days of implementation. Nurse Manager Y 12 12 Reduce unw anted alarms by: changing alarm parameter to fit patient physiological condition and replace electrodes w ith better quality that do not become detached Alarms w ill be broadcast to Central Station w ith retransmission to pagers provided to care staff. Yes Y Y C Alarms w ill be broadcast to the central station w ithin 4 months; complete by mm/dd/yyyy Set alarm volume on isolation Immediate; w ithin 2 room equipment such that the w orking days; low est volume threshold that can complete by be adjusted by staff is alw ays mm/dd/yyyy audible outside the room. See 3B1b See 3B1b Enable equipment feature that w ill alarm in adjacent room(s) to notify caregiver or partner(s). Immediate; w ithin 2 w orking days; complete by mm/dd/yyyy Yes Biomedical Engineer N Outcome Measure Yes Biomedical Engineer N N Proceed? Haz Score Probability Catastrophi Severity c Frequent Frequent Occasional 12 N N Actions or Rationale for Stopping Biomedical Engineer 3B1d Didn't hear; alarm volume too low 16 N Occasional 3B1c Didn't hear; care giver left immediate area 16 Frequent 3B1b Ignored alarm (desensitized) Occasional 3B1a Catastrophic 3B1 Don't respond to alarm Catastrophic Catastrophic Catastrophic Catastrophic Potential Causes Evaluate failure mode before determining potential causes HFMEA Step 5 - Identify Actions and Outcomes Decision Tree Analysis Single Point Weakness? Existing Control Measure ? Detectability Failure Mode: First Scoring Yes “Blow-up” of One Line Failure Mode: 3B1a - Crucial Alarm Ignored and Patient Decompensated Failure Mode Cause Severity Frequency Ignored alarm Catastr Frequent (desensitized) ophic Action Reduce unwanted alarms by: changing alarm parameter to fit patient physiological condition and replace electrodes with better quality that do not become detached Outcome Measure Unwanted alarms on floor are reduced by 75% within 30 days of implementation HFMEA & RCA Similarities Interdisciplinary team Develop flow diagram Systems focus Actions & Outcome measures Scoring matrix (severity/probability) Triage questions, cause & effect diag., brainstorming Differences Preventive v. reactive Analysis of Process v. chronological case Choose topic v. case Prospective (what if) analysis Detectability & Criticality in evaluation Emphasis on testing intervention