The Development of Emergency Department Patient Quality/Safety Indicators

On September 21, 2011, Patrick S. Romano made this presentation at the 2011 Annual Conference. Select to access the PowerPoint® presentation (3 MB). Plugin Software Help.

Slide 1

The Development of Emergency Department Patient Quality/Safety Indicators

Patrick S. Romano, MD MPH
UC Davis Center for Healthcare Policy and Research
AHRQ Annual Conference
September 21, 2011

Slide 2

Overview

• Goals and scope of current project.
• Conceptual frameworks.
• Literature review.
• Matrix of potential indicators.
• Specification.
• Preliminary results.
• Recommendations and next steps.

Slide 3

AHRQ Quality Indicators (QIs)

• Developed through contract with UCSF-Stanford Evidence-based Practice Center & UC Davis, maintained and extended through contract with Battelle.
• Use existing HCUP (hospital discharge) data, based on readily available data elements.
• Incorporate a range of severity adjustment methods, including APR-DRGs* and comorbidity groupings.
• Disseminate software and support materials free via http://www.qualityindicators.ahrq.gov.
• Provide technical support to users.
• Continuous improvement through user feedback, annual coding updates, validation projects.

* All Patient Refined—Diagnosis Related Groups

Slide 4

Evidence-based Indicator Development

• Literature review user suggestions.
• Initial empirical analyses and definition.
• Panel evaluation (modified Delphi Process).
• Further empirical analyses refined definition.
• Further review? Final definition.

Slide 5

The HCUP Partnership: A Voluntary Federal-State-Private Collaboration

40+ states.
90% of all discharges.
24+ states submit ED encounters.

Image: A map of the United States is shown.

Slide 6

Types of HCUP Databases

State Inpatient Databases (SID).

• Nationwide Inpatient Sample (NIS).
• Kids' Inpatient Database (KID).
• Nationwide Emergency Department Sample (NEDS).

State Ambulatory Surgery Databases (SASD).

State Emergency Department Databases (SEDD).

Slide 7

Goals and Scope

• Goals:
  • Develop quality/safety indicators, using HCUP data, that are applicable to the emergency department setting of care.
  • Set the stage for future incorporation into publicly available AHRQ QI software.
• Scope:
  • Implement the established AHRQ QI measure development process.
  • Adapt existing AHRQ QI to ED setting when possible.
  • Identify and evaluate new candidate indicators based on established measurement concepts.

Slide 8

IOM Committee on the Future of Emergency Care in the United States Health System (2007)

• Called for "a standard national approach to the development of performance indicators."
• "The measures developed should include structure and process measures, but evolve toward outcome measures over time... (and) should be nationally standardized so that comparisons can be made across regions and states."
• "Measures should evaluate the performance of individual providers within the system, as well as that of the system as a whole... (and) be sensitive to the interdependence among the components of the system."
• "Performance data should be collected on a regular basis from all of the emergency care providers in a community" and then publicly disseminated..."

Slide 9

Conceptual framework for prioritization:
Institute of Medicine, 2007

Slide 10

Conceptual Framework for Prioritization: Institute of Medicine, 2010

Image: The Conceptual framework for prioritization: Institute of Medicine, 2010 is shown.

Slide 11

Conceptual Framework for Prioritization: ICES/Alberta Quality Matrix for Health, 2010

Slide 12

Conceptual Framework for Prioritization: American College of Emergency Physicians, 2009

Slide 13

Application of Conceptual Framework

Slide 14

Literature review: strategy

Search goal:

• To find studies that introduced or used quality of care measures to assess patient safety in hospital emergency departments.

Search strategy using MESH headings in PubMed:

• ("Quality Assurance, Health Care"[Mesh] OR "Quality Indicators, Health Care"[Mesh] OR "Quality of Health Care"[Mesh] OR "Health Care Quality, Access, and Evaluation"[Mesh] OR "United States Agency for Healthcare Research and Quality"[Mesh] OR "Outcome Assessment (Health Care)"[Mesh])
  AND "Emergency Service, Hospital"[Mesh]
  AND ("Medical Errors"[Mesh] OR "Malpractice"[Mesh] OR "Safety"[Mesh] OR "Equipment Safety"[Mesh] OR "Safety Management"[Mesh]).

Validation using title and/or abstract keywords:

• "patient safety" OR "adverse event" OR "avoidable condition"
  AND "quality"
  AND ("emergency room" OR "emergency department").
• For the most important papers, we searched for 'all related articles.'

Slide 15

Literature Review: Process

• PubMed®:
  • 1,050 abstracts, decreased to 687 when limited to human subjects, English language, date within 10 yrs.
  • All abstracts were reviewed for relevance (i.e., describing one or more measures of ED quality/safety).
• Similar review by Alessandrini et al. for PECARN.
• Organizations and Web sites:
  • National Quality Measures Clearinghouse™ (AHRQ).
  • National Quality Forum.
  • Federal: AHRQ QIs and CMS/QualityNet.
  • ED: ACEP and SAEM.
  • AMA: Physician Consortium for Performance Improvement.
  • Other developers: NCQA and The Joint Commission.
  • Institute of Medicine/National Academy of Sciences.
  • Canada: Institute for Clinical Evaluative Sciences, Canadian Institute for Health Information.

Slide 16

Literature Review: Key Themes
40 Journal Papers, 23 Documents and Reports

• Some TJC Core Measures address processes of care in ED management of pneumonia or myocardial infarction.
• Critical trauma or shock care, generally based on detailed "peer" review of medical records to assess appropriateness and timeliness of diagnostic and therapeutic interventions.
• Time-based measures of waiting time, total LOS in the ED, ED disposition time for admitted/transferred patients.
• Appropriate prescribing and avoidance of medication errors for common conditions: asthma, bronchiolitis, acute GE, laceration.
• Appropriate use of imaging studies, laboratory, ECG.
• Appropriate assessment of pain, oxygenation, cognition.
• "Left without being seen" or "left AMA" (premature ED discharge).
• "Missed diagnosis" identified by return within defined time window for a serious condition.
• Revisits to ED within time window for same or related condition.

Slide 17

Matrix of Potential Indicators: Inclusion/exclusion Criteria

• Identified from published source:
  • Literature review (40 journal articles).
  • Organizations and Web sites (if a consensus-based approach and/or modified Delphi approach was used).
• Address the domains of effectiveness and/or safety:
  • A few measures of timeliness were included because the measure developer characterized them as having implications for safety in the ED.
• Focus on care provided in the ED (not pre-hospital care).
• Clinical guidelines, standards of care, and ED decision rules were not included unless operationalized as indicators.
• Can be implemented in at least one HCUP partner state using available HCUP data.
• When ≥2 indicators addressed the same outcome, only the more recent and/or more clearly specified indicator was retained.
• Excluded measures that were evaluated and discarded or rejected through a consensus-based expert panel process.

Slide 18

Matrix of potential indicators
Application of existing inpatient PSIs

• Foreign body left in.
• Iatrogenic pneumothorax.
• "Postoperative" hip fracture.
• "Postoperative" hemorrhage or hematoma.
• Accidental puncture or laceration.
• Transfusion reaction.

Slide 19

Matrix of potential indicators
35 new candidate indicators

• Age range:
  • 12 for children only.
  • 10 for adults only.
  • 13 for both children and adults.
• Donabedian's typology:
  • 11 process.
  • 17 outcome (or proxy outcome such as revisit).
  • 6 hybrid ("missed serious diagnosis").
  • 1 patient experience or health risk behavior ("left AMA").
• Developer(s):
  • 20 Institute for Clinical Evaluative Sciences, specified in ICD-10-CA.
  • 3 ACEP and/or PCPI.
  • 3 CMS.
  • 4 other organizations.
  • 5 researchers.
• Endorsement—6 endorsed by NQF.

Slide 20

Matrix of potential indicators
35 new candidate indicators

• Revisits—13:
  • 4 within 24 hours (1 specified as 24 hrs or 72 hrs).
  • 3 within 48 hours (2 specified as 48 hrs or 72 hrs).
  • 6 within 72 hours (1 specified as 72 hrs or 1 week).
• Missed serious diagnoses—7:
  • 1 unanticipated death within 7 days following ED care.
  • 6 admission for missed diagnosis (AMI/ACS, SAH, ectopic pregnancy, traumatic injury, appendicitis).
• Appropriate use of diagnostic test or imaging—5.
• Acute complications of ED procedures—3.
• Time within ED awaiting definitive care—3.
• Appropriate admission for inpatient care—2.
• Appropriate use of treatment or intervention—1.
• Left "against medical advice"—1.

Slide 21

Challenges in specification and testing with HCUP data

• Identification of patients "at risk":
  • What procedures place patients at risk for hemorrhage, pneumothorax, or accidental puncture/laceration?
• Timing:
  • Only 5 states (GA, MA, MN, NJ, TN) have POA in SEDD; only MA and TN also have PNUM.
  • In SID, POA means "present at the time the order for inpatient admission occurs" and ED dx may be lost.
• Low frequency with "true" frequency unknown:
  • Unable to choose "best" specification.
• Uncertain validity of utilization flag variables to identify patients who had specific procedures (US, ECG, CT).
• Unable to operationalize all specifications:
  • Exclusion of "planned" (or "invited") return visits to ED.
  • All presenting symptoms for "missed diagnoses."

Slide 22

Iatrogenic Pneumothorax 8 States, 2005-2008

Image: An sample chart titled:Iatrogenic pneumothorax 8 states, 2005-2008 is shown.

Eligible patients—central venous catheterization (including transvenous pacemaker); thoracentesis; pericardiocentesis; paracentesis; insertion of ETT with or without mechanical ventilation.

Slide 23

Postoperative Hemorrhage or Hematoma 8 States, 2005-2008

Image: A sample chart titled: Postoperative hemorrhage or hematoma 8 states, 2005-2008 is shown.

Eligible patients—central venous or arterial catheterization; aspiration or I&D; thoracentesis or thoracostomy; pericardiocentesis; paracentesis; insertion of ETT or rectal tube; LP; nerve block, fecal disimpaction.

Slide 24

Image: EDPSI #4: Potentially Missed Acute Myocardial Infarction (AMI) Numerator and Denominator is shown.

Slide 25

Variation Across States and Years

Image: A bar graph labeled: Figure 1 Percentage of eligible patients with Unanticipated Worsening of Acute Coronary Syndrome (ED-PSI #4) is shown:

Blue = 2007.
Red = 2008.
CV = 0.38-0.49.
Year-year corr = 0.94.

Slide 26

Image: EDPSI #5 Electrocardiogram for Non-Traumatic Chest Pain Numerator and Denominator is shown.

Slide 27

Variation Across States and Years

Image: A bar graph labeled: Figure 2 Percentage of eligible patients Electrocardiogram for Non-Traumatic Check Pain (ED-PSI #5) is shown:

Blue = 2007.
Red = 2008.
CV = 0.01-0.03.
Year-year corr = 0.79.

Slide 28

Image: EDPSI #5 Electrocardiogram for Stroke Numerator and Denominator is shown.

Slide 29

Variation Across States and Years

Image: A bar graph labeled: Figure 3 Percentage of eligible patients Electrocardiogram for Stroke (ED-PSI #6) is shown:

Blue = 2007.
Red = 2008.
CV = 0.12-0.14.
Year-year corr = 0.99.

Slide 30

Variation Across States and Years

Image: A bar graph labeled: Figure 4 Percentage of eligible patients Electrocardiogram for Stroke (ED-PSI #6) is shown:

Blue = 2005.
Red = 2006.
Gold = 2007.
Cyan = 2008.
CV = 0.02-0.03.
Year-year corr = 0.77-0.98.

Slide 31

Image: EDPSI #7 Electrocardiogram for Syncope Numerator and Denominator is shown.

Slide 32

Variation Across States and Years

Image: A bar graph labeled: Figure 5 Percentage of eligible patients Electrocardiogram for Syncope (ED-PSI #7) is shown:

Blue = 2007.
Red = 2008.
CV = 0.03-0.04.
Year-year corr = 0.76.

Slide 33

Variation Across States and Years

Image: A bar graph labeled: Figure 6 Percentage of eligible patients Electrocardiogram for Syncope (ED-PSI #7) is shown:

Blue = 2005.
Red = 2006.
Gold = 2007.
Cyan = 2008.
CV = 0.02.
Year-year corr = 0.51-0.999.

Slide 34

Image: EDPSI #8 Electrocardiogram for Asthma Revisit Numerator and Denominator is shown.

Slide 35

Variation Across States and Years

Image: A bar graph labeled: Figure 7 Percentage of eligible patients Electrocardiogram for Asthma Revisit (ED-PSI #8) is shown:

Blue = 2007.
Red = 2008.
CV = 1.44-1.66.
Year-year corr = 0.999.

Slide 36

Image: Electrocardiogram for Missed Subarachnoid Hemorrhage Numerator and Denominator is shown.

Slide 37

Variation Across States and Years

Image: A bar graph labeled: Figure 8 Percentage of eligible patients Electrocardiogram for Missed Subarachnoid Hemorrhage (ED-PSI #9) is shown:

Blue = 2007.
Red = 2008.
CV = 0.37-0.41.
Year-year corr = 0.54.

Slide 38

Image: EDPSI #11 Electrocardiogram for Chest Pain Revisit Numerator and Denominator is shown.

Slide 39

Variation Across States and Years

Image: A bar graph labeled: Figure 9 Percentage of eligible patients Electrocardiogram for Chest Pain Revisit (ED-PSI #11) is shown:

Blue = 2007.
Red = 2008.
CV = 0.29-0.43.
Year-year corr = 0.89.

Slide 40

Expert Work Group

• General interest in quality and safety-related concepts that can theoretically be captured using HCUP data:
  • Utilization of indicated services to optimize patient safety.
  • Revisits for related conditions after a high-risk index ED visit.
  • Prior ED visits for symptoms or signs of a potentially missed dx.
• Limited ongoing quality improvement efforts in EDs, typically based on EMR or manual record review .
• Some skepticism about whether gaps in care actually exist for ECG utilization.
• Questions re face validity of observed variation in the rates of some proposed ED-PSIs across states.
• Great interest in preliminary results.
• Strong support for further testing and refinement, with a focus on bringing states up to a common standard.

Slide 41

Recommendations and Next Steps

• More states should be included in the next round of empirical analyses, to better understand patterns of variation and the potential scope of use of ED-PSIs.
• Timeliness indicators are not promising, due to the easier availability of "time stamp" data from other sources.
• Utilization-related and revisit-related indicators appear to be most promising for further development and testing.
• Validation work should be undertaken to confirm that administrative data sets consistently capture:
  • Relevant service utilization, such as electro-cardiography.
  • Relevant revisits.
• Complications based on current PSIs represent important concepts, but there are severe limitations due to confusion and inconsistency in reporting of POA status:
  • Collaboration with HCUP partners needed.

Slide 42

Acknowledgments

• UC Davis team:
  • Banafsheh Sadeghi (epidemiologist).
  • David Barnes and Aaron Bair (emergency physicians).
  • Yun Jiang and Daniel Tancredi (programming and analysis).
• External advisors:
  • Jesse Pines (GWU), Michael Phelan (Cleveland Clinic), Emily Carrier (HSC), Evaline Alessandrini (CCHMC), Astrid Guttmann and Michael Schull (ICES), Jeremiah Schuur (Brigham & Women's), Drew Dawson, MD (Federal Inter-agency Committee on Emergency Medical Services), Rollin Fairbanks (National Center for Human Factors Engineering in Healthcare, MedStar), Andrew Garrett (ASPR), Charlotte Yeh (AARP).
• AHRQ CDOM staff:
  • Pamela Owens and Ryan Mutter (ED task).
  • Jenny Schnaier and Carol Stocks (HCUP).
• HCUP partners:
  • AZ, CA, FL, HI, IN, NE, SC, TN, UT.