Challenges and Approaches to Measuring Hospital Emergency Preparedness

On September 28, 2010, David Chin and Sheryl Davies made this presentation at the 2010 Annual Conference. Select to access the PowerPoint® presentation (404 KB). Free PowerPoint® Viewer (Plugin Software Help).

Slide 1

 >

Challenges and Approaches to Measuring Hospital Emergency Preparedness

Sheryl Davies, MA
Stanford University
Center for Primary Care and Outcomes Research

David Chin
University of California, Davis
Center for Healthcare Policy and Research

AHRQ Annual Meeting; September 26-29, 2010; Bethesda, MD

Slide 2

 >

Acknowledgements

Project team:

• Kathryn M. McDonald, MM¹
• Sheryl M. Davies, MA¹
• Tamara R. Chapman, MA¹
• Christian Sandrock, MD, MPH²
• Patrick Romano, MD, MPH²
• Patricia Bolton, PhD³
• Jeffrey Geppert, JD³
• Eric Schmidt, BA¹
• Howard Kress, PhD³
• James Derzon, PhD³

¹Stanford University; Center for Health Policy, Freeman Spogli Institute for International Studies; Center for Primary Care and Outcomes Research, School of Medicine
² University of California at Davis School of Medicine
³Battelle

AHRQ: Mamatha Pancholi, Sally Phillips, Kelly Johnson

ASPR: RADM Ann Knebel, Margaret (Peggy) Sparr, Ibrahim Kamara, Torrance Brown

This project was funded by a contract from the Agency for Healthcare Research and Quality (#290-04-0020)

Slide 3

>  

Outline

• Approach to development
• Content Domain Development
• Evaluation of existing measures
• Challenges specific to EP
• Adapting measurement approaches to EP
• Solutions to challenges
• Example measures and validation

Slide 4

 >

Measure Development and Validation Process

Image: A flow chart shows the measure development and validation process. Under the heading of "Sources" are three text boxes:

• Literature
• Actual Use
• Concept

Arrows point from each box to a large oval labeled "Candidate Indicators"; an arrow points from this oval to a text box labeled "Evaluation," and another arrow points from "Evaluation" to a diamond shape labeled "Selection." A series of arrows lead from "Selection" back to "Sources."

Notes: This is a simplified graphic that illustrates the general process of developing and validating measures, both for this set and past sets. The feedback loop shows the iterative nature of the process by which information from literature reviews, users' evaluation and experience is used to select the most promising measures and then refine definitions.

Slide 5

>  

Obstacles to Usual Developmental Framework

• Broad swath of potential indicators
• Available data inconsistent
• Many standards without evidence base to guide selection
• No standardized data collection system
• Challenges specific to measuring EP

Slide 6

 >

Revised Measure Development Framework

Image: A chart displays the following framework process. An arrow points down from each item to the one below it:

• Identify Potential EP Topics
• Identify Ideal Set of EP Topics
• Identify and Specify Measures within Topics
• Develop Implementation Guidelines
• Data Collection (Feasibility and Validity of Measures)
• Re-Specification (if needed)
• Validation of Measure Set

To the right of these steps is a cog captioned, "Ongoing Feedback, Refinement, and Reassessment"; an arrow indicates that the cog is turning.

Slide 7

 >

Our Process: Where Have We Been?

• Literature Review
• Review of Existing Guidelines
• Expert Panel Evaluation of Indicator Topics
• MMWG Feedback:
  • Focus on Functionality and Outcomes
• Indicator Development and Justification

Image: A series of arrow shapes pointing from left to right are captioned with the following text:

• 900+ Potential Indicators
• 179 Indicator Topics in Initial Evaluation
• 47 Indicator Topics in Conference Calls
• 42 Indicator Topics in Final Evaluation
• Priority Levels 1-4

Slide 8

 >

Panel Methods

• Review to identify guidelines, checklists, etc.
• Group together like guidelines to identify general topics
• Topics evaluated by expert panel (nominal group technique):
  • 43 panelists assigned to 3 duplicative panels
  • Rated topics on importance to include in report, participated in call, then re-rated subset of topics
  • Each call summarized and shared with other panels
  • Only highest rated topics moved to next step
  • Final rating also included set-building task
  • Ratings used to prioritize topics (priority level 1-4)

Slide 9

 >

Priority One Topics

Slide 10

 >

Priority One Topics (Continued)

Slide 11

 >

Priority Two Topics

Slide 12

 >

Priority Two Topics (Continued)

Slide 13

 >

Priority Two Topics (Continued)

Slide 14

>  

Results: Concept Areas Covered by Highest Priority Topics

• 6 out of 15 covered in highest priority.
• Priority 1 and 2 covered all except:
  • Staff and volunteer management
  • Fatality management
  • Disease reporting and surveillance
• Did not favor topics derived from guidelines from any single source, or from multiple sources.

Slide 15

 >

Database Analysis

Image: A chart displays the following process. An arrow points from each step to the next:

• Database search
• Database narrowing
• Linkage of dataset measures across databases
• Linkage of dataset measures to priority areas
• Validation and outcome analysis

Slide 16

>  

Search for EP Data Sources

• Literature Review using Web-based aggregators:
  • ISI Web of Knowledge, PubMed, Google Scholar
• State and federal sources:
  • ASPR, DHS
• National and regional EP expert feedback [Image: An arrow points down]

547 initially selected, however 44 unique EP data sources identified

Slide 17

 >

Database Identification and Narrowing

• Inclusion Criteria:
  • Must focus on hospital or healthcare system in part or whole
  • Must be state, regional data, (or aggregate of data above state level if available)
  • Data must be available and accessible (some EP data lost or "secure")
  • Must have available data dictionary
• Identified: 44
• Data sources which met criteria: 11

Slide 18

>  

Database Selection: The final 11

• Databases with EP information:
  • Price Waterhouse Cooper's Public Health Emergency Preparedness (PHEP)
  • The Joint Commission
  • Government Accountability Office 2003
  • Government Accountability Office 2008
  • American Hospital Association Health Forum
    • Annual Survey, TrendWatch
  • National Hospital Ambulatory Medical Care Survey (NHAMCS) Pandemic & Emergency Response Preparedness Supplement 08
  • National Hospital Discharge Survey
  • South Bay Disaster Resource Center at Harbor-UCLA Medical CTR
  • Veteran's Health Administration Data
  • Hospital Preparedness Program, ASPR
  • The Pediatric Preparedness of Emergency Departments: A 2003 Survey

Slide 19

 >

Identifying Links

• 11 databases evaluated in detail:
  • Characteristics
  • Quality
  • Size
  • Temporal
  • Relationship
• Indicators identified with relevance to MMWG:
  • ONLY 16:
    • National Hospital Discharge Survey
    • The Hospital Preparedness Program
    • American Hospital Association
    • The Joint Commission

Slide 20

 >

Example of Identified Link

• HPP SI 25
  • Number of participating hospitals statewide that have access to pharmaceutical caches sufficient to cover hospital personnel (medical and ancillary), hospital based emergency first responders and family members associated with their facilities for a 72-hour period.
• Links TJC EC.4.14.3
  • The organization plans for replenishing medical supplies that will be required throughout response and recovery, including access to and distribution of caches (stockpiled by the hospital or its affiliates, local, state, or federal sources) to which the hospital has access.

Slide 21

>

Example of Identified Link

• HPP SI6 Drills
  • Number of drills conducted during the FY 2005 budget period that included hospital personnel, equipment or facilities.
• Links to TJC EM.03.01.03, EP 3
  • For each site of the hospital that offers emergency services is a community-designated disaster receiving station, at least one of the hospital's two emergency response exercises includes an escalating event in which the local community is unable to support the hospital.

Slide 22

>

Example of Identified Link

• HPP SI 26 A3
  • Number and level of PPE statewide to protect current and additional health care workers during an event
  • Possess sufficient numbers of PPE to protect both the current and additional health care personnel deployed in support of an event.
• Links to TJC EC.4.11.9
  • The organization keeps a documented inventory of the assets and resources it has on site that would be needed during an emergency (at a minimum, personal protective equipment, water, fuel, staffing, medical, surgical etc.

Slide 23

>

Links to Priority Areas 1 and 2

• Only 7 represented in the linkages between databases to the major topic areas in Priority 1 and 2 (n = 16).
• None of the major function areas were represented (surge capacity).
• None of the patient care areas were represented.
• Unable to provide any link between databases and the priority areas determined by group.

Slide 24

>

Issues with Linkages Between Databases

• Lack of clear definitions
• Lack of similarity
• Extensive assumptions required
• From an EP perspective, indicators from databases do NOT accurately reflect EP function

Slide 25

>

Additional Database Problems

• Most linkages between only 2 datasets: HPP and TJC
• Data for measures collected and recorded differently:
  • HPP: mixed (continuous, categorical, rank)
  • TJC: binary (compliant, non-compliant)
• Do not measure EP function or outcome during a clinical or simulated situation
• Thus, data are inconsistent within and between datasets

Slide 26

>

Data Quality Example

Correlation:

• American Hospital Association Survey 2008
  • AHA: Total licensed beds—the total number of beds authorized by the state licensing (certification agency)
• Hospital Preparedness Program Survey 2006
  • HPP: Number of beds statewide, above the current daily staffed bed capacity that awardee is capable of surging beyond within 24-hours post event

Note: These variables differ from an EP perspective but collected from same agency (L & C) in state.

Slide 27

>

Data Correlation

• Number of hospital beds available
  • Rho = 0.8179
  • t* = 9.7456 >> 3.496 (95% Confidence)
• State population
  • Rho = 0.9948
  • t* = 66.96 >> 3.496 (95% Confidence)

Rho: Spearman Rank Correlation Coefficient

Slide 28

>

Summary For Databases

• Unable to identify existing measures:
  • No outcome or function analysis
  • Hypothetical, not real patient care events
  • Limited in scope of EP
  • Few measures in major priority areas
• Unable to perform validation of existing measures:
  • Lack of adequate linkages across datasets with similar data
  • Inconsistently defined data
  • Absence of patient outcome data

Slide 29

>

Challenges in Measuring Preparedness

Slide 30

>

Challenges in Measuring Preparedness

Slide 31

>

Challenges in Measuring Preparedness

Slide 32

>

Conceptual Models Related to Measurement

• Donabedian Model of Clinical Measurement:
  • Structure:
    • Material, human resources, hospital characteristics
    • Lack evidence linking structure with outcome
  • Process:
    • What you do: includes planning and response
    • Doing the right thing well
    • Includes functional measures
    • Assumed to be associated with outcomes
  • Outcomes:
    • True outcomes are difficult to measure
    • Approaches to estimating outcomes during exercise not established
    • Risk adjustment required

Slide 33

>

Guiding Principals to Address Challenges

• Aim to measure functionality.
• Identify a goal outcome.
• Seek continuous outcomes.
• Constrain the focus to hospital.
• Consider the potential data and distributions.

Slide 34

>

Potential Approaches to Measurement

• Survey of preparedness activities:
  • Example: Elements included in Emergency Operation Plan
• Exercise based measures of functionality:
  • Example: Time to establish a functional security checkpoint.
• Exercise + modeling based measures:
  • Example: Time to evacuate a hospital, based on small demonstration evacuation and modeling to extrapolate time to evacuate the entire hospital.

Slide 35

>

Steps Undertaken to Develop Measures

• Identify potential ways to measure topics:
  • Review existing metrics and concepts.
  • Identify most salient functionality reflected in topic.
  • Consider how well metric fits topic area.
  • Consider potential performance.
• Draft specifications:
  • Consider feasibility of implementation.
  • Consider how well the metric reflects actual functionality.

Slide 36

>

Steps Undertaken to Develop Measures, Cont.

• Define each component:
  • Consider alternative interpretations of specification.
• Justify choices based on literature and case studies.
• Define how to move from hospital based data collection to aggregate measures at state-level.
• Iterative process.

Slide 37

>

Example Indicator
Functional Measures

Topics:

1. Hospital has a plan for alternative means of communication or backup communication systems.
2. Hospital has a plan for coordinating all levels of communication, including both intra- and inter-organizational communication, as well as required technology.

Proposed Measure 1: The time to relay a field asset request or critical field information to a non-hospital-based emergency operations center (EOC) during an exercise. (Repeated for secondary and tertiary communication modalities.) [Preliminary recommendation for state level reporting: Mean time for all hospitals.]

Slide 38

>

Example Indicator
Modeling + Measure Based Indicators

Proposed Measure: The time to evacuate the hospital. [This time is to be based on the time to evacuate a sample of X patients, the time for planning evacuation, and the subsequent extrapolation to the entire hospital.]

Preliminary recommendation for state level reporting: Mean time for all hospitals.

• Modeling helps to reduce measurement burden
• Potential to reduce measurement bias
• Requires extensive development and validation

Slide 39

>

Example Indicators
Using multiple approaches

Topic:

• Hospital has a plan for safety and security of people, including staff and patients, and supplies, which may involve partnering with local law enforcement agencies.

Proposed Measure 1: The time to establish a functioning security screen checkpoint during an exercise, according to the hospitals EOP.

Proposed Measure 2: Does the hospital have an MOU or MOA with a security agency for security support?

Slide 40

>

Evaluation Criteria Based on National Quality Forum

Slide 41

>

Proposed Indicators
Known Evidence Base

Slide 42

>

Validation Recommendations

• Step 1: Establish consensual validity through structured panel review process.
• Step 2: Develop data collection processes.
• Step 3: Develop methods to assess feasibility.
• Step 4: Develop methods for assessing proxy outcomes in an exercise (optional).
• Step 5: Identify a representative sample of hospitals to pilot test measures.
• Step 6: Collect pilot data, including test-retest reliability, inter-rater reliability, and measure performance.
• Step 7: Assess the distribution of performance and relationship between measures.