System Level Performance Measurement: Using HMIS to Drive Change
Darlene Mathews, The Community PartnershipLianna Barbu, Community Shelter Board
September 23, 2008
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development2
Learning Objectives
1. Learn the fundamentals of performance measurement.
2. Learn how to develop and use HMIS-based performance measures to support local system planning and policymaking on homelessness.
3. Identify strategies to use HMIS-based performance measures to improve program and system performance.
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development3
Why HUD is Interested in Performance Measurement
Accountable to Congress, OMB, the White House, and taxpayers
HUD’s overall objective: to “end chronic homelessness and to move homeless families and individuals to permanent housing”
At least 71.5% of formerly homeless individuals will remain housed in HUD permanent housing projects for at least 6 months
At least 63.5% of homeless persons will move from HUD transitional housing into permanent housing
At least 19% of persons exiting HUD homeless assistance projects will be employed
CoC’s accountable for achieving & documenting results
Performance Measurement
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development4
Performance Measurement Fundamentals
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development5
What is Performance Measurement?
Performance measurement is a process that systematically evaluates whether your Continuum of Care’s efforts are making an impact on the
clients you are serving.
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development6
Multiple Levels of Performance Measurement
Local Service Provider
1. Program Level APR or Other Performance Measurement Tool
CoC
2. CoC/System LevelCoC Application or Other Performance Measurement
Tool
3. State Level State-Specific Performance Measurement Tool
4. National Level GRPA and PART Reviews
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development7
Purpose of System Measurement
• Are actions achieving intended goals?– What is the purpose of the system? Goals?– Does the system work?– If yes, what makes it work?– If no, what part doesn’t work, and how do you fix it
to make it work?
• Note that you may have system goals that only relate to certain types of clients or parts of the system (e.g., different goals for severely disabled persons)
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development8
Performance Measurement Building Blocks
• Inputs include resources dedicated to, or consumed by, the system—e.g., money, programs, facilities.
• Activities are what the system does with the inputs to fulfill its mission, such as providing shelter, feeding the homeless, or providing job training.
• Outputs are the direct products of system activities. They usually are presented in terms of the volume of work accomplished—e.g., number of participants served, number of service engagements, number of shelter nights.
• Outcomes are benefits or changes among clients during or after participating in system activities. Outcomes may relate to change in client knowledge, attitudes, values, skills, behaviors, conditions, or other attributes.
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development9
Example: Single Adult Shelter System’s Impact Over Time
75% of clients will have a plan to obtain and maintain stable housing.
50% of clients who exit will obtain stable housing.
85% of clients who obtain stable housing will not return to shelter within 12 months (43% of all exits).
400 single adults will receive shelter and access to services.
350 clients will receive individualized housing information and referral assistance.
180 clients will receive weekly check-up calls and problem solving assistance, as needed, after exiting to stable housing.
Short-term
Long-term
Outputs Outcomes
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development10
Framework for Converting System Goals into Outcomes
Who is the base for measuring
results?
Step 1
What do you hope to achieve
with this population?
Step 2
Within the base, how many persons achieved it?
Step 3
Who is the base population for measuring results?
Within the base, how many persons achieved it?
Step 4
Outcome (%)
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development11
Converting System Goals into Outcomes: Example
Program Goal: Exiting single adult shelter clients into “stable” housing
Persons who exited during the past year
(N= 400)
Step 1
Obtain stable housing
Step 2
200 persons obtained stable
housing
Step 3
400
200
Step 4
50% obtainedstable housing
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development12
Look Out for Ambiguous Concepts
For example, what do we mean by…
Developing and measuring performance outcomes often invites ambiguous concepts into the process.
• …obtaining stable housing?
• …obtaining employment?• …increasing income?• …accessing services?• …becoming more self-sufficient?
Which data elements and responses will count?
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development13
Key Underlying Assumptions
• Staff are collecting the required information at entry and exit for each client.
• Staff are recording this information into the HMIS in a timely fashion (e.g., 4th business day of every month for prior month).
• The target population is clearly defined.
• The program goals are clearly defined for the target population.
Outcomes can be calculated consistently for all clients if, and only if:
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development14
Developing Performance Measures: Key Elements
• For each measure (output or outcome):
a) Purpose: identifies the reason for the measure
b) Application: identifies the programs and/or system for which a measure applies
c) Definition: definition for the measure
d) Goal-Setting Methodology: used by program or system to set goals around the measure
e) Reporting Methodology: detailed description of how measure is calculated
• Use HMIS data elements
• Develop and test before implementation
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development15
Example: Successful Housing Outcome
a) Purpose: Indicates program’s success in ending homelessness by linking households to appropriate next step housing which includes transitional and permanent housing. A higher number and rate are considered positive.
b) Programs: All
c) Definition: the number of distinct households that exit (i.e., latest exit for households with multiple stays during report period) to successful housing as defined in Table X and the percentage this represents of total distinct households exited during the report period.
d) Goal-setting methodology: Meet or exceed CoC goal or prior performance. Multiply the percentage goal by the projected number of exited households (i.e., latest exit for households with multiple stays during the report period).
e) Reporting methodology: Calculate the total number of distinct household exits and the total number of destinations that are considered successful housing outcomes. Divide the number of successful housing outcomes by the number of total exits during the report period.
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development16
Considerations When Developing Measurement Methodology
Specify all parts of query/calculation:• Programs
– Across a program type (e.g., all outreach, all PSH programs)– All programs– Programs that serve a specific subpopulation
• Clients (de-duplicated across programs)– All Exiters– All participants– All currently enrolled– Those who stayed > 6 months
• Data Elements– Same HMIS elements as program measures– Need rules to reconcile data across programs if client was enrolled in more
than one program
• Timeframe– Define date range– Specify date field (entry date, exit date, placement date, assessment date,
etc.)
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development17
Wherever Possible Use HMIS to Define Your Concepts
• Emergency shelter • Apartment or house that you own
Based on the Destination HMIS data element, we can define stable housing (narrowly) using the following response categories:
• Transitional housing
• Permanent housing
• Substance abuse facility or detox center
• Hospital (non-psychiatric)
• Jail, prison, or juvenile detention center
• Room, apartment, or house that you rent
• Don’t Know
• Staying/living with family
• Staying/living with friends
• Hotel or motel voucher paid for without ES voucher
• Place not meant for human habitation
• Other
• Foster care home or group home
• Refused
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development18
Performance Measurement Methods
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development19
Performance Measurement Methods
1. Measure Interim Outcomes by Applying Program Measures System-wide
2. Use Program Performance Results to Change or Manage the System
3. Measure System Results with HMIS-Based Impact Outcomes
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development20
1st System Performance Measurement Method
Measure Interim Outcomes byApplying Program Measures System-wide
• What: Examine how the system (across all programs or across part of a system) performs on a specific measure– Example: What percentage of all homeless persons have greater
incomes when they leave the homeless system?– Example: What percentage of all homeless single men leave the
shelter system for transitional or permanent housing?
• Use: Mid-term (e.g., annually)– To assess and communicate the overall system’s success in
helping clients achieve intended outcomes– To indicate interim success towards overall impact goal– To reinforce or adjust current expenditures and support new
funding requests
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development21
Establishing System Performance Measures
• Adopt system goals
– Past performance data
– Account for state and federal goals
– Account for local goals related to achievement of long-term impact
• Establish program performance expectations based on system goals
– Negotiate with agencies
– Incorporate into CoC/funding agreements
– Include quarterly, semi-annual, annual goals
– Response to low performance
• Performance reports
– Set frequency
– Determine audience/distribution
– Use data that meets quality standards
– System and program level performance
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development22
Case Study: Columbus, Ohio Performance Measures
Prevention Outreach Emergency Shelter
Resource Specialists
Rapid
Re-Housing
Permanent Supportive
Housing
Number Served Successful Housing Outcomes Average Length of Stay Average Length of Participation
Recidivism Successful Income Outcomes
(Tier II shelter)
Direct Client Assistance Utilization
Occupancy (Tier II
shelter)
Housing Stability Housing Retention
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development23
Case Study: Columbus, OHPerformance Reporting
• Annual Evaluation– System & program measures compared to period goal – Used to help determination annual funding– Data mostly derived from HMIS – Programs scored as:
• High: no less than one not achieved • Medium: half or more achieved • Low: less than half achieved
• Quarterly System & Program Indicator Reports– System & program performance measures compared to period goal– Used to identify system(s) and program(s) “of concern”, need for
intervention
• Annual Community Report on Homelessness – Annual and trend data – Point-in-time count data
• Reports issued to CSB Board of Trustees, CoC Steering Committee, funders, www.csb.org
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development24
Low Barrier and Severe Weather Shelters are not required to set measures for the baseline Performance Indicators listed above because they are not indications on the quality of services offered. However, like all other programs, they are responsible for ensuring that their data is accurate so that we may calculate Occupancy, Clients Served and Chronically Homeless numbers per site.
Self Sufficiency Temporary, Transitional and Permanent Supportive Housing Programs required to submit a Self Sufficiency Indicator will have to choose from Substance Abuse, Education, Mental Illness Assistance or Employment.
Case Study: Washington, DC Performance Measures
Clients Served (all)
Clients Served (chronic)
Facility OccupancyRate
Clients Engaged in Case Mgmt
Exits(Positive & Negative)
Clients w/ Income Upon Exit
Length of Stay
Self Sufficiency
Low Barrier & Severe Weather
Temporary Shelters Transitional Housing Permanent Supportive Housing
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development25
• Outputs/Efficiency Measures– Clients Served– Chronically Homeless Served– Occupancy: the rate at which program was used
• Interim Outcome Measures– Permanent Housing
• Positive client destinations at exit (TH programs)• Retain clients for 6+ months (PSH programs)
– Income: the amount of income or sources obtained– Self-Sufficiency: change in substance use, education,
mental illness or employment
• Measures apply differently to each program type and are supplemented with qualitative data for ranking purposes
Case Study: Washington, DC Performance Measures (con’t)
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development26
2nd System Performance Measurement Method
Use Program Performance Results to Change or Manage the System
• What: Compare performance results on selected program measures from one program to another. (Risk adjust for client differences.)– Our system is not as effective as we want in increasing client
incomes… which programs are working best or least well?
• Use: Mid-term (e.g., annual)– To help understand whether individual programs are working– To help decide how to allocate dollars or rank programs for
funding– To identify best practices or agencies that need TA – To determine how performance expectations should differ for
different populations, and how to refer clients most appropriately
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development27
Caveats of System Performance Measurement & Analysis
• May require a lag in time even beyond the client’s completion of program involvement– Measuring recidivism requires a certain period of
time (12 months?) after program exit in which return to the system might occur
• If the HMIS only includes data from a subset of programs, the system analysis will provide an incomplete picture– Perceived decline in recidivism may only reflect
absence from participating providers as opposed to not being homeless anymore
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development28
Apples to Apples: Risk Adjustment
• Problem: Comparing program results can encourage programs to “cream” to ensure strong results
• Solution: Risk adjustment allows differences to be accounted for in client populations when comparing results across programs
Results can be adjusted on the basis of…• Client characteristics, such as demographics, family size, disability• Client history, such as past eviction, criminal background • Client functionality or level of engagement/commitment to change
• To adjust for client differences, programs need to collect consistent data on clients to use during analysis of program results
– Agree on these standards beforehand
• Develop an analysis plan for how to adjust for client differences– Engage a researcher to help develop the plan
• Even if results not formally adjusted, acknowledge that different programs may have different outcome expectations based on differences in clients targeted and/or served
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development29
3rd System Performance Measurement Method
Measure System Results with HMIS-based Impact Outcomes
• What: Define a process to track the impact of the system’s efforts on the need/problem.
– Is the system reducing the time that people stay homeless, and therefore reducing the point-in-time prevalence?
• Use: Long-term (e.g., three years)– To determine if you need to continue actions, expand current efforts (add
resources to the current system), tweak current efforts, or fundamentally change the structure and approach of your homeless system
– To evaluate whether past actions (expanding the system or adding money) helped address the issue
– To communicate the overall system’s success (such as, when justifying funding requests)
• Theory: In the long-term, important to demonstrate that the collective actions of the system solve the problem.
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development30
Sample Impact Measures
• Incidence of homelessness - Is homelessness declining?
• Incidence of street or CH - Is street or chronic homelessness declining?
• Length of stay in system, across all homeless programs - Do people stay homeless for shorter periods of time?
• Rates of Recidivism – Are repeat occurrences of homelessness avoided or declining?
Cross-tabulate results by core characteristics to understand if/how results vary for different
subpopulations
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development31
Some notes of caution…
• There is more to performance measurement than conducting the analysis
– Educate, train, obtain buy-in
• Be careful about how data is interpreted and used
1. Jump in, but don’t be careless in how results are used
2. Look at the results within the context of all the outputs, interim measures and impact measures to validate the interpretation that’s being made
3. Vet the results before publicly releasing anything4. Appropriately caveat the limitations of the data and
analysis
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development32
Comparative Trends in Bed Availability and Families Served in the Temporary Shelter System
0
200
400
600
800
1000
1200
2002 2003 2004 2005 2006
Year Round TemporaryShelter Slots forFamilies
Families Served inTemporary Shelter
Look before you leap….
It appears that the rate of family homelessness is going down.
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development33
Comparative Trends in Bed Availability & Families Served in the Emergency Shelter System
1124
806
651597
442
184 171 171 153 171
6 6
8
12
18
0
200
400
600
800
1000
1200
2002 2003 2004 2005 2006
0
2
4
6
8
10
12
14
16
18
20
Families Served in Emergency Shelter
Year-Round Emergency Shelter Slots for Families
Length of Stay (months)
…Lesson Learned
Reality: Programs couldn’t accept new families as beds were still occupied by existing clients who had dramatically longer LOS.
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development34
Value of Performance Measurement
• Communicates more accurate description of population (vs. individual, anecdotal needs)
• Creates higher level of understanding of community problem (builds community support)
• Informs resource allocation (annual funding) • Informs program development/improvement
(CQI) • Informs policy options (more shelter vs.
housing)
2008 HMIS Training: Setting the Standard - U.S. Department of Housing and Urban Development35
Additional Resources
• What Gets Measured Gets Done: A Toolkit on Performance Measurement for Ending Homelessness, National Alliance to End Homelessness, www.endhomelessness.org
• The Community Partnership for the Prevention of Homelessness (Washington, DC) www.community-partnership.org
• Community Shelter Board (Columbus, OH) www.csb.org
Questions & Discussion