Upload
moses-gaines
View
245
Download
0
Tags:
Embed Size (px)
Citation preview
Concurrent Validity of Alternative CANS Outcome
MetricsWilliam A. Shennum
Julian LeiroDelisa Young
Five AcresAltadena, California
What is test validity?
• The concept of validity was formulated by T.L. Kelley in 1927, who stated that a test is valid if it measures what it claims to measure.
• For example a test of intelligence should measure intelligence and not something else, such as memory.
Simplypsychology.org/validity
Concurrent validity
• Concurrent validity is a measure of how well a particular test correlates with a previously validated measure of the same construct.
• Concurrent validity relies upon measurements that take place at the same time.
Explorable.com
Who we are
• Non profit organization serving greater Los Angeles community
• Research & Evaluation Department– Outcome & performance measurement– Quality & business process improvement– Program evaluation– Project management
Five Acres’ service programs
Why CANS?
• Need for a common outcome measure across programs for use in planning & CQI.
CANS roll-out• Training• Inspiration – Dr. Lyons’ visit• Technology• Reporting
Individual Score Report (Excerpt)Client: “I” “I”
CANS Date: 9/24/2013 2/3/2014
Life Domain Functioning Intake Discharge1. Family 3 02. Living Situation 2 08. Medical 0 010. Sexuality 0 013. School Achievement 2 114. School Attendance 1 0Youth Strengths15. Family 3 116. Interpersonal 1 118. Educational 2 1
What About Program –Level CANS Outcomes?
• Several scoring options available• Which ones work best?–Easily understood by stakeholders–Closely associated with other
outcomes important to the organization (concurrent validity)
Study Design
Three service levels were included.
1. Community based mental health2. Wraparound3. Residential treatment
Study Design - continued
• Youth intake and discharge CANS were compared (N=233)
• Five CANS outcome scoring metrics were computed for each youth
• Statistical associations of each CANS metric with concurrently measured discharge outcomes were examined.
Concurrently Measured Outcomes
Service Intensity
Program Program Outcome/Success Measures
Lower
Higher
Community based mental health
• % Service completion (EBP’s)• Avg. YOQ points improved
(caregiver)
Wraparound
• % Program completion/ graduation
• % In community placement at discharge
Residential treatment (RBS)
• % Program completion/ graduation
• % In less restrictive placement at discharge
CANS Outcome Metrics Examined in This Study
Metric 1. Intake-discharge change scores
Method: Compute the average change from intake to discharge, for total score and for each domain
• Strength- easy to calculate• Weaknesses – unit of measurement not
common-sense, not tied to clients’ specific needs
Metric 1 – Intake/Discharge Change Scores, Example
Youth #CANS Total
Score IntakeCANS Total Score
Discharge ChangeRBS_2 15.5 10.4 5.1RBS_3 8.1 5.2 2.9RBS_4 1.6 6.3 -4.7RBS_5 10.7 3.9 6.7RBS_6 13.3 12.0 1.3RBS_7 6.7 4.6 2.1RBS_8 7.4 4.1 3.3
RBS_12 11.1 6.6 4.6RBS_13 3.0 3.7 -0.7RBS_14 11.3 3.6 7.7
Total Score 8.8 6.0 2.8
Program Results for Metric 1
Community Based
Services
Wraparound Residential Treatment
Avg. Change 2.0 0.7 1.9
Significance P<.001 P<.004 P<.02
Metric 2 - Reliable Change Index Analysis
Method: Determine whether each individual’s change score is statistically significant, by comparing it with a Reliable Change cutoff value
•RCI Formula: change in an individual’s score divided by the standard error of the difference for the test.•Parameters in formula include Intake and Discharge scores, test standard deviations, and test (rater) reliability.•Result is a yes-no score for each youth, for CANS total score, and each domain.
– Strength: statistically sound– Weaknesses: difficult to compute and explain to stakeholders
How to Calculate the RCI
M. Horsill, www.psy.uq.edu.au
1. Calculate standard error of measurement
2. Calculate standard error of the difference
3. Calculate RCI
If RCI is 1.96 or greater, the difference is statistically significant (95% confidence level, 2-tailed). 1.65 = one-tailed test cutoff
Metric 2 - Reliable Change Index Analysis, Example
Youth #CANS Total Score
Change RCI Score>=1.96?
(1=yes, 0-no)RBS_2 5.1 2.23 1RBS_3 2.9 1.29 0RBS_4 -4.7 -2.08 0RBS_5 6.7 2.98 1RBS_6 1.3 0.59 0RBS_7 2.1 0.95 0RBS_8 3.3 1.44 0
RBS_12 4.6 2.03 1RBS_13 -0.7 -0.29 0RBS_14 7.7 3.40 1
% Exceed RCI Cutoff 40.0%
Program Results for Metric 2
RCICommunity
Based Services
Wraparound Residential Treatment
>=1.96 14.9% 10.2% 24.2%
>=1.65 24.1% 12.6% 30.3%
Metric 3. Reduction in Number of Actionable Needs
Method: Compare the number of actionable needs at intake and discharge, total and by domain.
–Strength- addresses standout needs–Weakness- not tied to an individual’s specific needs identified at intake.
Metric 3. Reduction in Number of Actionable Needs, Example
Youth ## Actionable Needs Intake
# Actionable Needs Discharge Change
RBS_2 23 24 -1RBS_3 19 13 6RBS_4 3 10 -7RBS_5 23 7 16RBS_6 32 29 3RBS_7 12 6 6RBS_8 15 2 13
RBS_12 30 14 16RBS_13 5 6 -1RBS_14 24 6 18
Total Score 18.6 11.7 6.9
Program Results for Metric 3
Community Based
Services
Wraparound Residential Treatment
Avg. Change 4.5 1.4 4.4
Significance P<.001 P<.12 P<.03
Metric 4 - % Actionable Needs Resolved
Method: Track specific actionable items identified at intake, and calculate the % of these that are no longer actionable at discharge, total and by domain.
–Strengths - tailored to individual, easy for stakeholders to understand–Weakness – somewhat complicated scoring
Metric 4 - % Actionable Needs Resolved, Example
CANS Domain: Youth Strengths Intake
Discharge
Need Resolved? (1=yes, 0=no)
15. Family 3 1 116. Interpersonal 1 117. Optimism 2 1 118. Educational 2 1 119. Vocational U U20. Talents/Interests 2 1 121. Spiritual/Religious 2 2 022. Community Life 2 1 123. Relationship Permanence 2 2 024. Resiliency 1 125. Resourcefulness 2 1 1
% Needs Resolved 75%
Program Results for Metric 4
Community Based
Services
Wraparound Residential Treatment
% Needs Met 72.8% 49.2% 61.6%
Metric 5 – Actionable Needs Improved
Method: Track specific actionable items identified at intake, and calculate % of these that show any improvement, even if still actionable at discharge, total and by domain.
– Strength - tailored to individual, easy for stakeholders to understand
– Weaknesses – includes needs still remaining at discharge, somewhat complicated scoring
Metric 5 – Actionable Needs Improved, Example
Youth Behavioral/Emotional Needs Intake
Discharge
Any Improvement? (1=yes, 2=no)
41. Psychosis 0 0
42. Impulse/Hyper 2 0 1
43. Depression 1 0
44. Anxiety 1 1
45. Oppositional 3 2 1
46. Conduct 2 1 1
47. Adjustment to Trauma 3 2 1
48. Anger Control 2 2 0
49. Substance Use 0 0
% Any Improvement in Actionable Needs 80%
Program Results for Metric 5
Community Based
Services
Wraparound Residential Treatment
% Needs Improved
75.3% 59.4% 66.6%
Concurrent Validity of the 5 CANS Program Outcome Metrics
• Are the metrics statistically associated with concurrently collected measures of program success?
All CANS Metrics Significantly Associated with Concurrent Program
Outcomes CANS Outcome Metrics
Association w/Concurrent Outcomes
1. Avg.
Change
2. RCI
Cutoff
3. Change
in # Needs
4. Needs
Resolved
5. Needs
Improved
Median Statistical Significance
p = .020
p = .043
p = .009
p = .013
p = .018
Specific CANS Domains were Positively Associated with
Concurrent Outcomes CANS Outcome Metrics for
Selected DomainsFunctioning Youth
StrengthFamily
Needs & Strength
s
Behavior &
Emotional Needs
Median Correlation w/ Outcomes
r = .23
r = .33
r = .22
r =.28
Some Differences Across Program Were Evident
CANS Outcome Metrics
Program / Outcome
Avg. Change
RCI Cutoff
Change in #
Needs
Needs Resolve
d
Needs Improve
dCommunity Based
Service completion
p<.001 p<.05 p<.001 p<.01 p<.05
YOQ Points p<.05 p<.05 p<.01 p<.01 p<.01
WraparoundProgram
completionp<.001 p<.001 p<.001 p<.001 p<.001
In community p<.01 ns p<.05 p<.10 ns
Residential (RBS)Program
completionp<.10 p<.05 p<.05 p<.05 p<.05
Less restrictive ns ns ns p<.05 p<.10
CANS outcomes for ‘successful’ youthCANS Outcome Metrics
Program
Avg. Change in Total Score
% Youth exceedin
g RCI1.65
Avg. Change in # of
Actionable Needs
Avg. % Actionable Needs
Met
Avg. % Actionable Needs
Improved
Community MH 2.3 29.0% 5.6 76.7% 78.9%
Wraparound 2.6 34.8% 7.0 68.9% 79.9%
Residential 3.1 36.8% 8.1 70.5% 74.6%
Average 2.7 33.5% 6.9 72.0% 77.8%
Trends• “Reduction in number of actionable needs”
and “% of actionable needs resolved” metrics tended to work best for most programs
• “Average change score” metrics worked well in program serving less severe youth
• Suggestive evidence favored youth strengths domain as a driver of overall improvement
• CANS outcome profiles for successful youth were similar across programs – benchmarks would assist in setting performance targets.
Take Aways• All CANS metrics examined showed evidence of
concurrent validity across a range of programs• All CANS domain improvements were related to
concurrent outcomes, with suggestive evidence that “youth strengths” had strongest relationship to success.
• Programs serving higher severity needs may benefit from measuring metrics related to actionable need resolution and improvement.
• Benchmark databases could assist programs in setting outcome targets for youth.
Contact Information
Bill ShennumDirector of Research & Evaluation
Five Acres(626)798-6793