The Economic Case for

Campus Mental Health ServicesSeptember 17, 2014

Presenters Include:

Daniel Eisenberg, Ph.D.

Associate Professor of Health Management and Policy

at the University of Michigan

Glenn Albright, Ph.D.

Kognito Co-founder and Director of Research

Agenda

Daniel Eisenberg

– Overview of Healthy Minds Network

– New data from 2014 Healthy Mind & Healthy Body Survey

– New intervention studies using technology and media

– Economic case

Glenn Albright

– Research findings on the effectiveness of game-based role-

play simulations

Lisa Tannenbaum– Q&A

© 2014 Kognito. All Rights Reserved.

Building an Economic Case for College Mental Health Services and Programs

Daniel EisenbergUniversity of Michigan

Webinar hosted by Kognito

September 17, 2014

Overview of Healthy Minds Network

(HMN)

Research-to-Practice Agenda

5

Healthy Minds Network (www.healthymindsnetwork.org)

Building a collaborative, international network

(1) produce knowledge (research)

(2) distribute knowledge (dissemination)

(3) use knowledge (practice)

Research projects

Surveys: Healthy Minds, Healthy Bodies

Intervention studies

Dissemination

Data reports, data sets

Research briefs, webinars

6

Healthy Minds Study(>100 schools, >100,000 students)

7

Healthy Bodies Study(11 schools, ~7000 students)

8

New Data from 2014 Healthy Minds and Healthy Bodies surveys

Source: data.healthymindsnetwork.org

HMS 2014 Data Report

(www.healthymindsnetwork.org/for-schools/data-reports)

2014 HMS Data Report (cont’d)

(www.healthymindsnetwork.org/for-schools/data-reports)

HBS 2014: Body Image

Among females

Closest to how you currently look?

Closest to how you would ideally like to look?

11.1% 30.9% 36.4% 9.4%

6.1% 2.6% 1.7% 1.2% 0.2%

0.4%

36.1%

48.2% 13.1% 1.2% 1.1% 0.3%

0.0%

0.0%

0.0%

0.0%

BMI<20 BMI 20-25 BMI 25-30 BMI 30+

HBS 2014: Help-seeking

� 77% of students with clinically significant eating disorder symptoms did not receive any treatment

� Reasons for not seeking help (among students with positive EDE-Q screens):

I have not had a need for counseling/therapy: 31%

I prefer to deal with issues on my own: 27%

I question how serious my needs are: 20%

I don’t have time: 17%

New interventions using technology and media

Brief Videos (search “Tinyshifts” on YouTube)

www.tinyshifts.com

Economic Case for College Mental Health Services and Programs

Why Might Mental Health Affect Academic Outcomes?

� Depression, anxiety, eating disorders, etc. may affect: energy, concentration, cognitive ability (e.g., memory and processing speed), sleep (amount and quality), optimism about the future (and willingness to invest)

� Result:

• Less time on schoolwork

• Lower productivity during time spent

• Less efficient allocation of time (e.g., all-nighters to catch up; missing deadlines and class)

Longitudinal Study: How does mental health predict academic success?

� Study: Eisenberg, D., Golberstein, E., Hunt, J. (2009). Mental Health and Academic Success in College. B.E. Journal of Economic Analysis & Policy 9(1) (Contributions): Article 40.

� Data: random sample of undergraduate and graduate students

� Baseline: 2005 (N=2,798)

� Follow-up: 2007 (N=747)

Strengths of our Study

� Rich data:

• Widely-validated mental health screens (e.g., PHQ-9)

• Academic outcomes (GPA, retention) for all survey respondents

• Detailed covariates including personal characteristics and past academic performance (cumulative college GPA, high school GPA, SAT/ACT scores)

� 3 year follow-up period (2005-2008)

� Focus on mental health symptoms, not use of counseling, to minimize selection bias

Key Results (1): Mental Health and Retention

� Depression (PHQ-9 score) is a significant predictor of dropping out

� 10 point lower PHQ-9 score

� reduction in risk of dropping out by a multiple of 0.6 (e.g., from 10% to 6%)

Key Results (2): Mental Health and GPA

� Depression (PHQ-9 score) also a significant negative predictor of GPA

• � 10 point lower PHQ-9 score = 9 point increase in GPA percentile

� Co-occurrence of depression and anxiety associated with a significant additional drop in GPA

� Symptoms of eating disorders also associated with lower GPA

� Replicated analysis at School of The Art Institute of Chicago and found similar results

� Currently conducting these analyses at 7 other institutions

Economic Case for Mental Health Services

Calculating Economic Benefits of Reducing Student Depression

� Benefits from student satisfaction (reputation and alumni donations) are hard to quantify

• But note in 2010 Healthy Minds (26 schools):

Highly depressed (PHQ ≥ 15): 50% satisfied w/ school, 18% likely to donate

Not highly depressed (PHQ < 15): 78% satisfied w/ school, 27% likely to donate

� We focus on more easily quantified benefits: tuition and lifetime earnings

Return on Investment (ROI) Calculator

Parameters (customizable): www-personal.umich.edu/~daneis/roi/

Student population (enrollment)

Percentage of students depressed

Institutional drop-out rate per year

Tuition rate

Outcomes

Number of drop-outs averted due to programs/services

Total additional revenue for institution

Total additional lifetime earnings (productivity) for your graduates

Example Calculation

Assumptions:

� Student population = 10,000

• Depressed: 10% (1,000)

• Non-depressed: 90% (9,000)

� Drop-out rates (per year)

• Depressed: 30% (300)

• Non-depressed: 18% (1,620)

• Overall: 19.2% (1,920)

� Average effect of treatment = 5 pt reduction in PHQ-9

� 5 pt reduction in PHQ-9 -> Reduces drop-out probability from 30% to 24% (halfway down to 18%)

Example (cont’d)

� Hypothetical program: deliver treatment to 500 depressed students (half of depressed population)

� Without program: 500 students -> 500*30% =150 dropouts

� With program: 500 students -> 500*24% = 120 dropouts

� Drop-outs averted = 30 students

� 30 retained students -> ~60 student-years of tuition (assuming 2 extra years per student)

� +$1.2 million in tuition (assuming $20K/yr tuition)

� +$3 million lifetime earnings (+$50K per college year)

Example (cont’d)

� Costs of program?

� <$500,000 (e.g., 1 psychiatrist FTE + 3 therapist FTEs)

� Conclusion: depression programs can be justified by “business case,”just from institutional perspective

� Even more so from societal perspective

� Business case does not account for most direct benefits (increased wellbeing, reduced suffering)

Key Caveats

� Still uncertainty about effect of mental health on retention

� Despite rich measures, there may have been unmeasured factors (“confounders”) that contributed to different outcomes between depressed and non-depressed

� Results from the University of Michigan data may not generalize exactly to other campuses

• When students are suffering with mental health issues

it’s not clear to them or their friends that it is okay to

talk about these issues, and that resources are

available for those who need professional help

• Students need to know how to talk to friends about

whom they’re worried and how to get them into

professional help

The Jed & Clinton Health Matters Campus Programs

• Many very symptomatic students do not see

themselves as having “psychiatric problems” (less

likely to seek treatment)

• Campus culture is open about mental health and the

value of help-seeking

• Gatekeeper and How to Help a Friend training: wide,

targeted and strategic

© 2014 Kognito. All Rights Reserved.

The Need for Outreach

Kognito Creates Immersive

Gatekeeper Conversation

Experiences with Virtual Humans

Goal is to learn to identify, talk to and if

necessary successfully refer students that

users are concerned about.

Users assume the role of a faculty/staff or

student and practice role-plays with

emotionally responsive virtual students in

psychological distress

Conversations with Virtual Humans

Instructional Benefits:

� Safe to self-disclose, experiment

� Increase in engagement, openness

� Decrease in transference reactions

� Decrease social evaluative threat

� React like real students

- Individual personalities

- Memory

- Emotionally responsive

The Neuroscience of Challenging

Conversations

The Emotional

SystemRapid judgments,

large amounts of

information

synthesized at

once

The Cognitive

SystemSlower, rule-

governed

deliberation

Emotional Self-Regulation

Emotional Regulation

Reappraisal Strategy

Empathy

…

Cognition

EmotionCommunicationMotivational Interviewing

Collaboration, Trust

Emphatic Listening

Pacing Discussion

…

Emphatic Accuracy

Mentalizing

… Skills + attitudes +

confidence + motivation +

knowledge to apply and

engage in real life

conversations to drive

behavior change

Targeted Skills in Kognito’s

Conversations

University/College Training Simulations

A Meta-Analysis

At-Risk for College

Students

Meta-Analysis

Sample Size T = 12,670

University N = 2,853

High School N= 6,474

Veterans N = 1,198

Students N = 1,004

Middle Schools N= 1,141

* Data currently being added

Veterans on Campus

(Faculty/Staff Training)

At-Risk: Faculty & Staff• SPRC/AFSP Best

Practice Registry

• Student Peer-to-Peer

in NREPP Registry

Student Veterans

(Peer Program)*

Why a Meta-Analytic Study

Effect Size – measure of the strength of a relationship between two variables

regardless of statistical significance or sample size

Aggregate results of numerous studies and is a better estimate of population in

determining the efficacy of gatekeeper training on attitudes: preparedness, likelihood,

self efficacy and behavior

In this study - Demonstrate potential of a new game-based role-play training modality

in impacting gatekeeper skills

Utilizing Virtual Human Role-Play Simulations to Train Users to Identify, Talk To and Refer Students in Psychological Distress Including Those At-

Risk for Suicide: A Meta-Analysis, (2014) Albright, G., Davidson, J., Goldman, R., Shockley, K., Eastgard, S. & Himmel, J.

© 2014 Kognito. All Rights Reserved.

Methodology (N=12,670)

1. Pre-Training Survey – 11-item Gatekeeper Behavior Survey (GBS)

- Preparedness (5 items)

- Likelihood (2 items)

- Self-Efficacy (4 items)

Gatekeeper Behaviors (3 Items) – Number of Students:

- Concerned about due to psychological distress

- Discussed concerns with

- Referred to appropriate services

2. Completed One of Five At-Risk Gatekeeper Training Simulations

3. Post Training Survey (GBS, demographic and general self-efficacy items)

4. Three Month Follow-up Survey – GBS and Gatekeeper Behavior Items

© 2014 Kognito. All Rights Reserved.

Preparedness

How prepared are you to:1. Recognize when a student’s behavior is a sign of psychological distress?

2. Recognize when a student’s appearance is a sign of psychological distress?

3. Discuss with a student your concern about signs of psychological distress they are exhibiting?

4. Motivate students exhibiting signs of psychological distress to seek help?

5. Recommend mental health support services (such as the counseling center) to a student exhibiting signs of psychological distress?

Likelihood

How likely are you to:1. discuss your concerns with a student exhibiting signs of psychological distress?2. recommend mental health/ support services (such as the counseling center) to a

student exhibiting signs of psychological distress?

ConfidencePlease rate how much you agree/disagree with the following statements:

1. I feel confident in my ability to discuss my concern with a student exhibiting signs of psychological distress

2. I feel confident in my ability to recommend mental health support services to a student exhibiting signs of psychological distress

3. I feel confident that I know where to refer a student for mental health support4. I feel confident in my ability to help a suicidal student seek help

11-Item Gatekeeper Behavior Scale

DEMO

© 2014 Kognito. All Rights Reserved.

DemographicsVariable N Reporting Percentage of SampleGender 12,410 Female 76.6%

Male 22.6%

Transgender 0.8%

Employment 11,514 Educator 53.7%

Role Staff/ Administrator 27.2%

Students, RA’s & Other 3.4%

Race 11,435 White/Caucasian 82.5%

Black/ African American 10.3%

Asian 4.0%

Native American/Alaska Native 2.4%

Other 0.7%

Age 9,189 Educators -43.8 years (SD = 11.3)

Students – 20.7 years (SD = 4.3)

Tenure 8,344 Years – 11.23 (SD = 8.98)

Received Prior 10,310 12.9%

Training

© 2014 Kognito. All Rights Reserved.

Effect Analysis for Difference Score between Pre-

and Follow-up At-Risk Training

Study ParticipantsEffect

Sizes95% Confidence Interval

N d Lower Limit Upper Limit

University Faculty &

Staff326 0.590 0.433 0.746

Veterans on

Campus306 0.734 0.570 0.898

Students 242 0.532 0.351 0.713

Q= 16.63 df=4 p >0.05 I2=75.9%

Test for overall effect z=14.566, p<.01

Conclusion

This meta-analytic study provides further evidence that

the use of online game-based gatekeeper training

simulations where users practice role-plays with

virtual humans has an impact on learner

preparedness, likelihood, self efficacy and gatekeeper

behaviors that are sustained over time.

© 2014 Kognito. All Rights Reserved.

Q&A

Further Questions?

Contact us:

© 2014 Kognito. All Rights Reserved.

• Daniel Eisenberg: daneis@umich.edu

• Healthy Minds team: healthyminds@umich.edu

• Glenn Albright: glenn@kognito.com

• Lisa Tannenbaum: lisa@kognito.com

• Kognito: info@kognito.com