Program Measurement and Evaluation Guide:

Program Measurement

and Evaluation Guide:

Core Metrics for Employee Health Management

2 015

Health Enhancement Research Organization

and Population Health Alliance

Acknowledgments ................................................................................................................................................................................3

Chapter 1: Introduction ....................................................................................................................................................................5

Matt Damsker, Michael Connor, DrPH, Edward Marc Framer, PhD, Beth Umland,

David Anderson, PhD, Geoff Alexander, Michael Brennan, MS, MBA, Jennifer Flynn,

MS, Jessica Grossmeier, PhD, MPH, Ben Hamlin, Iver A. Juster, MD, Gordon D.

Kaplan, PhD, Adam Long, PhD, Craig F. Nelson, DC, MS, LaVaughn Palma-Davis, MA,

Robert Palmer, PhD, MSN, RN, Prashant Srivastava, David Veroff, MPP, Jerry Noyce,

and Karen Moseley

List of Measures

Chapter 2: Financial Outcomes..................................................................................................................................................11

Iver A. Juster, MD, and Ben Hamlin

Chapter 3: Health Impact ............................................................................................................................................................. 26

Gordon D. Kaplan, PhD, and LaVaughn Palma-Davis, MA

Chapter 4: Participation .................................................................................................................................................................39

Robert Palmer, PhD, MSN, RN, and Prashant Srivastava

Chapter 5: Satisfaction ....................................................................................................................................................................42

Adam Long, PhD, and Geoff Alexander

Chapter 6: Organizational Support ........................................................................................................................................ 48

Jennifer Flynn, MS, and Michael Brennan, MS, MBA

Chapter 7: Productivity and Performance .........................................................................................................................56

Jessica Grossmeier, PhD, MPH

Chapter 8: Value on Investment Framework ...................................................................................................................66

Craig F. Nelson, DC, MS, and David Veroff, MPP

Appendix ................................................................................................................................................................................................77

A: Participant Satisfaction (PSAT) Survey

B: Client Satisfaction (CSAT) Survey

C: Organizational Support

Case Studies

Barry-Wehmiller

GlaxoSmithKline

Lincoln Industries

TABLE OF CONTENTS

www.hero-health.org www.populationhealthalliance.org

Collaborators

Aetna

Alere Health

AllOne Health

American College of Occupational

and Environmental Medicine

Corporate Health Improvement Program (CHIP)

Engaged Health Solutions

Findley Davies, Inc.

Geneia, Inc.

Health Dialog

H2U | Health to You, LLC

HealthFitness

HealthPartners

Hospital Alemao Oswaldo Cruz (Brazil)

Kaiser Permanente

Mayo Clinic

National Association of Worksite Health Centers

National Business Group on Health

Onlife Health, Inc.

Optum

Riedel & Associates Consultants, Inc.

StayWell

Truven Health Analytics

University of Michigan

Endorsers

Johnson & Johnson

MediFit Corporate Services

RedBrick Health

Viridian Health Management

HERO-PHA Steering Committee

David Anderson, PhD, StayWell

Michael J. Connor, DrPH, Alere Health

Edward Marc Framer, PhD, HealthFitness

Karen Moseley, PHA

Jerry Noyce, HERO

Beth Umland, Mercer

Subject Matter Experts

Steve Aldana, WellSteps

Judd Allen, Human Resources Institute, LLC

Robert Eisenberger, University of Houston

Kimberly M. Firth, PhD, Samueli Institute

Ron Goetzel, Truven Health Analytics

Allison Hess, Geisinger Health Plan

Cheryl Larson, Midwest Business Group on Health

Debra Lerner, MS, PhD, Tufts Medical Center

Joe Leutzinger, PhD, Health Improvement Solutions, Inc.

Ari Levy, Engaged Health Solutions

Amaya Ortiz, Engaged Health Solutions

Tom Parry, PhD, Integrated Benets Institute

Nico Pronk, PhD, FACSM, FAWHP, HealthPartners

John E. Riedel, MPH, MBA, Riedel and Associates

Consultants, Inc.

Seth Serxner, PhD, Optum

Bruce Sherman, MD, FCCP, FACOEM,

Employers Health Coalition

Shelly Wolff, MBA, Watson Wyatt

Leadership Group

Geoff Alexander, Onlife Health

Michael Brennan, MS, MBA, The Boeing Company

Jennifer Flynn, MS, Mayo Clinic

Jessica Grossmeier, PhD, MPH, StayWell

Ben Hamlin, NCQA

Iver A. Juster, MD, Healthagen

Gordon D. Kaplan, PhD, Alere Health

Adam Long, PhD, H2U | Health to You, LLC

Craig F. Nelson, DC, MS, American Specialty Health

LaVaughn Palma-Davis, MA, University of Michigan

Robert Palmer, PhD, MSN, RN, Alere Health

Prashant Srivastava, eVive Health

David Veroff, MPP, Health Dialog

ACKNOWLEDGMENTS

www.hero-health.org www.populationhealthalliance.org

Financial Outcomes Work Group

Co-leaders:

Ben Hamlin, NCQA

Iver A. Juster, MD, Healthagen

Jeff Dobro, MD, Redbrick Health

Josh Dunsby, PhD, Mercer

Jessica Grossmeier, PhD, MPH, StayWell

Erik Lesneski, AllOne

Kristin Parker, PhD, MPH, Mercer

Erin L. D. Seaverson, MPH, StayWell

Kelly M. Shreve, MEd, MCHES, CIC, Capital BlueCross

David Schweppe, MPH, CPHIMS, Kaiser Permanente

Health Impact Work Group

Co-leaders:

Gordon D. Kaplan, PhD, Alere Health

LaVaughn Palma-Davis, MA, University of Michigan

Marybeth Farquhar, PhD, MSN, RN, URAC

R. Allen Frommelt, PhD, MS, Nurtur Health

Sandi Greenawalt, RN, URAC

Vince Haue, DrPH, MPH, Alere Health

Fikry Isaac, MD, MPH, FACOEM, Johnson & Johnson

Erik Lesneski, AllOne

Jenna Williams-Bader, MPH, NCQA

Organizational Support Work Group

Co-leaders:

Michael Brennan, MS, MBA, The Boeing Company

Jennifer Flynn, MS, Mayo Clinic

Nicole Gaudette, MPH, MCHES, Capital Blue Cross

Rosie Gonzalez, MS, RD, LD, HealthFitness Corporation

Deborah M. Gorhan, MS, CHES, Johnson & Johnson

Global Health Services

Andriana Hohlbauch, Truven Health Analytics

Travis M. Lehman, CHES, Highmark Inc.

Joe Leutzinger, PhD, Health Improvement Solutions, Inc.

Participation Work Group

Co-leaders:

Robert Palmer, PhD, MSN, RN, Alere Health

Prashant Srivastava, eVive Health

Kailin Alberti, MS, FACW, CWWPC,

ActiveHealth Management

Helene S. Forte, RN, MS, PAHM, Aetna

Kurt Hobbs, Mayo Clinic

J. Douglas Knoop, MD, MHA, FACS, CPE, Healthstat, Inc.

Jennifer Nailor, RN, BSN, CCP, Capital Blue Cross

Erin Rademacher, MA, StayWell

Productivity and Performance Work Group

Leader:

Jessica Grossmeier, PhD, MPH, StayWell

Jack Groppel, PhD, Human Performance Institute

and Wellness & Prevention, Inc.

Iver A. Juster, MD, Healthagen

Travis M. Lehman, CHES, Highmark Inc.

Paul C. Mendelowitz, MD, MPH, ActiveHealth Management

David Schweppe, MPH, CPHIMS, Kaiser Permanente

Satisfaction Work Group

Co-leaders:

Geoff Alexander, Onlife Health

Adam Long, PhD, Health to You, LLC

Joseph Alexander, Ortho Clinical Diagnostics

Crystal Hemmenway, Nurtur

John E. Riedel, MPH, MBA, Riedel and Associates

Consultants, Inc.

Lisa Saheba, MPH, URAC

Value on Investment Work Group

Co-leaders:

Craig F. Nelson, DC, MS, American Specialty Health

David Veroff, MPP, Health Dialog

Susan Dorfman, PhD, Communications Media, Inc.

Rebecca Kelly, PhD, RD, The University of Alabama

Karen O. Marlo, MPP, National Business Group on Health

Kenneth R. Pelletier, PhD, MD, University of Arizona

School of Medicine and University of California

School of Medicine San Francisco

David Schweppe, MPH, CPHIMS, Kaiser Permanente

www.hero-health.org www.populationhealthalliance.org

The Health Enhancement Research Organization (HERO)

and Population Health Alliance (PHA) are pleased to present

Program Measurement & Evaluation Guide: Core Metrics for

Employee Health Management (herein referred to as “Guide”),

a core set of metrics for the evaluation of employee health

management programs. After two years and countless

hours of research and discussions by more than 60

members of both organizations and many outside experts,

HERO and PHA are responding to employers who seek a

greater level of clarity regarding the value of their wellness

efforts. Thus, we recommend an initial set of measures

to assess the impact of the health management programs

offered to employees. The results are better informed

business decisions and boardroom discussions.

HERO is dedicated to improving the health of the employee

population through research and education to create and

disseminate evidence-based research describing “best

practices” in employee health management (EHM). PHA is

acknowledged for years of work in consensus-driven and

evidence-based evaluation measures and methodology and

has a broad perspective which includes the health of the

entire US population, including the employee population.

HERO and PHA collaborated with more than 40 other

organizations in developing the Guide. Virtually all industry

segments were represented, including employers, health

plans, program providers, academic research centers, and

certication agencies.

THE GOAL FOR THE GUIDE

The goal of this collaborative project and the Guide is to

provide standard measures for the assessment of employee

health management. This project does not seek to be

prescriptive about the types of programs offered to an

employee population. Rather, the recommended metrics

can be applied to any program intended to improve the

health of a population. For example, some programs may

be focused on low-risk individuals with the goal of keeping

risks low, while others may be focused on employees

already at risk of future disease with the goal of risk

reduction. Still other programs may be designed to help

individuals with disease achieve better outcomes. The

Guide includes metrics and evaluation strategies that

apply to these and other focus areas.

At the project’s outset, the additional goal of developing

standard recommendations for the levels of performance

that wellness programs should be expected to attain was

considered. However, our conclusion, based on a review

of the literature, is that codifying expected program

outcomes would be premature. Therefore, the scope

of the project was limited to providing a common set

of standard measures and measurement methods.

As data based on these standard measures become

available, future plans for the project include developing

standards of performance and best practice. While the

initiative is focused on supporting employer programs,

our hope is that other stakeholders and communities

also will benet from this work.

STAKEHOLDER BENEFITS FOR GUIDE USERS

The use of a core set of standard measures is expected

to benet all EHM program stakeholders.

Employers/Benets Managers: For those faced with

decisions regarding the selection of health enhancement

programs, core metrics can facilitate comparisons and

provide a reasonable basis for the development of vendor

performance metrics and expectations. In addition,

employers can use these data to identify gaps in their

own employee health management programs.

Benets Consultants: Core metrics can be used across

EHM vendors and employer purchasers of EHM services.

When EHM program outcomes are based on standard

metrics, sharing these ndings can be expected to result

in industry norms. These, in turn, will provide consultants

CHAPTER 1: INTRODUCTION

Matt Damsker, Michael Connor, DrPH, Edward Marc Framer, PhD, Beth Umland,

David Anderson, PhD, Geoff Alexander, Michael Brennan, MS, MBA,

Jennifer Flynn, MS, Jessica Grossmeier, PhD, MPH, Ben Hamlin, Iver A. Juster, MD,

Gordon D. Kaplan, PhD, Adam Long, PhD, Craig F. Nelson, DC, MS,

LaVaughn Palma-Davis, MA, Robert Palmer, PhD, MSN, RN, Prashant Srivastava,

David Veroff, MPP, Jerry Noyce, and Karen Moseley

www.hero-health.org www.populationhealthalliance.org

with reliable comparative data for making vendor

recommendations and for negotiating health improvement

performance standards.

Health Management Program Managers: Core metrics will

provide data to ne-tune health enhancement interventions

as well as data for reporting success to C-Suite stakeholders.

Accrediting Organizations: These groups will be able to

use metrics endorsed by relevant stakeholders to evaluate

vendor and/or health plan compliance; they can also serve

as industry ‘clearinghouses’ for aggregated results.

National Health Management Policy Makers: Core

metrics will facilitate the development of benchmarks and

recommendations for use by federal/state policy makers.

Employee Health Management Services Vendors: Core

metrics will create a level playing eld for competitors and

encourage product improvements based on efforts to meet

or surpass benchmarks based on the standard measure

set. These metrics will also support industry-level research

demonstrating the value of EHM programs.

Employee Health Management Participants: Participants

will benet from product improvements stemming from

competition to meet and surpass benchmarks based on

these core metrics.

SCOPE OF THE GUIDE

Measures applicable to key health management programs

delivered to an employer’s population were considered.

These were categorized into the following measurement

domains:

• Financial outcomes

• Health impact

• Participation

• Satisfaction

• Organizational support

• Productivity and performance

• Value on investment

OUR COLLABORATIVE PROCESS

HERO and PHA drew on member experts, prior

research, and a strong project process for the Guide.

The collaboration was guided by a small steering committee

comprised of members from both organizations. Seven

work groups were assembled, each addressing one of the

respective domains listed above. The groups were staffed

by HERO and PHA members and other volunteers.

The co-leaders of each work group (largely drawn from

each organization’s research-related committees) formed

a leadership group that met regularly with the steering

committee to provide updates, discuss issues, review and

offer comment and feedback on the measure-development

work in each domain, and to assure consistency across

domains. Major steps in the process included:

• Review of the literature to discover what metrics

are currently used to measure the performance

of employee health management programs;

• Obtain guidance and advice from other subject

matter experts in the domain areas;

• Identify and/or develop recommended measures;

• Review the work with key stakeholders to obtain

feedback and consensus;

• Release the work through conference presentations,

publication, and other channels recommended by

stakeholders and others.

EHM VALUE CHAIN

Measuring the value of EHM programs is widely desired by

employers. Unfortunately, accurately measuring the value

of EHM is not straightforward. There is no practical “gold

standard” methodology by which to measure savings or

other desired outcomes. We could nd no cases where

different evaluation methodologies have been compared

against the same program or over the same time period.

Nonetheless, the science of EHM evaluation has evolved

to the point that we can provide useful guidance on what

metrics to select—and the methodologies that accompany

the use of the metrics. Moreover, the Guide offers

information about how various metrics t specic cases

differing in population size, data availability, and resources

available for program evaluation.

EHM programs vary in the types of health opportunities

addressed, the specic content, and the multiple ways

individuals can participate. The following steps may not

all occur in a linear fashion, yet the overall EHM value

proposition is largely similar across program types:

1. Assess all individuals in the population across the

health continuum to identify opportunities to

maintain or improve health, or to reduce the risk

for future illness.

2. Engage individuals with programs and tools

through which they can successfully address

these opportunities.

3. Continue engagement long enough for them to

acquire and sustain healthy behaviors.

www.hero-health.org www.populationhealthalliance.org

4. This sustained “effective engagement” results in

preventing or reducing lifestyle-related risk factors

(e.g., excess weight, high blood pressure, or

unhealthy cholesterol).

5. Sustained healthy behaviors and clinical outcomes

result in fewer ER visits, hospitalizations, and

procedures related to lifestyle-related risk factors

and poor clinical outcomes. Sustained healthy behaviors

may also directly improve employee productivity

and performance.

6. Fewer ER visits, hospitalizations, and procedures

yield medical, absenteeism, worker’s compensation,

and disability cost-savings; and increased productivity

and performance.

7. Improved employee productivity and performance

contribute to improved nancial outcomes for

individuals and organizations.

Understanding the EHM value chain provides guidance on

what to look for on your programs’ reports: It is important

to look for metrics about activities and results in the steps

that lead to savings.

1,2

Metrics related to the rst ve steps

in the value chain are often referred to as “value metrics”

or “plausibility metrics” and serve as a reminder to check

whether the EHM programs accomplished enough to make

the claim of savings plausible.

NEXT STEPS

The development and release of the Guide is a major

industry initiative, but in many ways it is just the beginning.

These core metrics and methods need to be further applied

by employers and other purchasers in assessing value

and improving performance of EHM programs. Through

practical application, the measures will be rened

and further standardized, enabling more robust

reporting across the industry and leading, eventually,

to normative benchmarks.

The HERO-PHA measurement collaborative will continue

its process of encouraging and assessing the adoption of

core metrics and facilitate the development of additional

metrics, particularly in the areas of organizational support,

productivity and performance, and value on investment.

CHAPTER 1 REFERENCES

Grossmeier J, Terry PE, Cipriotti A, Burtaine JE. Best practices in evaluating

worksite health promotion programs. American Journal of Health Promotion.

Jan/Feb 2010; 24(3):TAHP1–TAHP9.

Linden A. What will it take for disease management to demonstrate a return on

investment? New perspectives on an old theme. Am J Manag Care 2006;12:217–222.

engage enough

of them

with effective

programs…

Find people with

potential health

improvement

opportunities…

and continue for

enough time and

intensity.

*Effective (activated)

engagement

Results in:

Improved

clinical outcomes

Improved

utilization outcomes

Improved

nancial outcomes

www.hero-health.org www.populationhealthalliance.org

FINANCIAL OUTCOMES

Given the importance of nancial outcomes to employers

who invest in EHM programs, the Guide focuses on specic

nancial metrics and savings methodologies, as follows:

1. Directly monetized claims savings using one of ve

savings methodologies.

2. The monetized impact on rates of hospitalizations

(and ER visits and procedures) that are potentially

preventable by EHM.

3. Financial impact based on a model that links to what

occurred during the program (such as participation,

changes in lifestyle-related health risks, and clinical

outcomes) and characteristics of program participants

using published evidence and/or rigorous claims-based

studies of prior years of the program or a vendor’s

book of business.

The ve savings methodologies applied to directly

monetized claims are:

1. Cost trend compared with industry peers

2. Adjusted-expected compared to actual cost trend

3. Chronic vs. non-chronic trend comparison

4. Participant vs. non-participant trend comparison

5. Comparison with matched controls in a

non-exposed population

HEALTH IMPACT

This measurement domain assesses the impact of EHM

programs on the overall health and well-being of targeted

populations. Four dimensions of health were identied for

inclusion in the base set of measures.

1. PHYSICAL HEALTH IMPACT

A. BMI (height; weight)

B. Blood pressure (systolic/diastolic)

C. Cholesterol (Total; HDL; LDL)

D. Fasting blood glucose or HbA1c

E. Medical conditions

F. Perceived health status

2. MENTAL AND EMOTIONAL HEALTH IMPACT

A. Perceived stress

B. Depression

C. Anxiety

D. Perceived life satisfaction

3. HEALTH BEHAVIORS THAT IMPACT PHYSICAL/

MENTAL AND EMOTIONAL HEALTH

A. Physical activity (total amount)

B. Tobacco use (all types)

C. Alcohol use (total amount/risky drinking)

D. Fruit/Vegetable intake

E. Sleep (typical hours/night)

F. Daytime sleepiness

G. Safety restraint use

H. Drinking/Driving

I. Health screenings according to recommended

schedule (blood pressure; glucose/A1c;

cholesterol; colorectal, cervical and breast cancer)

J. Immunization status (u, tetanus/diphtheria,

pneumonia, varicella, HPV)

4. SUMMARY OF HEALTH MEASURES

(RISK STATUS INDICES)

A. Overall risk reduction—maintenance of low risk

status and net risk reduction

B. Individual risk reduction

PARTICIPATION

Ideally, “participation” would be dened as a level of

interaction between an EHM program and an individual

that has shown some evidence of producing an outcome.

The level of interaction would presumably vary based on

the program and the modality. Due to a lack of consistency

between interventions, levels of intervention, and the

outcomes in the literature, the approach recommended

is to use a range of participation measures based on general

themes we observed in the literature. These were not

themes associated with specic outcomes and/or programs

but, rather, were those observed across the modalities.

In-person contact was associated with the lowest number

of contacts able to produce a positive outcome, while online

contact was associated with the highest number of contacts

required for an outcome.

Thus a categorical reporting structure using ranges

is recommended rather than a prescriptive minimum

number of contacts. This recommendation is based upon

observations from the literature with regard to the number

of contacts associated with a positive health outcome.

Displaying a categorical range allows employers to interpret

and understand the continuum of what could be dened

as participation within their population.

LIST OF MEASURES

www.hero-health.org www.populationhealthalliance.org

Participation Metrics Summary

CHANNEL /

MODALITY

CONTACT CATEGORIES FOR

REPORTING PARTICIPATION

Telephonic

• 1–2 contacts

• 3–4 contacts

• 5+ contacts

Web-based

• 1–5 contacts

• 6–10 contacts

• 11+ contacts

In-person

• 1 contact

• 2 contacts

• 3+ contacts

SATISFACTION

This outcome domain provides a set of satisfaction measures

and methods to enable consistent and transparent reporting

for appropriate and relevant comparisons. The satisfaction

areas addressed are Client and Participant. ‘Client’ generally

refers to the purchaser or cost-bearing entity for the EHM

program. ‘Participant’ has several synonyms depending upon

EHM area (e.g., user, consumer, patient). The domains are

listed below by area in a roughly prioritized fashion, with

those most critical for near-term adoption ranked higher.

I. PATIENT SATISFACTION

A. Overall (including loyalty)

B. Effectiveness

C. Scope

D. Convenience

E. Communications

F. Experience

G. Cost

H. Benets

II. CLIENT SATISFACTION

A. Overall (including loyalty)

B. Effectiveness

C. Value

D. Scope

E. Member experience

F. Account management

G. Reporting

ORGANIZATIONAL SUPPORT

Organizational Support refers to the degree to which an

organization is committed to the health and well-being of

its employees. The formal and informal programs, policies

and procedures within an organization that make "the

healthy choice the easy and desired choice" are recognized

as deliberate steps to which a company is committed.

A healthy culture incorporates management policies and

practices that involve, empower, and engage the employee

in decisions about their work, health and safety, and the

direction of the organization. Such an environment makes

it easy, convenient, acceptable, and expected to engage

in healthy behaviors. Intentionally limiting our focus to

supportive efforts that can be performed in the workplace,

a thorough review of the literature and interviews with

experts resulted in the identication of eight key elements

of organizational support. These elements represent the

deliberate steps a company can take to support their

employees and leaders in their health and well-being.

ORGANIZATIONAL SUPPORT ELEMENTS

A. Company-stated health values

B. Health-related policies

C. Supportive environment (the physical or “built”

environment of the workplace)

D. Organizational structure

E. Leadership support

F. Resources and strategies (adequate EHM services,

budget, communication, etc.)

G. Employee involvement (employees have

opportunity for input and evaluation)

H. Rewards and recognition

It is recommended that employers measure both their level

of organizational support and the degree to which their

employees, managers and leaders perceive both that health

is a priority for the business and that they are supported

by their employer organization. To accomplish this, these

measures would include the assessment of:

1. Deliberate steps (organizational support elements)

the employer has taken to create an environment

that supports health and well-being.

2. Employee-perceived level of organizational

support (POS).

3. Leaders-perceived organizational support (POS).

PRODUCTIVITY AND PERFORMANCE

Below is a list of metrics that can be used to assess the

worker productivity and performance gains realized from

EHM. Using the broadest possible denitions of productivity

and performance, metrics would ideally quantify worker

presence at work and the execution or accomplishment

of job-specic tasks against pre-determined performance

standards. Some organizations are able to capture employee

sick time associated with poor health, fully leverage disability

and workers compensation data to manage time away from

work, and measure observed changes in work output to

optimize on-the-job productivity. However, most employers

must rely on self-report tools for at least some of these

www.hero-health.org www.populationhealthalliance.org

issues. The recommended metrics below provide options

for measurement for organizations to select from based

on the availability of appropriate administrative data

or self-report tools.

I. TIME AWAY FROM WORK (TAW)

DUE TO POOR HEALTH

A. Unscheduled absence

B. Workers compensation

C. Short term disability

D. Long term disability

E. Self-reported absence due to employee

poor health

II. PRODUCTIVITY LOSS WHILE AT WORK

(PLAW) DUE TO POOR HEALTH

A. Self-reported presenteeism

III. WORKER PERFORMANCE

A. Employee performance ratings

B. Objective measures of performance by job type

VALUE ON INVESTMENT

The proposed VOI formula uses a cost effectiveness analysis

(CEA) convention, which places the dollar investment or

resources used rst (the numerator) and the outcomes

second (the denominator). The outcomes may be specic

clinical measures (reduced rates of a particular disease

state), or in dollar amounts representing the monetized

value of the outcomes.

The numerator will represent all inputs and investments

in an EHM program as shown below:

I. DIRECT COSTS

A. Program fees (which may include case

management; medication adherence; biometric

screening; employee assistance programs; health

risk assessment; lifestyle coaching; on-site tness

facility or club discounts; decision assistance;

triage/nurse line; injury prevention program;

concierge services; on-site clinics: ergonomic/

back health program: cost transparency programs;

Provider support programs, etc.)

B. Incentive costs (to the extent they are

incremental costs to the purchaser)

II. INDIRECT COSTS

A. Employee time (biometric screening, etc.)

B. Communications/Print materials

C. Data systems and reporting

D. Contract personnel

E. Legal review

F. Facility space

III. TANGENTIAL COSTS

A. Employee morale

B. Company reputation

C. Legal challenges

D. Selection effects (on employee population)

www.hero-health.org www.populationhealthalliance.org

INTRODUCTION

Financial outcomes are key performance indicators for

most capital, system, or human resource investments.

An important feature of the employee health management

(EHM) value proposition is the idea that—by improving the

health and reducing health-risky behaviors of employees

and their dependents—these programs produce a positive

return on investment (ROI). This section considers the ROI

contribution from savings in healthcare claims.

We aim to reduce the often-expressed confusion over

EHM nancial outcomes reports: e.g. that the metrics used

are unfamiliar or are inconsistent amongst vendors and

consultants; that the methodologies used to calculate savings

and ROI are not transparent; and that different programs

cite an implausibly-wide range of ROIs.

A common barrier to understanding EHM nancial

metrics is that they don’t easily t with the ROI paradigm

familiar to business decision makers, where return is usually

thought of as revenue—money earned for investment

made. In contrast, the nancial value of EHM is counted

as savings—money not spent to due prevented events

(such as hospitalizations or ER visits). While ROI from the

more-familiar paradigm often does include some savings

(e.g. fewer accidents from improving safety), it’s important

to reorient perspective in order to fairly compare EHM’s

statements of ROI with those of competing (potential or

actual) investments.

This section of the Guide begins with a summary of nancial

metrics and guidance, followed by a deeper dive into the

rationale for our metrics and guidance.

FINANCIAL METRICS AND GUIDANCE

SUMMARY

HERO and PHA recommend the following metrics to

measure healthcare cost (claims) savings from EHM:

• Directly monetized claims savings, using one of ve

methodologies;

• Monetized impact on rates of hospitalizations

(and ER visits and procedures) that are potentially

preventable by EHM;

• Financial impact based on a model that links to what

occurred during the program (such as participation,

changes in lifestyle-related health risks and clinical

outcomes) and characteristics of program participants,

using published evidence and/or rigorous claims-based

studies of prior years of the program or a vendor’s

book of business.

In addition, HERO and PHA recommend reporting impact

on lifestyle-related health risk factors. While there is good

evidence that preventing or decreasing such risks is cost-

saving, current evidence is not sufcient to recommend

a monetization formula based on specic risk factors.

To best work with your analyst, consultant or EHM vendor

to report savings, it’s important to understand the link

between EHM and nancial outcomes (nancial value

proposition), when savings may be expected to occur,

some basics about how savings are measured, and how our

recommended metrics get at nancial outcomes. This linkage

was described in the section EHM Value Chain (p. 6).

When Should We Expect to See Savings?

Despite the common expectation that EHM should produce

an ROI of at least one dollar per dollar invested (greater

than 1 to 1) in its rst year, much of the research on nancial

impact demonstrates savings no earlier than the second year

of EHM.

Understanding the EHM value-production chain

with its “leading” and “lagging” indicators will enable you to

advance or accept realistic performance goals, and to be a

wise reader of ROI reports.

Recall how EHM produces nancial value—by preventing

costly events such as trips to the ER, hospitalizations, and

certain procedures.

For example, EHM can identify individuals with health

risk factors (such as smoking, poorly managed stress or

depression) that are known to raise healthcare costs.

For members with chronic conditions, EHM identies

individuals who are not receiving (or adhering to) best

CHAPTER 2: FINANCIAL OUTCOMES

Iver A. Juster, MD, and Ben Hamlin

www.hero-health.org www.populationhealthalliance.org

practices. For example, taking ACE inhibitor medication

reduces the rate of hospitalizations for worsening heart

failure by about one-third.

Because ACE inhibitors are

inexpensive and hospitalizations for heart failure are very

expensive, it is cost-saving if EHM improves adherence

to ACE inhibitors.

A more common example is attention to their blood sugars,

cholesterol, diet, blood pressure, medication, regular foot

exams, and attentive wound care for individuals with diabetes.

Over time, diabetics who adhere to these practices have

fewer complications, and better function and quality of life.

The cost of these practices is often less than those of the ER

visits and hospitalizations that result from poor adherence.

Key points about the plausibility of reported

EHM savings:

• EHM must exert a strong impact on preventable

service utilization to get to positive ROI.

• Ensure the value-chain indicators line up to

make it plausible that the program produced

approximately that level of savings when judging

a report of EHM savings. Consider how many

hospitalizations and procedures would have to

be prevented to break even in the rst year

(see Deeper Dive section, p. 14).

A set of “leading indicators” can tell you during the rst year

whether your program is likely headed for savings later on.

As shown in the EHM Value Chain section (p. 6), each link

in the chain—from identication to changes in risk factors

and clinical outcomes—can be associated to performance

metrics. If these metrics are doing well, you can forecast

that your program will produce savings in an appropriate

time frame. This is similar to the concept of leading and

lagging economic indicators; for example when claims for

unemployment decrease consistently (leading indicator),

the growth in Gross Domestic Product rises several months

later (lagging indicator).

Table 1 shows important EHM leading and lagging

indicators. Sustained high performance on Leading Indicators

forecasts high performance on the outcomes of value to

employers—the Lagging Indicators. Time Course indicates

time points at which impact on the listed indicator is

typically rst observed. Also, lagging indicators other than

cost are themselves leading indicators for future cost.

A Closer Look at the Recommended Financial Metrics

HERO and PHA recommend three categories of metrics

to evaluate the nancial value of EHM programs. The

rst is “directly-monetized” (calculated using costs from

claims); the second is the monetized impact on rates of

hospitalizations that are potentially-preventable by EHM;

Table 1: Leading and Lagging Indicators of EHM’s Financial Impact

LEADING INDICATORS EXAMPLES TIME COURSE

Identication, Stratication and Targeting (outreach) Count/% with risk factors...conditions…etc. Few months

Program enrollment and use of tools Initial enrollment by type of program or tool Few months

Continuing engagement or program completion 4 or more sessions; or (better) program completion 6–12 months

Behavior change (lifestyle risks) Physical activity, tobacco, nutrition, stress 6–12 months

Behavior maintenance 6- or 12-month rates of low lifestyle risk 12+ months

Processes of care % of diabetics with annual LDL testing Six months

Medication adherence % of people with CAD on statins with MPR 80%+ 6–12 months

Achieving clinical targets % of diabetics with LDL less than 100 Six months

Activation (survey or composite measures) Patient Activation Measure or composite performance Six months

Satisfaction with EHM Positive experience and high marks on usefulness 6–12 months

Well-being Gallup-Healthways Well-Being Index 6–12 months

LAGGING INDICATORS EXAMPLES TIME COURSE

Functional status SF-12/36, Activities of Daily Living Six months

Quality of life and well-being SF-12/36, Gallup-Healthways Well-Being Index Six months

Absenteeism and presenteeism Health-related absenteeism and presenteeism scales Six months

Morbidity (ER, hospital, procedures) Rates for ER, hospital, and preference-sensitive procedures 1–3 years

Healthcare claims cost Paid or allowed amounts as trends 2–5 years

KEY

CAD: Coronary artery disease; MPR: medication possession ratio (dened as the % of the days that a person should be taking their medication, that they actually

are as evidenced by count of days’ supply dispensed); LDL: Low density lipoprotein cholesterol; SF12 and 36: Standard measures of functional status and quality of;

ER: Emergency room

www.hero-health.org www.populationhealthalliance.org

and the third is the monetized impact on lifestyle-related

health risks (based on published evidence of avoided costs

from eliminating or preventing these risks.

For the rst

metric—depending on availability of data, time, resources,

and expertise—we recommend selecting one of its ve

versions; or use a model. Your analyst, consultant, or

vendor may make that decision for you, but it’s valuable

to understand the implications of their decision.

All savings reports should be accompanied by value-chain

(“plausibility”) metrics such as initial and sustained engagement,

initial and sustained improvements in risk factors and utilization.

METRIC 1—DIRECTLY-MONETIZED SAVINGS

(1) Cost trend compared with industry peer organizations

• Compares company’s trend with that of industry peers

(optimally those without EHM).

• Because of the imprecision inherent in comparing

trends in statistically small populations, this metric is

recommended only for relatively large organizations

with access to a database of peer trends. May require

consulting expertise to appropriately adjust peer

trend and to account for other impacts on trend.

• Because most large companies have implemented

EHM, it is becoming very challenging to use this

methodology in many industries.

(2) Inection on expected cost trend

• Compares expected to observed trend. Usually

trend is “decomposed” into components such as

demographics, non-demographic (i.e., clinical), service

utilization, price, and changes in benet design. Credit

is taken for EHM-impactible components (e.g., the

non-demographic part of risk and certain types of

utilization). Expected trend is established by adjusting

the non EHM-impactible prospectively-estimated

components to their observed year-end values.

• Recommended only for relatively large organizations.

May require actuarial or epidemiological input to

prospectively estimate components of trend and

to make appropriate adjustments after completion

of the performance year.

(3) Chronic vs. Non-chronic trends comparison

• Often used when disease management (management

of people with chronic conditions) is the only or

primary EHM service and it is not feasible or desirable

(due to analytic capabilities or resource cost) to use

a more rigorous methodology.

• Compares expected to observed trend. For each of

the measurement and comparison ("baseline") years,

the population is divided into Chronics (members

who have at least one of the program-managed

conditions) and Non-chronics (everyone else). The

expected Chronic trend is equal to the observed

Non-chronic trend, and savings is calculated from the

difference between the expected and the observed

Chronic trend.

• Basic assumption is that in the absence of EHM

the two trends would be equal (or bear the same

relationship to each other) over time. For this reason,

measuring pre-baseline trends is recommended if

sufcient data history is available.

• Recommended only for large companies. Because

Chronic and Non-chronic members have very

different costs, analysts should consider risk-adjusting

trends in an effort to neutralize the effect of clinical

differences on costs.

(4) Cost or trend comparison of program participants

(P) vs. non-participants (NP)

• Compares cost-trajectories of P and NP, usually with

procedures to neutralize the expected impact of

non-EHM differences on cost trajectories.

• Recommended for relatively large organizations though

may not need to be as large as for methodologies

1 and 2. Often requires signicant analytic expertise

and time.

(5) Comparison with matched controls in a non-exposed

population

• Compares cost-trajectories of members who meet

criteria for EHM program targeting in the employer’s

population, with trajectories of matched members

who meet criteria in a different “comparison”

population that does not have EHM programs.

There are variations on criteria for the comparison

population.

• Recommended for moderate to large organizations,

though smaller may be valid in programs with high

program participation, especially if high in members

with chronic conditions (high spends). Can require

signicant analytic expertise and time.

• HERO and PHA regard this methodology as the

most rigorous and least subject to bias and “noise”

(due to non-EHM impacts), but the methodology

is rarely feasible because untouched comparison

populations are rarely available and expertise and

cost is substantial.

www.hero-health.org www.populationhealthalliance.org

METRICS—MONETIZABLE

(6) Utilization (hospitalizations and ER visits) for which

EHM has an impact

• Monetizes a downward trend in ER and hospital visits

and procedures that can be prevented by EHM (varies

with the nature of the program).

• Generally straightforward to measure given accurate

utilization data. Only modest analytic expertise

required.

• We strongly advise reporting on utilization along

with any directly monetizable metric.

(7) Reduction or prevention of lifestyle-related

health risk factors

• A model (does not use claims) that relates reduction

in or prevention of lifestyle-related health risk factors

to published evidence on the economics of preventing

and reducing such risk factors.

• Generally straightforward to measure reduction, but

monetizing risks prevented requires a valid estimate

of the type and frequency of risk factors that would

have been acquired by the population since the last

measurement year.

• While there is good evidence that preventing

or decreasing such risks is cost-saving, there is not

currently sufcient evidence to recommend

a monetization formula based on specic risk factors.

RECOMMENDED FINANCIAL METRICS:

A DEEPER DIVE

As explained in the Summary section, understanding how

EHM programs produce value helps you evaluate the impact

of an existing program or to compare programs’ savings

reports. Accurately measuring the savings from EHM is not

straightforward. While there is no practical “gold standard”

methodology by which to measure savings, the science of

EHM evaluation has progressed to the point that we can

provide useful guidance on what metrics to select—and

the methodology used to measure them—to best t your

membership size, data availability, and resources available

for program evaluation.

It’s always important to keep in mind the EHM value

production chain—how EHM’s programs, services and

tools produce savings through identication, engagement,

and improvement in lifestyle-related risk factors, clinical

outcomes, and EHM-preventable utilization.

Should Savings be reported at the EHM program level

or at the population level?

Recommendations for reporting at the program-

versus population-level:

Understanding the advantages and disadvantages

of program- and population-level nancial reporting

will help you work with your analyst, consultant,

or vendor to design a reporting package to t your

evolving needs. A partial list of solutions (not

mutually exclusive) includes:

• Population pricing and reporting (the program or

vendor designs a set of coordinated program

components and tools to deliver a targeted

population health status and ROI);

• Reporting ‘natural’ population-level metrics

(see above) along with cost- and trend-drivers

by demographics, conditions and risk factors;

• Hybrids of whole-population and by-program

reporting (particularly useful during transitions

to a true EHM during which population-paradigm

metrics are being developed and tested).

In this program-centric model, savings from the component

programs are summed to yield total EHM savings, expressed

as total dollars, per employee per month (PEPM), or per

member per month (PMPM). Each program may report

savings as per participant per year or per month (PPPY

or PPPM), but when there is more than one program,

per-participant savings must be converted to savings spread

over the entire covered population. This “sum of the parts

model” might erroneously double-count savings, and it can’t

account for the synergistic action of multiple programs.

As EHM evolves from a collection of programs designed to

address specic needs to a paradigm that monitors and

supports the whole person over time, the by-program savings

model is less capable of capturing what’s taking place in the

entire covered population, because individuals may engage

with multiple tools and programs simultaneously or over time.

EHM value proposition:

Identify opportunities to 1) improve (or maintain)

health and 2) mitigate or eliminate current risks or

avoid future risks; and address these opportunities

with effective programs and tools to improve the

population’s health status, improve productivity,

and lower health-related costs.

www.hero-health.org www.populationhealthalliance.org

The EHM value proposition is about improving or maintaining

health—not about particular programs or tools. Improving

and sustaining health status over time is needed to achieve

and sustain savings.

Over the next few years, national professional organizations

(such as PHA and HERO) will develop recommendations

and standards for reporting in the population paradigm,

but for the near future we will continue to see hybrids of

program-level and population-level reporting. Some metrics

(such as many clinical measures, population cost trend, and

hospitalizations) inherently relate to the population model.

For any savings metric, ask: What is its measurement group?

Common examples of measurement groups include:

• The entire covered population;

• Age-restricted subset of the covered population;

• All participants (in any EHM program or component);

• All participants (in any program or component) with

chronic conditions versus all participants without any

chronic conditions;

• Participants in a specic program.

While a population-centric metric paradigm has intuitive

appeal, it has an important downside: By itself, population

savings doesn’t contain actionable information. For example,

we are given a result (“Your program saved $4.00 PMPM”)

but that doesn’t tell us how the EHM saved $4.00. That

is, what were the savings drivers? Was it individuals with

certain demographic characteristics (e.g., females age 25–44)

or specic conditions (such as people with multiple lifestyle

risk factors or with chronic conditions)? Engaging person-

to-person with a coach or online? What about duration,

intensity, or frequency of engagement with one or more

programs or tools? Type of health opportunity addressed?

Program-level results excel at helping us understand which

program components drive savings (or losses); this approach

ts with the common approach of pricing by program

component, since each program has a reported ROI.

An emerging hybrid approach combines reporting savings at

the population level with insight into program-level impacts

using metrics specic to various types of health improvement

opportunities (e.g., lifestyle risk factors and clinical outcomes).

When Should We Expect to See Savings? An Illustration

As discussed above, most published research nds that EHM

programs produce savings no earlier than in the second

year.

Understanding the EHM value-production chain with

its “leading” and “lagging” indicators will enable you to

advance or accept realistic performance goals, and to be an

informed reader of ROI reports.

A simple example will illustrate why it’s so challenging

to exceed break-even in the rst year. A signicant driver

of claims savings in EHM is prevented hospitalizations.

Suppose the annual without-EHM rate of non-maternity,

non-newborn hospitalizations per 1,000 members is 45

(referred to as “45/K”), and that the employer pays, on

average $25,000 per hospitalization (including facility

and professional fees and related events and services

after hospitalization).

If the EHM vendor fees are $1.50 PMPM, or $18,000

per K (per 1000 members per year), then the 2 to 1

ROI target is $36,000, and—if all savings come from

avoided hospitalizations—the program must reduce

the hospitalization rate by 1.44 per K to achieve the

ROI target:

1,000 Number of members

45 Expected hospitalizations/K

$25,000 All-in cost of a hospitalization

$1.50 Cost of EHM PMPM, fees

$18,000

Cost of EHM per K

ROI target (savings per $ on fees)

$36,000

Savings target per K members

1.44

Number of hospitalizations/K needed to reduce

However if not all savings come from avoided

hospitalizations, the number of hospitalizations (per

1,000 members) needed to reduce from the pre-PHM

(or no-PHM) state may be less than 1.44/K. There may

also be a reduction in ER visits and outpatient procedures,

substitution of generic for brand drugs, and overall wiser

use of healthcare. On the other hand, some costs increase

as individuals start on appropriate treatment, become more

adherent to their meds, and have recommended preventive

or screening services. But in relation to a hospitalization,

these costs are usually overshadowed. It is likely, then,

that in our illustration, a reduction of only 1 or 1.25

hospitalizations/K is needed to support ROI of 2 to 1

given the above assumptions. Generally, only about 5 to

10 of the total 45 hospitalizations concern conditions that

can be strongly impacted by EHM, so the impact on

‘EHM-impactible’ hospitalizations would have to be on

the order of 10–20%.

Remember the key points: (a) EHM must exert a strong

impact on preventable service utilization to get to positive

ROI; (b) when judging a report of EHM savings, make sure

the value-chain indicators line up to make it plausible that

the program produced approximately that level of savings.

www.hero-health.org www.populationhealthalliance.org

As discussed in the Value on Investment section (Chapter 8),

the ROI denominator should take into account the entire

cost of delivering EHM, such as vendor fees, employer’s

cost of communicating and managing EHM, consultant fees,

biometrics, and incentives. Also, EHM’s true value may

include tangible and intangible savings (or revenue) besides

medical and productivity.

When Should a Model Be Used In Place of Measurement

for Savings?

Key points and recommendations: Report savings

from a model or a measurement?

• It isn’t always best to insist on measuring savings

from claims. This is especially true with small-to-

medium population size, or when funds for

evaluation are limited.

• Modeled savings provide a line-of-sight between

what your EHM programs do and savings based

on well-conducted published studies.

• Unless your organization has the population size

and funds required for a valid measured-savings

study—especially if your program design parallels

that in the programs with savings reported in the

published literature—we strongly recommend

that you consider quantifying savings from your

programs with models run by experienced hands.

EHM savings may be calculated based on the nancial data

in health care claims, or on a model derived from the type,

quantity, and intensity of engagement of members who

participate in EHM programs. It may seem intuitive that we

should always prefer reporting on nancial outcomes based

on claims, but often a good model based on your program

data and actual measurements in other populations is a

good (or even better) alternative.

Example: If I walk into the store with $100 in my wallet and

spend $50, I should expect to count $50 remaining when I

leave. That’s a direct nancial measurement. A more relevant

(and less direct) example is that I walk into the store to

purchase a pair of shoes that normally cost $100 but nd

they are on sale at $50. I still leave with $50 in my wallet. I

count the difference between what I expected to pay and

what I actually paid as $50 savings. Savings from EHM is not

a direct measurement; it’s more like this example (expected

minus actual spend). And like the example, measurement is

based on assumptions and may be subject to bias (such as

accurately estimating the non-sale price or the probability

that I would have bought shoes in the rst place).

Suppose your analyst tells you that as a result of your EHM,

your company’s health care cost was $100,000 less than it

would have been without the EHM. If you trust the “would

have been” estimate, how do you know this “reduction”

in spend was due to the EHM rather than to the effect of

other factors such as more use of less expensive generic

drugs, less out-of-network care, general improvements in

health not directly associated with your EHM, reductions

in hospitalizations due to conditions not covered by your

program, or the random variation that occurs in the cost

of healthcare over time?

Health care savings measurements are based on nding the

difference between (1) expected (what would have been

spent without EHM), and (2) the actual amount spent with

EHM. The expected spend is an estimate based on a series

of assumptions. While these assumptions and methods

aim to provide an accurate estimate of the expected cost,

we still can’t attribute with certainty the entire difference

between expected and actual cost to EHM.

Savings models also incorporate assumptions, but have

the benet of relating what the EHM program does to

nancial outcomes. They aren’t as sensitive to assumptions

about non-EHM factors that could impact costs. The

factors that go into EHM savings modeling are based on

studies designed to control for these non-EHM factors.

For example, an important impact of EHM is to reduce

lifestyle-related health risks such as tobacco use, lack of

physical activity, or high blood pressure. EHM savings

models use published studies on the cost difference

of having vs. not having each risk factor, or having then

eliminating risk factors.

Savings models are based on published evidence or well-

designed internal studies that relate factors such as

participation rates, intensity and duration, participant

characteristics (demographics, presence of risk factors

and chronic conditions, gaps in care) and outcomes (short-

and long-term reduction or prevention of risk factors, gap

closure and clinical outcomes) to savings. Then, the specic

EHM program’s factors are matched to those in the

model’s, generating a savings report for your program.

Essentially the model relates known relationships among

participation, participant characteristics, outcomes and

savings to facts about your program.

Models have several important advantages: they require

only data typically generated through the program, such as

demographics, participation, risk factors, diseases, or gaps

in care. Financial data—which must undergo a complex

process involving data cleansing and logical manipulations

to be useful for analytics—is not needed. Models can be

www.hero-health.org www.populationhealthalliance.org

run with any desired frequency and they clearly relate how

the program works to the dollars invested.

As mentioned all models are built on assumptions, so it’s

important to understand those that can inuence the

conclusions of modeling savings due to lifestyle risk reduction

and prevention. Keep in mind that:

• The factors used to build the model should be as

close as possible to those in the studies upon which

the model is based. For example, there should be

consistency in terms of concept (e.g., blood pressure)

and risk threshold level (e.g., high-risk blood pressure

denition of 140/90).

• If possible, model the behavior of as much of the

population as possible. For example, sometimes

savings due to lifestyle risk reduction is calculated on

the 20% of the population that supplied appropriate

data. It’s assumed that the other 80% didn’t change

but if some of the people who didn’t supply risk

factor data worsened, and people who got worse

were less likely to report their data, that model would

overestimate savings.

• To the extent possible, the model should take into

account what would have happened without EHM.

An ideal comparison group is one that was not

exposed to EHM but this is often not possible.

Instead, there may be public data such as national-

level data from Centers for Disease Control and

Prevention, National Center for Health Statistics,

or National Institutes of Health that can help to

provide comparison.

• Determine in advance what savings are appropriate

to model for your situation. Health care savings are

always appropriate to include, but you may also

want to include savings in the realms of disability,

absenteeism, presenteeism, and employee turnover.

USE MODELED SAVINGS USE MEASURED SAVINGS

Total members in your

covered population

The smaller the population, the less accurate are

measured savings. There’s no concrete rule based

on member count, but many consider that models

should be used for populations of less than 25,000

(discuss with your analyst).

Statistically "large"—as a very general guideline,

more than 25,000 members. Some analysis designs

may support smaller populations.

Type of data available

Medical and pharmacy claims that are not fully

adjudicated, lab results, eligibility, data generated

by the EHM.

Includes fully-adjudicated claims for accurate

accounting for ER, hospital, and procedure use

and cost

Desired frequency

of reporting

Monthly or quarterly

Annually reported 5–6 months after close

of performance year

Ability of the model

to incorporate your

specic data

Model accuracy is improved when it incorporates

program engagement and specic lifestyle risk data;

and information on the prevalence of members with

chronic conditions and other health risk in your

population. Accuracy may also be improved through

adjustment to reect your annual healthcare trend

and average cost PMPM. For models for absenteeism

and presenteeism savings, consider incorporating

wages for various types of workers.

By denition, measured outcomes incorporate

your specic data

How developed

Based on high-quality, published evidence relating

key actions of your EHM to improvement in clinical,

utilization, and nancial outcomes.

Validated (or audited) by a third party; based on

sound principles of study design and analytics

“Fully adjudicated” claims have been cleansed and treated so as to eliminate duplicates, compress adjustments and reversals, and combine all claims related to

a specic encounter (e.g., ofce visit, hospitalization, or ER visit) into a single claim that designates the type of service (e.g., hospital, ER, lab) and provider identier.

Financial editing facilitates accurate analysis. This is the quality of data found in claims data warehouses.

Table 2: Using Modeled v. Measured Savings

Table 2 provides more guidance on when to use modeled versus measured savings:

www.hero-health.org www.populationhealthalliance.org

How Accurately Can We Measure EHM Savings?

Key points and recommendations: How accurately

can we measure EHM cost savings?

• For all savings metrics, the basic principle is that

savings is the difference between expected and

actual cost.

• Savings metrics differ in how “expected cost”

is calculated.

• It is rarely feasible to perform a “gold standard”

savings calculation (which is based on randomizing

people to EHM or no EHM).

• Work with your analyst, consultant, or vendor to

use a metric that is as close to the gold standard

as possible, recognizing the limitations of the time,

resources and data required.

• Regardless of metric, include value (plausibility)

metrics in reporting to bolster a claim of

program savings.

It is impossible to measure with certainty how much was saved

or lost by EHM. This has to do with the basis of EHM’s

impact: savings are due to preventing costly adverse events.

Because we can never know with certainty how many

events (such as strokes or hospitalizations for complications

of diabetes) were avoided, we must make an educated

guess. The difference between the educated guess (what

would have happened without the EHM) and what actually

happened is our estimate of the EHM’s impact:

EHM savings = ($ spend expected) – ($ spend actual)

All savings measurement methodologies begin with the question:

“How can we estimate what would have happened without the

EHM?” Each of the recommended metrics must answer that

key question.

Most analysts believe that the best way to know what

would have happened without the EHM is to conduct

a randomized controlled study in which people are

randomly selected to participate in EHM or to not

participate. Given the proper conditions, random

assignment to treatment versus no-treatment “controls

for” or neutralizes the personal, organizational, and social

characteristics that could inuence the outcome. By

accounting for these inuencing factors we can know—

from what we observe in the control group—what

would have happened absent EHM.

However, it’s rarely feasible to do a randomized study

because employers want EHM to include their entire eligible

population. As a result, we must depend on alternative

methodologies to estimate the expected cost. Common

methodologies used to estimate expected costs are

explored in the Metrics section.

The accuracy of measured savings depends on how alike

the comparison and EHM scenarios are. It is believed that

accuracy is best achieved when the comparison scenario

involves a population that is very similar to the EHM

population—with the crucial exception that that population

did not have the option of being exposed to EHM. For

example, a factory with EHM might be compared to

a factory without EHM.

Often an unexposed population is not available (e.g., when

the company implements EHM across all employees). In that

situation, it is typical to compare the cost trajectories of

those who do and those who do not participate in various

EHM program components. However, even with careful

study design (using techniques to render the two groups as

comparable as possible on their observable characteristics),

we can’t really know how alike the exposed and unexposed

(or participant and non-participant) populations are in the

factors that drive cost trajectory.

Does this mean that we should never trust savings reports?

No, as long as we remember to ask about how well-

designed the savings study was (e.g., the size of the

populations being compared, markers of how alike they

were prior to EHM implementation, and what was done to

render the groups comparable during the analysis). Equally

crucial are the plausibility metrics, such as engaging a

sufcient percentage of members with health-improvement

opportunities and showing sustained improvements in their

risk factors and clinical outcomes.

A Deeper Dive into the Recommended Financial Metrics

HERO and PHA recommend three savings metrics:

(1) Directly monetized: One (from a selection of

ve options) that is measured using the cost elds

on claims, so by nature are already monetized

(2) Monetized improvements in healthcare service

utilization: Based on a model that relates measured

reductions in EHM-impactible healthcare service

utilization to the known costs of these services

(3) Monetized improvement or prevention

of lifestyle-related health risks

www.hero-health.org www.populationhealthalliance.org

The directly monetized metric has ve options characterized

by the methodologies used to measure them. We

recommend selecting one of these methodologies.

Five options for the directly monetized metric include:

(1) Cost trend compared with industry peers

(2) Adjusted-expected compared to actual cost trend

(3) Chronic vs. Non-chronic trends comparison

(4) Participant vs. Non-participant cost comparison

(5) Comparison with matched controls in a non-exposed

population

HERO and PHA selected these ve savings metrics because

they are commonly used or advocated, embody safeguards

to improve their validity, are measured using the employer’s

medical and pharmacy claims, and are reasonably easy

to understand. Each of these metrics has advantages and

disadvantages. However, as mentioned, none are perfect

and—to our knowledge—none have been directly

compared (on the same EHM program) to the prospective

randomized controlled methodology, or even to each other.

It is therefore strongly recommended that results using

any of these methodologies be viewed together with the

program’s value-chain plausibility metrics.

For a detailed discussion of nancial measurement

methodologies (including illustrations showing the tradeoffs

in feasibility and validity), refer to the PHA Outcomes

Guidelines Report vol. 5, p. 27–34.

Remember that all of these savings metrics incorporate

ways of answering the basic question: “What would

have happened without EHM?” The answer to this

question gives the expected cost, to which the actual

cost is compared: EHM savings = the difference

between expected and actual cost.

Directly monetized savings Metric Option 1: Cost trend

compared with industry peers. Cost trend is dened

as the rate of change of cost between two time points,

usually a year apart. Thus:

Cost trend (Year 1 to Year 2) =

(Year 2 cost – Year 1 cost) / (Year 1 cost)

Usually the costs are as per member per month (PMPM).

For example, suppose 2012 cost was $250 PMPM and 2013

cost was $265 PMPM. Then:

Cost trend (2012-2013) = ($15)/ ($250) = 6.0%

The difference between expected and actual trend can be

converted to savings:

YEAR

ACTUAL CG

PMPM

ACTUAL

SG PMPM

EXPECTED

SG PMPM

2012 $250.00 $245.00 $245.00

2013 $265.00 $257.00 $259.70

Actual trend 6.0% 4.9%

EHM savings

$2.70

CG: Comparison group; SG: Study group; PMPM: per member per month cost

In the above example, the 2012 healthcare cost of the Study

Group (SG-covered members of the company with EHM)

was $245.00 PMPM and the comparison group (CG) cost

was $250.00 PMPM. In this metric option, the CG is made

up of industry peers, as described below.

The 2013 actual costs for the CG and SG are shown, and

2012–2013 trends are calculated. The CG’s trend was 6.0%

and the methodology therefore expects that the SG’s trend

would have been 6.0% absent the EHM. But in fact it was

only 4.9%, a trend reduction of 1.1 percentage points.

The trend impact can be monetized by rst calculating the

SG’s expected cost as $245.00 x (1 + 6.0%) or $259.70.

Subtracting actual from expected cost, savings for the SG

(that is, for the EHM) were $2.70 PMPM.

This illustration used only three years of data to produce

2 trends (2011 and 2012 to produce the 2012 trend, and

2012 and 2013 to produce the 2013 trend). More data

history (and therefore more consecutive trends) is preferred

because it gives a better understanding of the employer’s

health plan economics, but often only three years’ data

are available.

All trend-based savings metrics calculate an expected trend,

and then monetize the SG-CG trend difference in this way.

For Option 1 of the savings metric, the expected trend

measured in the CG is that of industry peers, such as

airlines, travel, banking, pharmaceuticals, or technology.

Thus, the peer industry trend is the expected trend to

which the specic study company’s trend is compared.

The underlying assumption is that organizations in the

industry peer group do not have EHM and that other

factors that drive trend are very similar.

www.hero-health.org www.populationhealthalliance.org

Directly monetized savings Metric Option 2: Adjusted-

expected compared to actual cost trend. Expected trend

for the performance year is developed in advance of the

study year. Upon completion of the study year, the

expected trend would be adjusted for factors that were

not considered impactible by the EHM, if those factors

turned out to be incorrectly forecasted before the

beginning of the performance year.

Typically, overall expected and actual trends are

decomposed into components, with some designated

as EHM-impactible, others not EHM-impactible.

A common list of components of trend includes:

• Demographics (age and gender distribution),

• Risk (net of demographics),

• Utilization units (net of price),

• Price per unit of utilization,

• Plan design (e.g., deductibles, copays, and

coinsurance amounts).

In this scheme, it is considered that risk (net of demographics)

and at least some types of utilization can be impacted by

EHM; the difference between their expected and actual

values is monetized.

This “adjusted-expected whole-population cost trend”

is compared to the actual trend and the difference is

converted to savings using arithmetic similar to that in

metric Option 1, in which:

• The expected trends of components that can be

impacted by the EHM (risk net of demographics and

utilization) are carried forward into the adjusted-

expected column.

• The adjusted-expected trend components that are

not impacted by EHM are set to equal their

retrospectively measured values.

• The total adjusted-expected and retrospectively

measured trends are compared, and that difference

in total trends is monetized.

In the illustration on the next page, we assume for simplicity

that all utilization services can be impacted by the EHM.

Before the start of the performance year, all ve trend

components are projected.

• The non EHM-impactible demographic changes

component was initially projected at 1.0%, but after the

end of the year, retrospectively, it was measured at an

actual value of 0.0%; therefore the adjusted-expected

trend for this component was set to actual value of 0.0.

• Similarly, the other two non-EHM impactible factors

(unit price and plan design) were initially projected, and

their adjusted-expected values were set to be equal

to their actual values as measured after the end of the

performance year.

• The EHM-impactible risk net of demographic changes

component was projected at 0.0%, so its adjusted-

expected value was set at its initially-projected value

of 0.0%. After the end of the year, this component

was measured at -1.2% (value in the Retrospective

Actual column set at -1.2%).

8.5%

8.0%

7.5%

7.0%

6.5%

6.0%

5.5%

5.0%

2009 2010 2011 2012 2013

PHM

INTRODUCED

Study Company

Peer Companies

www.hero-health.org www.populationhealthalliance.org

Before the performance year, the Prospective Expected

total trend was projected to be 8.1%; the Actuarially

Adjusted “Expected” trend was reset to 8.4%, as described

above; and the Actual trend was measured at 6.2%. The

gap between Expected (8.4%) and Actual (6.2%) can be

monetized as described for Metric 1.

Directly monetized savings Metric Option 3: Chronic vs.

Non-chronic trends comparison. This metric is often used

to calculate savings from a disease management program.

It is less commonly used to estimate savings for members

with chronic conditions in a more comprehensive

EHM program because the supposition of this metric’s

methodology is that the comparison group is not touched

by EHM. The underlying assumption is that, absent

disease management, the trend of the Chronic population

(members who have at least one of the conditions managed

by the program, such as heart failure, diabetes, or asthma)

and of the Non-chronic population (everyone else) would

be equal or bear the same relationship to each other

over time.

The measured Non-Chronic trend is therefore the

expected trend; the Chronic trend is the actual trend.

While Metric Option 3 remains in use for DM-focused

programs, it should be noted that its fundamental

assumption has been subjected to only a few studies and

that it may not be valid to assume it is true. Some experts

recommend risk-adjusting the Chronic and Non-chronic

trends to attempt to mitigate the concern that the two

populations may exhibit different trends (absent EHM)

because they are inherently different. Several adjustments

to the methodology have been described; it is important

to ensure that your analyst or vendor understands when

and how to adjust for the differing risks in the Chronic and

Non-chronic populations. Nonetheless, this methodology

remains popular (for programs focused on DM) because it

is more rigorous than Options 1 and 2 and the calculations

are more straightforward than for Options 4 and 5.

We recommend using Metric Option 3 only for EHM

programs that primarily address chronic conditions,

especially when a suitable comparison group is not available.

For further guidance on evaluation of nancial impact of

programs that address chronic conditions, please see the

PHA Outcomes Guidelines Report vol. 5, pp. 55–64.

Directly-monetized savings Metric Option 4: Participant

vs. Non-participant cost comparison. Two basic

approaches are used to calculate this metric; both rely

on the assumption that the cost trajectories of EHM

participants (P) and non-participants (NP) would be

equal absent EHM. Thus the NP cost trend is used as the

expected (comparison) trend to calculate expected costs

for the P.

The simplest version of the P vs. NP metric—P vs. NP

cohort methodology—compares the cost trends of the

two P and NP cohorts (i.e., groups of the same people

tracked over time). Those trends may optionally be

adjusted for the difference in risk (predicted cost based

on their clinical proles) between the groups.

For EHM

with multiple components (e.g., health risk appraisals,

biometric screening, care gaps, telephone or online

coaching for lifestyle risks or chronic conditions), separate

participant (P) groups can be developed for each

component.

The comparison NP group is developed

from members targeted for EHM who did not participate.

A more rigorous version—the P vs. NP multivariate

methodology—is similar to the cohort methodology

but goes further to ensure that the NP comparison

population is well-matched to the P population. The

purpose of matching is to try to isolate the impact

of EHM by neutralizing, or controlling for, non-program

factors that might drive differences in cost trajectories

between the P and NP.

Cost Trend

PROSPECTIVE

EXPECTED

PROGRAM IMPACTABLE?

ACTUARIALLY

ADJUSTED “EXPECTED”

RETROSPECTIVE

ACTUAL

Demographic Changes 1.0% No 0.0% 0.0%

Risk Factors (net of

demographics)

0.0% Ye s 0.0% -1.2%

Unit Prices 6.0% No 5.2% 5.2%

Utilization 2.0% Ye s 2.0% 1.3%

Plan Design -1.0% No 1.0% 1.0%

Total Trend

8.1% 8.4% 6.2%

www.hero-health.org www.populationhealthalliance.org

Several ways are used to match the NP to the P groups.

The goal is to get the differences in cost-driving

characteristics between the groups to be statistically or

clinically insignicant. This allows us to assume that, at

least on characteristics that can be observed in the data,

the groups would be expected to exhibit identical cost

trajectories were it not for EHM. However, the method

cannot eliminate unobservable cost-driving differences

such as the effect of volunteering (selection bias).

Directly monetized savings Metric Option 5: Comparison

with matched controls in a non-exposed population.

This is the most rigorous of the directly monetized

savings metrics, and, if properly done, the closest

simulation to the gold standard randomized controlled

study. This is so because it uses a population not exposed

to EHM but otherwise similar along characteristics

(e.g., demographics, chronic conditions and historical cost

patterns) to the EHM-exposed population to develop

expected cost.

Key to success with this metric is to ensure that the

comparison population is like the EHM population in all

ways that are believed to drive cost trajectory, with the

exception that members in the EHM population have the

opportunity to participate in EHM. There are statistical

methods available to accomplish this and to determine

whether observable characteristics of the comparison

population are sufciently similar to the EHM population.

While this methodology is considered to be the most

rigorous of our recommended metrics, it requires a large

number of individuals in the comparison population

without EHM to ensure that all EHM-program participants

(or program-qualied)

can be matched with like individuals

in the non-EHM population.

Monetizable metrics

• EHM-impactible utilization (hospitalizations,

ER visits and procedures that can be potentially

impacted by EHM)

• Reduction in or prevention of lifestyle-related

health risk factors

Monetizable savings Metric 6: EHM-impactible utilization.

‘Utilization’ refers to use of health care services such as lab

testing, imaging, emergency room, hospitalizations, drugs,

and procedures. Depending on the components of an EHM

program, some utilization may be affected by EHM. The

impact may result in increased or decreased use of such

services. While we focus on decreased impactible utilization

here, it is important to recognize that EHM should increase

the use of certain services, such as preventive and screening

services, certain chronic medications, and outpatient visits.

It is even possible to see a rise in ER and urgent care visits

as well-informed patients learn to get urgent medical care

when they experience early warning signs of stroke, asthma,

or heart attack.

Because the cost of these services is known from claims

data, changes in their usage rates can be monetized

by multiplying the number of service units gained or lost

by the average service cost. As with directly monetized

metrics, the change in usage rate for a given service is

the difference between the expected and actual rates.

Example focusing on hospitalizations: By reducing and

preventing risks for chronic cardiovascular conditions,

diabetes, and COPD, a successful EHM program that

engages people with chronic conditions should reduce

unscheduled outpatient, emergency, and hospital visits

related to these diagnoses. Over several years, EHM

should reduce the rate at which people newly develop

these conditions. This example will focus on “potentially

preventable” hospitalizations (PPH) for these conditions.

The two pre-EHM years are selected as the comparison

period, and the analysis is performed on the program’s

second performance year (PY2).

The report shows that the population grew slightly between

the comparison period and the second performance year,

and that rates of chronic conditions in the population (i.e.,

prevalence) rose as well: the percent of the population with

one of the target chronic conditions rose from 6.6% in the

two pre-program years to 6.9% in the PY2. However, the

PPH rate declined from 3.14 to 2.62 per 1,000 members—

a decrease of 0.53 per 1,000 members or 9.26 for the

entire population (0.53 x 211,000/12,000).

www.hero-health.org www.populationhealthalliance.org

The slight rise in proportion of members with the target

conditions from the comparison period to PY2 may imply

that all else equal, PPH should have risen slightly if not

for EHM.

In fact, all-cause hospitalizations except PPH

declined slightly, but not nearly as much as the PPH decline,

suggesting (but not proving) that the PPH decline could be

attributed to the EHM programs.

Reduction in EHM-impactible hospitalizations may be due

to factors other than EHM, such as improved treatment of

established disease from drugs, medical devices and surgery;

movement of some treatment to outpatient settings; or

general improvement in risk factors not specically due to

EHM. Your analyst, consultant, or vendor should provide

this background context with your report.

EHM-impactible utilization metrics are important plausibility

markers in the value-production chain and EHM’s impact on

these metrics is the step just before (and the cause of) savings.

Your report should be clear about which types of utilization

were included in the savings calculation. Typically reports

should include the following types of utilization:

• Expect to see decreases in: ER

and hospital use

for common chronic conditions (e.g., heart disease,

stroke, asthma, COPD, and diabetes); use of

generic medications.

• Expect to see increases in: primary care visits,

screening services (e.g., breast, cervical or colon

cancer) and immunizations.

Monetizable savings Metric 7: Reduction in or prevention

of lifestyle-related health risk factors. Several studies

4,5

have concluded that worksite “health promotion”

programs— which generally focus on reducing lifestyle-

related health risk factors such as tobacco use, poor

nutrition, overweight, poor stress management, and physical

inactivity—reduce healthcare and absenteeism costs. It’s

thought that lower costs result from short- and long-term

reductions in the consequences of having these risk factors

(i.e., EHM-impactible utilization).

COMPARISON PERIOD

PY2

Prevalence Hosp Hosp/K Prevalence Hosp Hosp/K

Member-months 210,000 211,000

Member count 18,100 18,200

IVD 2.1% 32 1.83 2.6% 25 1.42

CHF 0.2% 3 0.17 0.3% 4 0.23

Diabetes 2.4% 3 0.17 3.2% 6 0.34

Asthma 2.7% 13 0.74 4.3% 9 0.51

COPD 0.4% 4 0.23 0.6% 2 0.11

PPH 6.6% 55 3.14 6.9% 46 2.62

All-cause hospitalizations 702 40.04 685 38.96

All-cause except PPH 647 36.97 639 36.64

Savings estimation

Trend: PPH -17%

Trend: All-cause except PPH -2%

Saved PPH/K 0.53

Saved PPH for population 9.26

Weighted cost/PPH $22,500

Saved PPH cost $208,393

Saved PPH cost PMPM $0.99

IVD: Ischemic vascular disease (coronary artery disease, peripheral artery disease and stroke); CHF: congestive heart failure; Prevalence: Percent of the

population with the listed condition; Hosp: hospitalization count; Hosp/K: Rate of hospitalizations per 1,000 members (per 12,000 member months); All-cause

hospitalizations include all hospitalizations except those for pregnancy, delivery, or newborns.

The average facility and professional cost per PPH in the PY2 was $22,500, so the 9.26 prevented hospitalizations

resulted in $208,393 in savings:

www.hero-health.org www.populationhealthalliance.org

Evidence for savings from prevention and reduction of

lifestyle-related risk factors comes from carefully-conducted

published studies. However, it should not be assumed that

the savings from preventing, as compared with the savings

from reducing, a risk factor are equal. While PHA and HERO

believe that evidence supports savings from reducing or

preventing risk factors, monetization for specic program

results is not yet supported by the evidence. Nevertheless

because of the great interest in this topic we will summarize

the evidence to date.

Risk prevention: HERO maintains a large, longitudinal database

to which several large employers contribute de-identied

data on their members’ healthcare costs, lifestyle risk

factors, and program participation. The HERO database

has supported several studies relating presence of these

risk factors to healthcare costs. The original HERO study

concluded that most of ten risk factors were associated

with increased cost. These results were supported by a

second larger HERO study published in late 2012.

Many

analysts use the studies’ ndings to monetize changes in the

individual risk factors in a population year over year, or in a

cohort of individuals tracked from one year to the next.

Risk reduction: Strictly speaking the HERO ndings should

be used to monetize prevented risks because the studies

showed the incremental cost of having, versus not having,

each risk factor. There is less evidence on which to base

savings from net risk reduction (i.e., the difference between

increases and decreases in risk factors). Except for a few

factors, the current evidence is strong enough only to

monetize savings per risk reduced, or for an individual’s

movement from a higher to a lower risk category.

SELECTING FINANCIAL METRICS:

A DECISION AID

Here’s guidance on selecting nancial metrics to best

evaluate your EHM program. Two important caveats:

First, this section is provided to help you work with your

analyst, consultant, or vendor and to assess their metric

recommendations. Second, these are intended to provide

guidance only. For example, if your membership size is only

15,000, your analyst may have good reason to believe that

some of the metrics labeled as suitable only for populations

larger than approximately 25,000 are valid in your situation.

As discussed earlier, there are two basic strategies of

metrics: models and measurements. The latter requires

accurate, adjudicated, nancial data, but it’s important to

note that models should still be based upon data specically

about your membership, programs and results. It’s not

necessarily always preferable to insist on a measurement

over a well-constructed model.

In selecting nancial metrics, your analyst, consultant,

or vendor will ask:

1. Do we have enough baseline (pre-program) claims

data? If so, is it of high enough quality? Baseline

data should include a minimum of 12 months (or 24

months for a more solid baseline) of membership,

eligibility, medical and pharmacy claims—and

preferably lab test (or biometric) results—and,

if appropriate, risk factor (HRA) data. It must be

possible to link individuals in the baseline data set

to those in the program years’ data.

2. Do we have fully adjudicated claims? If not, a dollar-

based analysis is not possible, though with an

accurate utilization le, Metric 5 could still be run.

3. Is our membership size more than approximately

25,000?

4. Do we have the analytic resources (expertise and

time) available for a methodologically sophisticated

study? If we do, do we need a sophisticated study?

If so, why?

5. Which EHM components are we implementing

(e.g., lifestyle coaching, case management, gaps in

care, disease management and maternity)?

6. Is the structure of our EHM program reasonably

close to those in published savings literature?

7. Does our consultant have a large benchmarking

database that includes employers in our industry?

8. If considering a rigorous study based on Metrics

4 and 5: Do our leading indicators indicate the

program has achieved enough initial success to make

it plausible to detect a sizable enough savings to

demonstrate ROI? Minimums we recommend are

50% health assessment and/or screening participation

rates; 30% enrollment into targeted coaching; and

15% or more participation in a population-wide

health improvement program (e.g., pedometer

program). These are minimums; typically, two years

of these data are required for a rigorous study.

✔ If your membership size is much less than 25,000, it’s

generally not advisable to run Metrics 1-5—though there

are circumstances where a valid result can be obtained with

substantially fewer members. Metric 6 may lose validity with

smaller populations as well due to the usually low PPH rate.

✔ If your program has a strong population health

orientation and have sufcient analytics expertise, and

have good justication for doing a rigorous study, HERO

and PHA recommend running Metrics 4 or 5 (the latter

if you have a good comparison population not exposed

to the components in your EHM program).

www.hero-health.org www.populationhealthalliance.org

✔ If your EHM program structure is fairly similar

to those in published savings studies, consider using

a model based on engagement, rates of lifestyle risk

prevention and reduction, and improvement in

clinical outcomes.

✔ Trend-based Metrics 1 and 2 may require input from

an actuary and, at best, can tell you only if your program

was cost-saving, but not what caused savings or what

worked or didn’t work. Combine trend-based metrics

with value-chain markers to gain the insight you need to

evolve your programs.

CHAPTER 2 REFERENCES

Grossmeier J, Terry P, Anderson DR, Wright S. Financial impact of population

health management programs: reevaluating the literature. Population Health

Management, 2012;15(3):129-134.

The SOLVD Investigators. Effect of enalapril on survival in patients with reduced

left ventricular ejection fraction and congestive heart failure. N Engl J Med

1991;325:293-302.

Grossmeier J, Terry P, Anderson DR, Wright S. Financial impact of population

health management programs: reevaluating the literature. Population Health

Management, 2012;15(3):129-134.

Grossmeier J, Terry P, Anderson DR, Wright S. Financial impact of population

health management programs: reevaluating the literature. Population Health

Management, 2012;15(3):129-134.

Baicker K, Cutler D, and Song Z. Workplace wellness can generate savings.

Health Affairs 2010;29(2):304-311.

(HERO 1) Goetzel RZ, Anderson DR, Whitmer RW, et al. The relationship

between modiable health risks and health care expenditures: An analysis of

the multi-employer HERO health risk and cost database. J Occ Envir Med

1998;40:843-854.

(HERO 2) Goetzel RZ, X Pei, MJ Tabrizi, et al. Ten modiable health risk factors

are linked to more than one-fth of employer-employee health care spending.

Health Affairs 2012;31(11):2474-2484.

CHAPTER 2 FOOTNOTES

This metric is included because we nd solid, consistent evidence of increased cost

in individuals with risk factors, and consistent evidence of savings when individuals

eliminate risk factors—but insufcient evidence from which to develop a specic

cost-saving model.

The recommendation to gure in all sources of value and all sources cost when

estimating ROI is meant to help employers evaluate their EHM program’s value from

a holistic perspective. It is not meant to suggest that performance guarantee ROIs

(which usually are stated as medical or medical plus productivity savings divided by

program fees) do so.

Expert opinion on whether savings should be modeled from gross or net risk

reduction. Under the assumption that EHM doesn't worsen risk factors, the gross

model takes credit for the number of risks reduced and ignores risks that newly

arose during the measurement period. The net model takes credit only for (risks

reduced - risks added). The gross model's proponents claim it more accurately

reects the impact of EHM; the net model's proponents claim that it better reects

the program's impact on the employer's nancial position, since risks acquired will

become nancial.

PHA Outcomes Guidelines Report, vol. 5 recommends “the appropriate use of

adjustment to achieve comparison group equivalence” (p. 15), and gives examples

of how to use risk adjustment in pp. 73-82. However, the use of risk adjustment

in cohorts is somewhat controversial because a cohort is composed of the same

people tracked over time. If your analyst uses this version of Metric 4, you should

ask if risk adjustment was used.

Many analysts believe that it’s not appropriate to measure the independent savings

from health risk appraisals and biometric screening on the grounds that their role is

to identify people with opportunities to improve their health and engage them with

programs and tools whose nancial value is the appropriate measurement target.

For example, it may be that people who volunteer are those who were about to

take the initiative to improve their health anyway; or that those who volunteer are

inherently earlier or later in their cost-trajectory.

Often it is not the P who are matched to like members of the comparison group

but rather members who were qualied to participate, regardless of whether they

did. This process eliminates the possibility of selection bias.

Typically, cardiovascular conditions include ischemic vascular disease (coronary

artery disease, peripheral artery disease, and stroke) and congestive heart failure.

The term, “for the condition”, designates hospitalizations where the condition

appeared on the claim or hospital discharge record as either the principal diagnosis

(main reason for which the patient was hospitalized) or in some cases the secondary

diagnosis (principal diagnosis was a complication of the EHM-impactible diagnosis—

for example, a diabetic hospitalized for lower limb amputation might show ‘diabetes’

as secondary diagnosis but since the amputation was due to a complication of

diabetes, such a hospitalization would be counted as a PPH).

Whether PPH rates should be adjusted for changes in prevalence of the related

conditions is controversial, but it is certainly useful to know whether the prevalence

of such conditions in your population is rising or falling.

Because of this tight relationship, some analysts have proposed replacing measured

nancial savings with monetized changes in utilization. While this would reduce the

‘noise’ introduced by direct-measurement methodologies, it doesn’t resolve the

question of whether the EHM program caused the reduced utilization; other factors

such as changes in benet design, the economy, and advances in medicine may

inuence utilization rates.

Use of ER services is a controversial measure of EHM effectiveness. In some

circumstances ER services for PPH-type conditions could increase with effective

EHM as members learn to attend proactively to the warning signs of clinical

decompensation and are treated in ER and released—thus preventing an inpatient

stay. However, because an ER visit that results in a hospitalization is not billed

separately, it might appear that ER visits are increasing while in fact some of them

are simply becoming visible as claims.

www.hero-health.org www.populationhealthalliance.org

INTRODUCTION

The initial task was to identify all areas that might reasonably

be construed to demonstrate the impact that an EHM

program might have on the overall health and well-being of

the population being served. To address this task, HERO

and PHA considered various models for health and well-

being.

1,2,3

These typically include dimensions such as: physical,

mental/emotional, social, environmental, spiritual, intellectual,

and occupational. Based on our review, we identied the 4

dimensions of health for inclusion in our basic set of health

impact measures that are discussed below.

Process for Selecting Health Dimensions

Choice of the health dimensions was based on consensus

among the workgroup members. However, to be included

a dimension had to meet at least one and usually more of

the following criteria:

1. Clear relationship with health outcomes

(as determined by literature review).

2. Clear relationship with healthcare and/or productivity

costs (as determined by literature review).

3. Able to be affected by employers via their EHM

programs.

4. Industry consensus on the importance of the dimension

(as determined by review of: (1) existing guidance

documents and (2) inclusion in more than one

(generally most) of the most widely used HRAs

(including WebMD, StayWell, Alere, HealthFitness,

University of Michigan, CDC, RedBrick, and Mayo Clinic).

Based on these criteria, the nal dimensions chosen for

the initial set of measures for the health impact section

of standards included:

• Physical Health (biometrics such as blood pressure,

height, weight, etc., and existence of chronic conditions)

• Mental/Emotional Health

• Health Behaviors

• Health Status

• Summary Health Measures (indices relating to

risk status)

Other dimensions that were discussed but not included

in this health impact section were:

Environmental and Occupational Health: It was the

consensus of the workgroup that these dimensions would

be addressed as part of organizational culture indices,

which were covered by a separate workgroup.

Intellectual Health: Not included due to lack of adequate

demonstration of the relationships with health outcomes

and with healthcare and productivity costs. While there

is signicant evidence that level of education can inuence

health status, this is a broader issue beyond specic

actions that should be taken as part of an employee health

management program. Encouraging life-long learning

could contribute to well-being and life satisfaction, but this

dimension more appropriately falls under the organizational

support section in the future.

Social Health: While there is evidence showing that social

support systems are important and effective in inuencing

health and behavior change, it is still not adequate to

demonstrate a direct relationship with health outcomes

and healthcare and productivity costs. Again, this

dimension could be addressed under the organizational

support section.

Spiritual Health (Purpose/Meaning in Life): This area is

gaining recognition but the evidence for the relationships

with health outcomes and costs were not considered

adequate for inclusion in this version. There is also a lack

of consensus on the role or inuence that an employer

can/should play with this dimension.

Although the consensus of the group was that each of

these dimensions are potentially important in promoting

the total health of the individual, there was less consensus

around how to best measure them as stand-alone areas

which impact health status. In addition, we could not identify

satisfactory brief surveys that could be incorporated into

our minimum measure set that addressed each of these

dimensions adequately. They will be reconsidered for

inclusion in future versions of this document.

CHAPTER 3: HEALTH IMPACT

Gordon D. Kaplan, PhD, and LaVaughn Palma-Davis, MA

www.hero-health.org www.populationhealthalliance.org

Criteria for Selecting Measures Within Health Dimensions

Measures (items) within each health dimension were

selected for inclusion based on the following general criteria:

1. The measure has been recognized by the eld as

being an important determinant of health outcomes

and healthcare/productivity costs. This determination

was based on literature review, review of major

HRAs, review of currently available guidance

documents, and the collective experience of the

workgroup members as subject matter experts.

2. The measure is reasonable for employers to

implement. This includes availability and ease of

measurement, reasonable cost, and, where applicable,

direct measurement capability. For example, measures

available in the public domain were preferred over

measures that were proprietary and, therefore, had

additional costs for use.

3. The measure is or can be made comparable across

program providers/vendors.

4. The measure can be affected by the employee health

management programs provided by employers.

5. Assessment of the measure does not signicantly

increase the overall length of the measurement tool.

The workgroup recognized the need to keep the

measurement tool to a reasonable length to improve

completion rates and to not be overly burdensome

to users who may wish to enhance the basic question

set with unique additional items. Therefore, wherever

possible single item measures were preferred over

measures requiring multiple items.

The following process was used to determine the measures

selected for inclusion in the nal draft of the health impact

basic measure set:

1. Literature Review

2. Review of currently available HRAs

3. Review of existing consensus documents

4. Expert opinion/consensus on the nal measure set

Special Issues

Length of Question Set: Ideally should be able to be

completed by participant in 15–20 minutes. Therefore a

goal was set to have no more than 30 items.

Direct vs. Self-report measures: We acknowledge that

to maximize the validity of the measures that are collected,

where possible the preferred measurement approach would

be direct measurement. An example would be the onsite

measurement of biometrics (height/weight, blood pressure,

blood lipids, blood glucose) or biochemically validated

tobacco use status. However, in most cases, direct

measurement may not be practical or even possible.

Therefore self-report remains the primary method for

collecting most of the health impact measures. Although

the validity of self-report has been challenged in certain

areas—particularly for health biometrics—consensus

in the eld is that self-report may be used with reasonable

condence that it represents the health status of the

individual.

4,5

A number of factors can inuence the likelihood

that individuals might bias their responses to HRA survey

items; these are reviewed in Donaldson and Grant-Vallone.

In the design of EHM programs, it is important to note that

when self-report data are being included, care should be

taken that participants are not incentivized to misrepresent

their status. Incentives that are based on outcomes

may require direct measurement for employers to be

comfortable with results.

Apart from these issues, it should be noted that self-

reported data are not without value in their own right.

These data reect the participant’s perceptions of their

health risks and allow the assessment of preventive

behaviors that are difcult to assess through direct

measurement. These perceptions are part of the

constellation of health status that is important for EHM

programs to address. For example, mismatches between

actual and perceived risk provide opportunities for more

effective tailoring of interventions. When self-reported

biometric values are used, we recommend that a ag be

included in the data set that allows self-report and direct

measures to be distinguished from one another.

Timing of Data Collection: The varying processes of how

Health Assessments (HA) are administered and how

biometric data are collected raises issues related to timing.

When these health impact measures are being used to

evaluate the overall impact of a health enhancement

program, they should be collected as close as possible

to the beginning and end of the evaluation period. The

rollout of the Health Assessment at the beginning of the

evaluation period, however, may allow people to complete

the assessment over a period of months, and in many

cases assessments are available for completion throughout

a program year. In addition, when biometrics are being

measured directly as in a health fair, there may be

reasonable scheduling issues that result in these measures

needing to be collected over a period of time, and

possibly not at the same time as other Health Assessment

questions are answered. Given these issues, the workgroup

recommends: (1) whenever possible HA’s should be

completed as near as possible to the beginning and end

of the program evaluation period. A reasonable time frame

www.hero-health.org www.populationhealthalliance.org

would be within 3 months of either side of the beginning

or the end of the program evaluation period. The employer

might consider offering incentives that would encourage

completion of the HA within the desired time frame.

(2) It is desirable that directly measured biometric and

self-reported HA data are collected at the same time.

When biometric data are being collected at a different point

in time (e.g., due to scheduling restraints), a reasonable rule

should be applied so that these data are as synchronized

as possible. We recommend a 3-month rule for combining

biometric and HA data collected at different points in time.

(3) In evaluating an EHM program, pre and post HA’s should

be sufciently far apart to allow for change to have taken

place. We recommend that HA’s be at least 9 months apart

to accomplish this purpose.

We also recommend that, if at all possible, the timing of the

health assessment and other measurements be consistent

year over year.

Specic Item/Response Option Wording: To maximize data

comparability across employers, it is desirable for all users

to include identical items and response options for these

basic measures. Having said that, minor variations in the

wording of an item which would not be expected to impact

the meaning of the measure may be acceptable. Examples

are included in the list of recommended measures that follow.

Basic vs. Expanded Measure Sets: The basic measure set is

being recommended as the basis for common measurement

across all employee health enhancement evaluation efforts.

However, the recommendations of this group do not mean

that individual users cannot add additional items to the

basic set as they look for ways to further rene their

evaluation strategy. The nature of the EHM program being

assessed may call for additional items to allow for a more

robust analysis. This document is not suggesting that this

should not be done. In addition, if the program being

evaluated does not address specic areas included in this

recommended measure set, it may be acceptable to remove

non-relevant items.

Are all Health Risks Equal? The answer to this question

may seem obvious. Not all risks are equal either with regard

to their impact on health status or their impact on near

or even long-term costs. These facts are functions both

of the strength of the association of risk factors to health

outcomes and utilization and the timing of the impact of

these risks on outcomes and utilization. So when it comes

to making decisions about what interventions to emphasize

in an EHM program, these factors should be considered

along with one other—the individual or population’s

readiness to address the risks identied. Generally, among

modiable risks, elevated biometric risks tend to reect

a point further along the health continuum from optimal

health and vitality to chronic illness and death. Therefore

their impact on health outcomes (e.g., adverse health events)

and costs will tend to be closer in proximity to their

measured status than would be the case for risks that are

considered part of lifestyle such as tobacco use, excessive

alcohol use, or lack of physical activity. And for each

biometric risk, the degree to which it is elevated also makes

a difference. So to the extent that elevated biometric

risks can be effectively identied and addressed within a

population, such an approach is likely to yield nearer term

results in terms of reduced healthcare and productivity costs.

It is also well-known, however, that traditional lifestyle-

related risks are often the precursors of biometric risks

and have been linked to increased health costs on their

own. Therefore, they continue to be the focus of and the

rst line of attack for prevention efforts. It is also important

to note that research on sets of risk factors shows that the

total number of elevated risks and the change in multiple

risk status is strongly associated with cost savings.

7,8,9,10

A basic message coming out of this research is that the

most important risk to change (from a set of elevated

risks) is generally going to be the one which the individual

is most ready to change.

The third factor of an individual’s

readiness to change, therefore, should be given serious

consideration in program design and the priority of risk

management within a population.

In summary, from a practical point of view for the

employer, most risks can be considered relatively equal

from the perspective of what should be done. In planning

interventions, employers should consider the risk factors

with the highest prevalence in their population and the

relationship between risk factors (e.g., obesity, physical

activity and nutrition are all linked), with the goal of

improving and maintaining low risk status and reducing

high risk status overall.

LIST OF RECOMMENDED MEASURES

The following list represents the minimum set of measures

recommended upon which a basic evaluation of the effect

of employer-sponsored health enhancement initiatives on

the health of the populations being served can be made.

References for each measure can be found at the end

of this section.

www.hero-health.org www.populationhealthalliance.org

DIMENSION 1: PHYSICAL HEALTH IMPACT

These items/measures represent the minimum set of indices

that can be used to judge the impact of health enhancement

programs on participants’ overall physical health status.

1. BMI (derived from Height; Weight)

12,13,14,15,16,17

Method:

Direct Measurement and Data Entry (preferred)

Self-report (if direct measurement is not possible)

Suggested Item:

Please enter your height and weight below. (If you are

a female and are currently pregnant, please enter your

pre-pregnancy weight.)

Height (without shoes): ̱ ̱ ̱ ̱ ̱ ̱ ft ̱ ̱ ̱ ̱ ̱ ̱ in

Weight (without clothes): ̱ ̱ ̱ ̱ ̱ ̱ pounds

Notes:

Reasonable variant ways of asking for these measures

are acceptable.

Note that for certain populations, BMI may not

adequately represent risk. For example, athletes with

higher lean body mass may have elevated BMIs but not

be at risk due to low body fat levels. For such populations

it would be desirable to add a measure of body fat.

At Risk Denitions:

Not at Risk: BMI = 18.5–24.9

At Risk, Underweight: BMI < 18.5

At Risk, Overweight: BMI = 25.0–29.9

At Risk, Obese: BMI >= 30.0

2. Blood Pressure (Systolic and Diastoic)

18,19,20,21,22,23

Method:

Direct Measurement and Data Entry (preferred)

Self-report (if direct measurement is not possible)

Suggested Item:

If your blood pressure was checked within the past year,

what was it when it was last checked? Enter the value or

check one of the options listed below.

̱̱̱̱̱̱̱̱̱̱ /̱̱̱̱̱̱̱̱̱̱ mm Hg

̱ ̱ ̱ ̱ ̱ Low or Normal (Below 120/80)

̱ ̱ ̱ ̱ ̱ Borderline high (120/80 to 139/89)

̱ ̱ ̱ ̱ ̱ High (140/90 or higher)

̱ ̱ ̱ ̱ ̱ Don't Know/Not Sure

Notes:

If individuals do not know or remember their last blood

pressure, it is advisable to allow them to give their best

estimate using ranges such as shown in the suggested

item. Reasonable variant ways of asking for these

measures are acceptable.

At Risk Denitions:

Not at Risk: BP < 120/70

At Risk, Borderline High: BP 120–139.9/80–89

At Risk, High: BP >= 140/90

3. Cholesterol (Total; HDL; LDL)

24,25,26,27,28,29

Method:

Direct Measurement and Data Entry (preferred)

Self-report (if direct measurement is not possible)

Suggested Items:

If your total cholesterol was checked within the past

year, what was it when it was last checked? Enter the

value or check one of the options listed below.

̱̱̱̱̱̱̱̱̱̱ mg/dL

̱ ̱ ̱ ̱ ̱ Desirable (Below 200)

̱ ̱ ̱ ̱ ̱ Borderline high (200 to 239)

̱ ̱ ̱ ̱ ̱ High (240 or higher)

̱ ̱ ̱ ̱ ̱ Don't Know/Not Sure

If your HDL cholesterol was checked within the past

year, what was it when it was last checked? Enter the

value or check one of the options listed below.

̱̱̱̱̱̱̱̱̱̱ mg/dL

̱ ̱ ̱ ̱ ̱ Low (Below 40)

̱ ̱ ̱ ̱ ̱ Average (40–59)

̱ ̱ ̱ ̱ ̱ High (60 or higher)

̱ ̱ ̱ ̱ ̱ Don't Know/Not Sure

If your LDL cholesterol was checked within the past

year, what was it when it was last checked? Enter the

value or check one of the options listed below.

̱̱̱̱̱̱̱̱̱̱ mg/dL

̱ ̱ ̱ ̱ ̱ Optimal (Below 100)

̱ ̱ ̱ ̱ ̱ Near optimal/above optimal (100 to 129)

̱ ̱ ̱ ̱ ̱ Borderline high (130 to 159)

̱ ̱ ̱ ̱ ̱ High (160 to 189)

̱ ̱ ̱ ̱ ̱ Very High (190 or higher)

̱ ̱ ̱ ̱ ̱ Don't Know/Not Sure

Notes:

If individuals do not know or remember their last cholesterol

values, it is advisable to allow them to give their best

estimate using ranges such as shown in the suggested items.

Reasonable variant ways of asking for these measures

are acceptable.

At Risk Denitions:

Not at Risk: TC < 200 and HDL >= 40 and LDL < 100

At Risk, Moderate Risk: TC = 200–239; HDL >= 40;

LDL = 100–159

At Risk, High Risk: TC >= 240 or HDL < 40 or LDL >= 160

www.hero-health.org www.populationhealthalliance.org

4. Fasting Blood Glucose30,31,32,33,34

Method:

Direct Measurement and Data Entry (preferred)

Self-report (if direct measurement is not possible)

Suggested Item:

If your blood glucose (blood sugar) was checked within

the past year, what was it when it was last checked?

̱̱̱̱̱̱̱̱̱̱ mg/dL

Was this value taken after you had not had anything

to eat or drink besides water for at least 8 hours

(check Fasting) or not (check Non-Fasting)?

❍ Fasting ❍ Non-Fasting

If you do not know your last blood glucose value, check

one of the options listed below.

̱ ̱ ̱ ̱ ̱ Low (Fasting blood glucose less than 70)

̱ ̱ ̱ ̱ ̱ Desirable (Fasting blood glucose 70–99)

̱ ̱ ̱ ̱ ̱ Borderline high (Fasting blood glucose between

100 to 125)

̱ ̱ ̱ ̱ ̱ High (Fasting blood glucose 126 or higher)

̱ ̱ ̱ ̱ ̱ Don't Know/Not Sure

Notes:

If individuals do not know or remember their last blood

glucose value, it is advisable to allow them to give

their best estimate using ranges such as shown in the

Suggested Item:

If your A1c level was checked within the past six months,

what was it when it was last checked?

If you have diabetes:

̱ ̱ ̱ ̱ ̱ Desirable (Below 7.0)

̱ ̱ ̱ ̱ ̱ High (7.0–8.9)

̱ ̱ ̱ ̱ ̱ Very high (9.0 or higher)

̱ ̱ ̱ ̱ ̱ Don't Know/Not Sure

If you do not have diabetes:

̱ ̱ ̱ ̱ ̱ Desirable (Below 5.7)

̱ ̱ ̱ ̱ ̱ Somewhat High (5.7–6.4)

̱ ̱ ̱ ̱ ̱ High (6.5 or higher)

̱ ̱ ̱ ̱ ̱ Don't Know/Not Sure

Notes:

If individuals do not know or remember their last HbA1c

value, it is advisable to allow them to give their best

estimate using ranges such as shown in the suggested item.

Reasonable variant ways of asking for these measures

are acceptable.

At Risk Denitions:

Diabetics:

Not at Risk: A1c <7.0

At Risk (Moderate): A1c =7.0–8.9

At Risk (High): A1c >= 9.0

Non-diabetics:

Not at Risk: A1c < 5.7

At Risk (Moderate): A1c = 5.7–6.4

At Risk (High): A1c >= 6.5

5. Medical Conditions

39,40

Method:

Self-report (can augment with claims-based identication

if available but should not replace self-report)

Suggested Item:

Do you have:

Arthritis

̱ ̱ ̱ ̱ ̱ Never ̱ ̱ ̱ ̱ ̱ In the past ̱ ̱ ̱ ̱ ̱ Have currently

Asthma

̱ ̱ ̱ ̱ ̱ Never ̱ ̱ ̱ ̱ ̱ In the past ̱ ̱ ̱ ̱ ̱ Have currently

Back Pain

̱ ̱ ̱ ̱ ̱ Never ̱ ̱ ̱ ̱ ̱ In the past ̱ ̱ ̱ ̱ ̱ Have currently

Cancer

̱ ̱ ̱ ̱ ̱ Never ̱ ̱ ̱ ̱ ̱ In the past ̱ ̱ ̱ ̱ ̱ Have currently

Depression

̱ ̱ ̱ ̱ ̱ Never ̱ ̱ ̱ ̱ ̱ In the past ̱ ̱ ̱ ̱ ̱ Have currently

Diabetes

̱ ̱ ̱ ̱ ̱ Never ̱ ̱ ̱ ̱ ̱ In the past ̱ ̱ ̱ ̱ ̱ Have currently

Heart Disease

̱ ̱ ̱ ̱ ̱ Never ̱ ̱ ̱ ̱ ̱ In the past ̱ ̱ ̱ ̱ ̱ Have currently

www.hero-health.org www.populationhealthalliance.org

Heart Failure

̱ ̱ ̱ ̱ ̱ Never ̱ ̱ ̱ ̱ ̱ In the past ̱ ̱ ̱ ̱ ̱ Have currently

Hypertension

̱ ̱ ̱ ̱ ̱ Never ̱ ̱ ̱ ̱ ̱ In the past ̱ ̱ ̱ ̱ ̱ Have currently

Hyperlipidemia

̱ ̱ ̱ ̱ ̱ Never ̱ ̱ ̱ ̱ ̱ In the past ̱ ̱ ̱ ̱ ̱ Have currently

Lung Disease, other than asthma

(e.g. COPD, Chronic bronchitis, Emphysema)

̱ ̱ ̱ ̱ ̱ Never ̱ ̱ ̱ ̱ ̱ In the past ̱ ̱ ̱ ̱ ̱ Have currently

Chronic Insomnia

̱ ̱ ̱ ̱ ̱ Never ̱ ̱ ̱ ̱ ̱ In the past ̱ ̱ ̱ ̱ ̱ Have currently

Notes:

These are the minimum conditions that we recommend

should be assessed; others may be added if there is a good

reason for doing so (such as a particular initiative being

pursued by the employer in collaboration with a vendor).

Reasonable variant ways of asking about these conditions

are acceptable. The concept is to identify current

medical conditions. It is advisable to give denitions

for medical terms that may be unfamiliar to individuals

(this is not shown in the suggested item).

At Risk Denitions:

Not Really Applicable: Since wellness programs cannot

eliminate the presence of a condition. Within a value

context, this information is mainly useful to distinguish

among populations with one or more medical conditions

in order to see the impact of a wellness program on

the health of these subpopulations.

DIMENSION 2: MENTAL AND EMOTIONAL

HE ALTH IMPACT

These items/measures represent the minimum set of

indices that can be used to judge the impact of health

promotion programs on participants’ overall mental

and emotional health status. Mental health issues are

a signicant cost area for employers in health care,

absenteeism, productivity and disability. Collaboration

between an employer’s EHM and EAP programs can

be very useful in addressing these issues offering both

individual and organizational approaches.

6. Perceived Stress

41,42,43,44,45,46

Method:

Self-report

Suggested Item:

How often is stress a problem for you in handling such

things as your:

health

nances

family or social relationships, or work?

Answer options for each:

̱ ̱ ̱ ̱ ̱ Never or rarely

̱̱̱̱̱ Sometimes

̱̱̱̱̱ Often

̱̱̱̱̱ Always

Notes:

Consider "at risk" if individual answers "often" or "always."

Although harder to measure across individuals, most

authorities agree that a high level perceived stress is

likely to have a negative impact on health either directly

or indirectly (by affecting adherence to healthy lifestyles

or prescribed health management regimens).

There is no agreed upon gold standard for assessing

perceived stress, although most SMEs agree that it has

meaningful impact on health and/or adherence to a

health-promoting lifestyle. Proprietary scales exist

(e.g. Cohen, 1983) but have associated costs and also

add length to a survey.

Given that various simple self-report items have been

used in research demonstrating the relationship among

multiple risk factors and healthcare/productivity costs,

it seems prudent to recommend including such an item,

pending the emergence of an improved measure.

Reasonable variant ways of asking for this measure are

acceptable. Concept is to assess the extent to which

the individual perceives his/her current stress level

to be a problem.

At Risk Denitions:

Not at Risk: Never or rarely, Sometimes

At Risk: Often, Always

7. Depression

47,48

Method:

Self-report

Suggested Items:

Over the last 2 weeks, how often have you been

bothered by feeling down, depressed, or hopeless?

̱ ̱ ̱ ̱ ̱ Not at all

̱ ̱ ̱ ̱ ̱ Several days

̱ ̱ ̱ ̱ ̱ More than half the days

̱ ̱ ̱ ̱ ̱ Nearly every day

Over the last 2 weeks, how often have you been

bothered by little interest or pleasure in doing things?

̱ ̱ ̱ ̱ ̱ Not at all

̱ ̱ ̱ ̱ ̱ Several days

̱ ̱ ̱ ̱ ̱ More than half the days

̱ ̱ ̱ ̱ ̱ Nearly every day

www.hero-health.org www.populationhealthalliance.org

Notes:

Both items are required. This approach to depression

screening allows for scoring.

There are longer instruments

for evaluating depression (CESD; Beck; PHQ); using

these would signicantly increase the length of the

survey. Consider using a longer instrument as part of an

intervention program rather than as an initial screener.

At Risk Denitions:

Not at Risk: Score < 3

At Risk: Score >= 3

8. Anxiety

Method:

Self-report

Suggested Items:

Over the last 2 weeks, how often have you been

bothered by feeling nervous, anxious, or on edge?

̱ ̱ ̱ ̱ ̱ Not at all

̱ ̱ ̱ ̱ ̱ Several days

̱ ̱ ̱ ̱ ̱ More than half the days

̱ ̱ ̱ ̱ ̱ Nearly every day

Over the last 2 weeks, how often have you been

bothered by not being able to stop or control worrying?

̱ ̱ ̱ ̱ ̱ Not at all

̱ ̱ ̱ ̱ ̱ Several days

̱ ̱ ̱ ̱ ̱ More than half the days

̱ ̱ ̱ ̱ ̱ Nearly every day

Notes:

Both items are required. This approach to anxiety

screening allows for scoring.

At Risk Denitions:

Not at Risk: Score < 3

At Risk: Score >= 3

9. Perceived Life Satisfaction

Method:

Self-report

Suggested Item:

In general, how satised are you with your life (include

personal and professional aspects)?

̱ ̱ ̱ ̱ ̱ Completely satised

̱ ̱ ̱ ̱ ̱ Mostly satised

̱ ̱ ̱ ̱ ̱ Partly satised

̱ ̱ ̱ ̱ ̱ Not satised

Notes:

Perceived life satisfaction has been correlated with

health status and annual health care costs.

At Risk Denitions:

Not at Risk: Completely or Mostly Satised

At Risk: Partly or Not Satised

DIMENSION 3: HEALTH BEHAVIORS THAT IMPACT

PHYSICAL/MENTAL AND EMOTIONAL HEALTH

10. Physical Activity (Total amount)

53,54,55,56,57,58

Method:

Self-report (consider augmenting by direct

measurement using, for example, pedometer

or accelerometer data if available)

Suggested Items:

Consider any high intensity activity that you do either

at work or in your leisure time. In a typical week,

how many days do you get at least 20 minutes of high

intensity physical activity? You may count any high

intensity activity that you do that lasts at least 10 minutes

at a time. (High intensity activities are activities that

increase your heart rate, make you sweat, and may

make you feel out of breath. Examples include jogging,

running, fast cycling, aerobics classes, swimming laps,

singles tennis, etc.)

Answer options: 0–7 days

Consider any moderate intensity activity that you do

either at work or in your leisure time. In a typical week,

how many days do you get at least 30 minutes of

moderate intensity physical activity? You may count any

moderate intensity activity that you do that lasts at least

10 minutes at a time. (Moderate intensity activities are

activities that require more effort than is needed to

carry out typical everyday tasks. Examples include

brisk walking, gardening, slow cycling, dancing, doubles

tennis, etc.)

Answer options: 0–7 days

Notes:

There are several ways this can be done within the

context of an HRA. NCQA guidance suggests one

of two possible methods. We recommend that one

of these be used.

Method 1: Days per week of 20 (30) minutes of

vigorous (moderate) activity (see suggested item)

For survey items that ask respondents how many days

per week they got at least 20 (30) minutes of vigorous

(moderate) activity, count the number of days for each

type of activity level.

Respondent is considered at risk for physical activity

if NONE of the following are met:

• Number of days of vigorous activity for at least

20 minutes < 3

• Number of days of moderate activity for at least

30 minutes < 5

www.hero-health.org www.populationhealthalliance.org

Method 2: Time per week using minutes per day

of vigorous (moderate) activity

For survey items that ask respondents how much time

per day spent doing vigorous (moderate) activity in

either actual minutes or number of 10 minute intervals:

Calculate total minutes per week of vigorous activities

• Calculate total minutes per week of moderate activities

• Calculate total amount of metabolic equivalent

(MET) minutes per week (optional)

- Multiply total minutes per week of vigorous

activities by 7.5

- Multiply total minutes per week of moderate

activities by 3.0

- Add the two values to determine the total

MET minutes per week

Respondent is considered at risk for physical inactivity

if NONE of the following are met:

• Total minutes per week of vigorous activities < 60

• Total minutes per week of moderate activities <150

•

Total combined MET minutes per week < 450 (optional)

A similar approach is taken by the CDC in their

recommendation that people achieve a minimum

of 150 moderate minute equivalents.

At Risk Denitions:

Low Risk: See above per method chosen

At Risk: See above per method chosen

11. Tobacco Use (all types)

60,61,62,63,64

Method:

Self-report (consider validation by biochemical testing)

Suggested Item:

Do you currently use any of the following tobacco products?

1. Cigarettes

[Daily; Some days; Not any more; Never used]

2. Cigars

[Daily; Some days; Not any more; Never used]

3. Pipes

[Daily; Some days; Not any more; Never used]

4. Smokeless tobacco

[Daily; Some days; Not any more; Never used]

Notes:

Consider at risk if any current use of tobacco.

Reasonable variant ways of asking about current

tobacco use are acceptable.

Survey items must determine whether the participant

currently smokes cigarettes. NCQA recommends using

validated survey items. To ensure the comparability of

populations identied as current cigarette smokers, the

survey items used must be able to:

• Identify smokers who smoke cigarettes under certain

circumstances (e.g., social occasions) and who may

not consider themselves as “smokers.” For example,

avoid asking only, “Are you a cigarette smoker?”

• Not identify individuals who have tried cigarettes

(e.g., one cigarette, one “puff”) but would not be

considered smokers. For example, avoid asking only,

“Have you ever smoked cigarettes?”

• Differentiate between cigarette smoking and other

types of smoking such as a pipe or cigar. For example,

avoid asking only, “Are you a smoker?”

Since any use of tobacco is considered to put an

individual at increased risk, additional items beyond

current use such as amount used or pattern of use

are desirable but optional. If the individual has quit

tobacco use, the time since quit should also be assessed

because it is required for NCQA accreditation.

At Risk Denitions:

Low Risk: No tobacco use

At Risk: Any tobacco use

12. Alcohol Use (Total Amount/Risky Drinking)

65,66,67,68,69

Method:

Self-report

Suggested Items:

Total Amount:

How many drinks of alcoholic beverages do you have

in a typical week? (one drink = one beer, glass of wine,

shot of liquor or mixed drink)

̱̱̱̱̱̱̱̱̱̱ {enter value}

At Risk Denitions:

Low Risk: Males <= 14 drinks/week;

Females <= 7 drinks/week

At Risk: Males > 14 drinks/week;

Females > 7 drinks/week

Risky Drinking:

During the past year, on any single day how often have

you had:

For men: More than 4 standard drinks?

For women: More than 3 standard drinks?

̱̱̱̱̱ Never

̱ ̱ ̱ ̱ ̱ One day

̱ ̱ ̱ ̱ ̱ 2–3 days

̱ ̱ ̱ ̱ ̱ More than 3 days

At Risk Denitions:

Low Risk: Never

At Risk: One or more days

www.hero-health.org www.populationhealthalliance.org

Notes:

Reasonable variant ways of asking for this measure are

acceptable. The item(s) selected needs to be able to

distinguish males who routinely consume > 2 drinks/

day or > 14 drinks/week and females who routinely

consume > 1 drink/day or > 7 drinks/week. These are

generally considered "at risk" levels.

The risky drinking item is meant to screen for excessive

drinking (rather than "at risk" or "heavy" drinking for

which the item above is used).

13. Fruit/Vegetable Intake

70,71,72

Method:

Self-report

Suggested Item:

Think of the foods that are a part of your normal diet.

How many servings of fruits and vegetables do you eat

in a normal day? One serving = ½ cup fresh, chopped,

cooked or canned vegetables; 1 cup leafy greens;

medium piece of fruit or ¾ cup juice.

Less than one serving

1 serving

2 servings

3 servings

4 servings

5 or more servings

Notes:

Consider at risk if fewer than 5 servings of fruits/

vegetables.

Reasonable variant ways of asking for these measures

are acceptable, as is asking for fruit and vegetable

servings separately.

Most supporting literature suggests combining fruits

and vegetables into one item; however, to relate to

Surgeon General targets, it is necessary to ask about

each separately.

At Risk Denitions:

Low Risk: 5 or more servings/day

At Risk: < 5 servings/day

14. Sleep (Typical hours/night)

73,74

Method:

Self-report

Suggested Item:

How many hours of sleep do you usually get at night?

6 hours or less

7 hours

8 hours

9 hours or more

Notes:

Reasonable variant ways of asking for this measure are

acceptable as are expanded answer options. At risk is

dened as less than 8 hours for those aged 18 to 21

years and less than 7 hours for those aged 22 years and

older, on average, during a 24-hour period.

At Risk Denitions:

Low Risk: 7–8 hours

At Risk: < 7 hours or > 8 hours

15. Daytime Sleepiness

Method:

Self-report

Suggested Item:

In the past 7 days, how often have you felt sleepy during

the daytime?

Always

Usually

Sometimes

Rarely

Never

Notes:

Because individual sleep needs vary and because actual

hours and restfulness of sleep are different issues, it is

recommended that some measure of daytime (waking

hours) sleepiness/fatigue is assessed.

There is no consensus on a single item to assess this

measure. Best available scale is probably the Epworth

Sleepiness Scale but it is proprietary.

Reasonable variant ways of asking for these measures

are acceptable. Consider at risk if individual reports

being tired/ sleepy more than occasionally during their

waking hours.

At Risk Denitions:

Low Risk: Rarely or Never

At Risk: Sometimes (Moderate Risk); Usually or Always

(High Risk)

16. Safety Restraint Use

76,77

Method:

Self-report

Suggested Item:

How often do you buckle your seat belt when driving

or riding in a motor vehicle?

Always

Almost always

Sometimes

Seldom

Never

www.hero-health.org www.populationhealthalliance.org

Notes:

Reasonable variant ways of asking for this measure are

acceptable. Consider at risk if individual does not always

use a seat belt when driving or riding in a motor vehicle.

At Risk Denitions:

Low Risk: Always or Almost always

At Risk: Sometimes (Moderate Risk); Seldom or Never

(High Risk)

17. Drinking/Driving:

Method:

Self-report

Suggested Item:

Do you ever drive after drinking, or ride with a driver

who has been drinking?

{Yes/No}

Notes:

Reasonable variant ways of asking for this measure

are acceptable.

At Risk Denitions:

Low Risk: No

At Risk: Yes

18. Health Screenings According to Recommended

Schedule (Blood Pressure; Glucose/A1c; Cholesterol;

Colorectal, Cervical, Breast Cancer; and Tuberculosis

for selected work settings)

Method:

Self-report; May augment with claims if available.

Suggested Items:

How long has it been since you last had your blood

cholesterol checked?

Less than one year

1–2 years ago

3–5 years ago

More than 5 years ago

Never

Don’t know

Notes:

Reasonable variants are acceptable. The item should be

able to distinguish between those who meet current

screening recommendations and those who do not.

The stem and answer options will differ based on the

screening. Item should ask for date of last screening or if

screening has occurred within recommended time frame.

At Risk Denitions:

Low Risk: See National Guidelines

At Risk: See National Guidelines

19. Immunization Status

Method:

Self-report; May augment with claims if available.

Suggested Items:

Have you been immunized or received a shot for:

Flu (in the most recent u season) Yes/No

Tetanus/Diptheria booster in the last 10 years Yes/No

Notes:

Reasonable variant ways of asking for this measure are

acceptable. With regard to u in particular, it may also

be desirable to capture information about the timing

of u immunization since getting a u shot prior to

the onset of u season is the most effective way to

prevent inuenza.

Depending on the demographics and job types of the

population being surveyed, other immunizations might

be considered for inclusion such as:

Pneumonia—for populations aged 65 or older;

Varicella/Zoster (Chickenpox)—for populations

aged 60 or older; or

Tuberculosis—for populations working in healthcare

settings.

At Risk Denitions:

Low Risk: Yes

At Risk: No

DIMENSION 4: HEALTH STATUS

20. Perceived Health Status

80,81,82

Method:

Self-report

Suggested Item:

In general, would you say your health is:

Excellent

Very good

Good

Fair

Poor

Notes:

Perceived health status has been correlated with health

status and annual health care costs.

This is a seminal, well-documented item, related both to

health and costs.

At Risk Denitions:

Not at Risk: Excellent, Very Good

At Risk, Moderate: Good

At Risk, High: Fair, Poor

www.hero-health.org www.populationhealthalliance.org

21. Healthy Days—Physical

Method:

Self-report

Suggested Item:

Now thinking about your physical health, which includes

physical illness and injury, for how many days during the

past 30 days was your physical health not good?

{Answer options: 0–30}

Notes:

Widely used measure from the CDC. There is also an

index developed by the CDC that requires an additional

item: During the past 30 days, for about how many

days did poor physical or mental health keep you from

doing your usual activities, such as self-care, work,

or recreation? Employers wishing to use this index

will need to add this item, but it is not necessary to

demonstrate health impact.

22. Healthy Days—Mental

Method:

Self-report

Suggested Item:

Now thinking about your mental health, for how many

days during the past 30 days was your mental health

not good?

{Answer options: 0–30}

Notes:

See note for Healthy Days—Physical.

Note: One area considered but not included in the

basic measurement set were measures of function such

as those provided by the SF-12 (Role mental/emotional

function and role physical function). These are excellent

measures to include, but they require the use of a

proprietary tool and also increase the length of the

survey beyond the limits we were trying to achieve.

In addition, these measures are less commonly used

by employers.

DIMENSION 5: SUMMARY HEALTH MEASURES

83,84

The following indices are recommended for evaluating the

impact of an EHM on the health of the population.

Overall Risk Reduction; Maintenance of Low Risk Status;

and Net Risk Reduction

Overall Risk Reduction can be used to describe the overall

change in the number of elevated health risks in a population

over time. This metric can be based on whatever total

number of risks that an employer deems important, but

a standard set of 10 is recommended as a minimum. This

will allow comparisons across wellness programs. Each risk

factor is assigned a risk status based on national guidelines,

where available, or expert opinion, where not. There are 4

medical risks: BMI; Cholesterol (at risk if TC, HDL or LDL is

at risk); Blood Glucose; and Blood Pressure (at risk if either

systolic or diastolic blood pressure is at risk). There are also

6 lifestyle risks: Tobacco Use (any = at risk); Alcohol Use;

Physical Activity; Fruit/Vegetable Intake; Stress; and Seat Belt

Use. At risk denitions are given in the previous sections.

Overall risk reduction then becomes the change in the total

number of elevated risk factors (out of 10 possible) between

two time periods.

It is also possible to assign risk status levels (low, medium,

high) based on the total number of elevated risks. The

denitions for risk status may depend on the risk set

considered; however, as a general rule of thumb, people

with 0 or 1 risk may be assigned low risk status, while those

with 5 or more risks would be high risk. By doing this it is

possible to focus on the percent of individuals maintaining

low risk status in your population. Maintaining low risk status

has been shown to be important for controlling healthcare

and productivity costs.

Finally, it is well-known that people move in both directions

with regard to health risk; therefore, an even better

indicator of the impact of an EHM program on health risks

is net risk reduction. This is dened as the total number

of risks in the population that decreased minus the total

number of risks that increased.

Individual Risk Reduction:

The same approach can be used to create indices to

determine value for reducing specic risk factors. Using

BMI risk status as an example, the following three metrics

could be reported:

(1) the change in the total number of people at risk

for obesity (BMI >= 30) over time (e.g. obesity

decreased by x% in the population over a specic

time period);

(2) the net change in BMI risk in the population (total

number improving BMI risk status—total number

increasing BMI risk status) was x% in the population

over a specic time period; and

(3) x% of the population maintained their low risk status

with regard to BMI over a specic time period.

www.hero-health.org www.populationhealthalliance.org

CHAPTER 3 FOOTNOTES

As the eld of EHM matures, additional measures will be considered. Emerging

studies point to well-being metrics as important determinants of health outcomes

and healthcare / productivity costs. Well-being metrics include the measurement

of several interrelated elements such as sense of purpose, social relationships,

nancial security, relationship to community and physical health. Research from

Gallup and Healthways shows that high well-being individuals cost less and perform

better than others.

Employers may want to verify new compliance rules and guidelines around

outcomes based incentives.

CHAPTER 3 REFERENCES

O’Donnell MP. The face of wellness: Aspirational vision of health, renewing health

behavior change process and balanced portfolio approach to planning change

strategies. The Art of Health Promotion 2008 (November/December): 1–12.

Burton, J. WHO healthy workplace framework and model: Background and

supporting literature and practices.

http://www.who.int/occupational_health/healthy_workplace_framework.pdf

Pronk N. Worksite health promotion: A global approach. ACSM Health and

Fitness Journal; 2011, 15:48–50.

Bourgeois FT, Porter SC , Valim C, Jackson T, Cook EF, Mandl KD. The Value

of Patient Self-report for Disease Surveillance. Journal of the American Medical

Informatics Association; 2007; 14:765–771.

Noyce et al. Uncommon knowledge: The value of health assessment data.

http://www.shrm.org/hrdisciplines/benets/articles/pages/assessmentdata.aspx

Donaldson SI and Grant-Vallone EJ. Understanding self-report bias in organizational

behavior research. Journal of Business and Psychology; 2002; 17:245–260.

Edington DW. Zero Trends:Health as a Serious Economic Strategy. Health

Management Research Center, University of Michigan, 2009.

Goetzel RZ, Anderson DR, Whitmer RW, Ozminkowski RJ, Dunn RL, Wasserman

J, The Health Enhancement Research Organization (HERO) Research Committee.

The Relationship Between Modiable Health Risks and Health Care Expenditures:

An Analysis of the Multi-Employer HERO Health Risk and Cost Database. Journal of

Occupational & Environmental Medicine; 1998; 40:843–854.

Goetzel RZ, Pei X, Tabrizi MJ, Henke RM, Kowlessar M, Nelson CF, and Metz RD.

Ten modiable health risk factors are linked to more than one-fth of employer-

employee health care spending. Health Affairs; 2012; 31(11):2474–2484.

Kowlessar NM, Goetzel RZ, Carls GS, Tabrizi MJ, and Guindon A. The

Relationship Between 11 Health Risks and Medical and Productivity Costs for

a Large Employer. Journal of Occupational & Environmental Medicine; 2011;

53:468–477.

Prochaska JO, Norcross JC and DiClemente CC. Changing for Good. Avon Books:

New York, 1994.

CDC. Behavioral Risk Factor Surveillance System. http://www.cdc.gov/brfss/

Edington DW. Zero Trends:Health as a Serious Economic Strategy. Health

Management Research Center, University of Michigan, 2009.

Goetzel RZ, Pei X, Tabrizi MJ, Henke RM, Kowlessar M, Nelson CF, and Metz RD.

Ten modiable health risk factors are linked to more than one-fth of employer-

employee health care spending. Health Affairs; 2012; 31(11):2474–2484.

Goetzel, RZ; Staley, P; Ogden, L; Stange, P; Fox, J; Spangler, J; Tabrizi, M;

Beckowski, M; Kowlessar, N; Glasgow ,RE, Taylor, MV. A framework for patient-

centered health risk assessments—providing health promotion and disease

prevention services to Medicare beneciaries. Atlanta, GA: US Department of

Health and Human Services, Centers for Disease Control and Prevention, 2011.

Available at: http://www.cdc.gov/policy/opth/hra/.

Healthy People 2020. http://www.healthypeople.gov/2020/default.aspx

National Institutes of Health. Clinical Guidelines on the Identication, Evaluation,

and Treatment of Overweight and Obesity in Adults: The Evidence Report. NIH

Publication No. 98–4083, September 1998.

CDC. Behavioral Risk Factor Surveillance System. http://www.cdc.gov/brfss/

Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA, Izzo JL, Jr, Jones

DW, Materson BJ, Oparil S, Wright JT, Jr, Roccella EJ, and the National High Blood

Pressure Education Program Coordinating Committee. Seventh Report of the Joint

National Committee on Prevention, Detection, Evaluation, and Treatment of High

Blood Pressure. Hypertension. 2003;42:1206–1252.

Edington DW. Zero Trends:Health as a Serious Economic Strategy. Health

Management Research Center, University of Michigan, 2009.

Goetzel RZ, Pei X, Tabrizi MJ, Henke RM, Kowlessar M, Nelson CF, and Metz RD.

Ten modiable health risk factors are linked to more than one-fth of employer-

employee health care spending. Health Affairs; 2012; 31(11):2474–2484.

Goetzel, RZ; Staley, P; Ogden, L; Stange, P; Fox, J; Spangler, J; Tabrizi, M;

Beckowski, M; Kowlessar, N; Glasgow ,RE, Taylor, MV. A framework for patient-

centered health risk assessments – providing health promotion and disease

prevention services to Medicare beneciaries. Atlanta, GA: US Department of

Health and Human Services, Centers for Disease Control and Prevention, 2011.

Available at: http://www.cdc.gov/policy/opth/hra/