a. :0.
am. ,..

VA.

. . v 33 .
..n.~..mw}..€d

55

‘ . Vcéyw‘fi.
. w cut EVA,
a . ark.
. “Pb u.

aw

.
1.3:.

4.3.. £qu
:3; 7+”. .
ax}. ; hm

‘-
..

arr, ! m...
Lhrr§<
. . ..
2
. .L
. . .u
. ‘ .rIz

. Joanie:
V . (I. A n ‘ I:

. . .. .

.. ”5,515

 

.. f.

A
.
u”
.92... .H 9%“
$3.
V. .
A Q

t

k .
i. 1.. .
. ... . ..
, a 5%“ ”.353.

m, .5

..
9......»
hi...

5:

(“A *p , —
[a r.
. ‘ z
i«......:..
.35

d , J04.

2:. .5. K

. . , a

Pit"... #; l—
...L; n

if.
. I:

 

 

at . ‘ an .agnamnhauviw.‘
.X3 3.; . «:5;
,,.,,%..u§rm§fi 3%.

 

.
:1 .

 

. ‘4‘.
. fl.”1§\
. .x... I.
1
. ii. w
3.4.315. u:

440.! ll
(turn!
# ,

THESIS
I
2 CC {

This is to certify that the

thesis entitled

The Effect of an Incentive/Disincentive
Worksite Health Promotion
Program on the Workforce

presented by

Sana Khoury Shakour

has been accepted towards fulfillment
of the requirements for

M.S. Epidemiology

degree in

W.

Michael R. Rip

 

 

 

Major professor

Date August 8, 2000

 

0-7639 MS U is an Affirmative Action/Equa! Opportunity Institution

 

LIBRARY
Michigan State
University

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

35939976132

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

11100 mus-p.14

THE EFFECT OF AN lNCENTIVE/DISINCENTIVE WORKSITE HEALTH
PROMOTION PROGRAM ON THE WORKFORCE

By

Sana Khoury Shakour

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Epidemiology

2000

ABSTRACT

THE EFFECT OF AN INCENTIVE/DISINCENTIVE WORKSITE HEALTH
PROMOTION PROGRAM ON THE WORKFORCE

By

Sana Khoury Shakour

The aim of this thesis was to investigate whether a worksite health
promotion program that took place at a hospital in western Michigan, produced
employment selectivity of healthier employees. To address this question, three
groups of employees were compared: those who left the workforce (leavers),
those who joined the workforce (joiners), and those who stayed employed
throughout the five year period of the program (stayers). The main variables of
interest were: the results of an annual health screen called Health Quotient (HQ)
points, health care costs, absenteeism due to illness, and absenteeism due to
short-term disability. Logistic Regression analysis was used to obtain crude and
adjusted odds ratios. Low HQ points were associated with leaving the workforce
compared to joining the workforce, or staying employed. However the first
association did not reach statistical significance. Differences in medical costs and
absenteeism between the three groups did not reflect differences in HQ points.
Leavers had the lowest medical costs. In conclusion, the current evaluation does
not provide strong evidence of selectivity in employment, but is not in favor of

widespread use of the approach.

To Elias and Yasmeen

ACKNOWLEDGMENTS

I would like to express my gratitude to Dr. Aryeh Stein for his valuable
advise during the course of this research. I also extend my appreciation to Dr.
Michael Rip for his kind help during the writing of this thesis. I would further like
to thank the members of this thesis committee, Dr. Nigel Paneth and Dr.
Wanjiang Fu. Appreciation is also due to the Fat Cook Health Sciences
Research and Education Institute, Grand Rapids, Michigan for funding my
graduate assistantship.

Finally, sincerest thanks to my husband Elias for his understanding,
patience, and love, to my sweet daughter Yasmeen for spiritually strengthening

me, and to my mother for her love and support.

TABLE OF CONTENTS

ACKNOWLEDGMENTS ...................................................................................... IV
TABLE OF CONTENTS ....................................................................................... V
LIST OF TABLES ............................................................................................... Vll
LIST OF FIGURES ............................................................................................ VIII
CHAPTER 1 .......................................................................................................... 1
INTRODUCTION ................................................................................................... 1
The Butterworth Experience ....................................................................... 3
CHAPTER 2 .......................................................................................................... 6
LITERATURE REVIEW ......................................................................................... 6
Why the workplace? ................................................................................... 7
Health outcomes ........................................................................................ 8
Cost outcomes ........................................................................................... 9
Who participates? ..................................................................................... 11
Methodological critique ............................................................................. 14
The impact of incentives and disincentives on participation ..................... 15
Thesis rationale ........................................................................................ 17
CHAPTER 3 ........................................................................................................ 25
METHODS .......................................................................................................... 25
Groups of interest ..................................................................................... 25
Research questions .................................................................................. 25
Health screening and Health Quotient ...................................................... 26
Plan of thesis analysis .............................................................................. 29
Attrition and exclusion .............................................................................. 29
Description of variables ............................................................................ 30
Analytical methods ................................................... 32
CHAPTER 4 ........................................................................................................ 35
RESULTS ........................................................................................................... 35
Characteristics of the studied groups ....................................................... 35
Research question 1 ................................................................................ 37

Preliminary comparison ............................................................................ 37

Adjusted associations ............................................................................... 38
Research question 2 ................................................................................ 44
Preliminary comparison ............................................................................ 44
Adjusted associations ............................................................................... 44
Research question 3 ................................................................................ 50
Preliminary comparison ............................................................................ 50
Adjusted associations ............................................................................... 50
CHAPTER 5 ........................................................................................................ 56
DISCUSSION ...................................................................................................... 56
Strengths and weaknesses of the Heatthlus program ............................ 59
Limitations of this analysis ........................................................................ 60
CHAPTER 6 ........................................................................................................ 62
CONCLUSIONS .................................................................................................. 62
APPENDIX A ....................................................................................................... 64
BUTTERWORTH QUESTIONNIAR .................................................................... 64
APPENDIX B ....................................................................................................... 68
HEALTH QUOTlENT POINT SYSTEM ............................................................... 68
REFERENCES .................................................................................................... 71

vi

LIST OF TABLES

Table 2. 1 Summary of the literature review ....................................................... 21

Table 3. 1 Health Quotient point range for each of the risk-factors and its impact
on the benefits package. .............................................................................. 28
Table 3. 2 Description of variables ...................................................................... 31

Table 4. 1 Characteristics of the studied employment groups at Butterworth. ....36
Table 4. 2 Health-Related Practices of the Studied Employment Groups at

Butterworth ................................................................................................... 37
Table 4. 3: Research question 1. Comparison of leavers and stayers with respect
to HQ scores in 1993, and selected demographic characteristics ................ 40

Table 4. 4: Research question 1. Comparison of leavers and stayers with respect
to the liability amount in 1993, and selected demographic characteristics. ..41

Table 4. 5: Research question 1. Comparison of leavers and stayers with respect
to days lost to illness in 1993 and selected demographic characteristics. ...42

Table 4. 6: Research question 1. Comparison of leavers and stayers with respect
to days lost to short-term disability in 1993 and selected demographic

characteristics. ............................................................................................. 43
Table 4. 7: Research question 2. Comparison of stayers and joiners with respect
to the HQ scores in 1997, and selected demographic characteristics .......... 46

Table 4. 8: Research question 2. Comparison of stayers and joiners with respect
to the liability amount in 1997, and selected demographic characteristics. ..47
Table 4. 9: Research question 2. Comparison of stayers and joiners with respect
to days lost to illness in 1997 and selected demographic characteristics. ...48
Table 4. 10: Research question 2. Comparison of stayers and joiners with
respect to days lost to short-term disability in 1997 and selected
demographic characteristics. ....................................................................... 49
Table 4. 11: Research question 3. Comparison of leavers and joiners with
respect to HQ scores in 1993 and 1997, and selected demographic
characteristics. ............................................................................................. 52
Table 4. 12: Research question 3. Comparison of leavers and joiners with
respect to liability amount in 1993 and 1997, and selected demographic
characteristics. ............................................................................................. 53
Table 4. 13: Research question 3. Comparison of leavers and stayers with
respect to the number of days lost to illness in 1993 and 1997 and selected
demographic characteristics. ....................................................................... 54
Table 4. 14: Research question 3. Comparison of stayers and joiners with
respect to the number of days lost to short-term disability in 1993 and 1997
and selected demographic characteristics. .................................................. 55

vii

LIST OF FIGURES

Figure 3. 1 Butterworth (Spectrum Health) Workforce Flow Chart ...................... 34
Figure A. 1 butterworth questionniar ................................................................... 64
Figure B. 1 Health quotient point system ............................................................ 68

viii

Chapter 1

INTRODUCTION

Wellness is defined as 'a composite of: physical, emotional, spiritual,
intellectual, occupational, and social health'.25 Health promotion is the means to
achieve wellness. Health promotion programs (HPPs) mainly take place at the
workplace.

Some employers have used the 'carrot' approach to motivate employees
to participate in wellness programs by giving cash or other incentives. A few
employers have used a 'stick' approach, using disincentives for non-participants
or for negative health-related behaviors.

The combined incentive/disincentive approach is intended to distribute
health care costs based on peoples' lifestyles. The underlying philosophy is that
the health care costs of engaging in unhealthy lifestyles should not be shared
equally among all employees. However, the effectiveness of the
incentive/disincentive method and its effect on the workforce has not yet been
studied.

It is important to note, that worksite HPPs are not intended to produce
employment selectivity, by alienating unhealthy workers or penalizing them for
existing medical conditions, but are designed to help workers improve their
health by acquiring a healthy lifestyle.

The fairness of the incentive/disincentive approach is questionable for

several reasons. One critical issue is the voluntaries of health related behaviors.

Some critics argue that the decision to engage in behaviors such as smoking and
drinking might be influenced by social and psychological forces. Therefore, in
some instances, unhealthy actions are not completely free choices.““""6 Also,
lower socioeconomic status is known to be associated with higher prevalence of
most diseases and risky behaviors, and incentive/disincentive programs ignore
this fact. Another concern that has been raised, is the ambiguous definition of
health risky behaviors; While in incentive/disincentive programs employees are
penalized for smoking, drinking, and not wearing a seat belt, opponents of this
approach argue that there is a wide range of activities that could exacerbate
health risks like skiing.“’6 Concisely, all the above mentioned issues raise the
concern about possible discrimination against certain groups by using any sort of
penalty.

Moreover, Employers using disincentives risk violating the HIPAA (Health
Institute Probability and Accountability Act) of 1996 which states that 'a group
health plan may not require any individual to pay a premium or contribution that
is greater than such premium or contribution for a similarly situated individual
enrolled in the plan on the basis of health status'.45 In addition, incentive/
disincentive programs could be found discriminatory under the new Americans
With Disabilities Act, which was effective July 1992 and prohibits discrimination
on the basis of an individual's physical or mental disability in employment and

several other areas.

One strategy to test whether an incentive/disincentive worksite wellness
program is socially responsible is to monitor changes that occur in the workforce
during the program.

The aim of this thesis is to assess whether incorporating an
incentive/disincentive component into a worksite HPP is associated with
selectivity in employment. That is, selecting employees based on their health
status. A comparison of the characteristics and health status of employees who
joined and employees left the workforce, during the 5 years of the intervention, is
undertaken. This thesis is one of a series of evaluation reports of an HPP, which

took place in a hospital in western Michigan.

The Butterworth Experience

In 1993, as part of the health care reform that was being initiated both on
the national and local level, Buttenlvorth Hospital in Grand Rapids, Michigan (now
Spectrum Health) started a new approach in managing the health benefits
package. With the incorporation of the Wellness Center, a company equally-
owned by Butterworth Ventures, Butterworth Hospital introduced the HealthPlus
program. HealthPlus is an incentive/disincentive health promotion program (HPP)
that rewards staff members and their spouses for healthy lifestyles and provides
financial incentives to those who would like to improve.

The main drive for implementing the program was cost containment.
There was a 100% increase in health plan costs at Butterworth (Spectrum

Health) Hospital between the years 1988 and 1994. The HealthPlus program

was an attempt to moderate this increase in health insurance costs, while
promoting the health of staff members and their spouses.

HealthPlus integrated both health promotion activities and an annual
screening test. The activities included a wide variety of programs targeting
several modifiable risk factors (nutrition, smoking cessation, stress management,
fitness, etc). The screening test, which was conducted annually during the
summer, assessed eight risk factors - some of which were self-reported and
others were measured. The screening was evaluated by credits called Health
Quotient (HQ) points. According to the HQ score, employees could receive
credits or have credits deducted from their overall benefits package.

A previous report that investigated determinants of participation in the
program has shown that the average participation rate in the first 4 years of the
program was slightly less than 30% and participants were more likely to be
women, whites, full-time and managerial employees. The report also indicated
that Individuals with adverse scores of body fat, cholesterol level and blood
pressure were more likely to participate in activities that targeted these domains.
But on the other hand, employees who scored positively in the fitness test were
more likely to participate in exercise activities, than employees with zero and
negative scores. Further, the HQ score was found to be a determinant of
participation. Employees with low HQ scores were more likely to participate in
health promotion activities. However,- after the first year of other program this

association was attenuated.47

Medical claims costs have increased through out the five-year period of
the program. However, the effect of participation in health promotion activities on
the mean cost per employee was not consistent. In the third year of the program
an increase in cost among non-participants was observed, but in the following
year the mean cost for participants in health promotion activities was higher than
non-participants.

The incentive/disincentive approach is very uncommon in programs of this
kind. In typical health promotion programs in which participation is optional,
healthier workers are more likely to participate. Whereas in the HealthPlus
program, participation was mandatory in order to benefit from the health care
package. This raises the question of whether incorporating a disincentive
component into a program might result in selective employment, i.e., cause less
healthy employees to leave and seek another place of employment.

It is important to note that HealthPlus Health Quotient was not found to
affect the turnover of the hospital noticeably, and full-year employees who
obtained health insurance as an employment benefit increased from 1993 to
1996.

The effect of disincentives on health and cost outcomes has not been
investigated yet. Therefore, the aim of this analysis is to detect whether the
health status was a determinant of leaving or joining the workforce. Addressing
this question is important in order to be able to attribute reductions in medical
costs to a successful program rather than to selection of an inherently healthier

workforce.

Chapter 2

LITERATURE REVIEW

Worksite HPPs are becoming more common in the United States.1 In
1989, the National Survey of Worksite Health Promotion Activities (NSWHPA),
estimated that 65.5% of all worksites offered at least one activity. In the 1992
survey, 85% of worksites with 50 or more employees were found to offer at least
one health promotion activity, ranging from health education to aerobics classes.2
These programs have grown not only in number but also in variety, and have
evolved from programs that were characterized by a focus on a single
intervention toward more comprehensive programs.10 Small worksites are also
promoting their employees' health; in 1998, about one in four small worksites
have offered HPPs. However, the primary focus of HPPs at small worksites is
job-related hazards.”A summary of the reviewed articles is included in Table2.1.

This enthusiasm for health promotion is increasing as research continues
to suggest that primary prevention is more cost-effective than secondary or
tertiary prevention within managed health care.32 In addition, there is a mounting
body of evidence linking HPPs with positive health and cost outcomes.25 Further,
many employers are recognizing that approximately one half of the health-care
costs are a direct result of lifestyle-related illnesses.31

All programs cited in the literature target modifiable risk factors such as:

cholesterol management, weight control, exercise, tobacco use, blood pressure

management, alcohol use, motor vehicle safety, nutrition, and stress
management. As to the type of health promotion activities, the most available
activities are health education, screening tests and smoking cessation programs,
whereas programs that require more resources like exercise are among the least
offered programsa.

To measure health outcomes, some of the behaviors may be self-
reported, such as seat belt use, level of exercise, or nutrition practices. A series
of tests such as a blood test for cholesterol levels, blood pressure, or physical
fitness assessment, is usually required to verify measures. However, the type of

health outcomes, and the ways to measure them vary widely among programs.

Why the workplace?

The worksite is a logical place for health promotion marketing, and it
serves as a key channel for delivering health promotion interventions. More than
60% of adults in the United States can be reached at the workplace.30
workplaces tend to include diverse populations in terms of race, gender, age and
health status. In addition, communication is organized, and peers may have a
supportive and competitive impact.2

Moreover, employers pay an estimated 30% of the national healthcare bill,
and with health care spending still on the rise, employers are promoting ideas to
moderate the growth of health insurance costs.”27 It is evident that employees
with risk factors such as obesity and smoking have higher healthcare costs, more

illnesses and absenteeism.‘ Analyses of the financial effect of risk factors on

health care costs in DuPont company, revealed that smoking costs $960 per year
for each smoker, whereas alcohol abuse cost the company $389 per year per
employee, and high cholesterol levels incur $370 per employee per year?!"26
These lifestyle choices also impact the costs associated with lost productivity and
absenteeism. In short, health promotion seems to make sense to the employer

from many different perspectives, including: improving employees' health;

optimizing productivity; and most importantly, reducing health care costs.13

Health outcomes

During the last two decades, the value of worksite health promotion has
been widely acknowledged. A summary of the reviewed studies is included in
Table 1. Most of the evaluations reveal small but favorable impacts on health
outcomes. Some research suggests that positive outcomes can be best achieved
if comprehensive programs are provided with one-on-one counseling to high-risk
employees.20 The Live for Life program is a good example. It included a health
screen, health promotion activities and personal consultation in addition to a
newsletter. The Live for Life program has been shown to reduce employee health
risks and health care costs, both at Johnson & Johnson and at Duke
University.” 20:32 In the Working Healthy Project, a multiple risk factor
intervention implemented in 26 manufacturing worksites, participants significantly
increased their consumption of fiber from 8.3 grams per kilocalories at baseline
to 9.2 at the final assessment, compared with the control group (t=3.5;P<. 001 ).30

Similar changes in dietary habits were observed in the WellWorks study, a

randomized controlled trial that included 24 worksites, in which intervention sites
had consultation and educational activities.8 Intervention sites in the WellWorks
study experienced a 10% increase in fruit and vegetable consumption vs. a 4%
increase in the control sites, and a reduction of the percentage of calories
consumed as fat (2.3% vs. 1.5% kcal). In DuPont intervention, 48 intervention
sites were compared to 19 control sites. The level of health-related behaviors
was improved over a two-year period, and the percentage of employees with
three or more risk factors decreased by 14% at intervention sites.”37

Additionally, absenteeism due to illness was decreased to a greater degree in

intervention sites over a six-year period.20

Cost outcomes

As mentioned above, a principal reason for the employer's interest in
health promotion is decreasing or at least moderating the growth of health cost
expenditures. According to one survey, about one half of the health care costs
are a direct result of life-style-related illnesses.31 Research data on the cost
benefit of worksite wellness are promising, although conflicting and limited. For
instance, although an association between health promotion and lower health
costs was demonstrated in some programs,31 most programs have minimal
change in health care costs in the initial stages. In a three-year period program in
a major corporation in Cincinnati that included high-risk screening and one-on-
one counseling, health care cost reductions were only evident in the third year of

the program (29% lower in total and 36% lower in lifestyle-related costs). The

authors referred to the importance of commitment for long-term programs, and
suggest that three-years is the minimum time period needed for potential
lifestyle-related medical costs to be reduced.24

Cost analysis of the Johnson & Johnson Live for Life program over a five-
year period indicated that participants had lower rates of increase in medical
costs than the control groups. By the fifth year of the program at Johnson and
Johnson, the average inpatient cost per employee was $265 in sites were the
program was operating as oppose to $403 in sites with no Live for Life (P=. 005),
no significant difference in outpatient costs was found.13

Not all programs have been found to be cost-effective. A seven-year
evaluation of the Blue Cross Blue Shield of Indiana found that program
participation was not associated with reduced medical costs.32 In the Adolph
Coors brewery, even though respondents to a health hazard appraisal were at
lower risk and perceived their health as better than non-respondents, they had
significantly greater claims costs than non-respondents at any given percentile in
the distribution below the 90‘“. Interestingly, among employees falling in the 90th
percentile, non-responders had greater costs than responders.11 Apparently
those in good health tend to seek medical services regularly as a preventative
measure and confirm their positive health status.

Another measure for reduced costs is reduction in absenteeism days. A
number of studies suggest that health promotion at the work place may be
associated with changes in absenteeism.“""'""19 In a randomized trial that

included 32 worksites, the prevalence of illness-related absenteeism was

10

reduced by three to four percent over a period of two- years.16 Also, at Duke
University participants in the Live for Life had an average of 4.6 fewer absentee
hours than non-participants.17

A recent review of the literature found that the most consistent positive
outcome of the multi-component programs is reduced absenteeism.” The
authors explain this finding as an indicator of a reduction in the overall risk, since
absenteeism is a nonspecific indicator of well-being, and different employees

benefit from a multi-component program in different ways.

Who participates?

The question remains, whether these programs are reaching the at-n'sk
workforce. A primary concern is that the concept of wellness might alienate
workers who engage in unhealthy life styles, who could benefit most from HPPs.
Unfortunately, there is some evidence in the literature that participants are
healthier than non-participants and those who most need the programs are least
likely to participate. 6'9'11'12 Responders to a health risk appraisal at Adolph Coors
were less likely to smoke, and controlling for age and gender had lower systolic
blood pressure (2.8 mm/hg, P<. 01) and lower serum cholesterol (5.8mg/dl, P<.
01). However, this pattern is not consistent in all studies. A survey conducted at
the University of Oregon, in which employees expressed an interest in attending
a worksite HPP, revealed that both groups that intended and did not intend to
participate in the program had similar health-related characteristics.7 Another

study that analyzed the response to a pre-program questionnaire found no

11

difference between participants and non-participants in self—reported health
status and only slightly more positive health habits were noted among
participants.5

In a petrochemical research company in New Jersey, behavioral risks
were measured by a health risk appraisal and employees were offered a series
of on-site wellness programs. Evaluation of this program indicated that most
activities attracted 10% to 40% of the employees at increased risk for the health
behavior addressed by the activity.6 Fitness activities were found least likely to
attract employees at increased risk for fitness-related problems, whereas the
participants tended to be more fit and less obese. For educational programs, on
the other hand, (i.e. smoking cessation, weight management and blood pressure)
significantly greater participation of high-risk groups was observed.

Another concern is that worksite HPPs may not equally reach all
segments of the workforce. Participation patterns vary widely among employees'
sub-populations. Therefore, understanding the variables that influence -
participation is essential for assessing the effectiveness of current programs as
well as for the planning of future ones. One factor that influences participation is
gender. Women had higher participation rates in most studies,“‘7 reported more
positive health-related behaviors4 and the magnitude of post-screening positive
change or adoption of positive health behaviors was higher for women than for
men.

In the Total Life Concept of AT&T, 2"” women were more likely to join

weight loss programs than men even if they were not overweight, which suggests

12

that the motivations that influence participation differ by gender. In the above
mentioned petrochemical research company, the proportion of women in high
risk groups who participated was substantially higher than that of men at similar
risk .6 Excluding the fitness center, the average participation rate in wellness
programs for women in high-risk groups was 35% compared with 20% for men in
high-risk. However, this pattern is not consistent. In a recent investigation that
examined the association of individual and organizational variables with the
availability of and participation in worksite HPPs,3 participation differed very little
across most variables. The results indicated that the overall participation did not
differ by gender. However, an apparent interaction of gender with other factors
influenced participation. Participation varied slightly by the category of health
promotion with men reporting higher participation rates in exercise and screening
tests.

Occupational status and educational level, two highly correlated factors
were also found to influence participation. In the WellWorks study, participants
were less likely to have a college degree.8 Non-college graduates were less likely
to be recruited in a program for independent school district employees in Dallas,
Texas.23 In an HPP at the University of Oregon, participants were more likely to
be in classified positions than faculty members .7 Similarly, in the Michigan State
University wellness program, faculty members were least likely to participate.15

The effect of demographic differences on participation patterns is less
significant in more homogeneous worksites,12 and is influenced by organizational

factors.9 Although other demographic variables like age and race have been

13

cited in the literature, there is very little evidence that these variables influence

participation in HPPs.

Methodological critique

Worksite health promotion programs are not planned for research
purposes and therefore some studies lack methodological rigor. However, some
of the methodological limitations are common to all health promotion research
whether at the worksite or not.

Several limitations should be considered in the interpretation of worksite
health promotion research. Of some concern is the reliance on participants' self-
report as measures of program-related behavioral change. Health practices such
as nutrition, tobacco use, and alcohol use were self-reported in all the studies
using different types of questionnaires, and the validity of these measures were
not tested. In the Healthy Worker Project, the number of sick days was self-
reported too.

It is important to note that participation had diverse definitions. At the
Working Healthy Project,” the Total Life Concept at AT&T,36 and the Adolph
Coors program,11 employees were considered to be participants if they
completed a baseline health survey. In a New Jersey Petrochemical company
program employees were considered to be participants if they participated in any
one health promotion activity. On the other end of the spectrum, Procter &
Gamble Company program participants had to complete a health risk

questionnaire and participate in follow-up high-risk interventions. It is important to

14

note that the definition of participation did not take into account the frequency or
duration of participation.

As can be seen in Table 2.1, participation rates ranged from 30% to 75%.
Participants almost inevitably differed from non-participants. Participants are self-
selected in the vast majority of the studies. Many of the evaluations overlook this
fact, and lack any assessment of its impact on the results. Evaluations that focus
on behavior change in active participants overlook the fact that if participants are
different than non-participants, the findings may not be representative of all
employees.

Another limitation is the lack of a control group in most evaluations. This
presents a threat to the internal validity. On the other hand , a considerable
percentage of the quasi-experimental studies, which do include controls such as
the Working Healthy Project, have demonstrated positive results.

An additional limitation of health promotion research is the difficulty of
differentiating interventions effects from other variables such as secular trends
and changes in health policy.

On a positive note, a review of the recent literature reports that more
rigorously- designed evaluations revealed more favorable health and cost

outcomes and tended to support, rather than refute, previous studies.32

The impact of incentives and disincentives on participation

Although there is an increased interest in fitness and wellness among

Americans,21 personal motivation is needed to drive individuals to participate in

15

health promotion activities. One approach is to give monetary incentives for
meeting wellness goals, and penalties for failing to meet goals; the first method is
the more common one.

An incentive based health promotion program can be an effective means
of increasing participation. Sometimes, an incentive provides the motivation to
move an employee from a stage of thinking about a behavior change to a stage
of action,27 while the basic idea behind disincentives is to get individuals to take
responsibility for their negative health-related behaviors.

The percentage of employers giving some sort of incentive or disincentive
has risen from 14% in 1992 to 39% in 1997.” In the past, incentives have been
successfully used to enhance participation and desired behavior change,” 4° but
has not been found to be cost-effective.39 Also, a more recent quasi-experimental
study, that compared outcomes of smoking cessation programs with and without
cash incentives, revealed that incentives may help achieve higher quit rates in
the short term. In the long term, however, it was more cost effective to invest the
money in counseling and health promotion activities, rather than to give it away
as cash incentives.31 In 1993, Baker Hughes, an oil field equipment
manufacturer, with 12, 500 employee in the US. implemented a program that
ranged from a $120 penalty to a $100 reward for health lifestyles.31 Baker
Hughes estimated a $3 million cost avoidance savings produced by the
program.41 No further evaluations of the program were located.

Among the programs cited in the literature, very few incorporate a

disincentive component. Supporters of this method argue that giveaways and

16

cash incentives do not work and estimate that 80% of the health care bills are for
20% of the employees that are often high-risk individuals who neglect their
health.” Therefore, this financial burden should not be shared equally among all
employees, but should be borne by individuals who engage in such behaviors.”
The incentive/disincentive approach has not been looked at in the literature.

The literature search for this review included a bibliographic database
search, manual search of specific journals, a reference list search. Databases,
which were searched for the period 1980 to 2000, included Medline, Health Star,
and Eric. Among the key word combinations used in the database search were:
health promotion and worksite, worksite health promotion and participation,
health promotion and rewards, health promotion and incentives, health promotion

and penalty, health promotion and disincentives.

Thesis rationale

The aim of worksite HPPs is to contain health care costs while improving
the health of the employees. In order to attain this goal, employees in need of
health improvement have to participate. It is evident that worksite HPPs suffer
from low participation rates of high-risk employees. Some critics claim that such
programs "are preaching to the choir."7 Hence, employers are seeking alternative
strategies to attract more employees to participate and encourage them to
improve their health practices.

Buttenlvorth (Spectrum Health) was among the first employers to adopt an

incentive/disincentive approach. Although the literature indicates that the

17

disincentive approach has been used in few other workplaces, its consequences
have not been looked at.

A previous evaluation report has shown that the HealthPlus Health
Quotient program has yielded limited but desired results.38 However, the question
remains whether the incentive/disincentive approach used in the program
alienated less healthy employees, or whether this approach produced selective

employment. This thesis is an attempt to address this question

18

Table 2. 1 Summary of the literature review

 

.323. «5.83 m

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.o :anano Loam 52:50 «.5285 x33
.228 a. .o 2258.. -N .n 8.38:8 c8880 8295 63 3.2a
.Eececcaan oceans. a. can: 62.3.9... 9 .85ch 8 a. can .928 2%; 5 88% Bio;
cc 8295...... 2.. chance a» «a 3.933 as .3 8:282 $3. 855 3328 5 .o 8:3 < 5.8:
t:
.8» £3 :8 25
Efloflcoonm :0 E835 05 05.5: 2:: Eu E 3.58 3.29.5
05 ..o tote 05 2...:on ch «Ev Eflaoa >5 5 333.0me $8 93.... no 52095 9:8 930
3.298 05 :_ 8.559:
25:05 ucm 558.328 5.3...
.mcofloa 530.950: 3.2.30 8 5
$80 23 .9552: EoEo>an. 228:. 8... .6.— 9...
5.3: cc: an... 05 :2..an :33 .2028 5.3: ”.6 US$28 :85.
95:839. 05 20.96 o... wow. 2. 5.8: 52:. :w .o :32an0 $2. 6:59: 97.5555 03:852.
85.3 .08 .0 22.33%
683.8: 96:20 oEmoE :_
5533.3 AN deemed. a:.::m.a
3.0.... «35.25 a ucm $5805ng 68
5.3: m.oo>o.nEo :o £952: 59 :_ ucm 89 :_ .mmfianm 059:2: 5288 8:853 58:00 8...
as .c 885. a... 39.96 a» Re ..a... 5.3.. a to 85...an $3 an 3 ”Success 82.. a < .58 new:
3.29.35 05 Co 358 6:395.
as :90 30323 9:295 ucm 5:53: .325: .833:
can ..a... 5.2.8 .81.... cases 588... 9F 3.9.83 an. ocean.
:05 305:0 «.3222. >028 5.8: 938835;. on :_ EcoEoEE. >550...
5...; 8 sees as 2.598 a» Bow 2..an n .o 8:888 $8 8:522... can. ea: 29.5.: < 8.3.82,
«9:850 5.8: .o accumagm
on.» 38
sea-3.26 3 309.... 0.9:.“ :ozunBEa .0 52:50 fiasco-033a 5.3.. 53523:. 25%

 

 

19

Table 2.1 (cont'd)

 

0000.00. 00.... 000030000
.0 0000.. 00.... 0 000.0 80003000
0.0000 05 0. 0050.300

 

 

 

 

 

 

 

 

 

 

00 >08 >030 0.5 .0 0:000. .00 can: 08050.0 00000000
.000: .050 05 :0 0.000.800 >0 000.080 0. 0003.083 5.3 00.0.0800
08.0.8 00.. ..0>0.. .0... .o .850... 02.0.08 8 o. 88. 00... .0... 0.0280 .0 ...... o... 00
.88.... .0 00.8.8 0.0... a... .0... 000 .3 .u... 00.0 880.00.... ... 0......” 0.8... .00. 0... ... 2.0003 9.. ...... 03.908.
0. >030 05 .0 0005.003 >.0800 < >0 08000. 00:. 0>00 ..0.0 .0 00.00001 0.>~ 0>00 0.0.0 .0 .0080: 000000.800 mm .0 .0... 000800001
0.0080: 05.003 .0.
00.00.00 00:00 00000208.
:8 .v0. .00> 8.5 05 0. 0.03.083 05 .0>0 0.00... 00.00000
00.0000... .00: :05 0.00.. 00.00000 .030. 0.0 0008 0. 0000000 :0
3.0000 05 5.... 903.053 00.. 0.03.083 ..0>0....0... 0000.0 50.. 00.30.00 0.0.... 0800.00.00
.0 00000 05 .0 00003800 02 0. 00000.00. 0.000 00.00000 .0 .3802 0.> v 0.00.. 00.00000 .0 .3802 :0: 0:0 0.03.0800
...... 00
5.2. 0000.0 000 05:08 on
08.0 .288 0 8:0... 0.08 .0838 o. 0. ...... 8.0.0 80.8....
0. 00.030 00.000000. .0... 05 .0 0. 0000.050 8000.090 02 .. S .u... 05:08 00A 5.... 00:0
0:0 003 0:0 0000.000 2000.091 0.03.083 .0. 0.000 8003:. .033 0.50 0.08 8003.00 000 80030. 4.00.3800 0.0:. 0000.0 0
00.0000... 000.50 5.00..
0. 000.0 00003800 0... :05 .000... 003000.. 000.50 5.00..
000.0 00.80 000.0 >030 008 ..0.0.00_0..0 000 0.0.0008 .0008 5.0.0.08
.988 ... o. 8.888 8 .8 0.08 .08. 0:0 0580... 08... 2.0.00... .808 ......000... 9.2.0 5.0.0.9.. .808 .00..
0.00.. m 05 .0>0 000.0 00000280. 0. 00.02. 080.00.. .5 0080.00.00 0.0.: 9.9.0 $50008 .0000 000.. m 0 .0>0 000.0 .0008 0 0.
05 0. 000000 05 ow 00000208. >...0..08.>..0.0.08 00>00 .0 x.0 0. 000 5.0.0.08 0.0020 000.. 600.080 00.388 000.0 0800.00.00
05 .0 .00... .00. 05 .0. 0.00:0..0 000.080 0000800000 .0..00.< 80000.0 6000800000 .0..00.0 00.0.9.0 05 .0 00.0.. 5.00.. 0. 000000
>80 00.... 0.00 000.0 00003800 000... ..05 005.08. 000>0.08w 0.>m ..0>0. 6.200.000 65000.0 085 .0.0>.0:0 00.03.0000
00000000 000.080 .0 0000800000
.0. o. 0.08. ... 008. 0.0.... 08:90:...
oz .88 .v... 02.8.8.8 .8:
0:0 .208 .v... 000080008 0.00.009
0.16.0. 0.... ...... ..88 .v0. 830...... 8.998 00.8..
0. 000. 0. >030 0.... .0 0005.00... < 0. 0000.00. 8000.090 0 00.... 0.00... md 002080 800.500 5.2.00 .00.0>..n. ..00 0000.08 08.800001
00
$2.0. 3
00.5.5 3.000.. 0.0! 00.00.00 0800.000 28080.00 00.00.00 330.00 00:00.86

 

 

20

Table 2.1 (cont'd)

 

§C8

 

 

 

 

 

 

 

 

 

 

 

$9.... 2.3.0 cm 0. Sauna c. 6.
3.2.0 203 3228“ c3050.... E388 .m
an... 5.3: 2.»...0 .o 8.3» m ucm c.8o50boa
8.9.33 c. 533.028 5.3 3.2.0 32.58 «85.03 05 30.6.an .0 9...... 3.0.22.3 9.. 3&3
Banana.“ 998. 9.35:3. 3» .o 0:0 >5 .m 5:335. $3 9233:. .$.Eaam xv... 5.3.. < 262 <
35... 5.3..
new £03ch 5.3.. :0 8...“.
an... 9.9.63 m c. 322...»...
.0 $235 gaxo 9.3 “53:950.... 02. in... m 9.655 E 5020
30.6360 850...! 9.3.5200 mi. .623 on. @5533. $9. 5 .88.... 9.. co .623 < .o 3382::
253.0...3 .0 $555.28 .0 cognagw
€833-52. Em ”89:95.... ...... 28522:. .5
«...—3.03.3 5223 308 new .5... 3.2.2.0. ... 9.23.8.3 x»: cm... a?26..2 can 65.858 .335
593...... 28 .859: 36.9. new 2.25933 2.0 o. 9.0 .955833 2.6.... o.nEm0
$.38... new .39 9.55.500 «and x»... 5.8.. m .0 5:03.50 *3” 5.8.. m 33.2.. E960... 9.; a .202.
$08 28 5.3.. :0 USE.
.28 can 5.8.623 852.3
2.52% new .3353 xv... 3:33 5.850... 5.3..
5.8; m c. 23532.5: ucm new 9.528 5.3.. .o 3:263 35 in
£09.88. 5923 ..m: 5.3.. .389... m 9.53.0... Enact 5.8.60. £000
a. «8:22.... 9.. 28:3,... 8. m8 Emu... 5.8.. o. 86883 ......m xv... 8.8...9228 95:8... :< 5.8..
in an.
8.33.9. .0 82...... 3......» 5.3.2:... .o 8.258 5.8.0.5.. 5.8.. 5:55.... .63.»

 

 

 

 

 

 

 

21

Table 2.1 (cont'd)

 

22020 05 .6 00002000
330:0: 0.00.000 05 :0: x0... :0 0026.020
05 :0 $0... 2 $0: 020050 022020 .002

0200002
05 :0 0:0 :000 :0: 80:00

:0 :00000 .0000 20:00.8 00
0020020.. 003 830:0: 026.020

 

 

 

 

 

30.2000 0:0 28:30.0 :0: 203 00:00:00 0:0 000.. :3 0:0. 5.00:0 0:0 20:00.05 00.0: 0:030:00
0.0.: 5.00: 0 :. :002.:.00 5:002:00 :. 08:20:20 0003000 ..0 :. :02 .200020 000.: .3550: 0:0 020200200 :05 :0
:3 00000000 003 0.0.”. :05 0.00.0200 2 200... 202 203 :020>> 03 N .0005: .00050 .00: 80000:. 0200200 203 2:002:00

0000020 .0200 0.2.0

0:0 0:00: 30.0.0 0.2.0 6050:5050 .0520
0.2.0 00:002 00:20 30: 0: 0. 00.2008 0050006205 00.0.2
9 :9... $0. 08 $00: .2 $8. 088:0 5.00: 0:0 0.88008: :05
00.030 003 55000200 :0 00200200 :0:0.: 0 00: 0... 0 :. 02022:. :0 020028 203 00... 0 :.
.0300 :05 .050: :0: 203 0:3 00030.0Em .:0.:00.0.:00 5.00: .0::00 0:0 02022:. :0: 0:0 02022:.
0.00.0200 0: 30:02. :. .0202. 00002008 2003002 :0 02.5 03 .r - .5550: 00.2008 .00: 000000... 203 0:3 00:20 03: 0: .r
200» 025
..0: 0030.020 200005200
203 0:3 2:009:00 0:
000320 003 0.50.0200 000:00 0:0 000 020502 003 0.0220 0:...
-8: 0:0 0:80:00 8: 0303.00. .0: .v0. 0:20:80 85 88.: .50: 0.08 28 5.09. as
:0 500.0000 02 $00 203 00000 30.00:: 0200.00.00.82 0.3 m 0.000 200 5.00: .0:::< :0 0.0220 0500000020

0:000 .2020 00: 0200620000: 0200200

203 0.08 =0 :0 $2 .38.: as 8:3
.:0>030_.. :00» :0>.0 .80 :. 2000 30.00.: 200> 0 000320
5500.00.00 :0 2:02 0 00 00: 20: 0: :2... 202 203 2005000.. 05 :. 0:000 20.. 5.00:
000:0:0 0:030:00 5.00: ..05 2.0000 0500 2.8.0.20 050: 0:0 5.00: :0 :0000200
050:3 =0: 2 >03 0: 00:002 2:022:00 :0 000::0200 :0:0.: 5.00: :30 :0 00000200 0:0 .20: 5.00: 0050020205
0. 205 .0m 000320 003 < .Co .v0. 226. 20.00.05 2200 :030. .808 28 5.00: .0:::0 .200020 020200200 :0
00.0: 5.00: 2202020 .A 5 .v0: 200020 000.: 0030. 2003002 000.: 0:0 6050:5050 000000: 0200200 203 2:002:00
8 8028:... oz .8: 8:. 9.0: 5.00: 830: 00: 9:80:80 0:0 .0200 .828 .062. 000:: .600 -8: :8 9:80:00

0:
32.0: :0

0005.5 3.000.. 0.0: 500.50 00:00.26 2302200 020050 338:0 00:00.05

 

 

 

 

 

 

22

Table 2.1 (cont'd)

 

000.00
5.000 000000 0 00.0.0200
0200.00.00 005 000 00.000

 

 

 

 

 

00.: 000.003 0 0. 00.0. 003 00000000 5.000 000 00000.0 .00. .050...
00000.0. 000 2005.900. 5.... >..>000 .0 20.090 0000290 .8000
00000089000 0209002002000 202000000 5.000 0003.0. 0 .000.00 5.000 .000000000.
2 0000.00000 05 00.2006 0 .. 05.0 05.0000 .0 00020200 $00 05.0000 0 .0 0000.028 05 .00< 00:00
.0200 05 0. 00006.0 0.03
00000.0 00000000 5.000 000.000
0. .0>00..00.s 000:0 000>0_020 .00.
000>0.020 .0. 000000 003 0.00... .0000. >000200 0
.0..00020 00020 5000.00.00 .0200 0005.0; 00 000 200. 00.0.0.8 2020.000 003002002
.0 0000.20.00 00000.. «on 00. 0. 000000.000 5002. $00 $00... 00.020... .028 0005.0; < .8205
0. 0000.000 003 050000. 008
0305. 00.0 20.090 .0..2.0 0 20.090
000.0203 05 .00 00.... 2000028002 0 0000.000 A 5.
.0 00000000 000.020 .0 20.090 .00> 0.0..020 .0 003 000 8.0 00000003 0.0.0000 00.00028
20000200032 0 0. 0050000. 05 00000 02.. >00 >..>000 .0020: 000.0 0 0. 0000.000 0.03 020.090 00000900
.0 0000900000 00. 05020.0. 0:. z 08 >00 .0 000000.000 0.00.05 .02 800008 90.9.0 .0200... 0 0.50.000
8.0.003 .0..0 00 000 000. 0.03 205
202020.60 000 00.0000. .0 >002 .0.00..0>0 0.03 020.090
0 .0 00.... 0 0. 00.000.00.00 0000290 5.000 0.90200
5.... 00.0.0 5.000 000 005 5.000 0>0.02. 0. 000.0 6.
0.030000 5.000 000009.000 00:00.0 0. 00.0000 05.000000 50.000.
000 00000809000 0.000.00200 20.090 0000290 0 >0 003000. 00000000: 0900000.
2 0000.00000 05 00.298 0... 000 5.000 05 0. .002..90m $00 003 000.00 5.000 .002. 0< 205.02 ...m
00.0 0.0..
00.03.96 3 0000.00 0.0200 00.000.00.00 .0 00.0.0000 00.20.00.00— 00.000 00.20380. >030

 

 

 

 

 

 

 

23

Table 2.1 (cont'd)

 

3.00000 $5 .00500 000 .000.
00005 00000000 000 ..00A0000 .00
.00 $00 .mm+v$..~. 000 000.0000...
..N .00....3 $00 6.00.0 0000. 0.050
000.00 0000.. .. .00 ..00000 >0 00.0.

.000...0.00. 0. 0000.005 .0000.00.0

 

 

 

 

 

00 00.... 0.05 000052 .000... 0. 0000.0 00.00.000.00
00000.0. 000 .000...0.00. 0. 00.0.0. .0200... 0003.00
00000.0. 0.0 8000060000 00.00.00.000 020058 0.0.... 00.0.
.0 00000005. 05 0000000 >030 0...... .05 00.000000 0.00 00... 9.00.... or 000. .00000000 .000 ..00000 00000.0. 000 .000...0.00m
0.00 05 00 00000000
3000.00.00-50 000 002. 0.0.0000 .00000050
9000.00.00 0003.00 0000.00000.0 00002.00. 0.0.0000 0030.220.
00000.0...00 5.... .000 3000.0 ..00 .000000> 000 ..00000 .0.000 .0000.00>000 0.0.... 3000.00.00
00.0.00000 0.0.00. 05 00.50.98 0. 3.000 05.0 .000 .0. 00.000000 0.0>.000 000000. 00.0.0 5.00.. 0.0.000 -000 000 3000.00.00
00.... 0.0.0000 .00000500 0.0.0000 .0 000 0.. .. .900. .00.. 000.000. 0.000. .00 0...... .02 00200.00 0.00000 00200.00 .0 000.0 00.00.00 2.00000. <
600 .u0. E0520 0000.000 05
005 00.0. ..00 .0..0... 2.080.090
00.. 000.0 20000500050.
0.00.000 0.0000000. 0 0.00 0. 0.0. 0000000. 05.00 .000 00000058 0.50000. 05050.0
.50. >.0> 0 00.. 0000.00 05 .0. 00000000 05 000 0005 5000058035 00.... 05 .0 0000050000
00.... .05 >0>.00 < 0.9.00.0 .0 000.00 05 0003.00 0000. 00:. 80200.0 500.0 .00. 05 0. 000.050 05 0.00.08 0. 0000 00....
0.0.00 0. 0.0..0. 05 0. 0005.00... .20.... < .000...00.0 00 05000. N. .< .> w .00 00 000000 000.050 00.0.00 00.000 00.000.09.30an
00.030 0.0... ... 908002.... Edumo 0.000.000 a. 09... 0.0... 800000.00 .0 0.082
0. 0000000 00 00000.0>0 05000080 0.0.... 00....5> 0200.00.00.50 005 00006.0 00.00 .0000000000
.>.000000 00< 000.0...00. 0 000000000 000 0:00 .000 .00.... 0. >0... 00 00.00 05 000 00.0 00000.00 .00..
0.00.00 0.0000. 0 0. 00000000000 .0..0.00 00.. 0.0.... 0500.00.00 .000 ..00 000.800 00.... 0.00 0.0005
0 00.0000 >0 00000000000 0... >0 00.00.60 .000 .0. >00 000000.090 .800080 00.00 00.0000... 0.000
00000020. 05 0000000.. .>..0..u. 00.000005 000000. 000 .00.... 0.0.... 00000.00000 05 0. .000 00.... 0.030000
....o... 0.0 05002.0 2... .8... .00. 0.... ... o... .3 800000.00 .200... 02.82 63.82 2.0.8 o... 0.00.. 00.0.... 5.00.. ....0 030.0
0005.00... 000 00.... .00000 .0. 00000.00 .02 5.... 00.0.00000 0.0.... 0000.. >5.00... .02 >500... 000 x000. 0000. >000 5.00.. 00 00000000000 <
.....0..0. .0
000....0 8.000. 0.01 0000.00 00.00.03 0.800.800 00.3.00 30.0.00 00000.05

 

 

 

 

 

 

24

Chapter 3

METHODS

The first part of this chapter includes a description of the annual health
screening. In the second part, the chapter includes a detailed description of the

methods used in the thesis analysis.

Groups of interest

In order to address the thesis question, three groups are defined (See Figure
3.1):
1) Stayers: Those long-term employees who were employed the whole five
years.
2) Leavers: 1993 whole year employees who were not employed in 1997.

3) Joiners: 1997 whole years employees who were not employed in 1993.

Research questions

Specific questions this thesis seeks to answer are:
1) Did the leavers have more adverse HQ scores and/or higher medical
costs than the stayers?
2) Did the joiners have better HQ scores and/or lower medical costs than
present time employees?
3) Did the joiners have better HQ scores and/or lower medical costs than

the leavers?

25

Answers to these questions are important for present and future assessment of
the use of the incentive/disincentive method. It will be of special importance for
planning future HPPs.

We expect that the group of leavers is the least healthy while the group of joiners
is the healthiest. But, we hypothesize that any differences in the health status of
these groups would be attributed to differences in demographic characteristics,

especially age.

Health screening and Health Quotient

Every year during the summer, the Wellness center staff members conducted a
health screening. The screening assessed eight risk factors. Risk factors were
chosen by consensus of the planning team, which included staff members from
the Wellness Center, (a company equally-owned by Buttewvorth Ventures), a
representative from a consulting company (Gelman Consulting, Inc., Southfield
MI), and Butterworth (Spectrum Health) representatives. Some of the risk factors
were measured and some were self-reported. Self-reported data were collected
by a questionnaire that was filled before or at the time of the screening (see
Appendix 1). Following is a brief description of the eight risk factors assessed by
the screening:

1) Nutrition: a self-report of food servings from 11 different foods was

considered to reflect nutritional behavior. The HQ scoring reflected

adherence to the Food Guide Pyramid.

2) Alcohol consumption: a self-report of weekly consumed amount of
alcohol.

26

3)

4)

5)

5)

7)

8)

Tobacco use: a self-report of daily use of tobacco products over the
previous 12 months.

Motor vehicle safety: a self report of percentage of time wearing a
seat belt and helmet use, points on driving record and drunk
driving.

Exercise: a self-report measure in the first 2 years of the program.
Beginning in 1995, it was determined by a Fitness Walking Test.

Body fatness: skin fold and bioelectrical impedance analysis were
used as methods to measure body fatness. The lower of the two
measurements was used as an estimate for the percentage of body
fatness.

Blood pressure: the lower blood pressure reading of the two arms
determined the HQ score.

Total cholesterol/HDL ratio: non-fasting serum lipids.

Each of the components above was weighted based on:

1.

Objectivity: self-reported variables were not weighted as heavily as
measured ones.

lts contribution to disease burden derived from a model that
predicted 'disease burden'.

The above-mentioned model was developed for the HealthPlus program.

The planning team weighed the risk factors and determined risk stratification

based on previous literature. The HQ point range for each component and its

impact on the benefits package are listed in Table 3.1.

The points earned for each of the eight components reflected the deviation

from the recommended healthy value. For example, the cutoff for a negative

blood pressure score was 140/90; the common determinant of hypertension. A

more detailed description of the HQ point system is included in Appendix 2.

27

The sum of the scores across components determined the Health
Quotient (HQ), which in turn affected the benefit package by $1 per HQ points
per 2-week pay period. Although the theoretical range for HQ score is -80 to +25.
ButtenNorth set -25 as the maximum number of credits to be deducted per pay

pefiod.

Table 3.1 Health Quotient point range for each of the risk factors and its
impact on the benefits package.

     
 

    
 
 
 

V) l

 
 

Point-range

l
“W“ W,

  
 
 

      
  
        
          
      
   
   

  
 
  
         

W)

  

in W W‘ :3" ‘
.=' llmWM IW
WWW“W WWWWWMIWWWW WNWWm
" ' ll

‘ -12 to+ MW
WWW \“Wll WW‘ WWW“) WWW“ WWWW ‘SWNSN ‘1 ‘

       

\ (WM

 

’0'
’

  
 

  
   

l NW 1.
WW. w. W -12to +5 ‘I “ +$12o
“ I Wm In“, W -12to +5 +$120
”‘“ W -» ' -8 to +5 +$120

-8 to +1

—12to+1

 

l‘ V ‘ III) W) L .
*The physicalI fitness measure was initiated' In 1995. In the first two years of the
program, a self-reported physical activity measure was used.
Participation in the screening was strongly recommended for all benefit-
eligible staff members (those who worked 32 hours or more per pay period) and
their spouses. Benefit-eligible members who did not participate in the health

screening were assigned the score of -25.

28

To avoid penalizing employees for existing medical conditions, the HQ
score was adjusted to reflect such conditions. In the first two years of the
program, employees with medical conditions had the choice of accepting their
HQ score or changing it to neutral, which is an HQ of ”0.” Starting in 1995, each
medical condition was evaluated individually on its' own merit, and an HQ score

was determined according to health history and previous HQ scores.

Plan of thesis analysis

Attrition and exclusion

Three groups of interest were included in the analysis:

1) Stayers: defined as employees who worked continuously for five full years
1993-1997 years and received benefits (n=1681).

2) Leavers: employees who worked a full year in 1993 and received benefits and
were not employed in 1997 (n=206).

3) Joiners: employees who joined the workforce at any time between 1994 and

1997 worked a full year and received benefits (n=1008).

29

In order to simplify the interpretation of costs and days absent from work, part-
year employees were excluded from the analysis. Also, benefit eligible
employees who undenrvent the screening and chose not to receive the benefit
package were not included in the analysis. Spouses were also excluded, for they

were not of interest for this analysis.

Description of variables

All variables included in the analysis are listed in Table 3.2. For this
analysis, the HQ score was used as a proxy measure for health status. The initial
HQ score is the sum of the points across all components of the screening. The
final HQ score is the initial score adjusted for existing medical conditions. Liability
amount reflects the medical care costs.

Employment status is a variable that was created in order to classify the
employees according to their hire and term dates. Employment status is a
categorical variable with four categories: whole year, hired in the year, terminated
during the year, hired and terminated during the year.

Over 90% of the workforce was white, in all years. Therefore, race was

categorized as white/non-white for the purpose of this analysis.

30

Table 3. 2 Description of variables

 

Variable name

Categories

 

 

 

 

 

Age <=25, >25, <=35, >35, <=45, >45, <=55, >55
Gender male female

Race white, non-white

Part/full time part time, full time

Employment grade exempt, non-exempt

 

Employment status

Whole year employee, hired in the year, terminated in the year,

hired and terminated in the year

 

Illness days

0, 1-5, 6+

 

Short term disability

days

0, 1-5, 6+

 

Liability amount

0, 1-99, 100-499, 500+

 

Insurance status

None, HMO, other

 

 

 

 

 

Benefit status Any benefit, no benefit
Tobacco points <-1, 0, +1

Alcohol points <0, >=0

Motor vehicle safety <4, -1, 0, +1

points

Nutrition points <-1, -1, 0, +1

 

Blood pressure points

<0, +1 -+3, +4-+5

 

Fat points

<-3, -3-+1,>+1

 

Fitness points

<-4, -4-1, 0, +1-+3,

 

Cholesterol points

<-2, -2-1, +1-+2, >+3

 

Initial HQ score

-25-11, -10-3, -2-+2, +3410, >=+11

 

 

Final HQ score

 

-25—-11, -10-—3, -2-+2, +3-+10, >=+11

 

31

 

All continuous variables were made into categorical variables. Grouping of the
variables was data driven, based on the frequency of the values. However, in the
comparison of stayers with joiners in the second question, age was used as a
continuous variable, since only 7 stayers were under the age of 25. Also, all

sequence variables were converted into numerical.

Analytical methods

The analysis was performed in two phases. An initial phase included
description of the demographic and employment characteristics of the three
study groups, in addition to preliminary comparison between them. The second
phase addressed the thesis questions adjusting for all potential confounders.
Analysis were performed in SPSS, version 7.5

Routine summary statistics were used to identify the demographic and
health-related characteristics of the three groups of interest. Cross tabulations
were used to obtain proportions. Chi-square tests were performed to test for
significant differences in proportions and t-tests for comparison of means.

Logistic regression models were developed to address the thesis
questions. The independent variable in all the models was the employment
group, according to the two groups compared in the question. For example in the
first question the independent variable was leavers vs. stayers.

The logistic regression analysis was performed in two steps. First, crude
odds ratios (ORs) were derived from univariate models. Each one of the

univariate models included one of the following exposure variables: age, sex,

32

race, initial HQ scores, liability amount, number of illness days, number of short
term disability days, part/full-time status, and employment grade. The second
step was four multivariate logistic regression models, each one included one of
the main variables of interest (HQ score, liability amount, number of illness days,
number of short term disability days) adjusted to age, ‘sex, race, employment
status, and employment grade. In addition, a test for trend for the crude and the
adjusted odds ratios was undertaken. Similar models were developed for each
one of the three research questions.

The employment group was coded as stayers=1, leavers=2 in the first
question, stayers=1, joiners=2 in the second question, and leavers=1, joiners=2
in the third question. So the results are interpreted as the odds of leaving in the
first question and the odds of joining in the second and third questions.

The first question was based on the 1993 data, the second question was
based on the 1997 data and for the third question data was combined from both

years 1993 for leavers and 1997 for stayers.

33

 

I“ Butterworth workforcel 997

 

‘“ lllllll“ “““I I“I“III“I“III'I‘““l ““I“:“:l““ll““““““I“I“I“\“lll:I““““ “““I“IIlll“““ “.“Illllll” I::F““III“llll““I l

I“I“IIll“l““l“““ “\“Illlll“““““lll' ““ W“ l“ W“: lll“““““I“I“llll““““““I“Il“ IIIIll “WW“ III

‘“ Wm“ “ill I I “WW II lll“““ ll I N=6296
W“ II ‘“ W“ III I“ I III I Will IIIlll‘IIllll‘ IIW“ ‘

II IIIlIIIIIIllWW “Ill““lII IIIIIlWIIIIlIII I'IIIIIIIll lIll‘“““IIl Il““““-IIII IIIIIln IIIIIWIII “IIIIll l“. I . I I

 

 

 

“I“
IN“ IIIIW W
W?“ IIIllllWIuIIWWI

 

 

 

 

I H V Benefit eligible “ SIM“ Bvenefitreligible“MM“
I “ \I

‘. N=2348 (56%) . I" N=3699 (82%)

l W!“

 

 

 

 

 

 

 

 

 

 

I.” “W Slayers“
I. Employed 93- 97
j.“ N=1681
I
Leavers “ H “ Joiners
“,' Not employed in 1997 I Not employed in 1993

 

N=206 (8.7%) I I N=1008 (27%)

 

Figure 3. 1 Buttenlvorth (Spectrum Health) Workforce Flow Chart

Chapter 4

RESULTS

This chapter presents the results of the previously described analysis. The
chapter will start with a general description of the three studied groups in term of
their demographic characteristics, their employment rank, and their health-related
practices. Following the general description, the results of the three comparisons

between the studied groups will be presented.

Characteristics of the studied groups

As shown in Table 4.1, the Buttenlvorth workforce consisted of mostly
whites, females, full-timers, and non-managerial (hourly positions) employees.
A total of 2897 employees were eligible for inclusion in this analysis, 1681
stayers, 208 leavers and 1008 joiners. Clearly the three groups had some
differences. Joiners were the youngest group. Joiners also included a higher
proportion of non-exempt employees and a higher proportion of non-white
employees than the other two groups. Stayers were found to be the oldest group,
with the lowest percentage of non-whites, and highest number of absenteeism
days. Leavers had the highest percentage of females, highest percentage of

exempt employees, lowest mean HQ score and lowest mean liability amount.

35

Table 4. 1 Characteristics of the studied employment groups at
Butterworth.

 

 

 

 

 

 

 

 

 

 

Leavers Stayers Stayers Joiners P-value
(1993) (1997)

Mean age 36.7 37.1 41.1 33.54 0
% Female 78.2 77.4 77.4 77.2 NS
% White 91.3 93.1 93.1 87.9 S
% Full time 64 83.5 78.2 70.9 S
% Exempt 69 70.3 76.9 85.7 S
Mean HQ -2.19 1.59 0 1.189 0
Mean liability 869.867 1226 1648.74 1237.5 .663
Mean illness 2.5 2.94 2.6 2.26 0
days
Mean short 2.44 3.84 2.87 2.42 0
term disability

 

 

 

 

 

 

 

 

* When comparing stayers to leavers 1993 data was used, and for the comparison of stayers with
joiners 1997 data was used, in order for the groups to be more comparable.

'For comparisons of means one-way ANOVA was used, and for the comparison of proportions
Chi-square tests were used.

*S=significant differences between the groups, NS=non-significant differences between the
groups.

Table 4.2 compares health-related behaviors among the studied groups.
In general, the three groups reported similar health behaviors. A slightly higher
percentage of stayers smoked, and a higher percentage of leavers did not wear
seat belts all the time. However, the major difference between the groups was
fruit and vegetable consumption. Leavers had the lowest intake of fruits and
vegetables and stayers seemed to have increased their intake of fruits and
vegetables between 1993 and 1997.

36

Table 4. 2 Health-Related Practices of the Studied Employment Groups at

Butterworth.

 

Health-related
practices

Leavers

Stayers
(1993)

Stayers
(1997)

Joiners
(In 1997)

 

Alcohol
consumption
(drink/week)

<=

100

99.1

99.1

99.8

 

>7

.9

.9

.2

 

Smoking
(% Smoked in
the last year)

11.1

15.3

11.3

13.4

 

Seat belt use
(% Time
wearing seat
belt)

<100°/o

11.1

6.4

3.9

6.5

 

100%

88.9

93.6

96.1

93.5

 

fruits

Consumption of

(Servings/week)

19

15

35

28

 

vegetables

 

Consumption of

(Servings/week)

 

 

13

 

14

 

35

 

28

 

Research question 1

Did the leavers have more adverse HQ scores and/or higher medical costs than

the stayers?

Preliminary comparison

Results of the analysis of the first research question are summarized in

Tables 4.3 - 4.6. Leavers and stayers were of different age groups. Leavers had

a higher proportion of employees under 34 and over 55 years of age than did

37

 

stayers. Two thirds of the stayers were between 25 and 44 years of age. The two
groups had comparable HQ scores. But, they differed in the distribution of their
medical costs. Slightly less than one-half of the leavers had no medical
expenditures in 1993 as opposed to less than one third of the stayers.
Consequently, stayers had a significantly higher proportion of employees in each
one of the remaining three categories of liability amount. Also, stayers had a
higher proportion than leavers of employees who had more than 6 illness days,

but this difference did not reach statistical significance (p=. 09).

Adjusted associations

HQ score
Overall, HQ scores were not associated with leaving the workforce. Wald score

P-value for this variable was not significant (p=. 21). However, it is note worthy
that subjects in the lowest category of HO scores were more likely to be leavers
(OR=1.77) (Table 4.3), which was not statistically significant, but had a tight
confidence interval with a lower bound of .96 (95%Cl=. 96-3.25). In addition to
that, the trend test indicated that there was some negative association between
HQ scores and the odds of leaving the workforce (p-value=. 05).

Liability amount paid

Having any liability amount was negatively associated with leaving the workforce.
Owing to the fact that around one half of the leavers had zero liability amounts in
1993, having any liability amount was negatively associated with leaving the
workforce. Both the wald score and the trend test p-values were significant (p<.

001).

38

Number of illness days

In the adjusted model, the number of days lost to illness in 1993 was not
associated with leaving the workforce. This lack of association was also
confirmed by the test for trend (p=. 127).

Number of short-term disability days
Similarly to the number of days lost to illness, the number of short-term disability

days was not associated with leaving the workforce (p=. 25), and there was no

significant trend in this variable.

Demographic and employment factors
Although there is a slight variation in the effect of demographic and employment

factors in the different models, employees over the age of 55, non-whites, and
part-time employees were more likely to be leavers in the four models. Subjects
in the oldest group were twice more likely to leave, non-whites had an odds ratios
that ranged from 1.5 to 1.86, part-time employees were two to three times more

likely to leave the workforce than white employees.

39

Table 4. 3: Research question 1. Comparison of leavers and stayers with
respect to HQ scores in 1993, and selected demographic characteristics.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable Categories Leavers Stayers Crude Adjusted 95% Cl Tre
s (7.) (%) P OR OR P nd
p
Age <25 11.7 7 1.5 1.88 98-359
2534 42.7 38.5 - - -
35-44 233* 34.7* .5* 57* 35.93
4554 13.5 15.9 .002 .77 .92 53-159 .00
55+ 87* 3.9* 2* 25* 1.34-5 02
Sex Male 21 .8 22.5 - - -
Female 78.2 77.4 30 1.04 1 .64-1.58 -97
Race White 91 .3 93.1 - - -
Other 8.7 5.9 '33 1.29 1.85** 1.03-34 -03
Employ Full -time 64 83.5 - - -
"‘9'“ Part-time 35* 155* <-°°1 2.83* 1.8** 1.13-2.8 -01
Status 2
Employ Exempt 31 29.7 - - -
"‘6'“ Non- 59 70.3 59 .93 .51** .41-.91 -01
Grade Exempt 6
HQ Adjusted 4.5 7.2 .74 .51 23-155
score -25—1 1 19.8 14.3 1.62 1.77 .96-3.25
93 -10—3 20.15 17.5 1.35 1.34 732.44
-2-+2 15.8 19.5 - - - .21 .05
+3-+10 25.7 27.8 .40 1.12 1.08 .51-1.9
>=+11 11.5 13.4 1 1 52.01

 

 

 

Crude ORs are derived from univariate logistic regression model.

Adjusted ORs are derived from a multivariate logistic regression model that included all the
variables listed in the table.

(-) Reference category.

*Univariate logistic regression model, p<. 05

”Multivariate logistic regression model, confidence interval for the OR does not include one

40

Table 4. 4: Research question 1. Comparison of leavers and stayers with
respect to the liability amount in 1993, and selected demographic

 

 

characteristics.
Variables Cate Leavers Stayers P Crude Adjusted 95% Cl P Trend
gorie (%) (%) OR OR P
3
Age <25 1 1 .7 7 1.5 1 .66 98-28
25- 42.7 38.5 - - - <.001
34 .00
35- 23.3" 34.7“ 02 .6* .54" .37-.80
44
45- 13.6 15.9 .77 .80 .50-1.28
54
55+ 87* 3.9* 2* 2.09" 1 .15-
3.78
Sex Male 21 .8 22.6 - - -
Fern 78.2 77.4 30 1.04 .89 .51-1.3 ~55
ale
Race Whit 91.3 93.1 - - -
e .33 .08
Othe 8.7 6.9 1.29 1.6 .93-2.7
r
Employ Full 64 83.5 - - - <.001
ment - <.0
Status time 01
Part- 36* 16.5* 2.83" 2.72” 1 94-38
time
Employ Exe 31 29.7 - -
ment mpt .69 .14
Grade Non- 69 70.3 .93 .77 .55-1 .09
Exe
mpt
$0 47.6 29.1 - - -

Liability

 

41

Table 4. 5: Research question 1. Comparison of leavers and stayers with
respect to days lost to illness in 1993 and selected demographic

 

 

 

 

 

 

 

characteristics.
Variabl Categori Leave Stayer Crude Adjuste 95% Cl Tren
es es rs s P QR d P d
(%) (%) OR P
Age <25 1 1.7 7 1 .5 1 .59** 1.01 -2.83
25-34 42.7 38.5 - - -
35-44 233* 34.7* .0002 5* .58” .39—55 .0001
45-54 13.5 15.9 .77 .82 52-13
55+ 87* 39* 2* 2.11" 1.17-3.8
Sex Male 21.8 22.6 - - -
Female 78.2 77.4 -30 1.04 .85 58-125 ~43
Race White 91 .3 93.1 - - -
Other 8.7 5.9 ~33 1.29 1.52 .89-2.61 ~12
Emplo Full — 64 83.5 - - -
yment time <.00 <.001
Status Part— 35* 155* 1 253* 2.9** 2.08-4.05
time
Emplo Exempt 31 29.7 - - -
yment Non- 59 70.3 ~69 .93 .82 .58-1.15 25
Grade Exempt
Illness None 89.8 85.2 - - -
days 1-5 1.9 2.5 33* .87 52-123
93 6+ 8.3 12.2 '05 .63* .69 .43-1.10 '31 '127

 

 

 

 

 

 

 

 

 

 

 

Crude ORs are derived from univariate logistic regression model.

Adjusted ORs are derived from a multivariate logistic regression model that included all the

variables listed in the table.

(-) Reference category.

*Univariate logistic regression model, p<. 05

“Multivariate logistic regression model, confidence interval for the OR does not include one

42

 

Table 4. 6: Research question 1. Comparison of leavers and stayers with
respect to days lost to short-term disability in 1993 and selected
demographic characteristics.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable Categorie Leave Stayer Crude Adjusted 95% Cl Trend
s 9 rs s P OR OR P P
(%) (%)
Age <25 1 1 .7 7 1 .5 1.51 292.59
25-34 42.7 38.5 - - -
3544 233* 347* .0002 .5* .57** .39-.83 .001
45-54 13.5 15.9 .77 .83 52-13
55+ 87* 39* 2* 2.11** 1.17-3.80
Sex Male 21.8 22.6 - - -
Female 78.2 77.4 ~30 1.04 .83 57-121 ~34
Race White 91.3 93.1 - - -
Other 8.7 5.9 ~33 1.29 1.5 .87-2.58 -13
Employ Full - 64 83.5 - - -
ment time <.00 <.001
Status Part-time 35* 155* 1 283* 3.04** 2.18-4.22
Employ Exempt 31 29.7 - - -
”lent Non- 59 70.3 ~69 .93 .79 551.11 ~17
Short None 89.8 85.2 - - -
term 1-5 1.9 2.5 33* .77 .27-223
disabil't .20 .25 .105
93 ' 5+ 8.3 12.2 53* .54 .37-1.09
y

 

Crude ORs are derived from univariate logistic regression model.

Adjusted ORs are derived from a multivariate logistic regression model that included all the

variables listed in the table.

(-) Reference category.

*Univariate logistic regression model, p<. 05

”Multivariate logistic regression model, confidence interval for the OR does not include one

43

 

Research question 2

Did the joiners have better HQ scores and/or lower medical costs than

stayers?

Preliminary comparison

As can be seen in Tables 4.7 - 4.10, joiners were a younger group with a
higher proportion of non-whites, a higher percentage of part-timers and hourly
employees than stayers. The two groups had comparable HQ scores, except for
a higher percentage of stayers with adjusted HQ scores. Overall, joiners had
lower medical costs than stayers, 40.9% of them had no medical costs in 1997
as opposed to 22% of the stayers. In each one of the three categories for liability
amount the percentage of joiners was significantly lower. Joiners had also less

illness and short-term disability days than did stayers.

Adjusted associations

HQ scores
The HQ score was a Significant determinant of being a joiner (p=. 01). However,

there was no trend in this variable (p=. 34). Looking at the stratum specific Odds
ratios, it seems that joiners were less likely to have an adjusted HQ score (OR=.

73).

44

Liability amount paid
The liability amount variable reached statistical significance both in the crude and

the adjusted models (p<. 001), and had a Significant negative trend (p<. 001).
Given that a Significantly higher proportion of the joiners than stayers had no
liability amount, subjects in the remaining three categories of the variable were
less likely to be joiners (Table4.8). The odds ratios were .47, .4, and .38 for the
first, second and third category respectively.

Number of illness days

Subjects that had one or more illness days were less likely to be in the joiners
group (p=. 0025)(Tab|e 4.9). This association was also confirmed by the trend
test (p=. 028).

Number of Short-term disability days

Similarly to the number of illness days, the number of absenteeism days due to
short-term disability was negatively associated with joining the workforce (.0037)
(Table 4.10). The test for trend was significant too, (p-value=. 0037).
Demographic and employment factors

Older employees were less likely to be joiners, this association was weak
(OR=.9), but reached statistical Significance in all four models. In addition to age,
race was another Significant factor, non-whites were twice more likely to join the

workforce than whites.

45

Table 4. 7: Research question 2. Comparison of stayers and joiners with
respect to the HQ scores in 1997, and selected demographic

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

characteristics.
Variabl Categori stayers joiners Crude Adjusted 95% Cl Trend
es es P OR OR P P
Age 41 .1* 3354* <90 .91 * .9* .89-.91 <.001
1
% %
Sex Male 22.6 22.8 - - - .82
Female 77.4 77.2 39 .82 1.02 81-13
Race White 93.1* 879* - - -
Other 5.9* 12.1* <00 185* 215* 154-3 <-°01
1
Emplo Full — 78.2 70.9 - - -
yment time <00 .82
Status Part- 21 .8’ 29.1“ 1 1 .47* 1 .02 .8-1 .3
time
Emplo Exempt 23.1 14.3 - - -
yment Non- 759* 857* <00 1.79* 1.2 .93-1.55 -15
Grade Exempt 1
HQ Adjuste 13.2* 8.7* .63* 1.54* .37-.8
score d
93 -25— 8.1 8.3 1.33 1.2 .8-1.8
11 .0054 .01 .34
-10—3 17.9 15.5 .83 .9 .65-1.25
-2-+2 19 19.7 - - -
+3-+10 30.7 36.2 1.13 .76-1.32
>=+1 1 1 1.2 11 .6 1 .9 63-128

 

 

Crude ORs are derived from univariate logistic regression model.

Adjusted ORs are derived from a multivariate logistic regression model that included all the
variables listed in the table.

(-) Reference category.

*Univariate logistic regression model, p<. 05

”Multivariate logistic regression model, confidence interval for the OR does not include one

46

Table 4. 8: Research question 2. Comparison of stayers and joiners with
respect to the liability amount in 1997, and selected demographic
characteristics.

 

 

 

 

 

 

 

 

Variables Catego Stay Joiner Crude Adjusted 95% Cl Trend
ries ers s P OR OR P P
Age 41.1 33.54 <.001 .91* .91" .9-.92 <.001
% %
Sex Male 22.8 - - -
22.6 .89 .50
Femal 77.2 .82 1 .08 .85-
e 77.4 1.36
Race White - - -
93.1 87.9* <.001 .0001
Other 6.9* 1 .85* 1 .9“ 1.39-
12.1" 2.59
Employment Full— 78.2 70.9 - - -
Status time <.001 .84
Part- 21 .8 29.1* 1 .47* .97 .78-
time * 1.22
Employment Exem 23.1 14.3 - - -
Grade pt <.001 .11
Non- 76.9 1 .79* 1 .22 .95-
Exem * 85.7“ 1.56
pt
Liability $0 22 40.9 - - -
3mm“ 51-99 14.3 122* .45* .47** .35-53
9 <.001 <.001 <.00
$100- 29.9 225* .4* .42** .33-.55 1
499 *
$500+ 33.8 24.3‘ .38* .43" .34-.55

 

 

 

 

 

 

 

 

 

 

Crude ORs are derived from univariate logistic regression model.

47

 

Adjusted ORs are derived from a multivariate logistic regression model that included all the
variables listed in the table.

(-) Reference category.

*Univariate logistic regression model, p<. 05

“Multivariate logistic regression model, confidence interval for the OR does not include one
Table 4. 9: Research question 2. Comparison of stayers and joiners with

respect to days lost to illness in 1997 and selected demographic
characteristics.

 

 

 

 

 

 

 

 

Variable Categori Staye Joiner Crude Adjusted 95% Cl Trend
s as rs s P OR OR P P
Age 41.1" 33.54 <.001 .91“ .91* .89-.92 <.001
% %
Sex Male 22.6 22.8 - - -
Female 77.4 77.2 -89 .82 .94 751.18 ~53
Race White - - -
93.1” 87.9’ <.001 <.001
Other 6.9* 1 85* 2.07" 1 .52-
12.1* 2.81
Employ Full — 78.2 70.9 - - -
ment time <.001 .1 3
Status Part— 21 .8* 29.1' 1 .47* 1 .18 .95-1 .46
time
Employ Exempt 23.1 14.3 - — -
mam Non- 76.9* ‘0‘“ 179* 188* 1.07- 012
Grade Exempt 857* 1.79
Illness None 35.3 41.1 - - -
days 1-5 488* 444* 78* 7* .56-.86 “025 023
93 6+ 159* 145* '01 78* 57* .51-.9

 

 

 

 

 

 

 

 

 

 

 

Crude ORs are derived from univariate logistic regression model.

Adjusted ORs are derived from a multivariate logistic regression model that included all the
variables listed in the table.

(-) Reference category.

*Univariate logistic regression model, p<. 05

”Multivariate logistic regression model, confidence interval for the OR does not include one

48

Table 4. 10: Research question 2. Comparison of stayers and joiners with
respect to days lost to short-term disability in 1997 and selected

demographic characteristics.

 

 

 

 

 

 

 

 

Variables Categori Staye Joiner Crude Adjuste 95% Cl Trend
es rs s P OR dOR P P
Age 41.1* 33.54 <.001 .91* .91“ .90-.92 <.001
% %
Sex Male 22.6 22.8 - - -
Female 77.4 77.2 -39 .82 .94 .75-1.18 53
Race White — - -
93.1* 87.9* <.001 <.001
Other 69* 1 .85* 2.05" 1 52.8
12.1*
Employ Full — 78.2 70.9 - - -
ment time <.001 .04
Status Part- 291* 1 .47* 1 .24“ 1-1 .54
time 218*
Employ Exempt 23.1 14.3 - - -
"‘6'“ Non- ‘0‘” 179* 1.22 .96-1.55 -09
Grade Exempt 759* 857*
Short None 87.2 90.5 - - -
term 1-5 25* 12* .45* .43* .20-.89
:Eab'my 6+ 10.3 8.3 '01 .78 53* .45-85 '00” '0037

 

 

 

 

 

 

 

 

 

 

Crude ORs are derived from univariate logistic regression model.

Adjusted ORs are derived from a multivariate logistic regression model that included all the

variables listed in the table.
(-) Reference category.
*Univariate logistic regression model, p<. 05

”Multivariate logistic regression model, confidence interval for the OR does not include one

49

 

Research question 3

Did the joiners have better HQ scores and/or lower medical costs than

leavers?

Preliminary comparison

Tables 4.11-4.14 summarize the results of research question 3. It is
evident that Joiners were a younger group than leavers. A Significantly higher
percentage of joiners were under the age of 25, and a higher percentage of
leavers were over the age of 55. Joiners were also more likely to be hourly
employees than leavers. Some differences in health related variables were also
evident. A significantly higher percentage of leavers had a low score of -25—1 1,
but at the same time a higher proportion of joiners had liability amounts of $500
or more. NO differences were evident between the two groups in respect to

absenteeism.

Adjusted associations

HQ scores
Although Wald score p-value (p=. 78) and the trend test for the HQ score variable

(p=. 18) were not significant, subjects that had a score Of -25 to -11 were less

likely to be joiners (OR=. 38) (Cl: .19-.75).

5O

Liability amount paid
Having a liability amount over $500 was associated with being a joiner, (OR=1.6)

(Cl=1-2.57). Also, the test for trend indicates a positive trend in this variable (p=.

0003)

Number of illness days
No association was Observed between the number of illness days and joining the

workforce (p=. 35), and the test for trend did not yield Significant results (p=. 16).
Number of Short-tenn disability days

Likewise, the number Of short-term disability days was not associated with joining
the workforce (p=. 7). The test for trend yielded a p-value of .70.

Demographic characteristics and employments factors

Subjects in the youngest group were more likely to be joiners; the Odds ratios
ranged from1.53 to 1.71 in the different models. Hourly employees were three
times more likely to be joiners. Employees over the age of 55 and part-time
employees were less likely to be joiners, with Odds ratios of .28 for the first and

.58 to .82 for the second.

51

Table 4. 11: Research question 3. Comparison of leavers and joiners with
respect to HQ scores in 1993 and 1997, and selected demographic

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

characteristics.
Variables Categorie Leaver Joiners Crude Adjust 95% CI Trend
s s (%) P OR ed P P
(%) OR

Age <25 1 17* 207* 1 85* 1 .53 .80-2.92

25-34 42.7 41 - - -

35-44 23.3* 24.8 <.00 1.1 1.32 .79-2.20 .009

45-54 13.5 11 1 .84 .75 .40-1.38 6

55+ 87* 2.5* .29* .28 .12-.82
Sex Male 21.8 22.8 - - -

Female 78.2 77.2 76 .94 .70 .42-1.15 -17
Race White 91 .3 87.9 - - -

Other 8.7 12.1 -17 1.43 1.15 51-219 10
Employ Full- 64 70.9 - - -
ment time .05 .79
Status Part-time 36 29.1 .73 .82 .50-1 .34
Employ Exempt 31 14.3 - - -
"‘9'“ Non- 59* 857* <00 258* 3.15* 1.97-5.02 -000
Grade Exempt 1 5
HQ Adjusted 4.6 8.7 1 .62 1 .67 .62-4.47
score 93 -25—11 19.8' 8.3' .355 .38 .19-.75

-10—3 20.16 15.5 .64 .69 .36-1.31

-2-+2 16.8 19.7- .92 - - - .78 .18

+3-+10 26.7 36.2.6 1 .15 1 .24 .66-2.20

1
>=+1 1 1 1.5 1 1.6.6 .86 .88 .42-1.84
9

 

Crude ORs are derived from univariate logistic regression model.

Adjusted ORs are derived from a multivariate logistic regression model that included all the

variables listed in the table.
(-) Reference category.
*Univariate logistic regression model, p<. 05

”Multivariate logistic regression model, confidence interval for the OR does not include one

52

 

Table 4. 12: Research question 3. Comparison of leavers and joiners with

respect to liability amount in 1993 and 1997, and selected demographic

 

 

 

 

 

 

 

characteristics.
Variable Categorie Leavers Joiners P Crude Adjuste 95% CI P Tre
s s (%) (%) OR d OR nd
p
Age <25 1 17* 207* 185* 1.71** 1.03-2.83
25-34 42.7 41 - - -
3544 23.3* 24.8 <.001 1.1 1.22 81-183 .000
45-54 13.5 11 .84 .87 .53-1.45 2
55+ 87* 25* 29* .29" .15-.58
Sex Male 21.8 22.8 - - -
Female 78.2 77.2 -76 .94 .81 .54-1.22 -64
Race White 91.3 87.9 - - -
Other 8.7 12.1 -17 1.43 1.25** .82-2.18 ~20
Employ Full — 64 70.9 - - -
ment time .05 .013
Status Part-time 36 29.1 .73 53** .43-92
Employ Exempt 31 14.3 - - -
"'9'“ Non- 59* 857* <-001 258* 2.87** 1,944.25 ~04
Grade Exempt
Liability $0 47.5 40.9 - - -
3mm” 31-99 92* 12.2 1.53 1.42 .80-2.53
93 $100-499 282* 22.6 '024 .93 .79 52-12 '03 '03
$500+ 15* 24.3* 187* 150" 1-257

 

 

 

 

 

 

 

 

 

 

 

Crude ORs are derived from univariate logistic regression model.

Adjusted ORs are derived from a multivariate logistic regression model that included all the

variables listed in the table.

(-) Reference category.
*Univariate logistic regression model, p<. 05
“Multivariate logistic regression model, confidence interval for the OR does not include one

53

 

Table 4. 13: Research question 3. Comparison of leavers and stayers with
respect to the number of days lost to illness in 1993 and 1997 and selected
demographic characteristics.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable Catego Leaver Joiners P Crude Adjusted 95% Cl P Tren
s ries s (%) OR OR d
(%) p

Age <25 1 1 .7* 20.7" 185* 1.67“ 1.01-2.77

25-34 42.7 41 - - -

35-44 23.3 24.8 <.001 1.1 1.19 .79-1.78 <.001

45-54 13.6 11 .84 .83 .50-1.36

55+ 87* 2.5‘ .29* .28” .14-.56
Sex Male 21 .8 22.8 - - -

Femal 78.2 77.2 -75 .94 .82 55122 ~54

e
Race White 91.3 87.9 - - -

Other 8.7 12.1 -17 1.43 1.24 72-215 ~35
Employ Full — 64 70.9 - - -
ment time .05 .001
Status Part- 36 29.1 .73 .58" .40-.83

time
Employ Exem 31 14.3 - - -
ment pt <.001 .001
Grade Non- 69* 85.7* 4.52 3.09" 203-4.7

Exem

pt
Illness None 89.8 41.1 - - -
days 1-5 1.9 44.4 -90 1.05 .93 .54-1.34 -35 ~16
93 6+ 8.3 14.5 .98 .75 .45-1.25

 

Crude ORs are derived from univariate logistic regression model.

Adjusted ORs are derived from a multivariate logistic regression model that included all the
variables listed in the table.
(-) Reference category.

*Univariate logistic regression model, p<. 05

“Multivariate logistic regression model, confidence interval for the OR does not include one

54

 

Table 4. 14: Research question 3. Comparison of stayers and joiners with

respect to the number of days lost to short-term disability in 1993 and 1997

and selected demographic characteristics.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable Categories Leavers Joiners Crude Adjust 95% CI
s (%) (%) P OR ed OR P P
Age <25 11.7 20.7 1.85’ 1.68‘ 1.01-
' 2.77
25-34 42.7 41 <.001 - - - <.001
35-44 23.3" 24.8 1.1 1 .20 .80-
1.81
45-54 13.6 11 .84 .84 .51-
1.38
55+ 87* 2.5 .29* .28“ .14-.57
Sex Male 21.8 22.8 - - -
Female 78.2 77.2 -76 .94 .82 .55- ~62
1.23
Race White 91.3 87.9 - - -
Other 8.7 12.1 ~17 1.43 1.23 71- ~33
2.13
Employ Full —time 64 70.9 - - -
"'6'“ Part-time 35* 29.1 -05 .73 .50** .42-85 0023
Status
Employ Exempt 31 14.3 - - -
”'9'“ Non- 59 85.7 <~°°1 4.52 289* 1.95- <-°01
Grade Exempt * 4.26
Short None 89.8 90.5 - - -
“3"" 1+ 10.2 9.5 -75 .92 .92 54- -7 -70
disabilit 1.56
y93

 

Crude ORs are derived from univariate logistic regression model.

Adjusted ORs are derived from a multivariate logistic regression model that included all the
variables listed in the table.

(-) Reference category.

*Univariate logistic regression model, p<. 05

“Multivariate logistic regression model, confidence interval for the OR does not include one

55

 

Chapter 5

DISCUSSION

The aim of this thesis is to investigate whether incorporating an
incentive/disincentive approach in a worksite health promotion program (HPP)
produced any selectivity in employment. The strategy used to address this
question was a parallel comparison of three groups: those who left the workforce
after implementing the HPP, those who joined the workforce, and those who
stayed employed throughout the five-year period of the program. The undertying
assumption was that if the program produced any selectivity in employment,
employees with poor health would be more likely to leave the workforce and
joiners would be in better health status than the existing workforce.

Four main outcomes of interest were considered to reflect the health
status of the employees: 1) HQ score derived from the annual health screening
2) the liability amount paid by the hospital for the health insurance, which
reflected health care costs of the employees 3) number of absenteeism days due
to illness 4) number of absenteeism days due to short-term disability.

The first research question was: did the leavers have more adverse HQ
scores and/or higher medical costs than stayers?

The analysis revealed that when adjusting for all other variables, low HQ
scores had a weak association with being a leaver (OR=1.77). However, this
association did not reach statistical significance (95%Cl=. 96-3.25). As for the

medical costs, leavers did not seem to have higher medical costs than stayers.

56

On the contrary, high liability amounts were negatively associated with leaving
the workforce, the number of absenteeism days due to illness or short-term
disability was not associated with leaving the workforce. Age was a significant
predictor of leaving the workforce. The youngest and oldest employees were
most likely to leave. Race is another demographic characteristic that was
significantly associated with leaving the workforce; non-whites were more likely
to be leavers. Part-timers were also more likely to leave the workforce, whereas
hourly employees were less likely to leave the workforce.

The second research question was: did the joiners have better HQ scores
and/or lower medical costs than stayers?

The two groups had similar HQ scores, except for a higher proportion of
stayers with adjusted HQ scores. Further, it is evident that joiners had lower
medical costs than leavers. In the multivariate model, having any liability amount
from $1 to over $500 was negatively associated with joining the workforce. The
numbers of illness days and short-term disability days were negatively
associated with being a joiner. Non-whites were twice as likely to join the
workforce, and age had a weak but statistically significant association with joining
the workforce (R=. 9).

The third research question was: did the joiners have better HQ scores
and/or lower medical costs than leavers?

Subjects that had low HQ scores of -25 to -11 were less likely to be joiners
(OR=. 38). On the other hand, having a liability amount in the highest category

($500+) was also associated with joining the workforce. Employment grade was

57

another factor that differentiated the two groups; Joiners were 4 times more likely
to be hourty employees than leavers. No relation was observed between the
number of absenteeism days and joining the workforce.

Overall, the results indicate that leavers had poorer health than the other
two groups, which was reflected by a high percentage of leavers having low HQ
scores of ~25 to -11. However, medical costs differences between the three
groups were not correlated with their HQ scores. On the contrary, leavers had
the lowest liability amounts. One potential explanation for this finding is that
employees who are undergoing a medical treatment, for example, are more likely
to stay employed than others due to their need for health insurance benefits.

The number of illness days is another measure of general well-being that
was used in many evaluations of worksite HPPS, and was found to be the most
consistent positive outcome of comprehensive worksite HPPS."'2° In this
analysis, The only significant difference in absenteeism was between joiners and
stayers; joiners had significantly lower numbers of illness and short-term
disability days, which was consistent with having lower medical care costs than
stayers. Further, absenteeism is a measure of costs, but in the first and the third
comparisons there was no correspondence between the effects of the two
variables.

Medical costs were found to be reduced by participation in worksite HPPS
in many studies.”24 Yet, some studies indicate that health care costs are not
always correlated with health status; a person with good health may seek

medical care frequently to confirm his health status, a phenomenon described as

58

the “worried well syndrome.” Which might explain in part the conflicting results
of the current analysis.

The main drive for conducting this analysis was the concern that a
disincentive approach might alienate employees with health problems, or who
engage in negative health practices, instead of urging them to improve. But the
three groups had comparable health-related behaviors and HQ scores, and not
only did the leavers not have higher health care costs, they had the lowest mean
health care costs, despite higher mean HQ score than the stayers.

One potential explanation for the lack Of correlation of HQ scores with
medical costs is that risk-factors such as high cholesterol levels or unhealthy
nutrition may not be manifested as diseases within a five year period as opposed

to motor vehicle safety factors that might have an immediate impact.

Strengths and weaknesses of the Heatthlus program

The literature indicates that, extending the lengths of follow-up was
associated with a positive measurable impact of interventions.”"42 HealthPlus
had a five-year follow-up period, which is a reasonable length for assessing the
effect of the intervention.

Another strength of the HealthPlus program was the inclusion of a
biometric fitness test among other objective measures. The vast majority of
programs reviewed relied on self-report of physical activity as a measure of
fitness. Further, unlike many of the programs, HealthPlus was evaluated

independently from the planning team.

59

Nevertheless, it is important to note that HealthPlus had some
weaknesses too. The reliability of the measurement methods used in the
screening was not tested, nor was the validity of the self-reported data.

Moreover, the use of a complex prediction model to determine point value
of the measured risk factors was very new to risk-rated programs. However, the
model failed to take into account the ease of modification. For example,
modifying the blood pressure is harder than wearing a seat belt or a helmet, but
the two components had the same amount of maximum annual deductions (see

Table 3).

Limitations of this analysis

Several limitations Should be considered when interpreting the results of
the current analysis. A major problem was the lack Of screening data for the
group of leavers. A total of 63% of leavers had valid HQ scores in 1993. It is not
clear whether the unavailable data were truly missing or simply reflected non-
participation in the health screening. The data that was sent to us from the
Wellness Center in Grand Rapids did not include non-participants in the
screening. In addition, subjects that elected not to participate in the screening
were defaulted to a score of -25. Yet, replacing the missing HQ scores with -25
would have seriously biased the results; the HQ score was considered to reflect
the health status of employees. Therefore, a low score of -25 may not reflect the
health status of non-participants. Therefore, for the purpose of this analysis

invalid HQ scores, for any of the subjects, were dealt with as missing data.

60

An additional limitation of the current analysis is the lack of a control
group. It is difficult to attribute any observed changes in the workforce to the
HealthPlus program, without comparing the results to a similar workforce over
the same period of time. Which presents a threat to the internal validity of the

study.

61

Chapter 6

CONCLUSIONS

The three studied groups clearly differed in health-related factors, but the
results are conflicting. The data suggests that Leavers had poorer health
evaluations than stayers and joiners. Nevertheless, these differences in health
status were not reflected in health care costs or the number Of absenteeism
days. The data further indicates that Joiners had lower medical costs and less
absenteeism days but did not have higher HQ scores.

Given the current results and the limitations mentioned in the previous
chapter, one cannot draw firm conclusion. Then again, one cannot reject the
hypothesis that the incentive/disincentive approach may result in selectivity in
employment. Thus, this aspect of the HealthPlus Health Quotient program Should
be further studied- using a control group- to determine which of the observed
differences are truly due to the program.

The current results as well as previous reports on the HealthPlus Health
Quotient program47are not in favor of widespread use of the
incentive/disincentive approach. However, this approach needs to be studied

more extensively in order to resolve what is ethical and what is not.

62

APPENDICES

63

APPENDIX A

BUTTERWORTH QUESTIONNIAR

 

Please note: The numbers on the following two
pages should be entered from top to bottom. M
digits must be complete, If your answer is "3” for a INCORRECT

Z—digit answer, you must record “03". See example.

 

 

 

 

 

CORRECT

 

 

 

 

 

 

 

l. AGE (41 ofjdnuary I ofnextyear) .0, ~‘_ 1‘ 5 ‘_ 5 ‘5 7 5 +9 2_ GENDER \ Male

0‘ 3'1;2.3‘ 1., ,5" 5.; 37. '3-19 .7 Female

 

 

 

 

 

 

 

 

 

3. TOBACCO USE

How would you describe your use of tobacco products? (Complete either "a" or “b"; not bulb.
lnclude all forms of tobacco use: cigarettes. cigars. pipes and smokeless tobacco).

 

.1. I do not use tobacco products. 6. I presently use tobacco products or have recently quit
‘. l have never used tobacco products or quit using within (In past month.
tobacco products for at least one or more 5:“ 5. On the average, how many cigarettes. cigars, pipes and/or
(7 1 recently quit using tobacco products (quit for at smokeless tobacco do/did you smoke/chew per day?
least 1 month, but not more than 12 months). .0; (1.2; 3 4 ,5. 5‘7, .- 9‘,

 

 

 

 

.0- 'T} = a; (3', .3,_.s.. 517 .8“. ‘s‘t

/

 

 

 

4. ALCOHOL USE

How many drinks of alcoholic beverage do you consume in a typical week? "you never drink, use “00": ifyou only drink
once in awhile or less than once per week. use "01 (Note: one drink equals one )2 oz. can of beer, one 12 07.. wine cooler,
one 6 01. glass of wine or one ounce of. hard liquor).

Average drinks per week:

 

 

 

 

 

 

 

S. MOTOR VEHICLE SAFETY

 

 

 

 

a. What percentage ofthe time do you wear your c. How many points do you currently have on your

scatbelt when riding in a motor vehicle? driving record?
(Example: 100 = 100% of the time; 095 = 95% (Examples: speeding l l-IS mph over limit: 5 pts: 15+
of the time; 050 = 50% ofthe time). mph over limit: 4 pts. "you have questions regarding
Percentage: point values for specific offenses. guidelines will be

0 ‘1 2 3‘ 4 5 6 7 a 9 available at the time of your screening).

0 1 2...; 4 s 5 .7, a 9 Points:

0123456789 0123455739

 

 

 

 

 

 

 

 

 

 

d. How man ' times Qiiwjygra gittollth do you drive or ride
_ l?- ‘ .
b_ [)0 you ride a motorcycle? Yes -. ' N0 in a motor vehicle when the driver has been under the

 

 

 

 

 

 

 

 

 

 

If yes. do you wear a Mm“? Yes 1 N0 influence of alcohol and may be impaired? (Generally, this
' ' ' would be defined as 2 or more alcoholic drinks within one
hour ol‘driving.)
Times: _.
0 l 2 3 4 5 6 7 a 9
o (3)3)3. 4 5 5 7 s 9
I - I 2

64

Figure A.1 (cont'd)

 

Elgase ngte: The numbers on the following two
pages should be entered from top to bottom. All
digits must be complctg If your answer is “3" for a INCORRECT

Z-digit answer, you must record “03". See example.
. . 0 05.32.? 311' 5 s 7129‘- .97
CORREC T Ti} (3,, (a. (3.53;..3‘; 1%?!) 1'3),

 

 

 

 

 

 

 

 

 

 

 

 

 

l. AGE (4: ofjanrmry 1 ofnexrycar) 10.1.1 2 3714' .51.; .‘7 (a; .9 2_ GENDER 9. 3; Male

(“0 l ('1‘. ’2", "j, 4",. 5, ’9“. "7‘ "a; ‘79 ,3 “ Female

 

 

 

 

 

 

 

 

 

3. TOBACCO USE

How would you describe your use of tobacco products? (Complete gither "a" or "b"; [Lo_t __9(h.
Include all forms of tobacco use: cigarettes, cigars, pipes and smokeless tobacco).

 

a. I do no! use tobacco products. b. l present/3y use tobacco products or have recently quit
r l have never used tobacco products or quit using wig/u), (I); past month.
tobacco products ‘0' a! least one or more years. On the average, how many cigarettes. cigars. pipes and/or
;' ..l i recently quit using tobacco products (quit for at smokeless tobacco do/did you smoke/chew per day?
least 1 month. but not more than 12 months). .0; ..1. :2. 3' . .5. 6, 7. a 9,

 

9.?"1‘: 23 slit/w 7.“ '0" is)

 

 

 

 

 

 

4. ALCOHOL USE

How many drinks of alcoholic beverage do you consume in a typical week? ll. you never drink, use "00"; if you only drink

once in awhile or less than once per week. use “0i". (Note: one drink equals one 12 oz. can of beer, one I). 02.. wine cooler,

one 6 02. glass of wine or one ounce of hard liquor).

Average drinks per week:
01,2314 5 s 7- 8' 9

 

 

 

 

 

 

 

 

S. MOTOR VEHICLE SAFETY

 

 

 

 

a. What percentage of the time do you wear your c. How many points do you currently have on your

seatbelt when riding in a motor vehicle? driving record?
(Example: 100 = 100% ofthe time: 095 = 95% (Examples: speeding 11-15 mph over limit: 3 pts; 15+
of the time; 0'30 = 50% ofthc time). mph over limit: 4 pts. If you have questions regarding
Percentage: point values for specific offenses. guidelines will be

0 ‘ 2 3 3 5 6 7 8 9‘ available at the tlme of your screenlng).

Vo.=1 2‘I3.‘ 4 s or 7' ,3 9' Points:

0123456789 0123455789

 

 

 

 

 

 

 

 

 

 

d. How many times ig an_gy_e_r;q;c;_ltgo_n_t_b do you drive or ride
b. [)0 you ride a motorcvcle? Yes ‘ No in a motor vehicle when the driver has been under the

 

 

 

 

 

 

 

 

 

 

If? yes, do you wear a helmet? Yes - ,’ No influence of. alcohol and may be impaired? (Generally, this
' ' would be defined as Z or more alcoholic drinks within one
hour of driving.)
Times:
0 1 2 3 ‘ 4 5 5 7 s 9
I o 03:181.?) ~ 5 . r . .
I I I Z

65

Figure A.1 (cont'd)

. _— -~—._...-

 

6. NUTRITION

. O O O I C I I O U C O O I

How often do you consume foods in the following categories? For each of the categories listed, please estimate to
the best of your ability the number of servings you eat er day 9_r per week. The sample food items and servingm
are intended to be guidelines only. Other food items t at you may consume from each of the groups should be

included in your estimations.

 

 

 

 

' Sample Fuosthems

 

Serving Sizes

 

 

Elumber of daily or weekly

 

Is this the

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

‘ servings: amount you eat
I i .Esunrls: _ _ __ Pcr- -
I O 1‘ 2 3 4 s s 7 s a W
. . >
I I 8 'o 1 2 3 4‘3's-va ’7‘.’s~ Day? “'k‘
a. Bread. cereal, rice, noodles. 1 slice of bread, ”2 cup cooked rice 3
crackers. pretzels. dinner rolls, or noodles, ”2 cup cooked cereal, I *3 10- .‘ I 9 2 ‘ 3' 4 '»5 0 7 8‘ -9 C . ( ,)
bagels, potatoes. flour tortilla, ounce ready-taveat ceral, I roll. 3 ’0‘ ‘ > '5'? ' '3 ' "i" "5" '0 7 ’8 '9
air or lite popcorn cups popcorn
b. I-ruits or 100911 fruit juices l small piece fresh, Ill cup canned. 3
IM cup dried, 5M cup juice 0 ‘ 2 3 ‘ 5 5 7 3 9 x
e 1‘ 3 2 . '3‘ ’4 s a 7 s 9
c. Vegetables ”2 cup chopped raw or cooked. 1 cup
leafy raw 3 o _ 1 2 3 4 s o 7 a 9
-o 1 '2 3 4 s ‘s 7 s a
d. Iowfat dairy products (skim. l/2”‘o 1 cup of milk or yogurt. I5 —2 ounces
or l'l'o milk. lowlat yogurt or frozen i (ll-LIIL‘CSC. l/l cttp ice milk 0 ‘ 3 3 ‘ 5 5 7 a 9
yogurt. ice milk. lowlat cheese) 5 o 1 2 :I 4 s a 7 s 9
I .
e. Lean meat. skinless poultry, fish. 3 ounces cooked meat, poultry or fish,
dry beans. egg substitute. lean & ”2 cup cooked beans, I egg substitute _ ° ' 2 5 ‘ 5 5 ’ ° "
trimmed beef and pork. tuna. wild 0 t 2 3 4 5 6 7 0 9
_Hgamc _ __,___.____-_.__._.-_-_________.___. __ ___________ - _F .,1__,_ ___ _ --. __ ---.
f. W’hole milk dairy products (whole : l cup of milk or yogurt. l5 - 2
or 2"‘o milk or yogurt. ice cream. ounces ol‘cheese. ”2 cup ice cream 0 ‘ 2 3 ‘ 5 5 7 ° 9
cheese. pudding) 0 , ‘ 3 3 ‘ 5 5 ’ 0 ‘ 9
.__-_.._. _....._-.-,.. _ -~__- -... .-.. .. .. _ __.__. .._-._._. _._ “if __ __c -_ ___-“ _ __ _ -
g. Eggs 8: high fat meats (sausage, 3 ounces cooked meat. poultry or fish,
luncheon meats, salami. bologna. 1 egg, 1 tablespoon peanut butter _ 0 ‘ 2 3 ‘ 5 ‘ 5 7 3 "
corned beef. hot dogs, hamburger, 0 ‘ ? 3 l 5 0 7 0 9
.....- Emil-£532?" b.3955: . ___ ___.
It. Deep fried foods (meat. poultry. 5 ounces cooked meat. 840 l'rics __ «__ ,_ __ _fi (___- ' _~' “ '
Iish. vegetables. potatoes) 3 ° ‘ 7 1 ‘ 5 5 7 3 9 '
o' 1. 2 s 4 vs a 7 s 9
i. Butter. margarine. oils. sour cram. l teaspoon butter. margarine or oil;
dressings. and cream cheese (not 1 tablespoon dressing. nuts or seeds: --..e. 0 ‘ 3 3 ‘ 5 ° 7 ' °
including fat free), bacon. gravy, 1 slice bacon 0 ‘ 2' 3 4 5 6 7 0 9
nuts/seeds j J
j. (Likes. cookies. pastries. ”8 pie. 2 inch square III-Lake, l ’ ‘ _ l
doughnuts. chips. regular popcorn haudlul ol‘chips (approx. I5), 2 ° ‘ 2 3 ‘ 5 5 7 ° 9 ’
(incltttlc “lat l'ree" snacks! cookies 0 1 2 J 4 s e 7 a 9
l
r It. Candy. sugared drinks (pop. 12 ounce sugared pop or drink
3 KooI-aid. liruit punches) 0 ‘ 2 3 ‘ 5 5 7 3 9
o 1~ 2 J 4 s s 7 s 9

 

 

 

 

 

 

 

 

66

Figure A.1 (cont'd)

 

 

The information on the preceding pages accurately and honestly reflects my health and lifestyle.

misrepresentation on my part will result in an adjustment of my Health Quotient.

 

Signature

 

(must be in ink)

I understand that any

 

Date

 

 

 

 

 
 
 

STOP

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

K hid ' ELLA
7. EVALUATION 8. BLOOD PRESSURE 9. BODY COMPOSITION
DATE:
Month Day Year SBP Ulll’ Skinfnld BIA Height W'eight
9 “‘0 9‘6 ins
a, or o ,0 go o .o o o.t1‘_ o 0,. pl 0; 0 or o o 01.0. \qfioj 03.9)
11; 17 141' 71 1 ("1,11 1‘1: 1 1 1, 1 1 1_M\1, 1'11} 1111
{2 2 »2. r2 2 :2“. 2‘ ‘2 2 2 2, 2' 2 :2. 2 2 .;2',t- 21:22
.'3_- 3 '3 (31 l 3' 3' '31 3 3 3 3 3 3' 3 3 :1“ 34:3}:
:41 4': (4 1 '4" 4 g4 4 4 -_4,4 4.14 41 .4 i4" “'iiil'iv‘.
'5': '5 .s l ‘5 '5 s s .s. 515‘ s '51 5 s ‘5 .s sfiiwsfis:
'6" 6 s e 6 a 6 a s 6 a j" '6 s 3‘ 4} 6-
=7". 7; :7, l 7 7 f7 7 .7: 7 :‘7‘ 7' 7 7 r7")- 7»
'a; a a l at s a '3 a 2 a a; .3 is, a'
79 31.4 I ‘11 9 9 9 .9' 9 3 9i ,9 ‘9 9:.
10. FITNESS WALK
I n. CHOLESTEROL , 12. OTHER
Mile 'l'ime Heart Rate ‘ i PHYSICIAN COMPANY
min. sec. (hpm) l TONI ”UL l COIN“. (r()l)li
r._ . ,
1 l
_ l I o o 6
o o o o o o o l o o o o o o ! o o o o .. 0 Employee
1 1 1 1- 1 1 1 i 1 1 1 1 1‘ 1 l 1 1‘1 1 1‘1 1 -Spousc
l 1
2 2 2 2 2 .2 2 l 12 2 2 2‘ 2 ‘ 2 27 2 2 2 42: 2 V Dependent
,3 3 .1 3. 3 3 i 3 3 3 :1 3 :1 3 3 3 3 3. 3
|
4 4 4 4 4 4 4 4 4 4 i 4 4 4 4 4 4 4
s 5 5 s s . s s s s s 5 s 5 s s 5 s 5
a s 5 s o l 6 6 s 5 i s 6 s a o 6 .
:7 7 7. 7 7 5 7 7 7 7 7‘ 7‘ 7 7 7 7 7
a a ‘11 a a l a a a a ; a a‘ a a s a a
9 9 79- 9 -9. _. 9 9 9 9 ; 9 9‘ 9 9 9 9 9

 

 

 

 

 

 

III

 

 

 

 

 

 

 

 

 

 

 

DO NOT COMPLETE THE NEXT SECTION. THESE TESTS WILL BE MEASURED DURING:
" THE HEALTH SGREENING PROCESS BY WELLNESS CENTER PERSONNEL.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

PLEASE DO NOT MARK IN THIS AREA

 

 

 

 

 

 

 

 

67

Printed in U SEA.

 

J
, I

 

lbs

Mark Hello: ' by NCS MM‘O4667~3 321

00135

 

 

APPENDIX 8

HEALTH QUOTlENT POINT SYSTEM

iiigéggpfi.
EEE£§.35-Iao.£
oiiEE-Ioiag-
Eggbafiflsiéuxioii
w...

L
A

,iE£bE£.§E§l§8-.

.2355:
Entries-3.3.533.
EEEoSE-«igma
.11-ginning...
.Eh§.§;§£
a§£s§§.!3

gum—£523

zoo-a

.IEIEE‘E-
EE-EEBnEESSq-igbg
.iog§.§.313§hi

«p.

r
‘

 

 

.Bguggggi 5.1113333 .5185928832982925
_n __ a.
.1!
.lplltiluagesoagg Egigszg .Elgiugogbgfg

PM a or
its... ... ~ ...-.5 51% a ... 33.. 25. 5:.
glow .85. Stash... gel-55:.
in... ES :8 38653.
12313 .85 5:23 385.22..
1.3.2.5.” ES 331.: r838...
12%}: .83. 51.83... 535:.
«358.3 .82... Sheet. .823?
13.12 55 5‘22“ 3:3.
.8123 5.... 551$ 335$.
ruins b n 55 551 a w 3pm.
geek 55 sale. 915.

e. a ..

 

525. a: E S: SE2:

31:32 ..c:

32am 2:; ES 2:332

555 ES.— EEEE =55:
5.552... 32:55: 95335::

 

E.
:25- ...—.3

3.. 93:2

2285328
___—.33.

355:-
.22-:-

 

2353 a:

Figure 8.1 Health quotient system

68

Figure 8.1 (cont'd)

 

.855... 8 «85:8 E058... Saga: 2. a. 8!: 3 E .856 p.98. :8— 5? 33.08.»... was—EB 2... En E85 .58
01!: SuzaBSugEofigwaSEEuS-fififlgaeg .8.83§.895oz§¢u8_88g§ 53w.

 

 

 

 

 

 

 

      

 

 

 
 

aw.
.....i y a
tan.
3+ 22.: 2::
it: a as H...-:3
It. .a a — 5.3 ad 612‘. g: b o n ”-3.2 2:: .3:
In .3 o w “33.. w v 31.2 NF. hos—mu3—G
...fi _. r
.202 a o: 9.825
Is. .a 3 23:5 3 . 8 35:5 ...
ca- 2: ...Sz .5 n2 N2.3%»
is. b c: 33% m9 . 8. H“:55 N... :32: SS-
m+ a
29: .a :9 28. .a .5.” .3
31.5% 31.333 3153mm 31.3.5. fights behn avg”
3.23.2 .1332 .3582 .133: 285:: 9.23:8 8.
int: .3735. 3.72.; .133. II.- I: ..-
..i- .... ...... In N... 3.2—: :3
m+ _— L
3.2:.
6815 9- H.» «.3 3.5... l. .o Egg—52.353
:58 £2508 ..S .u .95 63.5 :5 .33 a .22 .83.. >355:-
4 , u . 5.25.
“+1 E On
325. E. .... ......u wan—m...— EE—zz ...... 32¢: a: E ...... 3.532 2355 E.

69

REFRENCES

7O

References

. Feilding JE. Frequency of health risk assessment activities at US. worksites.
American journal of Preventative Medicine 1989;5(2):73-81.

. Mc Ginnis. 1992 National survey of worksite health promotion activities
summary. American Journal of Health Promotion 1993;7z452-64.

. Frosch JW, Alterrnan T, Peterson R, Murphy LR. Worksite health promotion
programs in the U.S.: Factors associated with availability and Participation.
American journal of Health Promotion 1998;13(1):36-45.

. Stonecipher LJ, Hyner GC. Health practices before and after a work-site
health screening. Journal of Occupational Medicine 1993;35(3):297-306.

. Stange KC, Strogatz D, Schoenback VJ, et al. Demographic and health
characteristics of participants and nonparticipants in a work site health-
promotion program. Journal of Occupational Medicine 1991 ;33(4):474-478.

. Lewis RJ, Huebner WW, Yarborough Ill. Characteristics of participants and
nonparticipants in worksite health promotion. American Journal of Health
Promotion 1996;1 1 (2)299-106.

. Zavela KJ, Davis LG, Cottrell RR, Smith WE. Do only the healthy intend to
participate in worksite health promotion?. Health Education Quarterly
1988;15(3):259-267.

. Sorensen G, Stoddard A, Hunt MK, et al. The effects of a health promotion
health protection intervention on behavior change: the WellWorks study.
American Journal of Public Health 1998;88(11):1685~1690.

. Sloan RP, Gruman JC. Participation in workplace health promotion programs:
The contribution of health and organizational factors. Health Education
Quarterly 1988;1 5(3):269-288.

10.Wilson MG, Holman PB, Hammock A. A comprehensive review of the effects

of worksite health promotion on health-related outcomes. American Journal of
Health Promotion 1996;10(6):429-435.

11.Lynch WD, Gilfillan LA, Jannett C, McGloin J. Health risks and health

insurance claims costs. Journal of Occupational Medicine 1993;35(1):28-33.

71

12.Conrad P. Who comes to work-site wellness programs? A preliminary review.
Journal of Occupational Medicine 1987;29(4):317-320.

13.Bly JL, Jones RC, Richardson JE. Impact of worksite health promotion on
health care costs and utilization. JAMA 1986;256(23):3235-3240.

14.Blair SN, Smith M, Collingwood TR, et al. Health promotion for educators:
impact on absenteeism. Preventative Medicine 1986;15:166-175.

15.Mavis BE, Stacknik TJ, Gibson GA, et al. Issues related to participation in
worksite health promotion:A preliminary study. American Journal of Health
Promotion 1992;7(1 ):53-60.

16.Jeffery RW, Forster JL, Dunn BV, et al. Effects of work-site health promotion
on illness-related absenteeism. Journal of Occupational Medicine
1993;35(11):1142-1146.

17.Knight KK, Goetzel RZ, Fielding JE. et al. An evaluation of Duke University's
Live For Life health promotion program on changes in worker absenteeism.
Journal of Occupational Medicine 1994;36(5):533-536.

18.Lechner L, de Varies H, Adriaansen S, et al. Effects of an employee fitness
program on reduced absenteeism. Journal of Occupational and
Environmental Medicine 1997;39(9):827-831.

19.Wilbur CS. The Johnson & Johnson program. Preventative Medicine
1983;12:672-681.

20. Heaney CA, Goetzel RZ. A review of health-related outcomes of multi-
component worksite health promotion programs. American Journal of Health
Promotion 1997;1 1(4):290-307.

21.Pencak M. Workplace health promotion programs. Nursing Clinics of North
America 1991 ;26(1):233-240.

22.Spilman MA, Goetz A, Schults MS, et al: Effects of a corporate health
promotion program. Journal of Occupational Medicine 1986;28:285.

23.Brill PA, Kohl HW, Collingwood TR, et al: The relationship between
sociodemographic characteristics and recruitment, retention and health
improvements in a worksite health promotion program. American Journal of
Health Promotion 1991;5(3):215-221.

24.Goetzel RZ., Jacobson BH., Aldana SG., et al. Health care costs of worksite

health promotion participants and non-participants. Journal of Occupational
and Environmental Medicine 1998; 40 (4): 341-346.

72

25.Reardon J. The history and impact of worksite wellness. Nursing Economics
1998; 16 (3):117-121.

26.Naas, R. DuPont links wellness program to reduced absenteeism. Business
and Health 10 (9): 35—37.

27.Sharkey P., Bey JM. Designing an incentive based health promotion program.
AAOHN Journal 1998; 46 (3).

28.Jaffe G. Corporate carrots, sticks out health bill. Wall Street Journal, February
3, 1998.

29.Wilson MG., DeJoy DM., Jorgensen CM., et al. Health promotion program in
small worksites: results of anational survey. American Journal of Health
Promotion 1999; 13(6):358-365.

30.Emmons DM., Linnan LA., Shadel WG. et al. The Working Healthy Project: a
worksite health-promotion trial targeting physical activity, diet, and smoking.
Journal of Occupational and Environmental Medicine 1999; 41(7):545-555.

31.Matson D., Lee JW., Hopp JW. et al. The impact of including incentives and
competition in a workplace smoking cessation program on quit rates.
American Journal of Health Promotion 1998;13(2):105—111.

32.Pellier KR. A review and analysis of the health and cost-effective outcome
studies of comprehensive health promotion and disease prevention programs
at the worksite: 1994-1995. American Journal of Health Promotion
1996;10(5):380-8.

33.Baun WB., Bemacki EJ., et al. A preliminary investigation : effect of a
corporate fitness program on absenteeism and health care cost. Journal of
Occupational Medicine1986; 28 (1 ):18-22.

34.Brill PA., Kohl HW., Togers T., et al. The relatioinship between
sociodemographic characteristics and recruitment, retention, and health
improvements in a worksite health promotion program. American Journal of
Health Promotion 1991 ;5(3):21 5-221.

35.Goetzel RZ., Kahr TY., Aldana 86., et al. An evaluation of Duke university's
Live for Life health promotion program and its impact on employee health.
American Journal of Health Promotion 1996;10(5):340-5.

36. Holt MC., McCauley M., Paul D. Health impacts of AT&T's Total Life Concept

(TLC) program after fice years. American Journal of Health Promotion
1995;9(6):421-5.

73

37.Bertera RL. Behavioural risk factor and illness day changes with workplace
health promotion: two-year results. American Journal of Health Promotion
1993;7z365-72.

38.Stein AD., Karel T., Zuidema R. Carrots and sticks: impact of an
incenive/disincintive employee flexible credit benefit plan on health status and
medical costs. American Journal of Health Promotion 1999;13(5):260-267.

39. Fowles J., Gibbs JO., Pearce HG. Consumer cash incentives: can they work?
Business and Health 1987;4z22-24.

40.Jason LA., Jayaraj 8., Blitz CC., et al. Incentives and competition in a
worksite smoking cessation intervention. American Journal of Public Health
1990; 80(2);205-206.

41.Naas R. Health promotion programs yield long-terrn savings. Business and
Health 1993, 10(13);41-42.

42.Pelletier KR. Areview and analysis of the clinical and cost-effectiveness
studies of comprehensive health promotion and disease management
programs at eh worksite: 1995—1998 update. American Journal of Health
Promotion 1999, 13 (6);333-345.

43.Kingery PM., Ellsworth CG., Corbett 38., et al. High-cost analysis. Journal of
Occupational Medicine 1994, 36 (12):1341-7.

44.Spencer LS, Pratt 08., Hausman A. Using health benefits as an incentive to
change employee health risks: preliminary program results. Employee
Benefits Journal, June 1996 226-36.

45. Newman B. Penalties, incentives, and wellness programs after HIPAA.
Employee Benefits Journal. March 1999: 29-32.

46.Priester R. Are financial incentives for wellness fair? Employee Benefits
Journal. March 1992: 38-40.

47.Stein AD., Shakour SK., Zuidema R. Financial incentives, participation in
employer-sponsored health promotion, and changes in employee health and
productivity: Heathlus Health Quotient program. Submitted to the Journal of
Environmental and Occupational Medicine.

74

   

lllljllljg‘lljjjjjllljljil