This is to certify that the
thesis entitled

THE EFFECT OF COLOR CONTRAST OF TEXT ON THE
NOTICEABILITY AND LEGIBILITY OF PRESCRIPTION
DRUG WARNING LABELS

presented by

RAGHAV PRASHANT SUNDAR

has been accepted towards fulfillment
of the requirements for the

MS degree in PACKAGING

 

 

jor Professor’s” Signature

7/2/0‘i

Date

 

MSU is an Affirmative Action/Equal Opportunity Employer

---c-0-o----o-n-n-t-I-o---o-I-o--o—

 

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5108 K.IPrq/Aco&Pres/CIRCIDateDue.indd

THE EFFECT OF COLOR CONTRAST OF TEXT ON THE
LEGIBILITY AND NOTICEABILITY OF PRESCRIPTION
DRUG LABELS

By

Raghav Prashant Sundar

A THESIS

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

MASTER OF SCIENCE

Packaging

2009

ABSTRACT

THE EFFECT OF COLOR CONTRAST OF TEXT ON THE LEGIBILITY AND
NOTICEABILITY OF PRESCRIPTION DRUG LABELS .

By

Raghav Prashant Sundar

Prescription drug warning labels (PWLs), small, colorful, auxiliary stickers
applied by pharmacy personnel on prescription vials, are used to provide important, drug-
specific information about medications. At present, the required information and the
presentation of said labels is not standardized in the US, and they are not regulated by the
FDA (Ault 2007). This is despite the fact that, these labels contain information that is
critical to the safe and effective use of the products contained within.

Thirty-three subjects from two populations (18-29 and 50+) were characterized in
a variety of ways and tested to objectively examine the performance of PWLs in five
color contrasts. Tests of label performance included: eye tracking, recall and recognition
and legibility. Data were evaluated in various ways.

Statistical analysis was performed using mixed models. There was not enough
evidence to conclude an effect of color contrast on the noticeability of the PWL’s,
although findings indicated that subjects were more likely to notice and spend more time
on the large white pharmacy label as compared to the cap and the PWL. It was revealed
that subjects from the younger population were more likely to recognize the PWL’s they
had seen. It was also identified that certain categories of information (dosage, patient
information) were recalled more than the others. Legibility findings were consistent with

previous studies with the black on white color contrast being the easiest to read.

Copyright by
Raghav Prashant Sundar
2009

For my Madhumitha,
You are nothing short of my Everything

iv

Acknowledgements

“. .. and when you want something, all the universe conspires in helping you achieve it”
- Paulo Coelho in his 1988 book, The Alchemist

Despite its staggering achievements humanity still lacks the ability to contain the
universe on this sheet of paper. Nevertheless, I would surely like to bring to light the
efforts of several remarkable individuals without whose inputs this thesis would have
been impossible to write.

I would like to start off by thanking my advisor, Dr. Laura Bix for her support and
guidance throughout my time as a Masters student. It was only due to her supervision and
valuable advice that I was able to take up this research project, let alone write this thesis.
Dr.Bix is a fiend to her students and always encourages us to think “out of the box” and
come up with innovative solutions to our problems. It is only through her positive attitude
and approach to learning that most of us find our graduate level work fulfilling.

I would also like to thank the other members of my academic committee,
Dr.Rafael Auras and Dr.Theresa Mastin, who with their insights, have time and again
propelled my work to greater heights.

I would like to thank Mr. Tom Ball fi‘om the Oil Health Center who supplied us
with all the vials and labels that were used in this study. Mr.Ball was always available to
provide us with his inputs on our research from a pharmacy perspective. This was
instrumental in setting up many of the experimental methods in this study.

All my fellow students from Dr.Bix’s Packaging HUB (Human Factors, Universal
Design and Bio Mechanics) Research group were indispensable to the smooth working of

this project. The theme for this entire study came off a Plan B project by Voramanus

Viraporn. Without her systematic report and literature review, this study would not have
materialized.

My dear friends Chris Steckler and Kirit Rosario helped me design and build the
eye tracking chin rest and table. Through this, we were able to achieve a high degree of
accuracy using our instruments and also keep older consumers comfortable during
testing.

None of the mind-numbing statistics seen in this work would’ve been possible
without Nora Bello, our statistician from the Statistical Consulting Center at CANR. With
her suggestive and innovative approach, we were able to make sense of all the data that
were had collected.

I would also like to extend my heartfelt thanks to the Center for Food and
Pharmaceutical Packaging Research (CF PPR) for their financial support. Additionally,
their feedback to me during our periodic meetings was enlightening.

I wouldn’t have gotten far in the USA if it weren’t for my supportive and
understanding employers and co-workers at the Division of Engineering Computing
Services, College of Engineering. Working for them as a graduate assistant web
developer / graphic designer did a lot more than just pay the bills.

Even though they were 8000 miles away, my family and friends in India have
always stood by me and made a difference with their love and prayers. My Mother
Jashoda, my father Sundar and my sister Poorvaja have always stood by me in the
toughest of times. Their contribution to this work, though unseen, is monumental. Last,
but certainly not least, a big hug to everyone belonging to the Indian community here at

MSU for always making me feel at home, loved and pampered.

vi

Our research group is forever indebted to the good people of East Lansing who

spared their time and energy to appear as test subjects for all our studies.

vii

TABLE OF CONTENTS

List of Tables ........................................................................................................................ x
List of Figures ....................................................................................................................... xii
Chapter 1: Warnings and their Effectiveness ................................................................... 1
Warnings and Us .................................................................................................................... l
Hazards, Risks and Control .................................................................................................... 2
Models and Their Importance ................................................................................................ 3
Warnings Effectiveness Research .......................................................................................... 6
Warnings and the Law ........................................................................................................... 10
Warnings and Packaging ........................................................................................................ 12
Chapter 2: Prescription Drug Warning Labels (PWL’s) ................................................. 14
Patients and Medical Information .......................................................................................... 14
Auxiliary Labels and PWL’s ................................................................................................. 15
Studies on PWL Effectiveness ............................................................................................... 18
Chapter 3: Aging and the Warning Process ...................................................................... 21
America’s seniors and Medical Information .......................................................................... 21
Warnings and Aging — Physiological and Cognitive changes ............................................... 22
Physiological Variables .................................................................................................... 22
Comprehension ................................................................................................................. 24
Cognitive Factors .............................................................................................................. 25
Chapter 4: The Human Eye and Eye Tracking ................................................................ 31
The Human Eye ..................................................................................................................... 31
Eye Movements ..................................................................................................................... 32
Methods to study Eye Movements ......................................................................................... 34
Eye Tracking — Applications and Spheres of Interest ............................................................ 36
Closing Statements ................................................................................................................. 38
Chapter 5: Print Legibility .................................................................................................. 40
Introduction ............................................................................................................................ 40
Factors that affect legibility ................................................................................................... 41
Measurement Techniques ...................................................................................................... 42
Legibility Measurement in this study .................................................................................... 44
Chapter 6: Materials and Instrumentation ....................................................................... 45
Materials ................................................................................................................................ 45
Instrumentation ...................................................................................................................... 5 1
ASL 501 HeadMounted Optics Eye Tracker .................................................................... 51
Lockhart Legibility Instrument ......................................................................................... 53

viii

Chapter 7: Methods ............................................................................................................. 5 6

Working Hypothesis .............................................................................................................. 56
Test Procedure ....................................................................................................................... 56
Consent ............................................................................................................................. 56
Eye Tracking ..................................................................................................................... 56
Recall and Recognition Test ............................................................................................. 58
Collection of Subject Information .................................................................................... 58
Additional Testing ............................................................................................................ 59
Legibility Testing .............................................................................................................. 60
Compensation ................................................................................................................... 61
Data Processing ................................................................................................................. 61
Chapter 8: Results, Analysis & Discussion ........................................................................ 63
Data Encoding and Entry ....................................................................................................... 63
Subject Demographics ........................................................................................................... 63
Noticeability of PWL’s using Eye Tracking Data ................................................................. 67
The Probability of noticing a zone ................................................................................... 68
The Proportion of total time spent
per zone standardized by zone density ............................................................................. 69
The total number of hits on a zone ................................................................................... 78
Recall ..................................................................................................................................... 82
Recognition ............................................................................................................................ 85
Legibility Testing ................................................................................................................... 88
Chapter 9: Conclusions, Limitations and Future Research ............................................ 92
Noticeability and Attentive Behavior .................................................................................... 92
Recall of label and Vial Information ..................................................................................... 94
Recognition of Labels ............................................................................................................ 94
Legibility of PWL’s ............................................................................................................... 95
Other Salient Conclusions ...................................................................................................... 95
Limitations ............................................................................................................................. 95
Appendices ............................................................................................................................ 97
Appendix 1: Consent Form ................................................................................................. i ...97
Appendix 2: Recruitment Advertisement For Test Population .............................................. 100
Appendix 3: Data Collection Sheet ........................................................................................ 101
Appendix 4: Raw Data for Recall Test .................................................................................. 103
Appendix 5: Raw Data for Recognition Test ......................................................................... 105
Appendix 6: Average Legibility Indices ................................................................................ 107
Bibliography ......................................................................................................................... 108

ix

LIST OF TABLES

1. Person Variables that affect Warning effectiveness .................................................. 24
2. Warning message readability scores .......................................................................... 51
3. Example legibility test order ...................................................................................... 61
4. Subject Gender and Ethnicity (Data for Figure 14) ................................................... 64
5. Health Literacy and Red/Green Color Blindness (Data for figure 15) ...................... 65
6. Medication Consumption (Data for figure 16) .......................................................... 66
7. Frequency Distribution Table of Noticeability Binary Variable ............................... 69
8. Predicted Probability of Noticing a Zone (Data for Figure 17) ................................ 70
9. Zone density values for all label zones ...................................................................... 72

10. Bonferroni adjusted pair wise comparisons across zones for four levels of total

time spent per zone .................................................................................................... 75
11. Cumulative frequency table for the number of hits on each label zone ..................... 78
12. Expected number of hits on a zone (Data for figure 19) ............................................ 79
13. Examples of Recalled Information in the 12 categories ............................................ 83

14. Predicted Probability of Recall of Information in a particular category
(Data for Figure 21) ................................................................................................... 84

15. Number of PWL's correctly recognized (Data for Figure 23) ................................... 86

16. Estimated Least Square Means and Standard Error for Recognition Scores /5 and
/ 10 .............................................................................................................................. 88

17. Least Square Means for the averaged triplicates of Legibility Index (Data for
figure 23) .................................................................................................................... 90

18. Information recalled by subjects of the control population in 12 categories
in binary form (recalled/not recalled) ........................................................................ 103

19. Information recalled by subjects of the test population in 12 categories in binary
form (recalled/not recalled) ........................................................................................ 104

20. Recognition Data for subjects of the control population for all 10 labels
in binary form (circled/did not circle) ........................................................................ 105

21. Recognition Data for subjects of the test population for all 10 labels in binary
form (circled/did not circle) ....................................................................................... 106

22. Average of Legibility Index triplicates for all subjects. Subjects 1-15 belong to the
control population. Subjects 16 to 33 belong to the test population. ......................... 107

xi

LIST OF FIGURES

1. Consolidated - Human Information Processing (C-l-IIP) Model ............................... 6
2. A photograph of a vial used in the study. A common 10 dram prescription vial of

the l Clic type with two types of labels; A PWL and a White Pharmacy Label ....... 16
3. Examples of actual PWL's used in local pharmacies ................................................. l7
4. A schematic diagram of the structure of the human eye ............................................ 32
5. The Dow Corning Opthalrnics, Near Point Visual Acuity Card ................................ 45
6. Pseudo Iso Chromatic plates booklet for testing Red/Green Color blindness ........... 47
7. Photograph of a vial used in the study. A common 10 dram prescription vial of the

l Clic type with two types of labels; A PWL and a White Pharmacy label .............. 48
8. PWL's used in the study ............................................................................................. 49
9. An example of a White Pharmacy Label from the Olin Health Center that was

pasted on all vials ....................................................................................................... 50
10. ASL 501 Head Mounted Optics Eye Tracker ............................................................ 52

11. Custom Designed Eye Tracking Table with Glass pane and Chin rest arrangement 53

12. Lockhart Legibility Instrument .................................................................................. 54
13. Recognition Sheet ...................................................................................................... 58
14. Subject Gender and Ethnicity .................................................................................... 64
15. Health Literacy and Red/Green Color Blindness ....................................................... 65
16. Medication Consumption ........................................................................................... 67
17. Predicted probability of noticing a zone :I: estimated standard error. Letters (a, b,

c) indicate statistically significant differences (p < 0.01) .......................................... 71
18. Scatter Plots and Trend Lines relating Total Time spent on vials to the Pr0portion

of total time spent on each zone standardized by zone density ................................. 74
19. Estimated number of hits on each zone i estimated standard error. Letters (a, b, c)

indicate statistically significant differences (p < 0.05) .............................................. 80
20. Average Number of hits according to Gender ........................................................... 81

xii

21 . Average Number of hits according to population ...................................................... 82

22. Predicted probability of recall d: standard error. Letters (a, b 0) indicate
statistically significant differences (p < 0.05) ............................................................ 85

23. Number of PWL's correctly recognized ..................................................................... 86
24. Estimated legibility index and 95 % confidence interval for the five PWL color

contrast combinations. Letters (a,b,c) indicate statistically significant differences)
(p < 0.01) .................................................................................................................... 91

Images in this Thesis are presented in Color

xiii

Chapter 1: Warnings and their Effectiveness

Warnings and us

A closer look at our environment reveals that warnings are everywhere. We find
them on bottles of alcoholic beverages, on hair dryers and on car seats for children. All
around, manufacturers of equipment warn us of potential misuses and hazards. Similarly,
manufacturers of drugs warn of possible side effects and adverse events. Warnings are
diverse; ranging from those that caution about slippery roads to those that inform about
the health risks associated with smoking. There has been an increase in warning research

in the recent past (DeJoy and Laughery 1999)

This may be linked to the increasing costs of accidents and healthcare in the US
and the escalating importance of warnings in litigation (DeJoy and Laughery 1999). It is
also appropriate to mention here that government organizations like the Occupational
Safety and Health Association (OSHA), the Environmental Protection Agency (EPA) and
the Food and Drug Administration (FDA) have fiieled research in warnings in their
respective sectors (DeJoy and Laughery 1999). Some see warnings as a means to transfer
responsibility of safety to the users (DeJoy and Laughery 1999). That said, warnings are
an important means of informing people that hazards exist and actions are needed to
counter them. They are an important part of the people’s ‘right to know’. (Laughery and

Hammond 1999)

The sum total of all warnings can be seen as serving a set of three purposes.
Firstly, warnings are used to prevent dangerous practices like smoking in gas stations or

exposure to sunlight while taking certain drugs. The second purpose is to cause

behavioral changes. A warning educating researchers to wear protective gloves is one
such example. The third type of warning prompts us to make informed decisions.
Required cigarette and alcoholic beverage warnings represent examples of this type.

Warnings can also be classified based on the primary sensory route that they use
to warn. Visual warnings can be found on the labels of packages or as visual cues
(signboards, strobe lights) in the environment. Warnings may also be auditory; examples
of auditory warnings include warning tones, alarms and messages in machines. Olfactory
warnings may take the form of chemicals added to odorless gases to help detect leaks.
Tactile warnings include textured surfaces on curbs that warn visually impaired people to
move away.

Before analyzing warning effectiveness, it is important that the concepts of

hazards and their control are understood.

Hazards, Risks and Control

A hazard is defined as “a set of circumstances that may result in injury, illness or
property damage (Laughery and Hammond 1999) " Risk is the likelihood that a certain
adverse event will occur. Following the above two terms, danger is often expressed as the
product of a hazard and the likelihood that it will cause damage (Laughery and Hammond
1999). In the context of hazard control, (3) warnings are referred to by some authors as
the third line of defense, the first two being better (1) design and (2) guarding.
Elaborating, when a different design or practice affords a greater amount of prevention of
damage to life and property, it is always preferred over others. Guarding against the

hazard is the next line of defense. For instance, heavy machines always have metal plates

covering high risk fixtures like cylinder nips and gear trains. Warning about the hazards
is the third and final line of defense. (laughery and Hammond 1999)

After assessing the hazards that may be present in a system (even after effective
design and guarding), the likelihood of adverse events is evaluated to determine if the
risk is high enough to justify the incorporation of a warning into the system. However,
this is done only after thoroughly analyzing the source (the hazard itself), possible
channels (visual, auditory, etc) and the receiver (workers, maintenance staff, etc).
Previous research also recommends testing of the warning before incorporating it into the
system (Laughery and Hammond 1999).

It has been indicated in the literature that properly documented warning and
hazard prevention research is scarce because of difficulties in conducting these studies
(Wogalter, DeJoy et a1. 1999). For instance, it isn’t appropriate to expose test subjects to
real hazards in their actual setting. In the recent past, studies on warnings and their
effectiveness have been firelled by the development of better study techniques, methods
and models (W ogalter, DeJoy et a1. 1999).

Warnings are often, if not always, approached with a theoretical framework. The
following sections outline a few of the widely used theoretical models used in warning

research.

Models and their importance

There are many theoretical frameworks or models that allow researchers to
investigate a warning, evaluate its effectiveness or make predictions about the same.

These models, though not complete or fool proof in any way, do serve to provide some

insight into the effectiveness of warnings before they are deployed. The following
paragraphs describe some of these models.

Communication Models: Warnings have always been looked upon as forms of
communication. Consequently, the first theoretical models that were used to assess
warning effectiveness came from communications theory. In general, modern
communication theory has largely evolved from the publication titled, “A Mathematical
Theory of Communication” by Shannon, C (Shannon 1948). A communication model
analyzes the three main components of the warning process namely, the source, the
channel and the receiver (Wogalter, DeJoy et a1. 1999). In the case of a package, the
source may be the manufacturer of the product. The channel is the warning itself and the
receiver is the consumer.

Most forms of the communications model also take into account the effects of
noise and feedback. For example, a communications model would be able to take into
account extraneous distractions (noise) and also consider alterations in subject’s
responses based on external cues (feedback). Examples of noise include cluttered visual
fields for print warnings and road signs. Examples of feedback include subject responses
to auditory beeps/alanns. Certain fields of study that derive from general communication
theory like ‘persuasive communications’ are heavily focused on beliefs and attitudes of
consumers who are presented with warnings. An interesting set of studies on warnings on
alcoholic beverage bottles and cigarette packs have been carried out using this model
(Andrews 1995).

Information Models: Information models, on the other hand, focus on the stages

that the information contained in the warnings passes through while being transmitted

from the warning to the user. These models take into account facets of information
processing like attention, comprehension, attitudes and beliefs, motivation and behavioral
change (Laughery and Hammond 1999). Information processing theories often mention
that the information passes through several stages in serial fashion and that failure at any
stage will lead to failure of the warning process.
A commonly used theory of information processing indicates that the following
four steps must be completed for the warning to be effective. (Lehto and Miller 1988;
DeJoy 1991; Wogalter and Laughery 1996; Laughery and Hammond 1999; Rogers,
Rousseau et al. 1999)
o The information must be noticed
o It must be encoded into memory. For this to happen the message must be
gathered through perceptual systems, generally vision; i.e., the message must be
legible
o The encoded message must then be comprehended
o It must, finally, move the reader to action
C-HIP: Certain researchers stress the importance of consolidating concepts fi'om
both information processing and communication theories, creating a common framework
that takes into consideration both models. An example of this kind of model is the
Consolidated Communication -— Human Information Processing (C-I-IIP) model
(Wogalter, DeJoy et a1. 1999). The C-HIP combines concepts of both the information

processing and the communications models as shown below.

Source (Manufacturer)

Channel
(A Visual, Audible or Olfactory Warning)

Receiver
(A Consumer, Worker etc)

Noticea biIity' 8: Attention
Memory 8: Co'mprehension
Attitudes'8r Beliefs
Motivation
Behavior

Figure 1 - Consolidated - Human Information Processing (C-HIP) Model

They incorporate information processing into the receiver end of the
communications chain. This model is used widely to analyze warnings and organize
literature regarding warning effectiveness research. It is noteworthy that entire books

have been written on warnings that organize content based on the above C-HIP model.

 

Warning Effectiveness Research

The following paragraphs will examine some of the variables that are analyzed in
warning studies and will also provide an overview about some of the methods that are
used to assess the effectiveness of warnings. There are essentially two kinds of warning

studies. There are those studies that analyze the warning as it passes through the various

stages before it can influence behavior (Type 1) and there are those studies that analyze
behavior and behavioral compliance itself (Type 2) (Young and Lovoll 1999).

Type 1: As mentioned before, information models assume that a warning must
pass through various stages (attention, sensory routes, comprehension and compliance);
these stages are the focus of the first type of study.

The attention phase is analyzed either directly or indirectly. Eye movements, are
used as an objective measure of attention and parameters such as fixation time and
reaction time have been used to quantify the noticeability of warnings. Indirect measures
like recall are also used to measure noticeability. (Fischer, Richards et al. 1989;
Krugman, Fox et al. 1999; Crespo, Cabestrero et a1. 2007). Other studies based on
attention have focused on additions and enhancements to warnings like symbols (Cahill
1976; Swindell 1999), color contrast (Sumner 1932), borders (Ells, Dewar et al. 1980)
etc.

After being noticed, the warning must be legible enough to be encoded into the
user’s memory. This must occur through the different sensory routes. A visual warning
must be seen and read; an auditory warning must be heard and understood. In addition to
this, the consumer must have the cognitive resources to successfitlly encode information
that is present in a warning. In other words, a warning may be legible, but if the user’s
mind is already overloaded with other tasks and information, encoding may not occur
efficiently.

Comprehension is the next stage in the information processing hierarchy. Studies
involving comprehension analyze parameters such as the level of message “readability”

(Wong 1999) and the literacy levels of subjects in order to evaluate the likelihood that a

person will understand the message (Karwowski, Yates et a1. 1991; Davis, Wolf et al.
2006)

Other important concepts that are studied by warnings researchers include
attitude, perception and behavior. Research in risk perception aims to determine the
attitudes of people as they approach warnings and hazards in general. There have been a
variety of studies that involve questionnaires that employ Likert scales with responses
involving subjective measures, (F ischhoff 1977; Baber and Wankling 1992). The same
questions can also be examined with more objective measures. For example, studies
sometimes make use of existing real-world information sets like databases of accident
information to quantify risk perception among drivers.

Tm 2: As mentioned earlier, the ultimate goal of a warning is to bring about a
behavioral change in its intended audience. I

“If a warning is eflective at the behavioral stage, the warning is probably
adequate at the earlier stages. Indeed, behavioral data are so important that if only one
measure of warning effectiveness can be obtained, a compliance test is the best one to do.
(Wogalter and Dingus I 999)”

These studies may be divided into those that track actual behavioral compliance
and those that analyze behavioral intentions alone. There are several problems that
prevent direct measuring of behavioral compliance. Some of them are mentioned below

0 It is not ethical to expose participants to real hazards. Such studies are also
difficult to set up when the real hazards are unpredictable (natural disasters).
0 They are often time consuming and difficult to manage (Wogalter and Dingus

1 999).

Studies that examine direct measures of behavioral compliance require very
careful planning. Incidental exposure may be critical (tell the participant that the
study concerns something else).

Furthermore, there must be careful screening of participants. Researchers must
make sure that studies are conducted on subjects that are likely to benefit from the
incorporation of an effective warning, i.e. the test population must be carefully
selected and monitored.

Small nuances in the experimental design will affect the outcome of the study
and, therefore careful planning of the experimental design is required (W ogalter
and Dingus 1999).

During this literature survey, four broad categories of behavioral compliance

experiments were identified. They are described below with an example each.

Self reports of compliance generated through questionnaires and interviews are
sometimes conducted to examine the effectiveness of warnings. The difference
between these and studies that measure behavioral intentions, is that these
questionnaires are administered “ex postfacto” (W ogalter and Dingus 1999) , i.e.
the questionnaire is given to the subjects after the actual response has taken place.
Studies in warning effectiveness sometimes involve actual Observations of
behavioral compliance Examples include studies in chemical laboratories where
warning effectiveness and compliance behaviors of researchers has been analyzed
based on actual observations of their activities (W ogalter, Godfrey et a1. 1987).
Other studies that measure behavioral compliance involve innovative methods of

Physical Tracing. Studies of this type involve some kind of measure that is used

to detect whether the participant paid attention to the warning and complied with

its message. One such study assessed the effectiveness of a warning advising

usage of gloves while spraying cleaning solution. The gloves were inspected to
see if they had been worn while spraying by observing physical damage to the

fingertips of the glove from the spray nozzle (Hunn and Dingus 1992);

o Epidemological compliance studies concentrate on warning effectiveness across
large populations. Studies that have evaluated the effectiveness of driver training
programs in preventing accidents across populations are examples of this type
(Edwards and Ellis 1976).

Studies on behavioral intentions, as opposed to behavioral compliance, are more
widespread. Research indicates a strong link between intentions and compliance, i.e.
there is a link between what people intend to do and what they actually do (Ajzen and
F ishbein 1977). Behavioral intentions are often assessed using Likert scales and
questionnaire responses. The above summary of methods and experiments is by no means

exhaustive, but it does put things in perspective.

Warnings and the Law

As mentioned before, it is likely that recent warning research is motivated by
widespread lawsuits and increasing settlements related to warnings, their absence or
ineffectiveness. The concept of US. product liability says that a manufacturer who
designs and sells a product must warn users of potential problems that arise during its
foreseeable use and instruct them on its safe usage (Madden 1999). The duty of a
manufacturer to warn about a product is explained under two principles: Negligence and

Strict Liability.

10

Negligence: The principle of negligence says that when there is any foreseeable
harm that may occur during the use of a product, the failure to warn is termed as
negligence and is grounds for a lawsuit (Madden 1999).

Strict Liability: The principle of strict liability in tort states that a manufacturer
has the duty to warn when a normal person cannot perceive the dangers associated with a
product. This is to say that when the dangers of a product go beyond what would
normally be foreseen by a user, the manufacturer has strict liability in a tort litigation.
There also exists a rule that states that a manufacturer need not warn at all when the risks
and hazards are obvious. For instance a knife need not contain a warning about a sharp
edge (Madden 1999).

A full analysis and review of the various laws and practices related to warnings is
beyond the scope of this chapter. A limited list of exemplar issues in lawsuits involving
warnings follow.

0 Obviousness of risks and hazards play an important role in litigation. Case studies
indicate that lawsuits initiated by users of BB guns, darts and slingshots have been
rejected because the dangers of these products were obvious (Madden 1999).

0 An automobile manufacturer was once given a favourable judgement when a
person accused them of not providing a warning. The person had tried to prime a
carburetor with gasoline poured fiom a quart jar (Madden 1999). The
manufacturer was able to establish that the act was careless in itself and the user
would not have heeded the warning anyway.

0 There are cases where the charges are transferred to a third party. For instance, in

hearings that involve prescription drugs, it is common to transfer the liability from

11

the manufacturer to the physician who serves as a “learned intermediary.”
(Madden 1999).

0 Several court cases have shown that a manufacturer must be able to foresee
possible alternate uses of the product and warn against the same (United States
Court of Appeals 1974). One such example involved a housewife who stood on a
chair when the Chair’s backrest failed. The manufacturer was penalized for the
same (Madden 1999).

0 Manufacturers are expected to warn (FDA 2004) about ingredients in their
products that cause allergy in substantial proportions of populations, e. g. Peanuts.

0 There have been cases in which “For professional use only” warnings were not
heeded by household users (Madden 1999). There was once a case where a
professional hair care preparation was not used properly by a household user and
this resulted in injury. The verdict, however, was in favor of the manufacturer.
The hazards of the product would have been obvious to professional beauticians.

0 There have been disputes about warning effectiveness with reference to the design
and layout of the warning itself. A manufacturer of fumiture polish was found
guilty for printing the warning on its package in the same color and typeface as
the rest of the text. This resulted in the subsequent death of an infant who injested
the product. The warning in this case, was not considered adequately noticeable

and did not effectively convey the gravity of the hazard (Madden 1999).

Warnings in Packaging

Several authors mention that communication is one of the primary functions of

any package, along with protection and containment (Paine and Institute of Packaging.

12

1963; Soroka and Institute of Packaging Professionals. 1995). Paine (1963) mentions in
his writing,

“It is obvious that retail packages must communicate, for not only do they have to
identify the contents, but in many instances they have to sell them as well (Paine and
Institute of Packaging. 1 963) ”

With risks comes the need and requirement to warn. Food packages may contain
warnings about allergy causing ingredients, storage conditions, contamination indicators
and tamper evidence. Medical packaging often contains warnings about side effects, risks
and possible drug interactions. Warnings on packages may take many different forms,
ranging from simple labels and print warnings to information inserts that may be several
hundred words long. In the context of logistics and distribution, packages contain

warnings about storage and handling.

13

Chapter 2: Prescription drug warning labels (PWL’s)

Patients and medical information:

It is estimated that around 52% of all Americans take at least one prescription
drug every week. This figure rises to 82% when we add non-prescription drugs into the
equation. Older age groups have a higher per capita consumption of prescription drugs.
The Slone survey of drug usage in the American population for 2006 says that 17-19% of
the people in. the age group of 65+ take at least ten prescription drugs in a week
(Kaufman, Kelly et a1. 2006). As a result, drug regimens in older populations tend to be
more complex than their younger counterparts. Thus, there is great concern about the
management of these therapy regimens in older populations (Gryfebecker, Segal et a1.
1989)

Further, as healthcare shifts from the inpatient setting to the outpatient setting,
furnishing timely information and achieving patient compliance becomes more difficult
(Davis, Wolf et a1. 2006). Studies have shown that a significant proportion of patients do
not comply with instructions that are given about prescription drugs (Rehder, Mccoy et
a1. 1980; Boyd, Boyd et a1. 2006). It has been estimated that nearly 1.5 million
medication errors occur annually in the USA, and that a majority of them are in the
outpatient setting, where it is up to the patient to use the information provided (Bates
2007) to him or her about the drug. It is also indicated that 33% of these errors cite
packaging and labeling as the primary cause (Berman 2004). It can also be gleaned from

the literature that medication errors, in general, generate costs of almost 13 million

14

dollars annually (Aspden and Institute of Medicine (U .8). Committee on Identifying and

Preventing Medication Errors. 2007).

Auxiliary labels and PWL ’s:

Pharmacists have often employed verbal counseling to inform patients about
drugs, their side effects and dosage instructions. There are studies to suggest that this type
of verbal instruction increases patient compliance (Ascione and Shimp 1984; Johnston,
Clarke et a1. 1986). However, there are disadvantages to verbal counseling. Issues with
memory and patient comprehension can result in error, even when verbal instructions are
provided (Gryfebecker, Segal et a1. 1989). Thus, pharmacists depend on printed
information to supplement verbal counseling. These may take the forms of Patient
Information Leaflets (PIL’s) and patient packaging inserts (PPI’s) as well as ‘auxiliary’
labels that are affixed on prescription vials. Sometimes these auxiliary labels exist as the
small colorful, prescription warning labels (PWL’s).

In packaging as a whole, labels assume great importance from an information
perspective as they always stay with the packages and provide timely information about
the product (Miller 1978). Yet, studies have shown that the white pharmacy labels (see
Figure 2) attached to the vials do not provide adequate information about the drug when
used alone. This is one reason why the auxiliary label is important (Mazzullo, Lasagna et
al. 1974). Authors have mentioned that the prescription warning labels (PWLs), aim to
convey information about appropriate medicine use, potential drug interactions and side
effects in the simplest way at the least possible cost (Brown, Solovitz et al. 1988).

Prescription drug warning labels (PWL’s) contain warning messages (see Figure

2) such as, “Do not consume alcohol while taking this medication” or information about

15

routes of administration, viz. “For external use only." At present, the required
information for prescription drug warning labels and the presentation of said labels is not
standardized in the USA, and they are not regulated by the FDA (Ault 2007). This is
despite the fact that, in many cases, these labels contain information that is critical to the
safe and effective use of the products contained within. Furthermore, studies have
revealed that many consumers do not heed these warnings (Franklin 2005; Davis, Wolf et

a1. 2006).

  
 

White Pharmacy Label

Prescription Warning Label

Figure 2- A photograph of a vial used in the study. A common 10 dram prescription
vial of the l Clic type with two types of labels; PWL's and White Pharmacy Labels.

DO NOT TAKE THIS DRUG
IF YOU BECOME

PREGNANT
“ ron menu“. use
A ONLY.-.
5 roan-re I
Wear

Figure 3 — Examples of actual PWL's used in local pharmacies

OWN

It is crucial that these labels and other inserts (for e.g. PPI’s) are effective and
warn patients at the time of medicine consumption (Kalsher, Wogalter et al. 1996).
Pharmaceutical supply catalogues reveal that label printers and suppliers group labels
into the following categories based on information conveyed and color.
0 Administration Route
0 Drug interactions
0 Dietary information
0 Warning information
o Refill information
o Dosage information
0 Storage instructions
There have been several studies on these crucial warning labels and their

effectiveness. The following section provides an overview of some of them.

Studies on PWL effectiveness:

Over the years, there have been studies that have looked at prescription warning
labels from several aspects ranging from reading difficulty (Davis, Wolf et al. 2006) to
symbol presence or absence (W iederholt, Kotzan et al. 1983).

A study conducted as early as 1983 by Wiederholt et 31 (W iederholt, Kotzan et a1.
1983) looked at the effectiveness of these labels using a tachistoscope. A tachistoscope is
a mechanical device that was used to display images in studies that evaluated attentive
behaviors. In this study, the tachistoscope was used to display images of prescription
vials with PWLs pasted on them. The experiment measured recall through multiple-
choice questionnaires. The study looked at the relative effectiveness of color and
black/white labels and the presence and absence of a graphic symbol on the recall of
warning information. An interesting experimental condition in this study was the
incorporation of the warning text on the pharmacy label itself, i.e. the non-usage of a
warning label. Researchers concluded that there were significant increases in recall when
the PWL was used, but also elaborated that the study did not conclusively evaluate the
effects of color and symbols.

In the same year Wilson and Hogan conducted an analysis of PWL text and
concluded that 11.2 percent of the 110 words analyzed in this study were above the 86‘
grade reading level. They also suggested the use of a ‘word familiarity index’ to test the
readability of these labels as opposed to using the traditional readability tests. (Wilson
and Hogan 1983)

Ferguson et a1 (1987) conducted a study on warning labels that looked at the

various levels of severity in warning label text. Subjects were questioned on their

18

perception of different levels of strength/ severity of warning label text. Authors conclude
that patients could be made to comply with therapeutic regimens by increasing the
severity of warnings. Subjects also perceived the medium severity warning to be the one
most effective (Ferguson, Discenza et al. 1987).

Other early studies by Brown et al. (1988), looked at the effect of these labels on
the memory of the patient for drug related information. It was concluded that PWLs were
an effective means of educating patients about drug related hazards and were very
effective on a short term basis. They also discussed the phenomenon of generalization of
medical information that occurs with patients, i.e. a patient being told to avoid aged
cheeses may begin to avoid all kinds of cheese (Brown, Solovitz et al. 1988).

It has been mentioned before in this chapter that medication use is higher among
the elderly. Early studies in 1989 by Gryfe-Becker et al, investigated the knowledge that
elderly patients would gain fiorn auxiliary labels. The study specifically used auxiliary
labels as supplements to verbal counseling and compared the rates of drug information
recall among patients who received only verbal counseling while being given bags With
drug vials as opposed to those who were also provided with auxiliary labels on the vials.
The interviews to measure recall information were conducted 10—15 days after the
medication was picked up by the patients from the pharmacies. It was concluded that
those subjects that were given auxiliary labels were able to recall drug information at the
higher rate than those who were not (Gryfebecker, Segal et al. 1989).

A study published in the Journal of Pharmacy Technology in 1991 focused on
pharmacists and their patterns of PWL use in the state of North Carolina. The results of

this study showed high levels of non-usage of PWL’s among several pharmacies for

19

many drugs. The usage of PWL’s differed between pharmacists employed in chain stores,
pharmacists using label software, and those who worked in independent stores
(Dombrowski, Schuna et al. 1991).

A recent study on designs and pictorials in PWL’s analyzed tagged and fold-out
labels as alternatives to the existing designs. For instance, fold out labels required that
subjects unfold them from the side of the container and then down. This label had a
surface area that was 40 % greater than that of the conventional label. Subjects were
asked to rate designs after being exposed to them. The study also looked at the presence
of pictorials on these labels and their effectiveness. The study reported a higher
preference for the new designs and labels with pictorials among all age groups (Kalsher,
Wogalter et al. 1996).

A widely cited and quoted study on the reading difficulty of PWL’s used the
REALM literacy test to determine health literacy levels among subjects and LEXILE
scores to measure the ‘reading level’ of the label text. The study attempted to correlate
readability levels as measured using the LEXILE text and patient understanding of the
warnings themselves (Davis, Wolf et al. 2006). It was concluded in this study that those
subjects who were reading at a low literacy level were less likely to understand and
interpret the meanings of the PWL text correctly. Around one-third of the total number of
subjects tested were below a sixth—grade reading level.

In subsequent sections, other important aspects of this study such as eye tracking,

aging and health information shall be analyzed.

20

Chapter 3: Aging and the Warning Process

America ’s seniors and Medical Information:

In a 2006 report written for the National Institutes of a Health (NIH), it is
mentioned that the population of the United States of America over 65 years of age is
expected to double in the next 25 years. It is projected that 20% of the US population will
be aged 65 years and over by 2030 as compared to 12 % in 2003 (NIH 2006). A deeper
exploration of related literature suggests that there has been a concurrent increase in the
number of citizens seeking medical care. A 2008 article published in USA Today shows
that the number of medical visits increased almost 26% between 1996 and 2006. The
article goes on to mention that the aging population may contribute to increased medical
visits (Hellmich 2008). Data released by the US Census bureau suggests that the baby
boomer generation will further tax American healthcare as they age. Quoting fi'om the US
Census bureau source,

“Just as this generation had an impact on the educational system and the labor
market, this large cohort will strain services and programs required by an elderly
population ..... Between 1990 and 2020, the population age 65 to 74 would grow 74
percent under middle series projections, while the population under age 65 would
increase only 24 percent. (Bureau 1 99 7)”

Other sources indicate that the proportion of seniors living alone is also increasing
(Report 2003). As stressed previously in this document, the approach to healthcare
delivery is shifting from the use of inpatient to outpatient settings. All these factors point

towards the increased need for efficient design of health information and its delivery.

21

Warnings and Aging — Physiological and Cognitive changes:

Dejoy asserts that there has not been significant wamings-related research in
which age is studied as a direct variable of interest (DeJoy 1999). This section will
attempt to summarize some of the physiological and cognitive factors that vary with
aging and how they affect the process of warning. This section draws information from
and owes its structure to the book chapter, “Maximizing the effectiveness of the warning
process: Understanding the variables that interact with age” by Gabriel. K. Rousseau and
Wendy. A. Rogers (Rogers, Rousseau et al. 1999).

Rousseau and Rogers mention that there are ‘person’ variables and ‘warning’
variables that play a part in warning effectiveness. ‘Person’ variables include those that
depend on the receiver, age being one of them. Age in turn affects other person variables
such as vision, cognitive capability and memory. Warning variables include
characteristics of the warning such as color, typeface and type size (Rogers, Rousseau et
al. 1999).

Several studies reveal that there are important ‘person’ variables that change with
age. Some of them are mentioned in Table l. A short overview of each factor in relation
to warnings and age is mentioned in the following sections.

Physiological Variables

Vision: Studies have shown that several facets of vision decline with age. Existing
literature has documented declines in visual acuity, contrast sensitivity, acuity for
peripheral targets and color discrimination (Kline and Scialfa 1997). This reduction in
visual capability suggests that older individuals will have trouble noticing and

comprehending warnings. The basic visual acuity of subjects also declines sharply with

22

age (Bix 1998). Literature reviews of vision in senior citizens assert the need for
increased contrast in warnings in products and systems that are intended for older
audiences (Chamess and Bosman 1992). Studies looking at older adults have constantly
stressed the need for senior-fiiendly features in the warnings, such as matte surfaces to
reduce glare and use of fonts that are over l4-points in size (Backinger and Kingsley
1993).

There is some literature that attempts to quantify the changes in color perception
as human beings age. A guidance document released by the Department of Health and
Human Services titled “Recommendations for developing user instruction manuals for
medical devices for home health care”, mentions that the color blue is very difficult for
older adults to see and should be avoided (Backinger and Kingsley 1993).

Gender: It is acknowledged that gender does play an important role in influencing
an individual’s willingness to look for warnings. Studies by different researchers have
established that females have a greater tendency to look for warnings than males when a
product is perceived to be hazardous (Godfrey, Allender et al. 1983; Goldhaber and
Deturck 1989). Rosseau and Rogers mention that there is yet no literature available to
predict how these gender differences behave with age (Rogers, Rousseau et al. 1999).
However, interesting warning related phenomena like the ‘boomerang effect’ explained

later in this chapter have been found to depend on a combination of age and sex.

23

Table l - Person Variables that effect Warning effectiveness

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Physiological Variables
Vision Gender
Comprehension Capability
Reading Comprehension Symbol Comprehension
Cognitive Factors
. . . . Costs/Benefits of . .
Familiarity Hazard Perception Compliance Risk Taking Style
Attention Working Memory Prospective Memory
Comprehension

Reading Comprehension: Studies comparing college-going students to middle-

aged adults have assessed the “willingness to read warnings” at different ages (Silver and

Wogalter 1990). These studies have reported that the general willingness to read

warnings as reported by subjects does not vary greatly in between these two age groups,

but there are several significant declines in the ability of subjects to process information

gleaned from text. There are studies on general prose comprehension that report that

older adults are unable to make technical inferences and tie in information fi'om different

parts of the same text (Kemper 1992).

There are also important age related declines in functional health literacy.

(Williams, Parker et al. 1995). Tests have shown that between 48 to 80 % of adults over

24

 

60 years of age have inadequate fimctional health literacy (Williams, Parker et al. 1995).
In a study conducted in 1993 by Hill, findings indicated that 18% of a sample of 83
community dwelling, educated, elderly adults “were reading below a twelfth grade level
on the reading subtest of the Adult Basic Learning Examination (Level 3) ” (Hill 1993).
Other sources indicate that age related declines in reading comprehension are not very
significant for highly educated older adults (Meyer, Young et al. 1989; Harris, Rogers et
al. 1998).

Symbol Comprehension: Symbols, graphical and iconic representations of
concepts, actions or instructions are important in warning design and effectiveness. Some
authors choose to distinguish between symbol (icons) and pictorial (actual images)
comprehension, but most research studies do not distinguish between the two (DeJoy
1991). As mentioned before, symbols are very important in conveying warning
information when consumers are from varied linguistic backgrounds. In general, studies
have shown that the use of a symbol-text combination in a warning always yields better
compliance behavior among consumers than the usage of a symbol or textual warning
alone (Otsubo 1988; Janynes and Boles 1990; Young and Wogalter 1990). In terms of the
differences in symbol comprehension across different age groups, there is some evidence
to suggest that older adults do not recognize and understand symbols as well as younger
adults, but this is still very vaguely understood (Rogers, Rousseau et al. 1999) .

Cognitive Factors

Familiarity: Familiarity is discussed by Rosseau and Rogers under three subtopics

(Rogers, Rousseau et al. 1999)

0 General Experience

25

0 Schemata
0 Technical Knowledge

General Experience, as the term suggests, deals in past experiences with the
product or system. It is generally understood that when people have used a product or
system for a long period of time with no adverse events, they tend to overlook warning
information (Godfrey and Laughery 1984). Studies among different age groups (under
30, 30-50 and 50+) have revealed that most individuals wouldn’t look for warning
information if they had used the product before without harm. Subjects have sometimes
reported that they were more likely to look for and comply with warning information in
electrical products than non-electrical products (Wright, Creighton et al. 1982).

Schemata are general knowledge structures that individuals rely on to explain and
predict events in life. One particular study analyzed the schemata of different age groups
relating to medical information by instructing subjects to arrange random blocks of
medical information “into the [perceived] optimal order for efi’ective instructions The
optimum schema was determined for different age groups and these ordered instruction
sets were used with a different set of subjects to assess recall. The results of the study
mention that the ordered information sets resulted in increased recall as opposed to
original information sets for all age groups (Morrow, Leirer et al. 1996).

Studies have shown that warning information is better recalled when presented
first in a set of information (primacy) and last (recency). Primacy (presenting information
in the beginning) and Recency (presenting information at the end) of information has a
tendency to promote recall, even though it seems to go against ordering schemata as

analyzed above (W ogalter, Fontenelle et al. 1985).

26

Technical Knowledge of drugs and medical information, in general, is low among
older adults. “Expecting older adults to comprehend detailed technical information
presented on warning labels may be unreasonable (Rogers, Rousseau et al. 1 999)

Hazard Perception: A survey of existing warnings related literature reveals that
consumer attitudes and beliefs towards warnings vary with age (DeJoy 1991). There have
been studies that have looked at the attitudes of consumers fi'om different age groups
(college, high school and middle school) to “signal words” like: danger, and caution.
Study results suggest that the youngest consumers generally exhibited a higher perception
of danger than older consumers when these signal words are present (Silver and Wogalter
1990). These findings are consistent with others that relate signal word and age to hazard
ratings, with younger consumers assigning higher hazard ratings than their older
counterparts (W ogalter, Jarrard et al. 1994). A study by Wogalter et al (1994) asked
subjects (high school students, college students and shopping mall visitors aged 21-80) to
rank warning labels found in actual consumer products in terms of perceived hazard. The
results of the study showed that the rank ordering was consistent among different age
groups. Additionally, it has been indicated that when consumers perceive a hazard to be
very high, they are more likely to notice, encode, comprehend and comply with the
associated warnings (DeJoy 1991).

Cost/Benefits of Compliance: There are always certain ‘costs’ to compliance (e. g.
wearing safety glasses, locating them, wiping them clean and so on). It would appear that
older adults and seniors would not comply with warnings when the cost of compliance is

too high. In other words, seniors may not have the physical and cognitive resources

27

needed to comply with a warning that has a high and complex compliance cost (Rogers
1983)

Risk-Taking Style: Case studies have investigated the likelihood of younger adults
to voluntarily assume risk by subverting warning information (DeJoy 1991). This is
sometimes referred to as the ‘boomerang effect’, named after the Australian hunting
implement that returns to the user when thrown. A study by Goldhaber and deTurck
compared the likelihood that middle school and high school students would comply with
a ‘no diving’ sign near a swimming pool. In this study it was seen that males from high
school were more likely to indulge in the prohibited behavior (boomerang effect).
However, another finding from this study was that high school female students were
more likely to comply with the warning sign, thus indicating an effect of gender on the
boomerang effect (Goldhaber and Deturck 1989). Other studies comparing middle aged
to young adults indicate that middle aged adults (40+) were more likely to comply with
warnings in general (DeJoy 1999). So far, studies have shown that older adults very
rarely take risks voluntarily.

Attention: Even though there aren’t many studies that have looked at the
differences of attention in relation to warnings and aging, there is literature to suggest
that attention, by itself does decline with age. There are studies to show that both divided
attention (ability to multi-task) and selective attention (ability to gain cues from the
environment) show significant reduction with age (Rogers, Rousseau et al. 1999).

Working Memory: It has been mentioned before that working memory, or short
term memory, declines with age (Salthouse 1991), and that older adults have trouble

comprehending warning label text when it overloads the working memory. Rousseau and

28

Rogers (1999) mention that working memory overloads may cause problems in all four
stages of the information processing model. Warnings may not be noticed, may not be
encoded properly when noticed, may be difficult to comprehend and so on. One
particular project by Morrell et al (1989) studied complex medical regimens and their
relation to declines in the comprehension of medical information (Morrell, Park et al.
198 9).

When warnings and health information overload the subject’s working memory,
there is always a risk of the subject not being able to comprehend the information on the
warning. An example of warning text that overloads working memory is as follows,

“Consult a doctor if symptoms persist or get worse, if new ones occur, or if
sleeplessness persist continuously for more than two weeks because these may be
symptoms of a underlying medical illness (Rogers, Rousseau et al. 1999) ”

Analyzing this piece of text, it is obvious that the multilayer (if. . .or if. . .because)
structure of this sentence leads to an overload on the working memory of the subject and
causes problems in information processing.

Another study on comprehension and understanding by Zuccollo and Liddell in
1985 concluded (Zuccollo and Liddell 1985) that

“60% of elderly patients (mean of about 79 [years of age]) had trouble reading
their medication labels and only 23% had a clear understanding of all the instructions
(Rogers, Rousseau et al. 1999) ”

Prospective Memory: “Prospective memory means remembering to perform
actions in the future (Rogers, Rousseau et al. I 999) Certain waming-related tasks may

be event based or time based, requiring delayed action for proper compliance. Research

29

exists to show that older adults have a problem with time-based tasks. However,
Rousseau and Rogers mention that there is ‘remarkably little research’ on prospective

memory and its relation to warnings in general.

30

Chapter 4: The Human Eye and Eye Tracking

The Human Eye:

An understanding of the basic structure and working of the human eye is
necessary to understand how eye tracking works, and how the corneal reflection
technique uses the structure of the eye to its end.

The eyeball is surrounded by a tough and opaque membrane known as the sclera.
There is however, a small disc shaped transparent region in the front called the cornea. It
is through the cornea that light enters the human visual system. Underneath the
membranous structure of the sclera is a vascular layer that has a dense network of vessels
that supply the eye with blood and other fluids.

The sclera does not stop with the underlying vascular membrane. The vascular
membrane surrounds the innermost part of the eye known as the retina. This is the layer
that serves as the eye’s primary visual sensor. The retina is rich in light sensitive cells.
These cells are divided into rods (light sensitive) and cones (color sensitive). The retina
does not have an equal distribution of these cells throughout.

The point near the entry of the optic nerve into the eyeball is known as the macula
lutea and within this region is the fovea centralis. The fovea centralis is associated iwith
the point ‘of clearest vision’ and is composed mainly of cone cells. The fovea is about 0.4
mm in diameter (Y arbus 1967).

The retina itself is a highly complex and layered structure with multiple layers of

nervous, vascular and optic material. The nervous cells in the inner layers of the retina all

31

lead to the optic nerve that firnctions as the carrier of impulses between the brain and the
eye. This nerve is about 5 cm long and around 4 sq. m in cross section (Y arbus 1967).
Before light can reach the optic nerve via the retina, it must first be introduced
through the cornea. The cornea transmits light into a lens arrangement whose focal length
is controlled by a ciliary body. A paired set of muscles known as the iris controls the
aperture of the lens. This aperture of the lens is known as the pupil and diametrical
changes to the pupil are involuntary. The diameter of the entire structure of the eyeball is,
on an average, 24 mm measured in all meridians (imaginary lines drawn through the
center of the eye). As shown in the diagram, the rectus muscles control the movement of

the eyeball and enable focusing on objects in a wide field of view (Y arbus 1967).

Rectus Muscle

 

 

 

 

 

 

 

Sclera
Cornea Retina
Iris /
L ( I Vitreous Body
ens

 

 

 

 

\——-Optic Nerve

Figure 4 — A schematic diagram of the structure of the Human Eye

Eye Movements:

From an anatomical and psychological perspective, eye movements do more than

just shift the point of focus of the visual system. Research has shown that an important

32

requirement for vision is the presence of a non-stationary retinal image, i.e. an image on
the retina that is not constant (Y arbus 1967). Subsequently, there is always some
movement of the eye needed in order to achieve the shifting image. If by some artificial
means, a stationary retinal image is created, the eye loses its power of resolution and
ceases to see (Y arbus 1967). Voluntary and involuntary movements of the eye serve
various firnctions in the human visual system. The two most basic eye movements are
fixations and saccades. Along with fixations and saccades, drift and tremor are the two
other forms of eye movement. All four movement types are breifly introduced below.

Saccades: Saccades are rapid sweeping movements used to shift the point of
visual focus. Saccades take very little time to execute, lasting between lOmilliseconds
and lOOmilliseconds. During this movement the user is unable to see. Some special
saccadic movements are described as ‘smooth pursuit’ movements where the eye tries to
follow a moving a target and the saccade seems to match the velocity of the moving
object. Saccades can be both voluntary and involuntary. Saccades usually end in fixations
(Duchowski 2003).

Fixations: Fixations are stabilizing movements that enable to eye to focus on a
particular point in the visual field.

Drifls: It was mentioned in an earlier paragraph that there should be no stagnation
of the image that is formed on the retina for normal vision. Building on this concept, It
has been suggested that the eye never has a constant point of fixation (Y arbus 1967).
Instead, there is a fixation field that is created due to drifting movements of the eyes that
occur during fixations. These drifts serve to keep the retinal image non-stationary during

fixation and maintain the resolving power of the eye at an optimum level.

33

Tremors: Tremors are barely perceptible movements of the eyes that require very
highly sensitive eye tracking methods to resolve and observe. Tremors are characterized
by extremely quick and small movements of the eye and they occur because of
continuous internal nervous activity (Sheahan, Coakley et al. 1993).

Out of the above movements described, fixations and saccades are the most
important to visual gaze. F ixations are used determine points of attention and saccades

are used to study the gaze trail.

Methods to study eye movements:

Over the years, there have been many instruments and procedures used to study
eye movements. Some of the earlier methods were very intrusive and elaborate setups,
but most methods used for eye tracking today are non-intrusive and rely on arrangements
that have no direct contact with the eye itself. A brief description of some of these
methods follows:

Afler Images and Observational Studies: Early methods used to study eye
movements were those that used retinal ‘after-images’. Generation of these images on the
surface of the retina involved the use of blinding flashes of light. Once the after-image
was created, a trained subject could observe this image in his or her field of vision. By
studying the movement of this after image, eye movements could be monitored as these
images were strictly stationary with respect to the retina. Methods to study eye
movements based on simple visual observations were also practiced at this time. After
images are no longer used in eye movement research, and they have been mentioned here

merely for their historical significance (Duchowski 2003).

34

Mechanical Methods: There are several methods to record eye movements by
mechanical means. The earliest accurate eye tracking methods involved the use of suction
devices attached to pointers on the surface of the eyeball. As the eye moved, the pointer
would move and trace paths. There were some later studies that used a camshaft
mechanism on the cornea. When the cornea moved, the cam would be displaced and it
would move a pointer that traced a line on a sheet of paper. The advantages of these
methods are extreme accuracy and fine detail. Even barely noticeable eye movements
like short saccades and tremor were detected with ease. However, these methods are very
intrusive and cause a great deal discomfort to subjects. Furthermore, the preparation of
the equipment required for testing was tedious and elaborate (Duchowski 2003).

Electric Methods: A very small electric voltage is always present between the
outer and inner sections of the retina and as well as between the cornea and the sclera.
As the eye moves, this potential difference varies. Electro-oculography is a technique that
employs this phenomenon to track eye movements. Electrodes similar to those used for
ECG measurements are placed on specific points on the forehead and the potential
created is recorded (Duchowski 2003).

Optical Methods: The majority of inexpensive eye tracking methods in use today
rely on optical methods.

Many optical methods, such as the one used in this study, employ the
phenomenon of corneal reflection (CR). The cornea of the eye reflects a portion of the
light that falls on it, as does any transparent curved surface. The ‘bright spot’ that is
created on the cornea is slightly displaced from the center of the eye ball. The relative

positions of this CR spot and the eyeball are used in many techniques to study eye

35

movements. CR based eye trackers are not intrusive, are inexpensive and are easier to
setup and use than mechanical methods. However, one major disadvantage of using the
CR to perform eye tracking is that the CR also changes position with the head. Stillness
of the subject’s head is an important pre-requisite in setups that use the CR to track eye
movements (Duchowski 2003). This might be difficult for people with neurological or
musculoskeletal conditions that make it difficult to remain still. Other optical systems

include those that use the Purkinje images that are formed on the eye ball.

Eye Tracking - Applications and Spheres of interest.

Eye tracking has been employed in several fields to study attention, natural
behavior and design. Duchowski, in his book on eye tracking methodologies and practice,
divides the applications of eye tracking into interactive and diagnostic applications
(Duchowski 2003); diagnostic applications are those that help researchers gain some
insight into the attentive behavior of the subject. This section attempts to summarize
some of the major diagnostic applications of eye tracking and provides references for
further reading.

Neuroscience and Psychology: Eye tracking has found vast applications in these
fields. Studies in attentional neuroscience have researched the relation of eye movements
to the other sensory inputs such as hearing and smell (Asaad, Rainer et al. 2000). Other
studies involve analyzing neural activity in the brain in relation to eye movements like
saccades and drifts (Gamlin and Tweig 1996; Snodderly, Kagan et al. 2001).

There is also a widespread body of research on reading and visual cognition. Eye
movements have been used extensively to study how human beings read lines of text

(Rayner 1992). Studies on concepts like scene perception have also increased our

36

understanding of how people scan over scenes and glean information. Studies on eye
movements contribute to the knowledge of Schemata (sequences of steps) that are used
by people when they fixate over a scene filled with visual information. There are also eye
tracking applications in fine arts and visual media. A famous large-scale study at the
National Gallery of London used eye tracking to study subjects as they fixated over
digitized representations of paintings (W coding 2002). Similar studies have been canied
out on other media such as film (d'Yedwalle, Desmet et al. 1998).

Other diagnostic studies in psychology using eye tracking include studies on
illusions, problem solving, visual search and even brain damage and its effects
(Duchowski 2003).

Industrial Engineering and Human Factors: “Eye tracking aflers a unique
measure of human attentional behavior. This is particularly important in evaluating
present and future environments in which human do and will work (Duchowski 2003) ”

Eye tracking has been used many times in evaluating human performance in
modes of transport such as aviation and driving. In aviation, eye tracking is used in
studies ranging from flight simulation to evaluation of graphical displays. Control panels
of aircraft and their ease of use are also studied using eye tracking (Anders 2001). Loss of
visual attention is attributed as the cause of a majority of road accidents (Chapman and
Underwood 1998). Studies on visual attention of drivers using eye tracking are, thus, very
important.

In industries that rely on visual inspection like meat, poultry, electronics and
quality control, eye tracking has been used to study and determine optimum methods for

the same (Duchowski 2003) .

37

Marketing and Advertising: There are widespread applications of eye tracking in
the field of persuasive communications. A study by Lohse in 1997 applied eye tracking
effectively to the field of copy testing and analyzed consumer eye gaze while fixating on
print ads in the Yellow Pages TM (Lohse 1997). Eye tracking has also been used to study
print advertising and investigate phenomena like “wear-out” associated with repeated
exposure to print advertisements (Rosbergen, Wedel et al. 1990). A widely cited study
on warning labels in cigarette advertising used eye tracking to assess adolescent attitudes
towards them (Krugrnan, Fox et al. 1994).

The above paragraphs provide only a brief overview of the possible diagnostic
applications of eye tracking. These serve to place the method in perspective. The reader is
directed to the references section to gain further insight into the studies mentioned above.

Packaging: It is obvious that eye tracking can be applied in a variety of scenarios
important to packaging designers and engineers. In a retail environment, it is important to
understand how consumers ‘view’ supermarket aisles and how they scan packages before
they make purchases. Indeed, the potential of eye tracking as an important tool for
research in retailing has prompted major firms such as Unilever, Walmart, Kimberly-
Clark, Target and General Mills to invest in and conduct studies using eye tracking
systems. It is noteworthy that some of these setups have received coverage on popular

news channels (Ahlbaum 2008).

Closing Statements:

The information contained on the labels of pharmaceuticals, are critical to their
safe and effective use. This study attempts to use eye tracking to quantify the

noticeability and legibility of PWL’s. Other ongoing studies at the School of Packaging,

38

Michigan State University, USA attempt to study overt anti-counterfeiting features and

change blindness.

39

Chapter 5: Print Legibility

Introduction:

Several authors have stressed that communication is one of the primary functions
of packaging, (Paine and Institute of Packaging. 1963; Soroka and Institute of Packaging
Professionals. 1995) and that warnings are an important form of communication
(W ogalter, DeJoy et al. 1999). It is known that very often, packages have warnings that
are critical to the well being of the consumers. The four stage information processing
model that was used to explain warning effectiveness in an earlier chapter stresses that
warnings must be legible in order to be encoded into the memory of the receiver. In
visual warnings, this encoding can occur only if the printed message is legible (Rogers,
Rousseau et al. 1999) .

A quotation from McLean, as quoted by Pietrowski (1993), clearly indicates the
complexities associated with legibility (Pietrowski 1993).

“Legibility is a dangerous — and interesting - word. It is dangerous because it is
so often used as if it had a definitive or absolute meaning, which it does not. It is a
personal word neither scientific nor precise. If you say ‘that is legible ’ you mean only
that you can read it: you do not know whether I can (McLean 1980) ”

Legibility has fiequently been explained as a complex interaction of multiple
factors dependant on the design of the message (typeface, print design, layout), surface
(matte, glossy), viewing conditions (light levels) and reader characteristics (visual acuity,
color perception) (Pyke 1926; Tinker 1963; Bix 2003). Legibility has also been named

very differently in different sources. Some terms commonly used include legibility,

4O

reading ease, perceptibility, encoding ease etc. This document refers to all the above as
simply, ‘legibility’ for the purpose of clarity.

However, several authors agree that legibility is best defined as the ability of
readers to read the printed message with ease (Tinker 1963; Pietrowski 1993; Bix 1998;
Bix 2001). There are many factors that affect the ability of readers to read printed

messages. Three of the major factors are described in the following section.

Factors that affect legibility:

As mentioned previously, optimum legibility is the result of several factors and
seemingly disparate systems working together (Tinker 1963; Pietrowski 1993; Bix 1998;
Bix 2001).

T fireface and Message Design: Typographic design has an important bearing on
the visual impact and reading ease of printed matter. The morphology of a typeface
includes terms like: x-height, serifs, ascenders, descenders, uppercase, lowercase letters,
keming etc (McLean 1980; Bix 2001). Readers are directed to the aforementioned
references for visual representations of these factors on popular typefaces. Each one of
these factors can impact the legibility of text. Reading case is also affected by the size
and width of the message (Tinker 1963; McLean 1980; Bix 2001).

Other factors that impact the legibility of printed matter are characteristics of the
message itself, like line length, width and line-spacing (leading).

Printing Surface: In his widely cited report to the H.M.Stationary office in 1926,
Pyke summarizes several characteristics that printing paper should have in order to have
maximum legibility (Pyke 1926). The general recommendations were that printing paper

should have very little specular reflectance (gloss) and a high degree of diffuse

41

reflectance, in other words, a matte surface. From a paper-making perspective, the
surface of the paper depends on several factors like pulping process, additives and
coatings. These coatings and additives may increase the surface smoothness and create
higher levels of specular reflectance. There are also specific recommendations regarding
the opacity and thickness of the paper for optimum print legibility.

Contrast: Contrast here, refers specifically to the difference in color between the
print and the background. This has been found to impact print legibility to a great extent.
Several studies have shown that dark text on light or neutral backgrounds have high
levels of legibility (Tinker 1963; Bix 1998; Bix 2001).

FDA’s labeling regulations for OTC drug packages recommend that “the type
must be all black or one dark color, printed on white or other light, neutral color,
contrasting background (FDA 1 999) This regulation is consistent with the concepts
mentioned above.

Additionally, Tinker mentions in his book that the “use of white print on black

background is attention getting to a striking degree... (Tinker I 963) ”

Measurement Techniques:

Since legibility is a very complex concept resulting from the interaction of several
print and reader characteristics, measuring it is challenging, to say the least. Nevertheless,
there have been several measurement techniques that have been proposed as measures of
print legibility over the years. Some of these are explained below. A bulk of the
information presented here draws from a book chapter in the 1963 book, Legibility of

Print by Tinker (Tinker 1963).

42

Visibility Methods: These methods use photographic filters positioned like eye
glasses in front of the eyes. These filters are rotated in order to increase and decrease the
apparent brightness of the print sample that is being viewed. Based on the extent of
rotation of these filters, legibility has been measured. This principle has been used in the
design of the Luckiesh-Moss Visibility meter (Luckiesh and Moss 1938; Luckiesh and
Moss 193 9).

Distance methods: In distance methods, print samples are placed at different
distances from the reader’s eyes and the relative ease of reading between the different
print samples are evaluated (Sanford 1888; Tinker 1963).

Quick-Exposure Methods: Legibility is also measured by evaluating the speed at
which readers can read printed information. Previous experiments have been performed
by exposing test subjects to printed matter for very short amounts of time through use of
instruments like the Tachistoscope (Dockeray 1910). A tachistocope is an optical device
that allows researchers to quickly present stimuli to subjects. It flashes images at a
predetermined rate. These were essentially precursors to modern setups that use
computer software and projectors. Similar techniques include speed-reading
measurements where reading speed has been linked to legibility.

Focal Variator Methods: These are similar to the visibility meters, but they use
lenses instead of photographic filters. The rotation of the lenses can increase or decrease
the blur of the printed matter that is being viewed. Legibility is calculated based on how
soon a reader can recognize and read printed matter (Weiss 1917).

Eye Based Methods: Eye movements and blinking rate have also been used to

measure legibility. Researchers have performed direct counting and recording of blinks

43

and have tried to correlate this with legibility. The available literature indicates that
illumination, subject position and other ambient factors must be kept constant during this
whole process (Tinker 1946).

Chapter 4 on Eye tracking mentions cases where eye tracking has been used to
study reading behavior. Such measures have also been used in the past to evaluate
efficient message design and print design. These could be used to evaluate print legibility

as well (Rayner 1992).

Legibility Measurement in this study:

This study makes use of the Lockhart Legibility Instrument (LLI) which was
developed at the School of Packaging, Michigan State University (Bix 2001). Detailed
diagrams and the working of this instrument shall be presented in the Materials and
Instrumentation chapter.

The ASTM standard for comparative legibility now uses the LLI and this method
will be used in this study to evaluate the legibility of the various color combinations of

the PWL’s (ASTM 2006).

Chapter 6: Materials and Instrumentation

The first part of this chapter features an in-depth discussion of the materials and
supplies used in this study. The second part includes detailed descriptions of the

instruments used and their working.

Materials

Near Point Visual Acuity card: This is a laminated card which is printed with
text of gradually increasing type sizes from bottom to top (See Figure 5). The subject is
asked to hold this card at around 16” from their eyes. They are then asked to read the
lowest line possible. The subject’s visual acuity is determined based on the lowest line
that is read with case. This visual acuity card is manufactured by Dow Corning®

Ophthalmics and the visual acuities that can be measured range from 20/20 to 20/ 120.

spKHN OCVRz
Hozav DORNK

ONHDR ZKBOV

  
 

.......... Dow Comtng I
.......... o -—-
'phthotmlcta animals

  

Figure 5 - The Dow Corning Opthalmics, Near Point Visual Acuity Card

45

REALM-R Card: “Low literacy is a pervasive and under recognized problem in
health care. Approximately 21 % of American adults are functionally illiterate, and
another 27% have marginal literacy skills. Such patients may have difliculty reading and
understanding discharge instructions, medication labels, patient education materials,
consent forms, or health surveys. ” (Davis, Michielutte et al. I 998)

The above quote from an article by Davis et al. illustrates the important of health
literacy in today’s world of self-medication and out-patient care. It was decided to
include a test of health literacy along with the eye tracking study in order to account for
the effect of low health literacy on the noticeability and recall of warning label
information. For instance, subjects who did not recognize the terms used in the warning
messages might spend more time on them while examining the vials.

The Reahn-R (Bass, Wilson et al. 2003) is a shortened version of the REALM
test (Rapid Estimate of Adult Literacy in Medicine). The original REALM test comprises
a list of 66 words; the REALM-R consists of 11. Subjects are asked to read this list of
words and their health literacy is estimated based on how many words they are able to
read and pronounce correctly. The words shown below were printed on a sheet of white

paperboard using 20 pt sans-serif font.

Fat Flu Pill Allergic Anemia
Jaundice Constipated Colitis Osteoporosis

Fatigue Directed

Subjects were asked to read the words and say “blank” when they came to a word
they could not read. Subjects are scored based on their ability to correctly pronounce the

words. However, the words fat, flu and pill are not scored, serving as an acclimation

46

period. Subjects with a score of 6 or less are considered to be at risk for poor health
literacy.

Pseudo [so-Chromatic Plates: A booklet with pseudo ISO-chromatic plates
manufactured by Richmond products (Albuquerque, New Mexico) was used to quickly
screen subjects for red/green color blindness. The subjects were asked to flip the pages
and read the numbers (see Figure 6). Out of the 14 images, subjects are considered to

have normal color vision if they were able to read 10 or more.

f

 

Figure 6 - Pseudo Iso Chromatic Plates booklet used for testing for Red/Green
Color blindness

Vials: The vials used in this study were 10 dram vials of the l-Clic type (Rexam
Prescription, Ohio). Previous studies have shown that vials of the 10 dram size are the
most commonly used for dispensing prescription medication in the United States (De La

Fuente 2006). Vials were obtained through the campus pharmacy (see Figure 7).

47

White Pharmacy Label

 

Prescription Warning Label

Figure 7 - A photograph of a vial used in the study. A common 10 dram prescription
vial of the l Clic type with two types of labels; PWL's and White Pharmacy Labels

Labels: The Five color contrast combinations of PWL’s that were tested in this

study were,

Black text on Blue background
Black text on Red background
Black text on Yellow background
Black text on White background
Blue text on White background

The first four color combinations in the above list were chosen from

pharmaceutical supply catalogues as the most commonly commercially used

combinations (Pharmex 2009). A fifth combination (Blue text on White background) was

added based on the expert recommendation of Mr. Tom Ball at the Olin Health Center,

MSU. At the University pharmacy, the light blue on white contrast is being used on

PWL’s for vials used to dispense tablets for typhoid. They carry instructions that direct

48

patients to refrigerate the vials. Pharmacists indicated that many patients had failed to do
so, and suggested that this contrast be included in the study.

Labels were designed using Adobe® lllustrator® CS3 (See Figure 8) and were

printed using high fidelity digital printing. The labels measured, 4 centimeters by l

centimeter and were pasted to the vials as shown in the Figure 7. For the eye tracking

portion of the study, the vials were placed inside pharmacy bags and handed to subjects.
-‘W"‘t
‘ MEUCATIONSWALLOW Black on Red
manna AVOID arrows“ Black on Blue
WHILE TAKING THIS DRUG J
WARNING: use THIS DRUG

ONLY AS DIRECTED Black on Yellow
00 NOT DRIVE WHILE -
TAKING THIS MEDICATION Black on Wh'te
“ ‘ Blue on White

Figure 8 - PWL's used in this study

In addition to the PWL. each vial also had a White Pharmacy Label affixed as
shown in Figure 7. This label contained information about the pharmacy, dosage, patient,

doctor and drug. Figure 9 depicts one of the prescription labels used in this study.

49

:r: can HEALTH CENTER PHARMACY ‘ ' I ' .
,-:-:.:-:-:°o§,‘3;:3:' Michigan State University (517) 353'9 165
5,; ’ East Lansing, Michigan 48824-1037
TEE T652?"'m”°i§‘”"w°”“bb7‘6’87bb"m
YOUNG , SUB" DR . JACKSON , C .
340 RIVER DR 3351‘ LANSING, “1353‘9165

ran TWO TIASPOONFULS
EVERY 4 TO 6 HOURS

III DATA PROCESSING

REORDER FROM?

SM:
1-800-235-0767 LABEL STYLE; LP

180141..

PHENERGAN W/DM SYRUP
NO REPILLS

Discard This Medication One Year After Date Dispensed

Figure 9 — An example of a White Pharmacy Label from the Olin Health Center
that was pasted on all vials.

Warning Label Text: It has been mentioned that the composition of the text that a
person reads strongly affects the gaze trail. It was recognized early in the study that the
five pieces of warning label text used as stimulus material had to be scientifically
evaluated and standardized for reading ease.

The Flesch-Kincaid reading-ease score, introduced by Rudolph Flesch in 1948 is
widely used in general readability assessment (Flesch 1948) and comes bundled with
popular word processing packages such as Microsoft® Word. The reading ease score of a
piece of text can be calculated in Microsoft® Word by it and clicking on Word Count in
the Tools Menu. This method of readability evaluation was used to create five warning
messages that closely matched those used commercially for PWL’s. The warning
messages used on PWL’s are shown in Table 2.

A Flesch-Kincaid reading ease score of 66.7 falls in the range that is easily

readable by most 13-15 year old students (Dubay 2004).

50

Table 2 - Warning message readability scores

 

Flesch-Kincaid Readability

 

 

 

 

 

 

 

 

 

PWL Wamlng label Text Score
Do not chew this
Black on Red medication, swallow 66.7
whole

Warning: Avoid smoking

Black on 8'” while taking this drug 66'7
Wamlng: Use this drug

Black on Yellow only as dl I I 66.7

Do not drive while taking
Black on White thl s medication 66.7
Blue on White Shake well and keep in the 66.7

refrigerator
lnstrumen tation:

The ASL 5 01 Head Mounted Optics Eye Tracker

An Applied Science Laboratories (ASL) (Boston, MA), bright pupil eye tracker

(ASL 501) was used in this study (see Figure 10). The construction and working principle

of this instrument are described below.

Construction: The head mounted optics eye tracker consists of a headset that
houses two cameras and a monocle arrangement. The ‘Eye Camera’ is used to record
images of the eye and the ‘Scene Camera’, as the name suggests, is used to record the
subject’s field of view. The eye camera is coaxial to a lamp that shines a near infrared

beam of light onto the monocle. The monocle is coated with a special coating that reflects

51

 

the light beam. This beam falls on the eye and illuminates the pupil, thus the name
‘Bright Pupil’. A part of this light beam is reflected off the surface of the cornea and
forms a small bright spot of light known as the Corneal Reflection or CR. The positions

of this CR spot and the large bright pupil depend on the direction of gaze of the eye.

 

Figure 10 - ASL 501 Head Mounted Optics Eye Tracker

Calibration: The floating calibration procedure for the eye tracking system
involves directing subjects to look at 9 different points on a single plane in the field of
view. The positions of the CR and pupil are recorded at each of the nine points during
calibration. Based on the gaze information for these nine calibration points, the recording
software extrapolates the point of fixation of the eye for all points in the visual field.

Setup: As such, the ASL 501 requires that the subject’s head be stationary‘at all
times. In order to calibrate as accurately and precisely as possible, a custom calibration

system was built, which incorporates both a pane of glass and a chin rest (see Figure 10).

52

 

Figure 11 - Custom Designed Eye Tracking Table with Glass pane and Chin rest
arrangement

Data Recording: As mentioned, the eye camera records the position of the eye.
The scene camera records a video of the subject’s field of vision. The recording software,
Eye Response® Gaze Tracker® (Boston, MA), calculates the points of fixation and
superimposes a set of crosshairs over the scene video.

A short note on the process of analyzing eye tracking data will be provided in the
Methods chapter.
Legibility Instrument:

The Lockhart Legibility Instrument (LLI) was developed by MSU researcher,
Hugh Lockhart, and is used to quantify the relative legibility of printed matter. It is

mentioned in the literature that the LLI was initially used as a tool to study noticeability

53

(Bix 2001). However, the instrument was later redesigned and its methods reviewed to
enable objective measurement of legibility (Bix 2001). In this study, the LLI was used to
assess the comparative legibility of the five color contrasts in accordance with the
methodology specified by ASTM D7298 (ASTM 2006).

Construction: The LLI is a large viewing box that is 50” by 19” by 19”. A
movable easel inside the instrument is illuminated by two 25W incandescent flood lights.
The amount of current fed into these lamps is regulated by a rheostat actuated by a knob.
The light level is measured by a light meter built into the easel. The light meter displays

its output on a digital display mounted on the body of the LLI (see Figure 12).

 

Figure 12 - Lockhart Legibility Instrument

Operation: Subjects view the easel through a viewer that directs their gaze into
the instrument through a set of polarizing filters. At their initial position, the filters
polarize all light that passes through them and the inside of the box appears completely
dark to the subject. As the subject rotates a flywheel mounted below the viewer, one of
the polarizing filters rotates and allows light to pass through. It, thus, appears to the

subject that the label on the easel is progressively receiving more light. The subject is

54

instructed to turn the wheel until the first point where they can read the text on the label
with ease.

The angle through which the polarizing filter has rotated is known as a legibility
index (0-90°) and it represents the relative ease of reading of the printed message. The
legibility index values range from 0 to 90, with 90 being most difficult to read. This
legibility index reading is taken down from a digital display on the LL]. The LLI used in

this study was manufactured by Sycamore Industrial Services, Holt, MI.

55

Chapter 7: Methods

Working Hypotheses:

There were two working hypotheses while studying the PWL’s in this study.

0 The black on red color contrast combination is more noticeable than the black on
white, black on blue, black on yellow and blue on white color contrast
combinations

0 The black on white color contrast combination is more legible than the black on
red, black on blue, black on yellow and blue on white color contrast

combinations.

Test Procedure:

Individual subjects were tested using the following procedure.
Consent:

Researchers provided subjects with a verbal explanation and an overview of all
test procedures. All test subjects were shown a consent form which they were asked to
review and sign. A copy of this form is provided in the Appendix 1.

E ye Tracking:

A random order of presentation of vials was determined for each subject using a
custom written ActionScript® randomizer program on Adobe Flash CS 3®.

Subjects were seated in a chair with adjustable height at the eye tracking table that
was fixtured with a pane of glass and a chin rest. The head band of the ASL 501 eye
hacker was placed on the subject’s head and tightened in place while making sure that

they were in a comfortable position. Minor adjustments to the optics system and the chin

56

rest were made so that they were appropriate for the subject. Once this was done, subjects
were asked to rest their chin on the chin rest.

Researchers then asked the subject to hold as still as possible, while they
performed the calibration procedure using the floating calibration technique described in
the previous chapter. This technique calibrated the subjects to the pane of glass that is
permanently affixed to the desk. Fixing the distance between the subject and the viewing
field allows the eye tracker to precisely track the position of the subject's eye in space and
minimizes the potential for parallax error.

Each treatment (the five vials with warnings in different color contrast
combinations (see Figure 7) was placed in a bag from the pharmacy, and randomly
handed to the subjects who were instructed to remove the vial from the bag and examine
it as they pleased, for as long as they wished. Subjects were provided with the following
scenario,

"You have been delivered new prescription drugs from the pharmacy. Please do
what you would normally do." Researchers also instructed the subjects as follows. "If you
read anything on the vial, please press the vial against the glass while doing so."

The eye tracker system software outputs a video file with the gaze trail of the
subject. This allowed researchers to determine whether or not the subject noticed the
PWL. It also allowed researchers to quantify the time that subjects spent on each of the
label and package elements.

A control population of 15 subjects (aged 18-29) and a test population of 17
subjects (aged 50 +) viewed each of the five labels using the Applied Science

Laboratories (ASL) eye tracking device. As a precaution, a greater number of subjects

57

were tested in the older population to account for the possibility of unusable eye tracking
recordings.
Recall and Recognition Test:

After testing with the eye tracking equipment, subjects were given a plain sheet of
paper and a pen and were asked to recall everything they could about the prescription
vials that they just viewed. This was a test of recall. A recognition test was then
administered in which subjects were shown a sheet of paper with 10 labels, in 10
different color contrasts (See Figure 13).

Please pick out the exact 5 labels that you just viewed

 

 

' um u! m: owuo "
om As mcrw .
00 NOV DRIVE WHILE
TAKING ms MEDtCATION ] ”norm“
WDICAfiON SWALLOW
mu

”rams wrkrswm
ML! “Km "‘5 MUG

‘ etuaosuoxwe‘i
,vegetariansmsrnurojY

WARMNG USE MS DRUG DC HGT WT! ‘HN‘ILE
ONLVAS DtKECIED IAKl‘fij THIS MLLULA'ICM

l .11 '51,}; ..
Figure 12 - Recognition Sheet
Five of these were identical to the labels on the vials that subjects had just viewed
using the eye tracker. The other five had the same messages, in different color contrasts
(black on purple, black on orange, black on green, black on pink and dark green on light
green). Subjects were asked to pick out the exact five labels that they had just viewed.
Collection of subject information:
Data recording sheets were used to gather subject information such as age,

ethnicity, gender, educational background and number of prescription drugs taken per

58

day. Subjects were assigned a numeric code in order to provide strict anonymity. They
were referred to by this number throughout the data processing and analysis stage. A
sample data recording sheet is provided in Appendix 3.

Additional Testing:

Health Literacy Testing: The REALM-R card was held at a convenient reading
distance from the subject. Subjects were provided with the following verbal instructions.

“Sometimes in the health care system, medical words are used that many people
are not familiar with. I would like to get an idea of what medical words you are familiar
with. I would like you to read the words printed on this card aloud. When you have
trouble with a word, please say ‘Blank’ and move on to the next word”

Checklists were used to record the number of words correctly pronounced. If a
subject took more than five seconds to read a word, that word was considered a failure. A
score of lower than six indicated that subjects were at risk for poor health literacy. At risk
denotes the fact that patients may not be able to understand the words that they are
expected to while interacting with a physician in a primary care setting.

Color Perception: To examine color perception and screen subjects for red/ green
color blindness, subjects were asked to read the numbers from a booklet with a set of 15
pseudo-isochromatic plates manufactured by Richmond products. Subject responses were
recorded and tabulated.

Visual Acuity: The visual acuity of the subject was then evaluated using the Dow
Corning ® Opthalmics Near Point visual acuity card. Subjects were asked to wear their

glasses or lenses, bifocal or contact while testing for near vision. The card was held at 16

59

inches fi'om their eyes. The visual acuities of subjects were recorded and tabulated based
on the lowest line that subjects could read from the card
Legibility testing:

Subjects viewed the fifteen PWL’s in three sets (see Table 3 for example) using
the Lockhart Legibility instrument, each set containing the five PWL color contrast
combinations in random order. At the end of testing, the legibility index values were
averaged across the three sets. This provided researchers with 33 replicates for each label.

For each contrast combination, the subjects were instructed to m the wheel on
the LLI until they were able to read the text on the labels with ease. The angle through
which the wheel turns is known as the legibility index and this provides a measure of the
relative legibility of the piece of printed matter. All testing was done in accordance with
ASTM standard D7298-06, Standard Test Method for Measurement of Comparative
Legibility by Means of a Polarizing Filter Instrumentation (ASTM 2006).

It is known from previous studies that while testing on the LL], the initial
readings differ fi'om later ones (Bix 1998; Bix 2001; Bix 2003). This necessitated the
introduction of two dummy labels prior to the actual testing of the PWL’s. Their
legibility scores were not considered for analysis. Previous studies performed using this
instrument used an easel distance of 17.5” and a light level of 25 foot candles. This study

also used the same settings (Bix 1998; Bix 2001; Bix 2003).

60

 

 

 

 

 

 

 

 

 

 

 

 

Dummy Set Set 1 Set 2 Set 3
Black on Red Black on White Black on Yellow
Black on Yellow Black on Red Black on Blue
Subject 1 Two Dummy
Labels were
tested at the Black on White Blue on White Black on White
beginning of the
Test Black on Blue Black on Yellow Blue on White
Black on Blue Black on Blue Black on Red
Table 3 - Example legibility test order
Compensation:

Once testing on the LLI was completed, each subject was given a cash reward of

30$. Subjects were thoroughly debriefed.

Data Processing:

Eye tracking: The gaze trail videos obtained using the eye tracker were analyzed
using the Eye Response® Gaze Tracker® analysis software. The images of the vials in
the videos were broken down into areas known as “look zones”. The amount of time the
subjects spent on each of these look zones was calculated using the software. The specific
areas of the vial that were studied were:

0 The White Pharmacy Label
0 The Cap

0 The PWL

61

 

Other important information, such as the number of times that a look zone was
entered (hit count) and the total time subjects spent examining each vial was also
recorded.

Legibility: Legibility indices or degrees of rotation of the polarizing filter were

coded in a spreadsheet that included subject demographics and characteristics.

62

Results, Analysis & Discussion

Data Encoding and Entry:

As explained in the Materials and Methods chapters, there were several different
forms of data that were collected during the course of this study. Eye tracking data was
collected as video files that were later processed using the Gaze Tracker® analysis
software. The result of this analysis was a Microsoft Excel® spreadsheet which had zone-
specific metrics. Legibility data (degrees of rotation of the polarizing filter) and subject
characteristics (visual acuity, health literacy etc) were coded and manually entered into
spreadsheets. These items were bundled into a large spreadsheet in the ‘flat file’ form
suitable for entry into the SAS® Statistical Software.

This chapter includes summaries of data sets from this study and subsequent

mixed model statistical analyses using the SAS® Statistical Software.

Subject Demographics:

The control population was comprised of 15 subjects between the ages of 18 and
29 years (ave.22.8 :i: 2.33). The test population was made up of 17 subjects who were
over 50 years of age (ave. 62.2 :i: 8.10. Demographics of these two groups are provided in
the table below. Subjects ranged in age from 20 - 77 years.

The following tabulations and corresponding charts depict the distribution of
subject gender and reported ethnicity (see Table 4), health literacy and Red/Green color
blindness (see Table 5), prescription drugs per week (Table 6). Graphs are shown after

the respective tables to represent this information in a visual manner.

63

Table 4 — Subject Gender and Ethnicity (Data for Figure 14)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Male Female Caucasian Asian-
j American
Gender Ehnicity

 

 

Gender Ethnicity
Popln . . . # f
#of Avg #of Average Total# #of . Avg Asian Avg
Male Age Female Age of Caucasian Age 1 American Age
Subjects :30 Subjects 180 Subjects Subjects SD Subjects :80
Control .
22 t 23.7 :1: . 22.9 :1: 22.6 1
Population 8 7 15 10 5
(1849) 1.06 3.09 g 2.42 2.40
Subject 65.4
Population 5 1 12 :15; 17 17 682,201 0 0
(50+) 6.54 ' ‘ . » ‘
38.6
47.2 t - » 47.7 :t 22.6 1:
“‘3' 13 i 19 19.74 32 ' _ . .27 20.45 5 2.40
22.31
lControl
Population
(18—29)
'/.Test
Population
(50+)

 

 

Figure 14 — Subject Gender and Ethnicity

 

Table 5 — Health Literacy and Red/Green Color Blindness (Data for Figure 15)

 

 

 

 

 

 

Health Literacy Color Blindness
Population . . . . . . # of # of
# fig’t‘flrgg‘ka's # “Kt“:l‘giua's individuals individuals At
Not at risk risk
Control Population
(18-29) 11 4 15 0
Subject Population
(50 +) 16 1 1 5 2
Total 27 5 30 2

 

 

 

 

 

 

I Control
Population
(1 8-29)

'1 Test
Population
(50+)

Z
Z
/
%
%
%
é
a

 

 

 

 

Not at risk At Risk Not at risk At risk
Health Literacy Color Blindness

Figure 15 - Health Literacy and Red/Green color Blindness
Only one member of the test population tested to be at risk for low health literacy,
in contrast to 4 members from the control population (aged between 18 and 29 years).
The native language for all subjects from both populations was English. Only one subject
out of the 32 that were tested had less than a college degree. All other subjects ranged in

their level of education from at least two years of college to a doctoral degree. All

65

 

 

members of the control population tested as having normal red/green color vision while 2

members of the test population tested to be at risk for abnormal red/ green color vision.
The data on Table 6 and Figure 16 indicates that a majority of the subjects in the

control population (18-29 years of age) did not report taking any prescription drugs on a

daily basis at the time of the study. However, a majority of the test population (50+)

reported taking more than 2 prescription medications every day.

Table 6 — Medication Consumption (Data for Figure 16)

 

 

 

 

 

 

 

 

 

 

"”2533 Dfl'ugs c::;:rgi;c:am’t:la $1.122) # of Subjects wfilzlsme test population
or Day
None 10 1
1 4 4
2 1 3
3 O 3
4 O 3
5 O 2
Over 5 0 1
Total 15 17
:gfgstzgfg 0.4 :1: 0.63 2.94 1 2.16

 

 

 

66

 

—l
o

  

Number of
Subjects

 

I Control

wuhwmuooto

'/ Test
Population
(50+ years)

 

 

None 1 2 3 4 5 More

 

Number of prescription drugs per day

Figure 16 - Medication Consumption

Noticeability of PWL’s using Eye Tracking data:

Eye tracking data was collected in the form of video files and these files were
analyzed by zone (PWL, White pharmacy label, Cap - see Figure 7). Dependent variables

for analysis of the eye tracking data were both binary and variable:
0 The proportion of subjects who noticed a zone (yes/no; a binary attribute)

o The proportion of total time per vial on a zone, i.e. Time spent on a particular

zone divided by the total time spent examining the vial itself (continuous variable)

0 Number of times the zone was crossed by the subject’s gaze trail for each zone of

each vial (number of hits) (discrete variable)

67

Previous studies have quantified noticeability based on several different
dependant variables. The time taken until the first fixation in the zone of interest, number
of fixations and the amount of time spent in the zone of interest have been used in the
past (Krugrnan, Fox et al. 1994). In this study noticeability was defined and quantified

using three response variables.
0 The probability of noticing a zone
0 The proportion of total time spent on a zone standardized by zone density and
o The total number of hits on each zone of each vial by a subject

Further explanations and statistical analyses for each of the three dependant variables are
provided below.
The probability of noticing a zone:

The success/failure of a subject to notice a zone (cross into a zone in their gaze
trail) was modeled as a binary response variable (noticed / did not notice) with zone
being a seven level categorical variable. The seven levels were: the white pharmacy label,
the five PWL’s and the cap. This data was available for 32 subjects (15 controls, 17 test
subjects). The frequency distribution table of the number of subjects who noticed/did not
notice label zones is shown below (see Table 7). As can be seen from the table, all
subjects (100%) noticed the White Pharmacy label while at least 10 subjects failed to

notice each of the PWL labels.

68

Table 7 - Frequency Distribution Table of N oticeability Binary Variable

 

 

 

 

 

 

 

 

 

PWL White
. Notleed (yes/no) Black on Black on Blue on Black on Black on Cap Phlaangrealcy
Red Blue White Yellow White
Frequency 11 11 10 1 1 11 124 0
No
Percent (%) 34.375 34.375 31.25 34.375 34.375 77.5 0
Frequency 21 21 22 21 21 36 160
Yes
Percent (%) 65.625 65.625 68.75 65.625 65.625 22.5 100
TOla' F uen 32 32 32 32 32 160 160
number req cy
of
subjects Percent (%) 100 100 100 100 100 100 100

 

 

 

 

 

 

 

 

 

This probability of noticing a zone was modeled as a function of population
(control versus test), zone (PWL’s - in 5 contrasts, cap and pharmacy label), their 2-way
interaction, gender and total time spent on the vial. The model was fitted using a logistic
regression model with the GLIMMIX procedure of SAS (Version 9.1, SAS Institute Inc.,
Cary, NC). The effects of health literacy, number of prescription drugs per day, ethnicity,
age, color blindness, order of presentation of vials and visual acuity of the subject were
included in the statistical model during the initial stages of analysis, but were later
dropped based on the type 111 test p-values (p > 0.10). Pairwise comparisons were
conducted using Bonferroni adjustment to avoid inflation of Type I error rate.

An effect of label zone was identified on the probability of noticing it (p <
0.0001). Pair wise comparisons indicated that there was a significantly greater probability
of noticing the White Pharmacy Label compared to any PWL, regardless of color (p <
0.01 for all corresponding pair wise comparisons) and a significantly lower probability of

noticing the cap compared to any PWL regardless of color (p < 0.01 for all corresponding

69

 

pair wise comparisons - see Figure 17). There was not enough evidence to conclude an
effect of color contrast on the probability of noticing a specific PWL (p > 0.60 for all
corresponding pair wise comparisons). The data did not support a difference in the
dependant variable between control and test population results (p = 0.98). Averaged
across label zones it was also identified that females had a greater probability of noticing
any zone as compared to the males (p < 0.0001).

Table 8 - Predicted Probability of Noticing a Zone (Data for Figure 17)

 

 

 

 

 

 

 

 

 

Zone Probability of Noticing a Zone (%) Standard Error
White Pharmacy Label 100 0.0011
Blue on White 69.62 8.998
Black on Yellow 67.14 9.375
PWL Black on Red 66.91 9.382
Black on White 66.63 9.421
Black on Blue 65.2 9.132
Cap 17.1 1 3.301

 

 

 

 

7O

 

Probability of Noticing an Zone (%)

 

White Blue on Black on Black on Black on Black on Cap
Pharmacy White Yellow Red White Blue
Label
PWL

Vial Elements

Figure 17 - Predicted probability of noticing a zone 1 estimated standard error.
Letters (a, b, c) indicate statistically significant differences (p < 0.01)
The Proportion of total time spent per zone standardized by zone density:

The proportion of the total time spent per zone per vial (time in a zone/total time
per vial) was calculated during analysis with the Gaze Tracker® Analysis software.
Recognizing the fact that different zones had differing amounts of text, and therefore,
would require varying times, researchers attempted to standardize the proportion of time
by dividing by the “zone density.” The zone density was calculated by dividing the

number of typographical characters in each zone by the area (see Table 9 below).

71

Table 9 - Zone density values for all label zones

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Density (number
Number of Area of Zone of characters per
Zone Typographlcal (square square
characters centimeter) centimeter)
Black on Red 35 8.75
Black on Blue 38 9.5
PWL’s Blue on White 33 4 8.25
Black on Yellow 32 8
Black on White 35 8.75
Cap 15 7.06 2.123
,1 261 6.121
2 270 6.332
White
Pharmacy 3 273 42.64 6.402
Labels
4 275 6.449
5 277 6.496

 

 

 

 

 

This dependant variable (standardized time, or proportion of time/zone density)
was analyzed and it was revealed that a large number of subjects had gaze trails that did
not cross into several PWL’s and caps. This created several zeroes in the pool of
dependant variable values, causing a problem in the residual plots that were generated.

Subsequently, this data set was truncated to include only those values that were
not zero, i.e., those instances where subjects had not logged into a given label zone were
eliminated from analysis.

After this, the response variable “standardized proportion of time” was log
transformed and modeled using a general linear mixed model fitted with the MIXED
procedure of SAS (Version 9.1, SAS Institute Inc., Cary, NC). The model included the

fixed effects of treatment, population (control vs test), total time spent on the vial and all

72

 

2-way interactions. Also, the model included the random effect of subject nested within
pepulation in order to account for the split-plot structure of the data.

The effects of health literacy, number of prescription drugs per day, ethnicity, age,
gender, order of presentation and visual acuity were considered for model inclusion but
did not make a significant contribution to model fit based on Bayesian Information
Criteria (BIC). Thus, these explanatory variables were not included in the final model.
However, the fixed effect of color blindness was included in the final model based on
BIC.

Pairwise comparisons were conducted on the estimated least square means of the
dependant variable using Bonferroni’s adjustment to avoid inflation of Type I error rate.
Model assumptions were checked and considered to be appropriately met. The salient
findings from this analysis are provided below.

The evidence did not support an effect of pepulation (control vs test) on the
standardized proportion of time spent on each zone. (P = 0.3453). After adjusting for the
effects of population and treatment, an association was identified between color blindness
and standardized proportion of time (P = 0.0209). Subjects that failed the red/ green color
blindness test had larger standardized proportion of time across all zones regardless of
population (control vs test) compared to those that passed the color blindness test. 5

During the analysis, a significant interaction effect between zone and total time
spent on the vial was identified on the standardized proportion of time. As a result, the
main effect of zone on the standardized proportion of time should not be interpreted on
its own. Thus the dependant variable estimates were calculated at four values of total

time (10 seconds, 20 seconds, 30 seconds and 40 seconds)

73

The tabulation (see Table 10) provides the Bonferroni adjusted pair wise
comparisons for the estimated least square means of the dependant variable at each of the
four values of total time. In the table, the “standardized proportion of time” dependant
variable is not statistically different for zones that share the same letter designation. It can
be seen that in all groups, there was not enough evidence to conclude an effect of color
contrast of PWL on the standardized proportion of time spent (all PWL zones share the
same alphabet). At the same time, the proportion of total time standardized by zone
density for the white pharmacy label was higher than that of the PWL and the cap. In
other words, subjects spent more of the total time on the white pharmacy label regardless

of differences in density and area.

74

Table 10 - Bonferroni adjusted pair wise comparisons across zones for four levels of
total time spent per zone.

 

Total time spent per vial

 

Total time spent on the vial is 10 seconds

Total time spent on the vial is 20 seconds

 

 

 

 

 

 

 

 

Zone Pair wise Comparison Zone Pair wise Comparison
Black on Red A Black on Red D
Black on Blue A Black on Blue D
Blue on White A Blue on White D

Black on
Black on Yellow A Yellow D
Black on White A Black on White D
White
w""°|_';::{'“a°’ B Pharmacy E
Label
Cap A Cap D

 

 

 

Total time spent on the vial is 30 seconds

Total time spent on the vial is 40 seconds

 

 

 

 

 

 

 

 

 

Zone Pair wise comparison Zone Pair wise comparison
Black on Red G Black on Red J
Black on Blue 6 Black on Blue J
Blue on White G Blue on White J

Black on
Black on Yellow 6 Yellow J
Black on White 6 Black on White J
White
Whitengzrnacy H Pharmacy K
Label
Cap G Cap J

 

 

 

 

The graph on the following page plots the value of total time spent examining the

vial (x-axis) against the proportion of total time standardized by zone density for each

zone (y-axis) for both subjects of both populations. Seven scatter plots were created on

 

this graph, one for each zone (5 color contrasts of PWL’s, the white pharmacy label and
the cap). Each scatter plot was used to generate a colored trend line. The figure has seven
trend lines that provide an overview of how subjects spent their time on zones as the
amount of total time they spent examining the vials increased.

As observed in the graph and as revealed in the pair wise comparisons shown
above, subjects, on the whole, spent a lot more time in the white pharmacy label zone as
compared to the other zones. The graph also seems to indicate that as the amount of time
spent on the whole vial increased, subjects spent more time on the white pharmacy label.
It is speculated that when subjects spent a large amount of time examining the vial, they
were spending a greater proportion of their time reading the white pharmacy label and
trying to glean as much information as they could from it. Despite spending larger
amounts of time on the vial, subjects spent a smaller proportion of their total time on the

PWL.

76

Figure 18 - Scatter Plots and Trend Lines relating Total Time spent on vials to the

 

 

 

 

 

 

Proportion of total time spent on each zone standardized by zone density
0.16
’z ‘ O '0. ..
e . ' e . e
‘e' 9
0.14 . . i -
° . e ee ' .. /
.e ' " ° ° . ’ -
in e . /
c O . . ‘0
¢ . / ‘
g . ' . ' o PWL - Black on Red
= 0.12 .. , 1 ,1
8 -.':;-.,x ' ' - PWL - Black on Blue
3' . 1'
3 ° ”3 -°' PWL - Blue on White
a x ‘ " .
E ,s“ 1...: o " ° ~< PWL - Black on Yellow
'8 0.1 “We. 0 . ‘
g . . . . :- PWL - Black on White
2 2"
a; ’ 8° . ° 0 White Pharmacy Label
3 e 0 e. .e 2 + C
D- . ap
g 0.08 :t . e ‘
3 ,' '. — - - Linear (White Pharmacy Label)
8. . ' —Linear (PWL - Black on Red)
C
a — - Linear (PWL - Black on Blue)
Ul
.. 0.06 - .
_E_ i - - Linear (PWL - Blue on White)
g — - Linear (PWL - Black on White)
”2 i . ~ . Linear (PWL - Black on Yellow)
.2 0.04 9
‘5 . , . + —Linear (Cap)
0.
. 2
j o.

 

 

 

 

50 60 70
Total time spent on the vial (seconds)

77

The total number of hits on a zone:

The total number of hits was recorded for the 3 zones on each vial: namely the
white prescription label, the PWLs (5 contrasts) and the cap. Similar to the previous
variables, zone was defined to be a categorical variable of 7 levels: the White pharmacy
label, 5 color contrasts of PWL’s, and the cap. This data is shown below in the form of a
cumulative frequency table with the number of hits for each zone ranging from 0 to 4 (see
Table 11).

Table 11 - Cumulative frequency table for the number of hits on each label zone

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

PWL White
Number of Hits mack Black Blue Black Black on Cap 312i
0“ Red on on on White
Blue White Yellow
Freq 11 11 1o 11 11 124 0
o
”(2?“ 34.375 34.375 31.25 34.375 34.375 77.5 0
Freq 19 18 20 18 20 33 96
1
“a?” 59.375 56.25 62.5 56.25 62.5 20.625 60
Freq 2 3 2 3 1 3 58
2
P952?“ 6.25 9.375 6.25 9.375 3.125 1.875 36.25
Freq 0 0 0 0 0 0 4
3
Percent
M o o o o o 0 2 5
Freq 0 0 0 0 0 0 2
4
Percent
(%) 0 0 o o 0 0 1.25
TOW Freq 32 32 32 32 32 160 160
Number
of
subjects ”a?“ 100 100 100 100 100 100 100

 

 

 

 

 

 

 

78

This ‘total number of hits’ response variable was modeled as Poisson distributed
using a generalized linear mixed model. This model included the fixed effects of zone,
population (control and test), their 2-way interaction and the effects of gender and total
time spent on the vial. Also, the model included the random effect of subject nested
within population in order to account for the split-plot structure of the data. The model
was fitted using the GLIMMIX procedure of SAS (Version 9.1, SAS Institute Inc., Cary,
NC). The effects of health literacy, number of prescription drugs per day, ethnicity, age,
color blindness, order and visual acuity were considered but were not included in the
final model based on type HI test p-values (p > 0.10).

Table 12 - Expected number of hits on a zone (Data for figure 19)

 

 

 

 

 

 

 

 

 

Zone Predicted Mean Number of Hits Standard Error
White Pharmacy Label 1.3585 0.1101
Black on Red 0.7122 0.1496
Black on Yellow 0.6996 0.1472
Black on White 0.6712 0.1433
Blue on White 0.6645 0.1453
Black on Blue 0.6519 0.1447
Cap 0.2003 0.03793

 

 

 

 

The number of hits, as obtained fiom the gaze trail analysis differed by zone (p <
0.0001). There were a significantly greater number of hits on the White pharmacy label
when compared to the PWL, regardless of PWL color contrast (p < 0.05 for all
corresponding Tukey adjusted pair wise comparisons - See Table 12 and Figure 19). The
number of hits on the cap was less than all PWL’s regardless of their color (p < 0.01).
The evidence did not support an effect of color contrast on the number of hits on PWL’s

(p > 0.99 for all pair wise comparisons) (see Table 12 and Figure 19).

79

The estimated mean number of hits and corresponding standard error are provided
for each label zone (see Table 12 and Figure 19). Pair wise comparisons were conducted

using the Tukey Kramer adjustment to avoid inflation of the Type 1 error rate.

1.6

—l &
-| N 5

Estimated number of Hits
.0
on

 

0.6
0.4
0.2
0
White PL Black on Black on Black on Blue on Black on Cap
Red Yellow White White Blue
PWL
Vial Element

Figure 19 - Estimated number of hits on each zone :I: estimated standard error.
Letters (a, b, c) indicate statistically significant differences (p < 0.05)

As expected, longer time spent examining the vial was associated with a greater
number of hits. Averaged across populations and treatments, female subjects made more
visual hits than male subjects on any zone (p = 0.0099). Also, the number of hits
averaged across zones was greater for the control population than for the test population
(p = 0.0010). The graphs below (see Figure 20 and Figure 21) show the average number
of hits that were made by subjects of both genders and both populations on the seven
label zones (white pharmacy label, five color contrasts of PWL’s and the Cap). They

reflect the findings described above.

80

 

I Male
6 Female

 

 

 

Vitae/Jr?” as;

:o xosm

       

2% Bo=e>

co xoufi

g; as;

no can

7///////A////7//. 2.5

:o :35

.7////fi//////V/i as.

:o xoafi

as... be .3532 onflo><

Label Zone

Figure 20 - Average Number of hits according to Gender

81

 

I Control
Population (18 to
are

/. Test Population
(50+ yearS)

Average Number of Hits

’/
r t
2 g
a 2

RV

 

 

 

 

§§§ 8 513 E0 5% 5% 59
£53 in 35 £5 3:; E
n. in in in in

PWL
LabelZone

Figure 21 - Average Number of hits by Population (Control vs Test)

Recall:

Researchers encoded free form recall responses into 12 categories, namely: label
layout, dosage information, drug names, patient information, pharmacy information,
prescriber information, PWL color, PWL text, cap information, vial colors and
medication characteristics. Recall was coded in each category for each subject in binary
fashion (recall; Yes/No) and analyzed using a generalized linear mixed model. Fixed
effects included in the model were: population (control vs test), response category and
their 2-way interaction. Subjects within the two populations (control and test) were
considered as random blocking factors in the model. Table 13 provides examples of recall

information in each of the 12 categories.

82

Table 13 - Examples of Recalled Information in the 12 categories

 

 

 

 

 

 

 

 

 

 

 

 

Category Element Example
'There was some dosage information in the
Label Layout bottom”
Dosage Information “take two pills twice daily...”
White . . . . ,.
ll
Pharmacy Drug Names | thlnk one of the pl 3 was Wellbutynn
Label Patient Information “...i remember seeing an Amy on there ...."
Pharmacy lnfonnation '..all vials were from the Olin health canton.”
. . “..One of the doctors was called Amy
Prescnber Information something."
PWL Color "...One of the labels was red...”
PWL . . . .
PWL Text ..one label said 110:3}: dnnk when taklng the
Cap Cap Information “...press down and turn on all the mp5."
Other Vial Colors “...the vials were yellow in color..”
Medication Characteristics “. . .there were small white tablets in the vial.."

 

 

 

 

 

Gender, ethnicity, age, health literacy, color blindness, number of prescription

drugs per day and visual acuity were initially considered for inclusion into the model.

However,

none of

these variables

showed

a

p-value < 0.05 when included in the model and, thus, were excluded from further

analysis.

83

Table 14 - Predicted Probability of Recall of Information in a particular category

 

 

 

 

 

 

 

 

 

 

 

 

(Data for Figure 21)
Category Element ”33mg,“ Standard Error
Label Layout 51.82 10.2
Dosage lnfonnatlon 68.97 8.468
Whlte Pharmacy Drug Names 55.93 9.285
Label Patient lnfonnatlon 58.32 9.622
Pharmacy Information 55.22 9.903
Prescnber lnfonnatlon 40.68 9.088
PWL PWL Color 43.76 9.148
PWL Text 53.18 9.2
Cap Cap lnfonnatlon 21.31 7.43
Other Vlal Colors 29.94 8.867
Medlcatlon Characterlstlcs 10.82 5.952

 

 

 

 

 

Results are presented here in the original scale as ‘least square mean’ estimates of
the probability of recall and estimated standard errors at 95% significance (see Table 14).
Model fitting and parameter estimation was conducted using the GLIMMIX procedure of
the statistical software SAS (version 9.1, SAS Institute, Cary, NC).

When averaged across population (control vs test), the proportion of subjects that
were able to recall a given category was influenced by category itself (p = 0.0008). Post-
hoc pair wise comparisons were performed using Tukey-Kramer’s adjustment to avoid
inflation of type I error rate (See Table 14 and Figure 22). Figure 22 indicates dosage
information as the most frequently recalled category. A significant interaction between
population (control versus test) and recall category on the probability of recall (p = 0.21)

was not concluded. A breakdown of the raw data for this dependant variable is provided

in Appendix 4.

84

 

Estlmated Probablllty of Free Recall (100%)

 

Recall categories

Figure 22 - Predicted probability of recall + standard error. Letters (a, b c) indicate
statistically significant difference (p < 0.05)

Recognition:

The number of PWL’s correctly recognized by the subjects were noted and

counted (Table 15 and Figure 23).

It was mentioned in the materials and methods chapters that subjects were
presented with a recognition sheet with ten labels (see Figure 13) and asked to pick out
the five labels that they had just viewed on the prescription vials. On this recognition
sheet, five PWL’s were the actual PWL’s that subjects had viewed in the prescription
vials. The other five PWL’s were labels which had the same text, but different color

contrasts. The tabulation below indicates the number of PWL’s that subjects recognized

85

in both populations (Table 15 and Figure 23). A breakdown of the raw data has been
provided in Appendix 5.

Table 15 - Number of PWL's correctly recognized (Data for Figure 23)

 

 

 

 

 

 

 

 

 

Number of PWL’s correctly Number 0" 30b!“
recognlzed Control Populatlon ‘ Subject Populatlon
None 1 1
1 1 8
2 2 4
3 7 2
4 1 2
5 2 0

 

 

 

  
 

\\\\\\\\\\\\\\\\‘

 

 

 

 

"m; / 7
b] g g I gggrmfion
/ g (18 -29)
/ / A Pgmulation
g g (53:,
a 2

None 1 2 3 4 5
Number of PWL's correctly ncognlzed

Figure 23 - Number of PWL's correctly recognized

For the pmpose of statistical analysis, subjects were scored in this recognition test

in two ways.

86

 

The number of PWL’s out of the original five that were correctly recognized was
counted and subjects were given a score out of 5. This was one dependant variable. A
subject who had recognized all five PWL’s correctly would have a score of five and a
subject who had recognized none would have a score of O.

A second dependant variable was created by counting both the number of PWL’s
correctly recognized and the number of PWL’s correctly ignored. For example, a subject
who recognized all five PWL’s correctly, while ignoring the other five correctly received
a score of 10.

For both dependant variables explained above, a general linear model was deemed
to be an appropriate fit. This model included the fixed effect of population (control vs
test). In addition, the effects of health literacy, number of prescription drugs per day,
ethnicity, age, color blindness, gender and visual acuity were considered for model
inclusion. Neither of these explanatory variables made significant contribution to model
fit based on a Maximum-likelihood based Bayesian Information Criteria (BIC). Thus,
these explanatory variables were not included in the final model. Model assumptions
were checked and considered to be appropriately met. The model was fitted using the

MIXED procedure of SAS (Version 9.1 , SAS Institute Inc., Cary, NC).

87

Table 16 - Estimated Least Square Means and Standard Error for Recognition

 

 

 

 

 

Scores /5 and /10
Estimated Least Square Means
Population of recognltlon test scores Standard Error
Score/5 Score/10 Scorels Score/10
Control (18 to 29 years) 2.66 7.13 0.34 0.45
Test (50 + years) 1.77 5.11 0.32 0.42

 

 

 

 

 

Table 16 indicates the estimated lease square means and standard errors for both
dependant variables (/5 and / 10). Pair wise comparisons were conducted using Bonferroni
adjustment to avoid inflation of Type I error rate.

For the first dependant variable where subjects were scored out of five, a marginal
main effect of population (control vs test) was identified on the number of correctly
recognized PWL’s (p = 0.0540). Subjects from the control population correctly
recognized more PWL’s than the test population (i.e. younger subjects did a better job at
recognizing the PWLs).

For the second dependant variable where subjects were scored out of ten, a
significant main effect of population (control vs test) was identified on the number of
correctly recognized PWL’s (p = 0.0031). The control population correctly recognized

more PWL’s than the test population.

Legibility Testing:

The Lockhart legibility instrument provides a measure of legibility based on a
legibility index, i.e. the degrees of rotation of the polarizing filters in the legibility

instrument. These readings were collected in triplicate on a total of 33 subjects ( 15

88

 

control; 18 test). Please note that the total number of subjects (control + test) for all other
response variables is 32, whereas the total number of subjects for legibility data is 33.
This is because eye tracking and subsequent recall data was not available for one subject
in the test population. A breakdown of the raw legibility data is provided in Appendix 6.

Triplicates of legibility data were averaged for each subject by PWL color
contrast and were then log transformed to meet model assumptions. The transformed
response variable was then modeled as a function of the fixed effects of population
(control vs test), label color and their 2-way interaction using a general linear mixed
model. Subject, nested within? population, was fitted as a random factor in the model in
order to recognize the appropriate experimental unit for population (control vs test) and
the blocking factor for treatment. Visual acuity was included in the model as an
explanatory covariate. Age, in its continuous form, was considered for inclusion in the
model but did not improve model fit. It was thus removed from the final model. Model
fitting and parameter estimation was conducted using the MIXED procedure of the
statistical software SAS (version 9.1, SAS Institute, Cary, NC).

Regardless of the PWL color contrast, even after adjusting for reduced visual
acuity, the test population (50+) indicated a statistically greater degree of filter , rotation
(poorer legibility of prescription warning labels) than the control population (1829) (p <
0.0001). This is consistent with previous studies performed using this method (Bix 1998;
Bix 2001), and can be explained by the physiological effects of aging that in turn cause
older subjects to require more light. These include: the yellowing of the cornea, vitreous
humor and reduction in pupil size as people age. Visual acuity was also significantly

associated with filter rotation (p < 0.0016); people with poorer visual acuity scores

89

required a greater degree of rotation of the polarizing filter to read text, such that one unit

increase in visual acuity resulted in a 20% cumulative increase in filter rotation. This

relationship was apparent in all PWL color contrasts for both populations (control and

test).

Table 17 -- Least Square Means for the averaged triplicates of Legibility Index (Data

 

 

 

 

 

 

 

for figure 24)
PWL Color Contrast $3$t§sggffe€lballtllty Lower fimrdence C 0 1111353; Limit
ndex

Black on Blue 43.129 38.1082 48.8113

Black on Red 33.338 30.275 36.711

Blue on White 24.4183 22.1747 26.8888
Black on Yellow 22.4046 20.3461 24.6714
Black on White 20.3945 18.5207 22.4579

 

 

 

 

 

90

Legibility Index (degrees of rotation)

 

Black on Blue Black on Red Blue on White Black on Yellow Black on White
Color Contrast

Figure 24 - Estimated legibility index and 95 % confidence interval for the five PWL
color contrast combinations. Letters (a,b,c) indicate statistically significant
differences)

Averaged across both populations (control and test), PWL color had a significant
impact on degrees of filter rotation (P<0.0001). Results are presented in the original
scale (05) as least square mean estimates and estimated 95% confidence interval. Post-
hoc, pair wise comparisons were performed using Bonferroni’s adjustment to avoid
inflation of type I error rate (see Table 17 and Figure 24). It can be seen from Figure 24
that the black on blue color contrast is the least legible and the black on white color

contrast is the most legible.

91

Conclusions, Limitations & Future Research

Prescription warning labels, small colorful stickers found on prescription vials,
were studied for their noticeability and legibility. Five different color contrasts (black on
white, black on red, black on blue, black on yellow and blue on white) were analyzed
using an ASL 501 bright pupil eye hacker for noticeability. This was followed by tests of
recall and recognition, and testing culminated with a test of legibility using a Lockhart
Legibility Instrument.

32 subjects were broken into two populations; a control population of 15 subjects
aged between 18 and 29 years of age and a test population of 17 subjects aged over 50
years. Data was obtained from these subjects and analyzed using a mixed model
statistical analysis. For a complete and detailed analysis of data obtained, readers are
encouraged to go through Chapter 8: Results, Analysis and Discussion.

This chapter contains an overview of the several conclusions that can be drawn
from this study. The later part of this chapter contains a summary of the various
limitations and possible sources of error in the study.

The conclusions below are explained in the context of future studies that are

envisioned in similar topics.

Noticeability and Attentive Behavior:

Study findings do not provide evidence of color contrast promoting differences in
attentive behavior on PWLs. There was not enough evidence to conclude an association
between color contrast of PWL’s and noticeability (see chapter 8). It is speculated that

expanding the sample size and repeating this study will result in similar findings.

92

All the dependant variables used to quantify noticeability indicated that attentive
behaviors are directed toward the large, white prescription label. It registered a higher
number of hits across all populations and had a higher estimated probability of being
noticed (see chapter 8). There may be features inherent to this label (For eg: large size,
higher amount of information) that promote its noticeability. These must be explored
further.

It is speculated that changing the position of the PWL itself may improve
noticeability. A cursory visual inspection of the prescription vial reveals the presence of
the white pharmacy label, regardless of the angle from which it is viewed. This is not true
in the case of the narrow, vertically positioned PWL (see Figure 7). Redesigning the
layout and positioning of both the labels may contribute to better noticeability of the
important information contained within the PWL’s. Simple tests of vial examination and
recall may reveal valuable information about these new layouts.

A future study is also envisioned where warning information is integrated into the
white pharmacy label in the form of small blurbs. This would enable studying of other
warning features that may promote noticeability like borders and symbols as opposed to
color.

The analysis of noticeability data was performed in multiple ways. At the time of
publishing of this thesis, several more techniques are being studied and investigated. It is
envisioned that a future study will compare the effects of data manipulation and analysis
(time, time/character, time/label density, etc.) on outcomes, i.e. an analysis of the

empirical methods that can be employed for labels and packaging.

93

The information processing model that was the basis for this study describes
noticeability and legibility as two distinct and separate steps in the path to warning
effectiveness. Subsequently, eye tracking and legibility testing were performed separately
and different dependant variables were used to quantify the two. It is envisioned that
future studies will integrate these two steps by design and analyze these quantities
together. Studies shall also involve creating newer models that assess warning

effectiveness.

Recall of label and vial information:

As revealed in the analyses in Chapter 8, certain elements of the label like dosage
information, patient information and drug information have a statistically higher
probability of being recalled when subjects have examined prescription vials. It is
suggested that future studies investigate recall rates of warning information when they are
integrated with dosage and patient information.

With the increase in the availability of digital printers capable of performing
variable data Operations, personalized printing of warning information together with

patient information is also a field that is worth exploring.

Recognition of labels:

As revealed in Figure 22 in Chapter 8, data supports an association between the
age and the number of labels that were correctly recognized by subjects. There must be
more exploration into the management of medical information of older consumers,

specifically with respect to warnings. As mentioned, therapeutic regimens are becoming

94

increasingly complicated and increasing the cognitive loads at a time when cognitive

ability is degenerating.

Legibility of PWL’ :

The findings in this study are consistent with previous work that was done on
legibility of printed matter. The black text on white background color contrast has had the

best legibility scores for all age groups in several previous studies.

Other salient conclusions:

While analyzing eye tracking data, there was revealed to be an association of
gender and number of hits per zone (see Chapter 8). This opens potential avenues of
research. Gender differences on attention capture on warnings is an area that could be
researched in the firture. These findings also seem to coincide with those of Godfrey et al
in establishing that female subjects have a greater tendency to look for warning
information (Godfrey, Allender et al. 1983).

Eye tracking has several potential applications in evaluating package design and
placement. Further studies are envisioned in several areas including, but not limited to
anti-counterfeiting and shelf/aisle placement.

The section below highlights some of the limitations of this study.

Limitations:

Along with wearing the ASL 501 bright pupil eye tracker,subjects were made to
sit at a special eye tracking table while placing their chin on a chin rest. This entire setup

was not a representation of the normal position/environment in which subjects would

95

examine prescription medications at home. This may have affected the results in some
way.

The labels used in this study were totally devoid of any symbols in order to make
analysis easier. The only graphical element on the entire label set was the pharmacy logo
on the large white pharmacy label. Warnings in real life are almost always seen with
accompanying symbols.

The information processing model that was presented in Chapter 1 indicates that a
four stage process needs to be followed for a warning to be effective. It was also
explained that this study would try to gain insight into noticeability and legibility of
PWL’s. Optimum noticeability and legibility of a warning do not necessarily imply that
the warning will be easy to comprehend and comply. Further studies must investigate
comprehension and compliance of PWL’s in order to obtain a complete spectrum of
information about these labels in the context of the information processing model.

When subjects arrived for this study, it is possible that they had in their minds,
information about the labels on the prescription vials in their home. Delving into concepts
like prospective (long term) and working (short term) memory, it is possible that some of
the recall and recognition responses were influenced by these reserves of information.
Drawing parallels to Morrow et al’s work on medication information schemata (Morrow,
Leirer et al. 1996), it is possible that some recall responses may he been predetermined
by these schema and not actually associated with the vials that were viewed during the

eye tracking study.

96

APPENDIX 1: CONSENT FORM

Michigan State University

School of Packaging
INSTRUCTIONS AND RESEARCH CONSENT FO_RM - Prescription vial use experiment

You are being asked to participate in a research study. The entire study should last no
more than 2 hours. In exchange for your participation, you will receive $30. You may
discontinue participation at any time and still keep the $30.

As part of this research, we will record your gender, ethnicity, educational background,
age and the number of prescription and OTC medications that you take each day. This
information will be tied to a subject number; you will not be identified by name and
your privacy will be maintained to the maximum extent of the law.

We will also test your visual acuity (20/20, 20/30, 20/40, etc.) and color blindness.
These tests will be conducted by asking you to view a series of cards and asking you to
decipher images to the best of your ability. Following these tests, you will be then
shown a card with 11 commonly used medical terms. You will be asked to say these
words. This will give us an idea of your familiarity with some terms commonly used in
the field of healthcare.

Instrument 1 ASL E yetraclrer

After these tests have been run, we will begin the eye tracking portion of the study. The
eye tracker is a very sensitive eye movement monitor, which can tell us exactly where
your eyes are looking while you are viewing a package. A beam of light that cannot be
detected by the human eye will be shone into your eye. The instrument tracks the
movement of your eye by tracking the movement of the beam.

You will wear a light (4 ounce) camera on your head (attached to a headband). Once
the eye tracking equipment is adjusted, we will prepare the eye movement equipment.
You will be asked to sit at a desk. For a period of about 5 minutes you will be asked to
sit as still as possible with your chin on the chinrest, and move nothing but your eyes.
While holding your head and body as still as possible, you will be asked to look at certain
locations in space so that the researcher can calibrate your eye’s position. After we
have calibrated the equipment, you may take your chin off of the chin rest and move
until testing begins.

During the actual experiment you will be asked to rest your chin in the chin rest, while
you examine 5 different prescription packages. You will be handed a series of five bags,

97

one at a time, and asked to remove the contents (a prescription drug vial), and do what
you normally would do if you were viewing a new medication. If you read anything on
the vial, please press the vial against the glass while doing so. At the end of this test you
will be asked to answer a short questionnaire based on the vials that you just viewed.

If you have any questions at any time please ask.
Instrument 2 Lockhart Legibility

After we have tested you for color blindness, measured your visual acuity and had you
view a series of prescription vials using the eye tracker, you will be asked to read several
labels using an instrument called the Lockhart Legibility Instrument (LLI). The LLI is a tool
that provides a numerical value for how easy or difficult a message is to read. The
purpose of this research is to investigate the legibility of a variety of package labels. It is
our hope that this will ensure that the label information provided on a variety of labels
is sufficiently legible.

You will read a label placed inside the grey box. Look into the box through the screen on
the front. As you look through the screen turn the knob on the right side of the box until
you can easily read the words on the card in the box without straining your eyes. The
operator will record the value you get for each card. Once the value is recorded, turn
the knob back to its starting position so that the screen is dark again. The operator will
put a different card in the box for you to read.

Risks and Benefits

There is little or no risk associated with this research. it is possible that you will
experience some discomfort while you are asked to remain still with your head on the
chin rest, or when the researcher adjusts the head piece. Please let us know if you do
experience any discomfort at any time. There is no direct benefit to you for participating
in this study. However, we are hopeful that we can learn what constitutes well designed
prescription labels so that this information is used to create future label designs that are
more effective.

Confidentiality

Study results will be treated in strict confidence. Your privacy will be protected to the
maximum extent allowable by law. Within these restrictions, results of this study will be
made available to you at your request. Data will be stored on a password protected
computer in the Packaging Building and in a locked file cabinet in Laura Bix’s office for a
minimum of 3 years and a maximum of 10 years. Collected data will only be stored
subject number and cannot be tied to you identity.

98

If you have any concerns or questions about this research study, such as scientific issues,
how to do any part of it, or if you believe you have been harmed because of the
research, please contact the researcher Laura Bix 517-355-4556; 153 Packaging Building
East Lansing MI 48824 bixlira@msu.edu.

If you have questions or concerns about your role and rights as a research participant,
would like to obtain information or offer input, or would like to register a complaint
about this study, you may contact, anonymously if you wish, the Michigan State
University's Human Research Protection Program at 517-355-2180, Fax 517-432-4503,
or e-mail irb@msu.edu or regular mail at 202 Olds Hall, MSU, East Lansing, MI 48824.
I voluntarily agree to participate in the Eye Tracking and label legibility study.

 

Date:

 

You will be provided with a copy of your signed consent form.

99

APPENDIX 2: RECRUITMENT ADVERTISEMENT FOR
TEST POPULATION

Prescription Vial Use Experiment

Participants wanted for research concerning prescription vial use. In exchange for your
participation, you will receive $30. The study will take no longer than 2
hours. To participate in this study:

0 You must be able to provide your own transportation to the School
of Packaging at Michigan State University

You must be over 50 years of age

You must not be legally blind

Must not wear hard contact lenses

Be willing to provide a contact phone number or email so that researchers can contact
you to remind you of your appointment

 

You are being asked to participate in a study of prescription drug use being conducted by
graduate student Raghav Prashant Sundar. As part of his Master’s Program, Raghav is
investigating the use of prescription vials. He is using two instruments to do this. Instrument l is
an ASL eye tracker. If you choose to participate, you will be hooked up to the eye tracker which
will track your eye movements as you look at 5 prescription vials. You will also be asked to
answer a brief questionnaire about the prescription drug vials you just viewed. Instrument set up
and the test itself should take no longer than 45 minutes.

Prior to testing, your visual acuity (20/20,20/30, 20/40, etc.) and a test for color blindness will be
administered. These tests involve viewing a series of cards. You will also be shown a list of
eleven health terms and asked to read these terms aloud. This will give us an idea of your
familiarity with some terms commonly used in the field of healthcare. We will also ask you to
fill out a brief survey which includes information about your race, gender, age, educational
background and the number of prescription drugs and OTC drugs that you consume per week.

The final portion of the research involves a test of the legibility of the labels that you viewed on
the vials previously. This portion of the study is done on the “Lockhart Legibility Instrument”
(LLI), an instrument that was developed at the School of Packaging. For this portion of the
assignment, you will be asked to rotate a small wheel located to your right as you are seated in
front of the instrument until the first point that you can easily read the words on the label. This
portion of the test should take no longer than 25 minutes.

If at any time you are uncomfortable with the testing or wish to discontinue the data collection
process, you may discontinue participation without penalty.

If you are interested in pursuing this opportunity, please contact Raghav Prashant Sundar at
sundarra@msu.edu or cell phone 517-898-9029 to make an appointment.

If you have questions or comments regarding this study, please contact Dr. Laura Bix, Assistant
Professor or Packaging at Michigan State University at 517-355-4556 or bixlaura@msu.edu.

100

APPENDIX 3: DATA RECORDING SHEET

Human Factors & Medical Packaging Research Group, School of Packaging
Research Study (IRB IRB# 08-246! APfl l029464)

’ Level 1: Personal Information

Subject Id: Age:

 

Ethnicity: Highest level of education:

 

Number of Prescription Drugs per weelc

 

Level 2: Eye Tracking <Please turn over for Recall Test forms(free and
guided)>

Level 3: Pre-Testing

Literacy Score:

Hold document at convenient reading distance and say. 'I want to hear you read as
many words as you can from this list. Begin with the first word in List 1 and read aloud.
When you come to a word you cannot read, do the best you can or say, ’blank’ and go
onto the next word.” (6 or less has poor health literacy)

Fat Flu Pill Allergic Jaundice Anemia Fatigue

Directed Colitis Constipation Osteoporosis

 

Visual Acuity:

I want you to hold this card at about 16 inches from your eyes and try to read the
lowest line on this card.

20I100: SDKHN OCVRZ
20I150: HCZSV DORNK
20I100: ONHDR ZKSCV
21)/80: ZVRCK NOHDS
20I60: KNCSO VHZRD
ZOISO: RVDKC SHONZ
20/40: HNDCR VSKZO
20/30: ZCONS DHRVK
20120: CKDZH RNOSV

101

Result: 20/

Color Blindness Test:
6 42 56 57
5 3 56 27

86 15 74 47

75

89

10 and above means ”normal color vision"

Level 4: Legibility instrument

 

Label Trial 1

Trial 2

Trial 3

 

Dummy 1

 

Dummy 2

 

 

 

 

 

 

 

 

 

 

Level 5: Compensation

102

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

F F F F o F F o o o F o 3
o o F F F F F F o o o o 3
o F o F F F F F o o o F «F
o F o o o o o o o o o F «F
o F F F o F F o F o o o FF
F F o o F F F F o o o o 3
o F F o o o F o F o o F a
o o F F o F F o F o o o a
o o o F F F o o o o F F F
o F o o F F F F F o F F o
o F o o F o F o F o o F m
o o o o F o o o o o o o F.
F F F F F F F F o o o o a
o F F F F F F F F o o F N
o F o o F F o F F o o o F
83.0.588ch 598.. 3260 cam... seawater—c. oEmz seawater—c. 5.65.85 8:52 «35 co 5:3:er 9560
c3862). .33 may”; mc_Em>> common. Santana Egan. 33535 95 .3832 30 _m_> Foo—as»

 

 

53:30.. 2:55:39: 5.8.. DES :— noteuouee NF 3 nets—=3.— _eh=3 2: .Fe Boer—.5 ,3 3:39. nets—5&5 - 3 ~33.

“no... .33”. .5». Sun in”. "F. x=ucoaa<

103

 

an

 

N0

 

Fm

 

on

 

mu

 

nu

 

hm

 

0N

 

mN

 

1N

 

nu

 

NN

 

..N

 

cu

 

0F

 

OOOOOOOOOOOFOOOO

POOFFOOPOOOOOOPO

POPPFOOOPOPOOOOP

POOOOOOOOv-FOFx-PO

OOOOOOOOOOOOOOCO

OOPPOOOOOOPFOOOO

e-OOFOOOOOOFOOOOO

3.

 

O

O

o

PCs-FFOFOOx-OFv-OOOI-

FOFPFFFFOPFOFPPOO

VOOFFOOOOOOOOPPPO

POOOOOOOOFI-FOFF'PO

O

FOPFOOOPOPFOFPFFP

O

O

O

2.

 

83.50220
8.8.82

 

 

59.3
.33

 

20.60

.33
ac.Em>>

 

Fxo...
uc.Em>>

 

3:33....
09on

 

oEmz
63:3...

 

5:33....
289.

 

3.88.6.5
«Loatcmen.

 

3:62
9:0

 

m.m.> .0
.3E:z

 

3.5.52...
30

 

£0.00
.m.>

 

.83..»

 

 

55:39. 2:55:39: 3...... Can... 5 notewouao NF 5 zeta—.52— 28 05 Fe 383.5. .3 3:39. 33:83.:— - a 03:.

104

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

o m o F o F o F o F F o F
o m o o o F F F o F F o «F
o m o o o F F o F o F F 9
0F c o o o o o o o o o 0 NF
9 m o o o o o F F F F F FF
w v o o o o F F o F F F 3
o m o o F o F F o o F F a
o F o o o o o o o o o F a
o m o F o o F F o F o F F
m N o F o o o F o F o o o
F. o o F o o o o o o o o m
m m o o o o o F o o F F F.
m m o o o o o F F o F o n
3 m o o o o o F F F F F N
N N o o o o o F o o F o F
on o :02 one! o E: o 26 o
3:80... 3.80... ...om gen...“ :o :00 ...o o .26 a. an“; hangs can.“ :0 e.... > canoe“
:85 s. a x35 8.85 ..one ..osm _m s. a .35 c. E
.835
eo.ae_..e>

Feet—.309 «cone 5.2.588 :o 942:. 25 3063:» .3 35598 22> ...o .53“. 352.5
cos—emcee".

 

 

 

 

 

A295 3: BEE—3.9V 8.8.. Pan:— 5 33a— 3 E. 3.. zeta—=32— .ehaeo 2: Fe 333:... ..8 3.5 ect.:weuom . 3 033—.

.3... 32.3000”. .2 Eco Sam. I m 533.?

105

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m m o o o F F F o F o F an
c F o o o o o o o o F o N."
o F o o o o o F o o o o Fm
m o o o o o o o o o o c on
v N F F o F o o F F o o N
m N o o F o F F o o o F N
e o o F o o o o o o o o N
m m o o o o o o F o F F N
v F. F o o o o F F F F o N
F N o o o o o F o o o F N
v F o F o F o F o o o o ”N
n F o F o o o o o o F o N
N F F o F F F o o o F o FN
N F F F F o F o o o o F N
m N o o F o F o o F o F 3
o F o o o o o o o o F o 3
o F. o F o o o F F F F o F
3:50.. 2.88. odum x5... .8 condo sauna each—a 3...“; 2...; 03m .8 Bur; no:
xoafi .8 02m :35 .8 3.35
x33 :35 x35 :35 xoefi ..oufi
.838
oo.aatu>

Emu—880 83¢ 8.2.539. :o 94.5.. «in 333.8 3 35:88 32> go 9.5:. 852.5
8.2833.

 

 

 

 

 

Esau .2. 228.28
8.8.— .EaEa 5 82: 82:88.. 3. 8 23a. 3 =8 8.. 8.888.— 33 3F .8 383.; .8. 8.5 82:88: - FN 838,—.

106

Appendix 6: Average Legibility Indices

Table 22 - Average of Legibility Index triplicates for all subjects. Subjects 1-15
belong to the control population. Subjects 16 to 33 belong to the test population.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

PWL color contrast
Subject Age Black on Black on Blue on Black on Black on
Red Blue White Yellow White
1 21 26.7 34.6 19.3 16.2 13.96
2 20 15.96 20.86 1 1.93 10.53 10.23
3 22 18.6 24.26 13.13 11.83 10.96
4 23 28.13 34.9 21.63 17.66 18.33
5 21 16.2 21.33 12.56 12.96 11.4
6 21 25.13 30 16.13 16.66 16.6
7 22 24 38.03 16.86 14.93 16.03
8 24 26.83 35.73 20.13 18.66 17.46
9 22 15.86 15.43 1 1.43 10.26 9.43
10 22 22.73 27.9 15.33 12.7 12.76
11 24 22.36 29.3 17.23 14.5 13
12 21 18.7 15.96 14.8 14.3 11.93
13 26 25.03 45.5 20.73 16.06 15.53
14 24 26.6 41.63 19.26 18.46 15.4
15 29 15.46 18.76 11.8 10.86 10.3
16 70 52.33 62.5 32.66 25.53 21.26
17 71 63.23 83.33 44.46 47 38.16
1 8 70 90 90 64.83 56.8 55.6
19 53 34.93 89.4 31.96 31.16 24.9
20 54 34.5 48.23 23.46 21.13 18.9
21 64 90 90 52.56 55.8 45.9
22 58 35.73 87 30.53 27.6 25.1
23 51 48.3 44.5 36.76 35.86 31.63
24 77 90 90 90 90 90
25 74 33.56 38.26 23.2 20.6 20.46
26 51 63.9 90 45.8 46.36 37.73
27 71 33.03 39.06 22.46 22.7 16.53
28 63 49.03 87.4 28.13 27.66 26.56
29 61 60.5 82.36 39.2 37.66 31
30 55 51.66 53.13 38.86 32.93 31.66
31 64 34.83 49.9 27.23 26.03 23.23
32 64 67.43 90 57 52.1 46.1
33 59 41 .33 43.33 28.66 24 24

 

107

 

BIBLIOGRAPHY

Ahlbaum, A. (2008). "Eye Tracking breaks into shopper research with major Tobii
order." Retrieved Feb 4th, 2009, from

http://www.tobii.com/all news/eye tracking breaks into shopper research with major
_tobii order.aspx?profile=l 7741 &menu=l 7757&submenu= l 7888.

Ajzen, I. and M. Fishbein (1977). "Attitude-Behavior Relations - Theoretical-Analysis
and Review of Empirical-Research." Psychological Bulletin 84(5): 888-918.

Anders, G. (2001). PIlot's Attention allocation during approach and landing: Eye and
Head-tracking research in an A330 full flight simulator. International smsium on

aviation psychology (ISAP). Columbus, OH.

Andrews, J. C. (1995). "The Effectiveness of Alcohol Warning Labels - a Review and
Extension." American Behavioral Scienfis_t 38(4): 622-632.

Asaad, W. F., G. Rainer, et al. (2000). "Task-specific neural activity in the primate
prefrontal cortex." Journal of Neurophysiology 84(1): 45 1 -45 9.

Ascione, F. J. and L. A. Shimp (1984). "The Effectiveness of 4 Education Strategies in
the Elderly." Drug Intelligence & Clinical Pharmacy 18(11): 926-931.

Aspden, P. and Institute of Medicine (U .S.). Committee on Identifying and Preventing
Medication Errors. (2007). Preventing medication errors. Washington, DC, National
Academies Press.

ASTM. (2006). "Standard Test Method for Measurement of Comparative Legibility by
Means of Polarizing Filter Instrumentation." from
http://www.astm.org/Standards/D7298.htm.

Ault, H. (2007). "A warning about prescription drug labels." Retrieved 01/13/09, 2009.

Baber, C. and J. Wankling (1992). "An Experimental Comparison of Text and Symbols
for in-Car Reconfigurable Displays." Applied Ergonomics 23(4): 255-262.

Backinger, C. L. and P. A. Kingsley (1993). Write it right: Recommendations for
developing user instruction manuals for medical devices used in home health care.
Washington, DC, US Department of Health and Human Services.

Bass, P. F., J. F. Wilson, et al. (2003). "A shortened instrument for literacy screening."
Journal of General Internal Medicine 18(12): 1036-1038.

Bates, D. W. (2007). "Preventing medication errors: A summary." American Journal of
filth-System Pharmacy 64: S3-S9.

108

Berman, A. (2004). "Reducing medication errors through naming labelling and
packaging." Journal of Medical Smptoms 28(1): 9-29.

Bix, L. L. (1998). The Effect of subject on legibility. School of Packaging. East Lansing,
Michigan State University. Masters.

Bix, L. L. (2001). Towards a performance standard for typeface legibility: The Lockhart

Legibility Instrument. School of Packaging. East Lansing, Michigan State University.
PhD.

Bix, L. L. (2003). "The effect of color contrast on message legibility." Journal of Desiga
Communication(5).

Boyd, S. T., L. C. Boyd, et al. (2006). "Medication therapy management survey of the
prescription drug plans." Journal of the American Pharmacists Association 46(6): 692-
699.

Brown, C. S., B. L. Solovitz, et al. (1988). "Short-Term and Long-Term Effects of
Auxiliary Labels on Patient Knowledge of Precautionary Drug Information." Drug

Intelligence & Clinical Pharmacy 22(6): 470-474.

Bureau, U. C. (1997). Aging in the United States: Past, Present and Future.

Cahill, M. C. (1976). "Design-Features of Graphic Symbols Varying in Interpretability."
Perceptual and Motor Skills 42(2): 647-653.

Chapman, P. R. and G. Underwood (1998). Visual search in dynamic scenes: Event types
and the role of experience in viewing driving situations. Eve Guidance in Reading and
Scene perception. G. Underwood. Oxford, England, Elsevier: 369-394.

Charness, N. and E. A. Bosman (1992). Human Factors and Age. Handbook of aging and
cogpition. T. A. Salthouse and F. I. M. Craik. Hillsdale, NJ, Lawrence Embaum
Associates: 495-552.

Crespo, A., R. Cabestrero, et al. (2007). "Visual attention to health warnings in tobacco
advertisements: An eye-tracking research between smokers and non-smokers." Studia

Psychologica 49(1): 39-51.

d'Yedwalle, G., G. Desmet, et al. (1998). Film perception: The processing of film cuts.

Eve Guidance in reading and scene perceptionz. G. Underwood. Oxford, England,
Elsevier: 357-368.

Davis, T. C., R. Michielutte, et al. (1998). "Practical assessment of adult literacy in health
care." Health Edugrtion & Behavior 25(5): 613-624.

109

Davis, T. C., M. S. Wolf, et al. (2006). "Literacy and misunderstanding prescription drug
labels." Ann_als of Internal Medicine 145(12): 887-894.

De La Fuente, J. (2006). The use of a universal design methodology for developing child-
resistant drug packaging. School of Packaging. East Lansing, Michigan State University.
Masters in Packaging.

DeJoy, D. M. (1991). "A Revised Model of the Warnings Process Derived from Value-

Expectancy Theory." Proceedings of the Human Factors Sociey 35th Annual Meeting:
1043-1047.

DeJoy, D. M. (1999). Attitudes and Beliefs. Warnings and Risk Communication. M. S.
Wogalter, D. M. DeJoy and K. R. Laughery. London ; Philadelphia, Taylor & Francis:
189 - 213.

DeJoy, D. M. and K R. Laughery (1999). Preface. Warnings and Risk Communication.
M. S. Wogalter, K. R. Laughery and D. M. DeJoy. Philadelphia, PA, Taylor and Francis:
xiii-xvi.

Dockeray, F. C. (1910). "The span of vision in reading and the legibility of letters."
Journal of Educational Psychology 1: 123-131.

Dombrowski, R. C., A. A. Schuna, et al. (1991). "Auxiliary label use in pharmacists in
South Carolina." Journ_al of Pharmacy Technolggy 7: 64 - 70.

Dubay, W. H. (2004). Principles of Readability. Costa Mesa, CA.

Duchowski, A. T. (2003). Eye tracking methodology : theory and practice. New York,
Springer.

Edwards, M. L. and N. C. Ellis (1976). "Evaluation of Texas Driver Improvement
Training-Program." Human Factors 18(4): 327-333.

Ells, J. G., R. E. Dewar, et al. (1980). "An evaluation of six configurations of the railway
crossbuck sign. " Ergonomics 23: 359-367.

FDA (1999). Final rule - Over the Counter Human Drugs Labelling requirement.
FDA (2004). Food Allergen Labeling and Consumer Protection Act. USA.

Ferguson, J. M., R. Discenza, et al. (1987). "Increasing the odds of patient compliance
through prescription warning labels." Jouflrl of health can: marketing 7(1): 37-46.

Fischer, P. M., J. W. Richards, et al. (1989). "Recall and Eye Tracking Study of
Adolescents Viewing Tobacco Advertisements." Jama-Journal of the American Medical
Association 261(1): 84-89.

110

Fischhoff, B. (1977). "Cognitive Liabilities and Product Liability." Journal of Products
Liabiliy 1(3): 207-219.

Flesch, R. (1948). "A new readability yardstick." Journal of Applied Psychology 32: 221-
233.

Franklin, D. (2005). The Consumer: And Now, a Warning about Labels. The New York
Times. New Yor.

Gamlin, P. and D. Tweig. (1996). "CRI: A Combined Visual Display and Eye Tracking
System for High-Field fMRI Studies " Retrieved February 4th, 2009, from
http://www.nsfgov/alardsaarch/showAward.do?AwardNumber=9634422.

Godfi'ey, S. S., L. Allender, et al. (1983). Warning Messages: Will the consumer bother
to look? Human Factors Society 27th Annual Meeting, Santa Monica, CA.

Godfrey, S. S. and K. R. Laughery (1984). "The biasing effects of product familiarity on
consumers' awareness of hazard." Proceedings of the Human Factors SocietLZSLh
Annual Meeting: 388-392.

Goldhaber, G. M. and M. A. Deturck (1989). "A developmental analysis of warning
signs: the case of familiarity and gender." Proceedings of the Human Factors Society
33rd Annuai Meeting.

Gryfebecker, B. M., H. J. Segal, et al. (1989). "Effect of Auxiliary Prescription Labels on
the Elderly Ambulatory Patients Drug Knowledge." Dicp-Lha Annals of Pharmacotherapy
23(4): 324-329.

Harris, J. L., W. A. Rogers, et al. (1998). "Written language comprehension in younger
and older adults." Journal of Smech Langu_age and Hearing Research 41(3): 603-617.

 

Hellmich, N. (2008). Aging population making more visits to the doctor's. USA Today.

Hill, R. D. (1993). "Predictors of Reading Performance in College-Educated Older
Adults." Educational Gerontology 19(1): 21-30.

Hunn, B. P. and T. A. Dingus (1992). "Interactivity, Information, and Compliance Cost in

a Consumer Product Warning Scenario." Accident Analysis and Prevention 24(5): 497-
505.

Janynes, L. S. and D. B. Boles (1990). A bepavioral study of warning labels for copaumer
products: Perceived danger and u_se of pictoggapfi. Human Factors and Ergonomics
Society 34th Annual Meeting, Santa Monica, CA, Human Factors Society.

111

Johnston, M., A. Clarke, et al. (1986). "Facilitating Comprehension of Discharge
Medication in Elderly Patients." Age and Ageing 15(5): 304-306.

Kalsher, M. J ., M. S. Wogalter, et al. (1996). "Pharmaceutical container labels:
Enhancing preference perceptions with alternative designs and pictorials." International

Journal of Industrial Ergonomics 18(1): 83-90.

Karwowski, W., J. W. Yates, et al. (1991). Advances in industrial ergonomics and safeg
3 : proceedings of the Annual International Industrial Ergonomics and Safety Conference
held in Lake Tahoe, Nevada. 10-14 June 1991 : the official conference of the

International Foundation for Industrial Ergonomics and Safeg Research. London ;,
Taylor & Francis.

Kaufman, D. W., J. P. Kelly, et al. (2006). "Patterns of medication use in US adults and

children, 2005: A report from the slone survey." harmacoapidemiology and Drug Safeg
15: $65-$66.

Kemper, S. (1992). Language and Aging. Handbook of aging and cogpition. T. A.
Salthouse and F. I. M. Craik. Hillsdale, NJ, Lawrence Embaum Associates: 495-552.

Kline, D. W. and C. T. Scialfa (1997). Sensory and Perceptual functioning: Basic
research and human factors implications. Handbook of human factors and the older adult.
A. D. Fisk and W. A. Rogers. New York, Academic Press: 27-54.

Krugrnan, D. M., R J. Fox, et al. (1999). "Do cigarette warnings warn? Understanding
what it will rake to develop more effective warnings." Journal of Health Communication
4(2): 95-104.

Krugrnan, D. M., R J. Fox, et al. (1994). "Do Adolescents Attend to Warnings in
Cigarette Advertising - an Eye-Tracking Approach." Journalfiof Advertising Resea_r_c_h_
34(6): 39-52.

Laughery, K. R. and A. Hammond (1999). Overview. Warnings and Risk
Communication. M. S. Wogalter, D. M. DeJoy and K. R. Laughery. London ;
Philadelphia, Taylor & Francis: 3 - 12.

Lehto, M. and J. Miller (1988). "Effectiveness of Warning Labels." Journal of Products
Liabilig 11: 225-270.

Lohse, G. L. (1997). "Consumer eye movement patterns on yellow pages advertising."
Journal of Advertising 26(1): 61-73.

Luckiesh, M. and F. K. Moss (193 8). "Visibility and Readability of print on white and
tinted papers." Sight-Saving Review 8: 123-134.

112

Luckiesh, M. and F. K. Moss (1939). "Visibility and Readability of printed matter."
Journal of Applied Psychology 23: 645-659.

Madden, S. (1999). The law related to warnings. Warnings and Risk Communication. M.
S. Wogalter, D. M. DeJoy and K. R. Laughery. London ; Philadelphia, Taylor & Francis:
3 - 12.

Mazzullo, J. M., L. Lasagna, et al. (1974). "Variations in Interpretation of Prescription
Instructions - Need for Improved Prescribing Habits." J arna-Joumal of the America
Medical Association 227(8): 929-931.

McLean, R. (1980). The Thames and Hudson manual of typgmphy. London, Thames
and Hudson.

Meyer, B. J. F., C. J. Young, et al. (1989). Memory improved : reading and memog
enhancement across the life span through strategic text structures. Hillsdale, N.J.,
Lawrence Erlbaum Associates.

Miller, J. A. (1978). Labeling research : the state of the art. Cambridge, Mass, Marketing
Science Institute.

Morrell, R. W., D. C. Park, et al. (1989). "Quality of Instructions on Prescription Drug
Labels - Effects on Memory and Comprehension in Young and Old Adults."
Gerontologist 29(3): 345-354.

Morrow, D. G., V. O. Leirer, et al. (1996). "Medication instruction design: Younger and
older adult schemas for taking medication." Human Factors 38(4): 556-573.

NIH (2006). Dramatic Changes in US. Aging Highlighted in New Census, NIH Report,
National Institutes of Health.

Otsubo, S. M. (1988). "A behavioral study of warning labels for consumer products.

perceived danger and use of pictographs. " Proceedings of the Human Factors Socieg
32nd Annual fiMeetm: 536- 540.

Paine, F. A. and Institute of Packaging. (1963). Fundamerytals of packag1_n' g. London,
Blackie.

Pharmex. (2009). "Top 40 Pharmex English and Spanish Warning Labels." Retrieved
03/21/2009, 2009, from
http://wwwphannex.com/PharmacyWarningLabels/WamingLabels/wleng.asp.

Pietrowski, J. (1993). Development of a methodology to quantify package and label
legibility. School of Packaging. East Lansing, Michigan State University. Masters.

113

Pyke, R. L. (1926). Remrt on the legibility of print. London,, H. M. Stationery Off.
[printed by Johnson, at the University Press.

Rayner, K., Ed. (1992). Eye movements and visual cogpition: Scene percaption and
Reading. New York, NY, Springer-Verlag.

Rehder, T. L., L. K. Mccoy, et al. (1980). "Improving Medication Compliance by
Counseling and Special Prescription Container." American Journal of Hospital Pharmacy
37(3): 379-385.

Report (2003). Public Health and Aging: Trends in Aging -- United States and
Worldwide, Morbidity and Mortality Weekly Report.

Rogers, R. W. (1983). Cognitive and physiological processes in fear appeals and attitude
change: A revised theory of protection motivation. Socigal psychophysiology. J. Cacioppo
and R. Petty. New York, Guilford. 153-176.

Rogers, W. A., G. Rousseau, et al. (1999). Maximizing the effectiveness of the warning
process: Understanding the variables that interact with age. Processing of medical

information in aging patients : cogpitive and human factors perspectives. D. C. Park, R.
W. Morrell and K Shifren. Mahwah, NJ, Lawrence Erlbaum: ix, 332 p.

Rosbergen, E., M. Wedel, et al. (1990). Analysing visual attention to repeated print
advertising using scanpath theory, University Library Groningen, SOM Research School.

Salthouse, T. A. (1991). Theoretical persmctives on cogm'tive aging. Hillsdale, N.J., L.
Erlbaum Associates.

Sanford, E. C. (1888). "The Relative legibility of small letters." American Journal of
Psychology 1: 402-435.

Shannon, C. E. (1948). "A Mathematical Theory of Communication." Bell System
Technical Journal 27(3): 379-423.

Sheahan, N. F., D. Coakley, et al. (1993). "Ocular Microtremor Measurement System -
Design and Performance." Medical & Biological Engineering & Computing 31(3): 205-
212.

Silver, N. C. and M. S. Wogalter (1990). "Strength and Understanding of signal words in
elementary and middle school students." Proceedings of the Human F actora Society 35th
Annual Meeting: 580-583.

Snodderly, D. M., I. Kagan, et al. (2001). "Selective activation of visual cortex neurons

by fixational eye movements: Implications for neural coding." Visual Neuroscience
18(2): 259-277.

114

Soroka, W. and Institute of Packaging Professionals. (1995). Fundamentals of packaging
technology. Hemdon, Virginia, Institute of Packaging Professionals.

Sumner, F. C. (1932). "Influence of color on legibility of copy." Journal of Applied
Psychology 16: 201-204.

Swindell, J. A. (1999). Measuring visual search reaction time and accuracy for a product
warning label frmction of icon, color, column and vertical placement. Raleigh, North
Carolina State University.

Tinker, M. A. (1946). "Validity of Frequency of Blinking as a Criterion of Readability."
Journal of Exm’mental Psychology 36(5): 453-460.

Tinker, M. A. (1963). Legibilig of print. Ames,, Iowa State University Press.

United States Court of Appeals, F. C. (1974). "Terri Lee DREISONSTOK, an infant, by
her mother and next friend, Catherine A. Dreisonstok, and Catherine A.Dreisonstok,
Appellees, (Vs). VOLKSWAGENWERK, A.G., a/k/a Volkswagenwerk
Aktiegesellschaft, a foreign corporation, and Volkswagen of America, Inc., a New York
corporation, Appellants." Retrieved January 9th, 2009, from
http://bull_(.resource.org/courts.gov/c/F2/489/489.F2d. l 066.73-1074.html.

Weiss, A. P. (1917). "The focal variator." Journal of Exm’ ental Pachology 2: 106-
113.

Wiederholt, J. B., J. A. Kotzan, et al. (1983). "The Effectiveness of Auxiliary
Prescription Labels - a Pilot-Study." Drug Intelligence & Clinical Pharmacy 17(3): 216-
221.

Williams, M. V., R. M. Parker, et al. (1995). "Inadequate Functional Health Literacy
among Patients at 2 Public Hospitals." J ama-Joumal of the American Medical
Association 274(21): 1677-1682.

Wilson, J. and L. Hogan (1983). "Readability Testing of Auxiliary Labels." Drug
Intelligence & Clinical Pharmacy 17(1): 54-55.

Wogalter, M. S., D. M. DeJoy, et al. (1999). Organizing Theoritical Framework
Warnings and Risk Communication. M. S. Wogalter, D. M. DeJoy and K. R. Laughery.
London ; Philadelphia, Taylor & Francis: 15-23.

Wogalter, M. S., D. M. DeJoy, et al. (1999). Preface. Warnings and Risk
Communication. M. S. Wogalter, D. M. DeJoy and K. R. Laughery. London ;
Philadelphia, Taylor & Francis: xiii.

115

Wogalter, M. S. and T. A. Dingus (1999). Methodology for evaluating behavioral
intentions. Warnings and Risk Communication. M. S. Wogalter, D. M. DeJoy and K. R.
Laughery. London ; Philadelphia, Taylor & Francis: 53-77.

Wogalter, M. S., G. A. Fontenelle, et al. (1985). Behavioral effectiveness of warnings.
Human Factors Society 29th Annual Meeting, Santa Monica, CA.

Wogalter, M. S., S. S. Godfrey, et al. (1987). "Effectiveness of Warnings." Human
Factors 29(5): 599-612.

 

Wogalter, M. S., S. W. Jarrard, et al. (1994). "Influence of Warning Label Signal Words
on Perceived Hazard Level." Human Factors 36(3): 547-556.

Wogalter, M. S. and K. R. Laughery (1996). "Warning! Sign and Label Effectiveness."
Current Direction§ in Psychological Science 5: 33-37.

Wong, 1. C. K. (1999). "Readability of patient information leaflets on antiepileptic drugs
in the UK." Seizure-European Journal of Epilepsy 8(1): 35-37.

Wooding, D. (2002). Fixation Maps: Quantifying Eye-Movement Traces. Eye Tracking
Research and Applications (ETRA ). New Orleans, LA.

Wright, P., P. Creighton, et al. (1982). "Some Factors Determining When Instructions
Will Be Read." Ergonomics 25(3): 225-237.

Yarbus, A. L. (1967). Eye movements and vision. New York,, Plenum Press.

Young, S. L. and D. Lovoll (1999). Intermediate processing stages: Methodological
Considerations for Research on Warnings. Warnings and Risk Communication. M. S.
Wogalter, D. M. DeJoy and K. R. Laughery. London ; Philadelphia, Taylor & Francis:
x111.

Young, S. L. and M. S. Wogalter (1990). Judgements of Hazard, Risk and Danger: Do
they Differ? Human Factors and Ergonomics Society 34th Annual Meeting, Santa
Monica, CA, Human Factors Society.

Zuccollo, G. and H. Liddell (1985). "The Elderly and the Medication Label - Doing It
Better." Age and Ageing 14(6): 371-376.

116