ABSTRACT
Dissertation Title: “NOTHING TO BE WORRIED ABOUT YET”:
PERCEPTIONS OF COLORECTAL CANCER
AMONG INDIVIDUALS BELOW THE AGE OF
50.
Nora M. Mueller, Doctor of Philosophy, 2019
Directed by: Professor Barbara A. Curbow, Department of
Behavioral and Community Health
National guidelines recommend regular screening for colorectal cancer (CRC)
begin at age 50. Recent research however showed rates rising among adults below the
age of 50 despite declining overall rates of CRC. Screening guidelines have not been
updated to reflect new findings. Younger individuals may not be aware of their CRC risk
and providers may be unaware of the need to screen younger patients.
An online survey of individuals between the ages of 24 and 45 was conducted in
April and May of 2018 using Amazon© Mechanical Turk. Measures collected included
perceived CRC risk and susceptibility, screening knowledge, and understanding of
screening results. Analysis I examined these variables by generational cohort. Analysis II
randomized participants to receive one of two fictional colonoscopy result reports, the
first being the standard of care report provided by a medical center in Maryland and the
second a modified report containing a “gist” synthesis of findings. Mixed methods
examined participants’ interpretations of these reports. Analysis III randomized
participants to receive one of two statements describing trends in the rates of colon cancer
to examine participants’ ability to discern false and true information.
Analysis I found Gen-Xers more likely to report higher perceived risk (general
cancer and CRC) and severity (general cancer) and be aware of current CRC screening
guidelines. Both groups displayed a lack of awareness of CRC screening methods.
Analysis II showed that individuals who received the gist report correctly rated the report
as more serious and the importance of follow-up higher than those who received the
standard medical report. Analysis III revealed a lack of knowledge among participants to
differentiate between correct and incorrect statements regarding colon cancer trends, with
participants defaulting to accepting the validity of the statement.
Findings revealed gaps in knowledge of recommended screening activity and
available screening methods. Providers should be prepared to discuss available screening
options and accurate risk information; in addition, providers should discuss with patients
which medical information they read online and remind individuals to take a more critical
stance. CRC screening reports that contain a synthesis of findings may help patients plan
next steps accordingly.
“NOTHING TO BE WORRIED ABOUT YET”: PERCEPTIONS OF COLORECTAL
CANCER AMONG INDIVIDUALS BELOW THE AGE OF 50.
By
Nora M. Mueller
Dissertation submitted to the Faculty of the Graduate School of the
University of Maryland, College Park, in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
2019
Advisory Committee
Professor Barbara A. Curbow, Chair and Advisor
Associate Professor Faika Zanjani
Assistant Professor Evelyn C. King-Marshall
Professor Robert S. Gold
Associate Professor Xin He, Dean’s Representative
© Copyright by
Nora M. Mueller
2019
To Steve, Finn, and Celia,
who are everything
Acknowledgements
I have been very lucky to have had the unwavering support of talented (and patient)
women mentors, especially Rani Elwy, Barbara Bokhour, and Barbara Curbow. Very
little would have been possible without their encouragement and thoughtful guidance.
I would like to sincerely thank my committee members for their consistently thorough
feedback. A special thanks to Faika Zanjani for sticking with me despite moving away
and for always returning with doughnuts.
Thank you also to my BCH PhD cohort, who made this an enjoyable process, and the
previous cohorts, who were willing to share their experiences and wisdom.
Finally, thanks to Steve for everything (but especially the coffee), and my parents, for
always checking in.
ii
Table of Contents
List of Tables and Figures ..............................................................................................................i v
Chapter 1: Introduction .................................................................................................................... 1
1.1 Overview ................................................................................................................................ 1
1.2 Frameworks ........................................................................................................................... 3
1.3 Justification for Current Study ............................................................................................... 5
1.4 Aims and Hypothesis ............................................................................................................. 6
1.5 Definition of terms ............................................................................................................... 11
Chapter 2: Literature Review ......................................................................................................... 15
2.1 Colorectal Cancer................................................................................................................. 15
2.2 Health Literacy and Cancer Care ......................................................................................... 25
2.3 Fuzzy Trace Theory ............................................................................................................. 27
2.4 Mechanical Turk in Research .............................................................................................. 30
2.5 Conclusion ........................................................................................................................... 33
Chapter 3: Generational Differences in Colorectal Cancer Knowledge ........................................ 34
Abstract ...................................................................................................................................... 34
3.1.Introduction .......................................................................................................................... 35
3.2. Methods .............................................................................................................................. 37
3.3. Findings .............................................................................................................................. 40
3.4. Discussion ........................................................................................................................... 50
3.5. Conclusion .......................................................................................................................... 51
Chapter 4: “Nothing too outrageous” v. “I am highly concerned”: Variation in descriptions of
medical information when presented differently. .......................................................................... 52
Abstract ...................................................................................................................................... 52
4.1. Introduction ......................................................................................................................... 53
4.2. Methods .............................................................................................................................. 56
4.3. Analysis .............................................................................................................................. 59
4.4. Results ................................................................................................................................. 60
4.5. Discussion and Conclusion ................................................................................................ 68
Chapter 5: Awareness of Colorectal Cancer Trends Among Individuals Below the Age of 50 .... 70
Abstract ...................................................................................................................................... 70
5.1. Introduction ......................................................................................................................... 71
5.2.Methods ............................................................................................................................... 72
5.3.Findings ............................................................................................................................... 74
5.4.Discussion ............................................................................................................................ 78
Chapter 6: Summary ...................................................................................................................... 80
6.1 Overview of Findings and Implications ............................................................................... 80
6.3 Strengths and Limitations .................................................................................................... 86
6.4 Future Research Directions .................................................................................................. 87
Appendix A: Methods .................................................................................................................... 91
Appendix B: IRB Documentation ................................................................................................ 103
Appendix C: Questionnaire .......................................................................................................... 104
Appendix D: MTurk Interface ..................................................................................................... 130
References .................................................................................................................................... 136
iii
List of Tables and Figures
Figure 1A: Adapted Conceptual Model for the Uptake of Colon Cancer Screening..................…..4
Figure 1B: Colorectal Cancer Incidence and Mortality Trends …………………………..….…….5
Figure 2A: Diagram of the Colon and Rectum …………………………………………......….….15
Figure 2B: Common Types of Cancer in the United States……………..……………..….…...…..18
Figure 2C: Colorectal Cancer New Cases and Estimated Deaths…………..………….……..…...18
Table 1.1: Participant Demographics………………………………………………….…...……..41
Table 1.2: Group Comparisons of Continuous Variables……………..…………….….…………44
Table 1.3: Group Comparison of Categorical Variables….………..…...…………..……………..45
Table 1.4: Stepwise Linear Regression Models…………….……....………………..……….…...47
Table 2.1: Participant Demographics ………………………………………………..……………61
Table 2.2: Bivariate Analysis of Report Impressions by Report Type………………...…..………63
Table 2.3: Qualitative Responses to "If this colonoscopy report belonged to a loved one, and he or
she asked you what it meant, what would you say?...........................................................………...65
Table 3.1: Participant Demographics…………………...…………………………………....…...75
Table 3.2: Bivariate Comparison of Groups…………………....………………………..………..76
Table 3.3: One-way ANOVA of Statement Received and Response Type…………..……...…. ..77
Table 3.4: Two-Way ANOVA of Statement Received and Correct/Incorrect Response.............. 78
Table 3.5: Multiple Linear Regression Assessing Predictors of Confidence in Response………. 78
Figure A1: Boxplots of Health Literacy by Masters’ Qualification................................................97
Figure A2: Boxplots of Numeracy by Masters’ Qualification…………………………….………97
Figure A3: Map of participant location created using the latitude and longitude variable………..98
Figure A4: Interaction Effects of Statement Received and Correct Identification of
Results……………………………………………………………………………………….…..100
Table A1: Questionnaire constructs and corresponding question numbers…...............................102
iv
Chapter 1: Introduction
“Compared with adults born circa 1950, those born circa 1990 have double the risk
of colon cancer and quadruple the risk of rectal cancer.” (Rebecca L. Siegel et al.,
2017)
1.1 Overview
Increasing colorectal cancer (CRC) screening has been a focus of public health for
several years, hinged upon the fact that with regular screening, CRC can be prevented
through the removal of precancerous polyps, or treated before spreading (Baron et al.,
2008; Holt et al., 2013; Makoul et al., 2009; Sarfaty & Wender, 2007). If detected with
Stage I CRC, individuals have a 5-year survival rate of 92%; comparably, the Stage II
survival rates are between 62% and 87%, depending on the size and location of tumors,
while stage III rates are between 53% and 89%. If detected at Stage IV, CRC has likely
metastasized to other parts of the body, and the 5-year survival rate drops to only 11%
(ACS, 2017a). The importance of screening and early detection to reduce CRC morbidity
and mortality therefore cannot be overstated.
The number of people receiving CRC screenings worldwide has increased,
leading to a 3% overall decrease in CRC incidence in the United States between 2001 and
2010 (Edwards BK 2010). However, recent findings from a large retrospective cohort
study found rates of colon and rectal cancer in the United States increasing in adults ages
20-29 (2.4%, 3.2%, respectively) and 30-39 (1.3%, 3.2%) from the mid-1980s through
2013. Further findings included increases in rates of colon cancer (1.3%) and rectal
cancer (2.3%) for adults age 40-49, and increased rates of colon cancer (0.5%) and rectal
cancer (2.3%) in adults ages 50-54 over the same period. This is in opposition to trends in
older cohorts; adults age 55 and older experienced a marked decline in rates of CRC in
the last three decades (Rebecca L. Siegel et al., 2017).
1
While some of the increase CRC in younger populations may be due to better
detection because of an increase in the use of colonoscopy for screening purposes,
researchers posit that recent changes in exposure to carcinogenic risk factors are more
likely causes (Edwards et al., 2010a; Siegel, DeSantis, & Jemal, 2014; Rebecca L. Siegel
et al., 2017). Factors that have been linked to CRC include excess body weight, high
consumption of meat, low consumption of fiber, alcohol and tobacco use, and low
physical activity, possibly through increasing inflammation in the bowel and colon, or
through more complex genetic interactions (Aune et al., 2011; Haggar & Boushey, 2009;
Johnson et al., 2013; Rebecca L. Siegel et al., 2017).
For the average individual, current guidelines issued by the National Cancer
Institute endorse regular screening for CRC beginning at age 50 and continuing until age
75, at intervals determined by family history, comorbidities and previous findings. For
individuals with a family history or irritable bowel syndrome, it is recommended that
screening begin earlier, as determined by a gastroenterologist. Fecal occult blood test,
sigmoidoscopy, colonoscopy, virtual colonoscopy, and DNA stool test are all common
tests that vary by length of time it takes to administer, what part of the colon is examined,
and the amount of preparation required, among other factors (ACS, 2017a). While the
colonoscopy procedure has high sensitivity and specificity, it is often regarded as
invasive, risky, or embarrassing by patients (Leggett & Hewett, 2015). Patient
preferences and situational context should be accounted for when determining which type
of screening method will be used; however, research shows that colonoscopy is the most
prescribed screen and considered the most reliable test, resulting in frequent
recommendations for colonoscopy screening without proper patient consultation
2
(Sanguinetti et al., 2015). Subsequently, patient understanding of the colonoscopy
procedure is suboptimal, which has important implications for adherence to screening
recommendations and follow-up treatment (King-Marshall et al., 2016).
1.2 Frameworks
Conceptual Model
This dissertation uses the Conceptual Model for the Uptake of Colon Cancer
Screening (May, Whitman, Varlyguina, Bromley, & Spiegel, 2016) as a springboard for
examining patient factors impacting knowledge of CRC screening guidelines and
understanding of screening results (Figure 1). This conceptual model is based on a 2002
report by the Institute of Medicine, and includes patient characteristics, provider
practices, and system-level factors that predict or act as barriers to screening (IOM,
2003). This model has been adapted to fit the scope of this study and expanded to include
factors that facilitate or bar understanding of screening guidelines and results. Included in
the analysis are patient variables found in the model, which have been underexplored in
previous research, as well as new patient variables that have been shown to be significant
in previous research by the research team. These variables are shown in Figure 1A, with
starred variables indicating inclusion in this study.
Theory
This study is further informed by a Health Literacy (HL) and Numeracy framework, as
well as Fuzzy Trace Theory (FTT). Health literacy, or the ability of patients and health
professionals to provide, communicate, understand and process health information, is a
complex issue that is not well understood. Numeracy, which is a person’s ability to
3
understand and work with numbers, is considered a part of health literacy, as the
variables are associated with negative health behaviors (Kiechle, Bailey, Hedlund, Viera,
& Sheridan, 2015; Yee & Simon, 2014), and yet the impact of HL, and especially of
numeracy, on health outcomes remains unclear (Berkman et al., 2011).
Fuzzy Trace Theory (FTT) is a psycholinguistic theory developed by Valeria
Reyna and Charles Brainerd and has been used to examine medical decision-making and
information processing (Reyna, 2008, 2012; Reyna & Brainerd, 1995). The theory
contains two main constructs, verbatim representations and gist representations
(usually shortened to verbatim and gist). Individuals process new information using one
Figure 1A: Conceptual Model. Adapted from May FB, Whitman CB, Bromley EG,
Spiegel DM (2016).
of these two representations: verbatim, which is a literal and “surface-level” memory of
the information, or gist, which is an affective, meaning-based memory. Gist
representations are more likely to be remembered, and can be recalled and used to
interpret new information, unlike verbatim representations, which are usually quickly
4
forgotten and offer little assistance in decision-making (Reyna, Nelson, Han, &
Dieckmann, 2009). Based on previous work by the University of Maryland Department
of Behavioral and Community Health Colorectal Cancer Research Group, led by Dr.
Barbara Curbow, this study proposes to further examine the relationship between HL,
numeracy, and individuals’ ability to form gist interpretations of medical information.
1.3 Justification for Current Study
Figure 1B displays increases in CRC incidence and mortality among individuals
below the current recommended age of screening initiation in the United States. A recent
projection predicted continued increases in CRC incidence rates among adults under the
Figure 1B: Colorectal Cancer Incidence and Mortality Trends by Age and Sex,
United States, 1975-2014 (Sigel, et al. 2017).
5
age of 55 through 2040 (Araghi et al., 2018). When CRC is detected individuals below
the age of 50, it is usually found at an advanced stage and mortality rates are higher
(Campos, 2017). It is therefore critical to begin discussing CRC in younger populations
and possible actions to optimally prevent this pathology. For this to occur, an
understanding of how individuals in this population conceptualize CRC is vital.
Furthermore, it is important that these individuals are able to proceed with screening and
are provided them with appropriate medical information throughout the screening
process. To begin to address these issues, the analyses within this dissertation examined
how individuals below the current recommended age of regular CRC screening perceive
CRC risk and severity, as well as the importance of CRC screening. Furthermore, I
examined how these individuals understand results from a CRC screening report when
presented to them, and if small alterations to this report can lead to a better understanding
of findings. Finally, I tested individuals’ awareness of current CRC trends, their ability to
distinguish between false and true information regarding these trends, and the factors
associated with certainty of the accuracy of their response.
1.4 Aims and Hypothesis
This study recruited a sample of individuals below the current recommended age of
CRC screening using Mechanical Turk, an Amazon.com subsidiary company, to examine
differences in knowledge of CRC screening and understanding of colonoscopy screening
results. One aim was to examine differences in CRC knowledge and perceptions of risk
and severity between generational cohorts, that is, individuals classified as Millennials
and those belonging to Generation X. The Pew Research Center, acknowledging that
generational groupings are imprecise, defines Millennials as individuals who were born
6
somewhere between the early 1980s and mid- 1990s; for the purposes of this study, I
included older Millennials, or individuals who were between the ages of 25 and 35 at the
time of data collection. Generation X is the cohort prior to Millennials, and birth years for
GenXers generally range from the late 1960s through the early 1980s; for this study, I
included younger GenXers, or individuals who were between the ages of 36 and 45 at
data collection (Pew Research Center, 2014). These age ranges were chosen to examine
perceptions of susceptibility and understanding of screening among individuals who
currently are not of age of recommended CRC screening, as CRC incidence is increasing
for this population. The two age ranges allowed for a comparison between those who are
approaching the age of currently recommended regular CRC screening (36-45 years of
age), compared with those who are farther removed from this responsibility (25-35 years
of age).
Variables bolded in Figure 1A were examined to discern variation between and
within generational cohorts, as were their relationship with the dependent variables
(knowledge of CRC screening, understanding of screening results). Individual level
variables that were included in the original Conceptual Model for the Uptake of Colon
Cancer Screening, but were not included here, are fear of CRC diagnosis, fear of the
colonoscopy procedure, diet and cost, as these were considered to be either not yet
applicable to the age groups I examined (fear of CRC, fear of procedure, indirect cost of
procedure) or outside of the scope of the current survey (diet). New additions to the
individual-level factors were age (and generational cohort), race, numeracy and perceived
severity of CRC, which are variables of interest based in the literature. The below aims
and hypotheses guided this work.
7
Aim 1
Compare differences in perceived risk to CRC and severity of CRC diagnosis
between age groups.
H1: GenXers report higher levels of perceived risk to CRC, compared with
Millennials.
H2: Millennials are more likely to think a CRC diagnosis will result in death,
compared with GenXers.
Theory/Evidence-based Justification: Recent research has found rates of colorectal
cancer increasing in adults below the age of 54 (Rebecca L. Siegel et al., 2017); however,
national guidelines still set the average age of CRC screening commencement at 50, with
variations based on family history and lifestyle factors (ACS, 2017a). This may lead to
the perception that CRC is a disease related to aging, and younger individuals may not be
aware of the risk of CRC (Rabin, 2017). Furthermore, inadequate health literacy and low
numeracy is associated with higher levels of perceived risk (Reyna et al., 2009): I
therefore posit that Millennials are less aware of factors surrounding CRC and therefore
overestimate the severity of a diagnosis. These hypotheses will be tested using Questions
2-13 (health literacy and numeracy), Question 33 (perceived susceptibility to CRC) and
Question 34 (perceived severity of CRC diagnosis) of the questionnaire.
Aim 2
Assess knowledge of CRC screening methods and guidelines between age groups.
H3: GenXers score higher on measures of knowledge of CRC screening,
compared with Milllennials
Theory/Evidence-based Justification: Adults between the ages of 36-45 are
approaching the age at which guidelines suggest regular treatments should begin.
Individuals with a family history of CRC are recommended to begin screening at age 40,
while those without a family history of CRC are counseled to begin screening at age 50.
8
Conversations with providers about CRC family history, lifestyle factors, and screening
preferences optimally occur early in the patient-provider relationship to determine the
optimal age to begin regular screening. Low health literacy and numeracy have been
shown to be barriers to understanding medical recommendations and information (Seurer
& Vogt, 2013). African Americans shown to be less likely to undergo CRC screening or
diagnostics, and suffer from higher rates of CRC incidence and mortality, and at a
younger age, than Caucasians or Asians (R. Williams et al., 2016). Research suggests that
adequate health literacy may serve as a buffer (Reyna et al., 2009). These hypotheses will
be tested using Question 14 and 18 of the questionnaire (Knowledge of CRC Screening)
and Questions 2-5 (Perceived Health Literacy).
Aim 3
Assess participants’ relationship with a provider and the impact the relationship
with a provider has on CRC screening knowledge and perceptions of susceptibility
H4: GenXers more often report having a provider whom they see regularly,
compared with Millennials.
H5: Participants with an established relationship with a provider display more
CRC knowledge compared with participants without an established relationship.
Theory/Evidence-based Justification: The patient-provider relationship has been shown
to impact numerous facets of healthcare (Ha & Longnecker, 2010; Zolnierek & Dimatteo,
2009), including the likelihood of undergoing CRC screening: Research showed that
patients who felt they had enough time with a provider and received sufficient
information about the procedure were more likely to receive CRC screening (Carcaise-
Edinboro & Bradley, 2008). Similarly, patients with a more established relationship with
a provider are more likely to discuss their CRC risk and when to begin screening. Thus, I
posit that older participants, having had more time in to engage in the medical system, are
9
more likely to have an established relationship with a medical provider. This hypothesis
will be examined using questions 14 and 18 (CRC screening knowledge) and 74 of the
questionnaire (Provider Relationship).
Aim 4
Assess whether participants who receive a colonoscopy report formatted to contain
Gist information report different perceptions about the report than participants
who receive a standard colonoscopy report.
H6: Participants who receive Report 1 (verbatim) provide lower scores of
perceived seriousness, compared with participants who receive Report 2.
H7: Participants who receive Report 1 (verbatim) rate the importance of follow-
up lower than participants who receive Report 2.
H8: Participants who received Report 2 (gist) are more likely to agree that there is
a plan of care in place for the patient, compared with participants who receive
report 1.
Exploratory: How would patients describe the reports to their loved ones, and
how does this differ by report type?
Theory/Evidence-based Justification: Report 2 presented the participant with “gist”
information, i.e. a synthesis of pertinent information and follow-up steps that was
designed to required less interpretation and placed less cognitive burden on the patient.
Fuzzy-Trace Theory (FTT) guided the design of Report 2 (Reyna, 2008, 2012; S. G.
Smith et al., 2015). Information presented in medical jargon has been shown to cause
confusion and anxiety among patients (Herber, Gies, Schwappach, Thürmann, & Wilm,
2014; Schnitzler et al., 2017). These hypotheses will be tested using Question 23 of the
report, as well as Question 25 (If this report belong to a loved one, and he or she asked
you what it meant, what would you say?).
Aim 5
H9: When presented with a false statement about colon cancer trends,
participants with low HL are more likely to believe it is true, compared with
those with high HL
10
1.5 Definition of terms
Colorectal Cancer (CRC) Colorectal cancer is any cancer that begins in the colon or
rectum. These cancers are often grouped together because they share similar features and
similar methods of screening, diagnosis and treatment. Most colorectal cancers begin as a
growth on the inner lining of the colon or rectum, called a polyp.
Colonoscopy An outpatient procedure to examine the colon and rectum for precancerous
or cancerous polyps or lesion, or to monitor existing conditions. During a colonoscopy, a
small, flexible tube with a camera at the end is inserted through the colon to allow
providers to see the inside. Any polyps that are found can be removed and retrieved for
possible biopsy. The procedure itself takes about 30 minutes to complete. However,
preparation for the procedure usually requires fasting or a clear diet for one or more days
before, in addition to the consumption of a laxative in either pill or liquid form.
Colonoscopies are usually performed by a gastroenterologist and require sedation.
Complications are rare, but include perforation, bleeding and pain (ACS, 2017a).
Health Literacy (HL) One of the earliest definitions of HL described it as a patient’s
ability to apply literacy skills to health related materials like prescriptions, appointment
cards, medicine labels, and directions for follow-up care (Parker, Baker, Williams, &
Nurss, 1995). Since then, health literacy has had a variety of definitions (Nutbeam, 2000;
M. S. Wolf et al., 2012), but at their most simple, these define HL as the ability of
individuals to obtain, process, and understand health information. Definitions like these
are somewhat controversial, as they seem to place the burden of obtaining and
understanding health information on the patient. Some researchers therefore argue that
HL is actually the purview of health providers and health systems, in that they are
11
responsible for ensuring the availability of health materials, proper communication of
information, and access to services which allows for optimal HL in the patient (Lambert
et al., 2014; R. Rudd, AT, & D, 2012; R. E. Rudd, 2013). The Calgary Charter on Health
Literacy offers a definition of HL that attempts to incorporate the role of the health
professional:
Health literacy allows the public and personnel working in all health-related
contexts to find, understand, evaluate, communicate, and use information. Health
literacy is the use of a wide range of skills that improve the ability of people to act
on information in order to live healthier lives. These skills include reading,
writing, listening, speaking, numeracy, and critical analysis, as well as
communication and interaction skills (Coleman, Kurtz-Rossi, & McKinney, 2009).
HL is measured through either skill-based assessment or self-report. When asked
about existing health literacy measures, 88% of those who work with these tools stated
that they are inadequate and do not capture the complexity of health literacy; 90% stated
that new tools to measure health literacy are needed (Pleasant, 2014).
Numeracy Montori and Rothman (2007) define numeracy as follows:
The specific aspect of literacy that involves solving problems requiring
understanding and use of quantitative information. Skills include understanding
basic calculations, time and money, measurement, estimation, logic, and
performing multistep operations. Most importantly, numeracy also involves the
ability to infer what mathematic concepts need to be applied when interpreting
specific situations.
Even though numeracy is often seen as a part of literacy and health literacy, only one
measure exists that combines the two into a single score (The Test of Functional Health
Literacy in Adults, or TOFHLA). Numeracy generally lags behind health literacy:
national surveys indicated that approximately 36% of American adults have inadequate
numeracy skills (Reyna et al., 2009). While health literacy is correlated with educational
12
level, no such association exists for numeracy (Rothman, Montori, Cherrington, &
Pignone, 2008).
Fuzzy Trace Theory )FTT Fuzzy-trace theory is a dual-process model of memory,
reasoning, judgment, and decision making that contrasts with traditional memory
approaches in that it states that memorization does not equal knowledge. Rather, the
theory introduces the construct of “Gist representation”, which is the titular fuzzy trace
(Blalock & Reyna, 2016).
Gist This a main construct of Fuzzy Trace Theory, first proposed by Valerie Reyna and
Charles Brainerd (Reyna & Brainerd, 1995). It is also called gist representation or gist
memory, and is defined as the bottom-line, summative meaning. Gist memories can be
used to inform new situations, unlike verbatim memory. Gist stands in opposition to
verbatim.
Verbatim This is the second main construct of Fuzzy Trace Theory, and defined as the
word-for-word representation of a memory. It is considered more superficial than Gist, is
more easily forgotten and does not extend to new learning. Verbatim is also called
verbatim memory or verbatim representation.
Perceived Risk A subjective assessment of risk of developing a health problem. This is a
central construct of the Health Belief Model, which states that individuals who perceive
themselves to be at high risk of developing a health problem are more likely to engage in
behaviors to reduce their risk, compared with individuals who perceive themselves at
lower risk.
Perceived Severity This is another central construct of the Health Belief Model and
13
refers to the degree people deem a particular disease or condition serious. This includes a
weighing of the cost and consequences of treating or not treating the disease.
14
Chapter 2: Literature Review
2.1 Colorectal Cancer
2.1.1 Overview
Colorectal cancer (CRC) is the name given to cancers in any of the four parts of the colon
(the ascending colon, including the cecum, the transverse colon, the descending colon,
and the sigmoid colon; see
Figure 2A for reference)
or rectum. These cancers
are grouped together
because of shared etiology
and attributes. Colon is
another name for the large
intestine, colloquially
referred to as the bowel
(or, in combination with
Figure 2A: Diagram of the colon and rectum. From Centers for
Disease Control: the small intestine, the
https://www.cdc.gov/cancer/colorectal/basic_info/what-is-
colorectal-cancer.htm bowels), which leads to
CRC often being called bowel cancer. Anal cancers, despite their anatomical proximity,
are not classified as CRC because they originate from different cell types and do not
develop in ways similar to CRC (ACS, 2017a, 2017b).
CRC begins with the formation of pre-cancerous polyps in the lining of the
colon or rectum. When cancer forms within a polyp, it can begin to grow into the walls of
the colon or rectum, and from there spread to nearby lymph nodes and more distant parts
of the body. Not all polyps, however, are precancerous, and not all precancerous polyps
15
develop into cancer. The most common type of polyps are adenomatous polyps, or
adenomas. Approximately 10% of these adenomas develop into adenocarcinoma, a type
of CRC that accounts for 96% of all cases (ACS, 2017a).
2.1.2 Risk of CRC
The lifetime risk of CRC is approximately 5% (Sag, Selcukbiricik, & Mandel, 2016)c.
Worldwide, colorectal cancer is the fourth leading cause of death from cancer, and
incidence rates are projected to increase (Subramanian et al., 2016). The American
Cancer Society (ACS) describes an increase in the global incidence from 1992 to 2005
among adults between 20 and 49 years, specifically a 3.5% increase per years among men
and 2.9% per year in women. In the United States, the ACS identified a consistent
incidence increase from 1998 to 2007 (Rebecca L. Siegel et al., 2017), and similar trends
are seen worldwide (2007).
2.1.3 Risk Factors
Several risk factors for CRC have been identified, most notably familial risk. Heritable
causes are estimated to account for anywhere between 5% and 30% of colorectal cancers.
Most of these cases do not belong to a known inherited syndrome (Campos, 2017),
meaning that the genetic link to family history and development of CRC or pre-cancerous
polyps is not established. An individual’s risk of CRC increases two to three times with
the existence of a first-degree relative with CRC, compared to those without relatives
with CRC. This risk increases as the number of relatives with CRC increase.
Furthermore, an individual with a history of CRC themselves has an increased risk of
further CRC incidence (Campos, 2017; Johnson et al., 2013).
16
While a family history of colorectal cancer or inflammatory bowel disease is one
significant risk factor for CRC, lifestyle variables likely interact with these to account for
many other risk and protective factors (Johnson et al., 2013) and are the focus of most
behavioral interventions (Edwards et al., 2010a). The most cited life-style related risk
factors include obesity, cigarette smoking, heavy alcohol use, and consumption of red
meat; protective factors include increased levels of fruit and vegetable consumption and
physical activity (Johnson et al., 2013). One study found that switching rural South
Africans from their traditional high fiber, low-fat diet to a western-style diet high in fats
and low in fiber lead to inflammation in the colonic mucosa in about two weeks (O'Keefe
et al., 2015). This supports a classic study from 1999, which found that the risk of CRC
increased sharply within one generation of families moving from Japan to the United
States, likely because of an increase in the consumption of red meat (Marchand, 1999).
However, more recent studies have challenged the idea of a Western diet as fundamental
to CRC, especially in younger populations, and instead call for more thorough
investigations of genetic causes (Deen, Silva, Deen, & Chandrasinghe, 2016).
2.1.4 Burden of CRC
In 2002, colorectal cancer comprised 9.4% of the global cancer burden in both sexes and
was most frequent in North America, Australia, New Zealand, and parts of Europe,
resulting in a widespread belief that colorectal cancer is disease of the Western lifestyle
17
(2007). In the early
1950s, CRC was a
leading cause of cancer
death in the United
States (Siegel et al.,
2014). In 2016, CRC
ranked as the second
Figure 2B: SEER 2016: Common Types of Cancer in the
leading cause of cancer United States
deaths behind lung and bronchus cancer, and fourth most common cancer diagnosed
(Figure 3A). In the same year, CRC was estimated to account for just under 135,000 new
cancer cases and approximately 50,000 of all cancer deaths in the United States (Figure
4A). There has been a steady overall decline over the past decades in both CRC incidence
and mortality, most recently at a rate of approximately 3% per year from 2004-2013. This
has been attributed to monumental public health efforts, including screening, detection,
and treatment, as well changes in lifestyle factors such as a general decline in smoking
and alcohol consumption (Edwards et al., 2010b).
However, this decline is not consistent
across age-groups: Sigel, et. al (2017) analyzed
colorectal cancer trends in SEER data from 1974
through 2013 and found that while, as expected,
rates declined in individuals age 55 and older, the
Figure 2C: Seer 2016: Colorectal
rates of colon cancer incidence increased by 2.4% Cancer New Cases and Estimated
Deaths
per year in adults age 20-29, and by 1% in adults
18
age 30-39, between 1985 and 2013. Furthermore, around 1995, rates among individuals
age 40-49 and those 50-54, which had previously been declining, began to increase at a
rate of 1.3% and .5%, respectively. Rectal cancer rates increased even more steeply; from
1974-2013, rectal cancer increased by 3.2% per year in adults age 20-29, and, beginning
in 1980, at the same rate for adults age 30-39. Adults age 40-49 and 50-54 experienced a
rise in rates of 2.3%, beginning in the 1990s. The authors attribute these rising rates to
changes in behavioral factors, such as increased sedentary lifestyles, poor diet and
obesity.
Furthermore, the overall decline in CRC is not uniformly distributed across
populations and reflect larger social inequity in the US. Research shows that the burden
of CRC is greater among African Americans and those of lower socioeconomic status.
This is thought to be the result of several factors: behavioral risk factors that result in
earlier development of CRC, limited knowledge about the importance of regular
screening, a lack of access to preventive services, difficulty taking time off of work for
medical visits, or cultural differences in how testing is viewed (Clouston et al., 2017).
2.1.5 Financial Cost of CRC
Patients in the US face staggering costs for medical care. Cancer care may be especially
costly, as treatments can be lengthy and require the consultation of numerous specialists.
The AHRQ estimated that the direct medical costs for cancer in the US in 2014
were $87.8 billion, and the cost of CRC alone was previously estimated to be around $6
billion annually (AHRQ, 2016; Redaelli, Cranor, Okano, & Reese, 2003).
A case study of a Medicare patient receiving treatment for Stage IIb CRC found
that the total cost of treatment over four months was $124,057, of which the patient paid
19
approximately $8573. This expense covered treatment, including surgery and adjuvant
chemotherapy, but did not include follow-up care and out-of-pocket costs for additional
screening (ACS, 2017c). It furthermore did not include the indirect costs of cancer care,
such as time missed from work, transportation to and from the medical facilities, the
worry and stress associated with cancer diagnoses and treatment, the burden on friends
and family, and the impact on overall health.
A cost-effectiveness analysis for the five most common CRC screening methods
found that the cost of any of the methods, when compared to no screening at all, was
acceptable by the standards of an industrialized nation, as the number of life-years saved
justified the cost. Yet when individual screening methods were compared against each
other, no optimal screen emerged in terms of cost (Lansdorp-Vogelaar, Knudsen, &
Brenner, 2010).
2.1.6 Screening
If found early, the 5-year survival rate is over 90%; however, only about 35% of CRC is
diagnosed at Stage 1, and as described above, diagnosis is often at a later stage for
younger individuals (Campos, 2017; Taggarshe, Rehil, Sharma, Flynn, & Damadi, 2013).
For individuals of average CRC risk, the American Cancer Society recommends
screening by colonoscopy every ten years beginning at age 50, or screening with a
flexible sigmoidoscopy or barium enema every five years, also beginning at age 50
(ACS, 2017a). However, as early as 2011 calls have been made to lower this to age 40
(D. M. Davis et al., 2011)
Individuals considered to be at higher than average risk for CRC are those with a
family or personal history of CRC or polyps, or a personal history of chronic
20
inflammatory bowel disease, namely ulcerative colitis or Crohn’s disease. The most
common symptoms of CRC are blood in stool, abdominal pain, weight loss, alteration in
bowel habit, mucus in stool, a feeling that the bowel is not empty, and tiredness (Yardley,
Glover, & Allen-Mersh, 2000). Individuals with these risk factors or symptoms are
advised to speak with a gastroenterologist to determine an individualized screening
schedule (ASGE, 2014; Levin et al., 2008).
Black males are disproportionally affected by CRC, leading to recommendations
to start screening earlier and conduct it more frequently in this population. The American
Society for Gastrointestinal Endoscopy (ASGE) recommends that African Americans
begin screening at age 45, as both incidence and mortality are higher among African
Americans than any other population (ASGE, 2014)
As of 2015, at least 40% of individuals who were of age to begin CRC screening
were not concordant with guidelines, meaning that they had either not yet initiated
screening or were not getting screened at regular intervals. Further research found that a
provider recommendation to undergo screening was the most influential factor in
determining whether or not someone has received screening for CRC (Alberti, Garcia,
Coelho, De Lima, & Petroianu, 2015).
Several CRC screening methods exist. In the United States, the five most common
screens are: 1. Fecal Occult Blood test (FOBT), which is an at-home test that checks for
blood in feces; 2. Stool DNA test, which is an at-home test that detects certain genetic
changes associated with colon cancer; 3. Barium enema, which is an X-ray exam that can
detect abnormalities in the large intestine; 4. Flexible sigmoidoscopy, which consists of a
flexible tube inserted into the anus to examine the rectum and lower colon; and 5.
21
Colonoscopy, which uses a similar tube to examine the rectum and entire colon, and is
often considered the gold standard for detecting and treating CRC because of its ability to
both detect and treat CRC through the immediate removal of precancerous polyps (ACS,
2017a).
A method not yet prevalent in the United States but part of recommended regular
screening in other developed countries is the Fecal Immunochemical Test (FIT). The one-
sample FIT is easier to administer than the traditional FOBT, which requires three
samples, and as a result increases adherence to screening steps (Hassan et al., 2012;
Issaka, Avila, Whitaker, Bent, & Somsouk, 2019). FIT was also demonstrated to have
fewer false negatives compared to FOBT(Mousavinezhad, Majdzadeh, Akbari Sari,
Delavari, & Mohtasham, 2016) and exhibit moderate sensitivity, high specificity and high
overall diagnostic accuracy (Lee, Liles, Bent, Levin, & Corley, 2014).
2.1.7 CRC Screening in Younger Populations
The question of whether or not young people are being overlooked in screening has been
raised and should be considered. 60% of young patients are diagnosed with stage 3 or 4,
compared with patients over the age of 50, who are mostly diagnosed at stage 1. Most
younger patients had been symptomatic for over six months at the time of diagnosis
(Taggarshe et al., 2013). As a result, suggestions for educational and preventive
interventions focusing on younger cohort have been made. Additionally, providers are
cautioned to adopt a more suspicious attitude towards CRC symptoms in younger
populations (Campos, 2017). Lowering the screening age has been proposed, though
implications of this, such as cost and survival rates, have not yet been adequately
assessed and there currently is no consensus on this recommendation (Deen et al., 2016).
22
2.1.8 Colonoscopy
According to the World Gastroenterology Association (WGA) (2007), CRC
screening, compared with screening for other cancers, is particularly challenging, as
reflected in current low screening rates in most countries where there is a high risk for
colorectal cancer. Colorectal cancer screening is complex and presents with multiple
points at which decisions are required. To begin, there are multiple choices available for
screening, each requiring a different level of involvement on the part of the patient. Some
require requires sedation and a partner or caregiver (for example, for a colonoscopy
procedure, the patient is required to bring along a person responsible for transporting
them home). Further steps involve patient-provider conversations regarding the
availability and feasibility of screening, patient acceptance of the method, financial
coverage, overcoming barriers such as fear and discomfort, and appropriate follow-up.
The WGA states that for screening programs to be successful, these steps must occur
along a continuum, with failure at any point detrimental to the screening process.
Colonoscopy remains the most widely prescribed and most known CRC screen
(Sanguinetti et al., 2015). Research has identified several barriers to getting a
colonoscopy, including cost, access to health care, communication by providers, low
education and income, fear and embarrassment, inability to take time off work, and being
unable to find someone to accompany the patient to the procedure (B. Curbow, King-
Marshall, Mueller, Sultan, & George, under review; B. A. Curbow et al., 2015; Jones,
Devers, Kuzel, & Woolf, 2010; M. S. Wolf, Baker, & Makoul, 2007). Patient with
inadequate health literacy are especially prone to encountering these barriers, and face
additional obstacles such as confusion during medical consultations, inability to complete
23
medical forms or embarrassment during clinical encounters, all of which can lead to
medical non-compliance (M. S. Wolf et al., 2012).
Most CRC diagnoses are after evaluation by colonoscopy, and colonoscopy can
be viewed as a pivotal event in the process of preventing, detecting, and treating
colorectal cancer. If caught early through screening colonoscopy, colorectal cancer can
be prevented through the removal of precancerous polyps or detected at an early, easily
treatable stage (ACS, 2017a). Because it allows for screening, diagnosis and treatment
(through the removal of polyps) at the same time, colonoscopy is the most widely
recommended screen for CRC. Research that examined provider discussions about CRC
screening methods with patients found that providers consider colonoscopy to be the
most significant test for CRC screening and diagnosis, and discuss it most with eligible
patients (84% of the time, compared with FOBT at 49.4% of the time, and
sigmoidoscopy at 34.1% of the time) (M. S. Wolf et al., 2007). This is problematic
because numerous barriers to colonoscopy exist, and presenting other, less invasive CRC
screening methods to patients may result in increased adherence to screening
recommendations (Basch, Basch, Zybert, & Wolf, 2016; Klabunde et al., 2005).
Presenting other methods may also result in decreasing inequalities in CRC mortality, as
barriers to colonoscopy are compounded for the African American population (Atkin et
al., 2010; Clouston et al., 2017). However, one study found that conversations about
numerous CRC screening options left patients confused, and resulted in less adherence to
screening guidelines when compared with individuals who only discussed one option
(Jones, Vernon, & Woolf, 2010). This implies that there is not only a need to present
patients with the multiple options available to them, but to also optimize how information
24
is presented and to provide continued counseling through the decision-making process.
2.2 Health Literacy and Cancer Care
The Centers for Disease Control and Prevention (CDC), Health and Human Services
(HHS) and the Institute of Medicine (IOM) define health literacy as the ability to obtain,
process, and understand basic health information and services so as to make appropriate
health decisions {CDC, 2013}.
Health literacy levels have been associated with patient understanding of
screening procedures (Cho, Plunkett, Wolf, Simon, & Grobman, 2007), and research
indicates that health literacy is a critical factor in communication across cancer care (T.
C. Davis, Williams, Marin, Parker, & Glass, 2002). Health literacy has been shown to be
associated with health knowledge, adherence to medication regimen, and treatment
guidelines (Giaj Levra, Cuniberti, Rava, Vietti, & Sciascia, 2012; Miller, 2016; Muir et
al., 2006; Murphy et al., 2010; Rust & Davis, 2011), and numerous interventions have
succeeded in increasing health literacy among participants (Muir et al., 2006; Ntiri &
Stewart, 2009; Whitley, Smith, & Vaillancourt, 2013).
However, the mechanisms through which health literacy impacts health outcomes
are not yet defined (Al Sayah, Majumdar, Williams, Robertson, & Johnson, 2013). While
some propose that health literacy acts to improve health outcomes through increasing
adherence to provider recommendations, others see this view as too narrow and argue
that health literacy is fundamental in the empowerment of patients, for example by
increasing individual ability to overcome functional barriers to accessing health care
(Nutbeam, 2000). In such a view, increasing health literacy has implications not only for
improved health care and outcomes, but for improvements in other areas requiring critical
25
thinking and decision making, such as cognitive information representation.
Nutbeam (2000) identified three hierarchical levels of health literacy through
which individuals progress; 1. Basic or functional health literacy, which are everyday
reading and writing skills; 2. Communicative or interactive literacy, which interacts with
social context and allows individuals to extract personal meaning out of communication;
and 3. Critical literacy, which, again combined with social skills, allows for the use of
critical thinking and analysis to exert greater control over life and health. These
classifications, rather than being individual endpoints, lie on a spectrum. Individuals
move along the spectrum through cognitive development and exposure to different
messages, though this movement is mediated by personal and social skills (Nutbeam,
2000). Patients on the lower end of the health literacy spectrum report poorer health
status, have less understanding about medical conditions and treatment, may have
increased hospitalization rates, and have poorer performance on healthcare tasks (M. V.
Williams, Davis, Parker, & Weiss, 2002). Low health literacy may limit patients’
understanding of cancer screenings, symptoms, staging, and risks and benefits of
treatment options (Cutilli & Bennett, 2009), and compared to patients with high health
literacy, those with low health literacy take a less active role in health decision-
making (Adams, Smith, & Ruffin, 2001). While patients with low health literacy have
reported being familiar with CRC and CRC screenings, research indicates they may not
understand the benefits of early detection or the concept of cancer prevention (T. C.
Davis et al., 2001). Low or limited health literacy skills have also been found to be
associated with less knowledge about CRC and CRC screening and with more reported
barriers to completing FOBT and colonoscopy (N. B. Peterson, Dwyer, Mulvaney,
26
Dietrich, & Rothman, 2007). Davis (2001) found that patients age 50 and over were
familiar with but had little knowledge about CRC and the digestive system. Specifically,
they were unfamiliar with cancer-related terms such as “polyp”, “tumor”, “growth” or
“lesion” (T. C. Davis et al., 2001; T. C. Davis et al., 2002).
Patient understanding of screening guidelines have been examined, and tools to
increase this have been proposed (Allen et al., 2012; Hanoch, 2004). Allen, et. al.
illustrated the high confusion around revised breast cancer screening guidelines in 2009,
and called for the development of communication aides and shared decision-making tools
for providers to alleviate patient confusion (Allen et al., 2012). However, improving
patient understanding of test results has not been the focus of interventions, even though
research has tied patient confusion and misunderstanding of results to non-compliance
with recommended follow-up care (Jones, Devers, et al., 2010; Jones, Vernon, et al.,
2010). Previous research showed that almost 30% of patients who arrive for a
colonoscopy procedure are unable to correctly state the reason for the procedure, and
many gave incorrect descriptions of the anatomy involved, often stating that the
procedure was meant to examine the stomach rather than the colon (King-Marshall et al.,
2016). Since misunderstanding and non-adherence is linked with demographic indicators,
the need for tools to improve patients’ ability to interpret cancer test results is of vital
importance if we aim to eliminate health disparities created by suboptimal education and
lower socioeconomic status.
2.3 Fuzzy Trace Theory
Fuzzy Trace Theory (FTT) was developed as a grand theory in psychology in the
late 1980s and early 1990s to better understand sophisticated forms of human information
27
processing, especially the relationship between reasoning and memory (Reyna &
Brainerd, 1995). One early tenet was that mature reasoning uses synthesized, “gist”
interpretations of materials, which are formed by individual emotion, education, culture,
experience, worldview, and level of development, rather than memories of exact content
(“verbatim” representation) to make decisions or draw conclusions. This stood in contrast
to the belief of the time that memory capacity was positively associated with cognitive
reasoning skills. This idea was based on then-recent experiments that showed that adults
and children recalled what they learned from an experiment, rather than the actual
activity or text that was presented, and the term “Fuzzy Trace” was coined to describe the
representation used for reasoning (Bjorklund & Buchanan, 1989). FTT applies to written
as well as numerical material; findings suggest that most people, when presented with
quantitative information, translate it to a qualitative form, or gist representation. For
example, percentages are often translated to words like “most” or “less than half” for ease
of interpretation and decision-making.
The implications of FTT for medical information processing and decision making
are vast. FTT may explain why communicating precise information about risk may not be
effective in preventing certain behaviors, and why some health materials are more
effective in communicating a message than others. Implications for informed consent
exist. According to FTT, if a patient recalls an informed consent conversation verbatim,
i.e. repeats the facts or numbers presented, it is still unlikely that true consent is being
given, because the patient has not demonstrated a gist understanding, i.e. bottom-line
interpretation of how the procedure is likely to affect him or her (Reyna, 2008).
2.3.1 FTT and Health Literacy
28
FTT has been used to enhance patient understanding (increasing gist representations) of
CRC screening methods. Medical information is inherently presented in a verbatim
format, rather than a gist-based one, despite indications that patients prefer the latter.
Smith, et. al supplemented the usual information with a gist-based leaflet and tested
patient understanding of CRC screening information using simple true/false questions,
and revised the questionnaire until patients achieved an acceptable knowledge level
(Smith et al., 2013). A follow-up randomized controlled trial to test the effect of the
pamphlet on screening intention had limited findings, and the authors concluded that
while screening knowledge was increased, a ceiling effect limited the intervention’s
impact on screening intention. It was noted, however, that individuals with low health
literacy were more likely to read the supplementary gist pamphlet than the standard
pamphlet provided medical information, indicating that these individuals are searching
for more understandable information (Samuel G. Smith et al., 2015).
2.3.2 Numeracy and Fuzzy Trace Theory
Numeracy has been an under-explored construct in health care, although numbers feature
prominently in medical information. Reyna et. al. (2008) questioned the unspoken
assumption that patients understand and benefit from numerical health information, and
went further to question that adequate health literacy implies adequate numeracy. The
authors return to Fuzzy Trace Theory to suggest that like verbatim representations, which
are literal representations of information but contain little meaning and are therefore not
useful in medical decision making, the ability to “know” numbers, that is, be aware of
percentages of chance or risk, does not necessarily equate to the ability to use that
information to make informed decisions. As this research is still in its infancy, it is
29
suggested that further research on numeracy include examining the relationship between
numeracy and risk perceptions (Reyna et al., 2009).
The authors, in addition to other researchers, have furthermore tied inadequate
numeracy to an overestimation of risk and posited that this heightened risk perception
leads to avoidance of advantageous health behaviors such as screening, thus impacting
health outcomes (Gurmankin, Baron, & Armstrong, 2004; Lipkus & Peters, 2009).
Lipkus et. al. summarize the implications of these findings as follows:
The main challenge for public health and health care is to educate the individual as
well as create methods of communicating quantitative information that are
understood, accepted and used by the public to encourage adaptive health behaviors
and decisions. (Lipkus, 2009)
2.4 Mechanical Turk in Research
Mechanical Turk has risen in popularity among researchers since its inception in
2005 due to the ease and low cost of recruiting large numbers of participants (also called
Worker or Turkers), to complete tasks (also called Human Intelligence Tasks, or HITs)
released by researchers (also called Requesters). It is estimated that hundreds of academic
papers yearly rely on MTurk data (Ipeirotis, 2010a). Several assessments of MTurk data
exist and found that relative to other convenience samples, Turkers are generally more
diverse than traditional convenience samples, especially college surveys, which are
known to be homogenous in terms of age, education, life experience and political
engagement (Henry 20018). Furthermore, participants response to experimental tasks in
ways similar to other large samples, displaying accuracy within 10% of traditional
surveys (Bentley, Daskalova, & White, 2017; Cassese, Huddy, Hartman, Mason, &
30
Weber, 2013). These findings, combined with the exceptional cost saving offered by the
reduction in recruitment time and materials, have propelled MTurk to the forefront of
online crowdsourcing.
Amazon restricts exact demographic information about Turkers and is not
forthcoming about how Turkers are classified. For example, MTurk conveys a “Master
Qualification” on Turkers with many completed tasks and high approval ratings, but the
exact mechanism that allow for conveyance of the Master Qualification is unclear.
Requesters pay an additional fee (approximately $.50 per participant) to release their
HITs only to those with Master Qualifications. Amazon charges between 20%-40% on
top of payments to Turkers as a fee. Additional qualifications, such as specifying the
years of birth or marital status of participants, incur additional fees averaging between
$.25 and $.50 per participant. One can also specify the country of residence of
participants, but there is speculation about how this and other demographics are verified
and updated. In a 2016 blog post (available here:
https://forums.aws.amazon.com/message.jspa?messageID=717590), Amazon confirmed
that the location of US residents is not updated, and Turkers discuss that they themselves
are not able to update their location on MTurk, despite attempts to do so.
Since Amazon does not reveal or accurately track demographic information about
Turkers, researchers have taken it upon themselves to collect these data. The most
interesting collection of MTurk statistics belongs to Panos Ipeirotis at NYU, who posts
MTurk data on his blog Behind Enemy Lines: A Computer Scientist at a Business
School. Here he presents data on the number of tasks that are posted and completed each
day, as well as the size and variability of the MTurk worker pool. He also launched the
31
Mechanical Turk Tracker, which provides a daily synthesis of Mechanical Turk activity
and demographic make-up (Difallah et al., 2015 {Ipeirotis, 2010 #276; Ipeirotis, 2010a).
Ipeirotis describes the MTurk population as consisting of anywhere between 100,000 and
200,000 Turkers, with 50% of the worker population changing every 12 months.
Demographically, workers are primarily self-reported residents of the United States
(between 50%-80%) and India (between 20%-40%), with United States Turkers being on
average younger, more educated and more likely to be female than the general US
population. Other researchers have found these same trends and report Turkers as notably
younger and more ideologically liberal than the public (Berinsky, Huber, & Lenz, 2012).
Workers in the United States were found to use MTurk as a source of supplemental
income, while those in India viewed it as a primary means of financial support (Ipeirotis,
2010b).
Critiques abound for the MTurk platform itself. To post a task, Requesters are
required to build their own interfaces and workflow systems using HTML, CSS, or
JavaScript. This has been cited as a barrier to accessing MTurk and an unnecessary
source of potential data error (Difallah et al., 2015). To illustrate this, the code used to
post the HIT information for this task is posted in Appendix D. Furthermore, worker
exploitation in terms of the dismal pay allotted to some HITs have been reported in the
media (Semuels, 2018). A 2016 Pew Research Center survey found that 25 percent of
workers who earned money from online job platforms like Mechanical Turk and similar
sites did so because no other work was available, leading researchers to question the
ethics of using MTurk when workers are limited in their agency (Pittman & Sheehan,
2016).
32
As a final critique, researchers warn that by filtering all experiments through digital
mediums, hands-on experience with (and therefore first-hand knowledge of) the
participant is lost, which may turn out to be a detriment to social science (D. Peterson,
2015).
2.5 Conclusion
As stated above, a main concern that arises from the literature is that rates of CRC
are rising in adults below the age of recommended screening. The present research
examined perceptions of risk and severity of CRC diagnosis, understanding of CRC
screening methods, and interpretation of CRC screening results among individuals below
the age of screening. Health Literacy and Numeracy were examined to determine if these
variables influence perception and should be the focus of intervention. Subsequent
recommendations for health interventions to reduce CRC or pre-cancerous conditions in
this population are presented. To address previous critiques of Amazon MTurk, workers
were paid an average of $15.00/hour for completing the online questionnaire.
33
Chapter 3: Generational Differences in Colorectal Cancer Knowledge
Abstract
Introduction Current guidelines recommend that regular screening for colorectal cancer
(CRC) begin at age 50; recent research however showed that in the US, rates of CRC are
decreasing for those 55 and above but are increasing among adults below the age of 50.
The mechanisms behind the rise in CRC rates among younger adults are unclear, and
guidelines for screening have not been updated to reflect new findings, despite calls to do
so.
Methods 624 individuals under the recommended age of CRC screening participated in
an online survey collecting information about perceptions of risk and severity of CRC
and knowledge of current screening guidelines and available methods of screening.
Participants were divided into generational cohorts and group differences were assessed.
Multiple linear regression models examined possible predictors of ratings of risk, severity
and knowledge.
Results Findings indicate skewed perceptions of risk and a lack of awareness of CRC
screening guidelines and available screening methods among all participants. Those
classified as belonging to Generation X reported higher rates of perceived risk of CRC
and cancer in general, compared with individuals classified as Millennials. Having an
established relationship with a medical provider and high Health Literacy were positively
associated with high perceived importance of screening for both CRC and cancer in
general.
Conclusion Individuals below the age of 50 may not be aware of the risk of CRC.
34
Providers should be aware of the increased risk to younger patients and the need to screen
symptomatic patients. Providers should furthermore be prepared to discuss accurate risk
information and available screening methodologies. Continuous relationships with
medical providers may aid in addressing low rates of CRC screening.
3.1 Introduction
Overview Increasing colorectal cancer (CRC) screening has been a focus of public health
for decades, hinged upon the fact that with regular screening, CRC can be prevented
through the removal of precancerous polyps or treated before spreading (Baron et al.,
2008; Holt et al., 2013; Makoul et al., 2009; Sarfaty & Wender, 2007). If detected with
Stage I CRC, individuals have a 5-year survival rate of 92%; comparably, Stage II
survival rates are between 65% and 87%, depending on the size and location of tumors,
and stage III survival rates are between 53% and 90%. If detected at Stage IV, CRC has
likely metastasized to other parts of the body, and the 5-year survival rate drops to only
12% (ACS, 2017a). The importance of screening and early detection to reduce CRC
morbidity and mortality therefore cannot be overstated.
CRC Risk The number of people receiving CRC screenings has increased, leading to a
3% overall decrease in CRC incidence in the United States between 2001 and 2010
(Edwards, 2010). However, recent findings from a large retrospective cohort study found
rates of colon and rectal cancer increasing in adults ages 20-29 (rates of colon cancer
increased by 2.4%, and rates of rectal cancer increased by 3.2% in this age group) and
30-39 (1% for colon and 3.2% for rectal cancer) from the mid-1980s through 2013.
Further findings included increases in rates of colon cancer (1.3%) and rectal cancer
(2.3%) for adults age 40-49, and increased rates of colon cancer (0.5%) and rectal cancer
35
(2.3%) in adults ages 50-54 over the same period. This is in opposition to trends in older
cohorts, where adults age 55 and older experienced a marked decline in rates of CRC in
the last three decades (Rebecca L. Siegel et al., 2017). A projection of cancer-incidence
rates predicted continued increases in CRC incidence rates among adults under the age of
55 through 2040 (Araghi et al., 2018).
CRC Screening Methods Options for screening for CRC include fecal occult blood test
(FOBT), sigmoidoscopy, colonoscopy, virtual colonoscopy, and DNA stool test; these
common tests vary by length of time to administer, section of the colon examined, and
amount of preparation required (ACS, 2017a). Knowledge of available methods for CRC
screening has been shown to be suboptimal, with providers mostly recommending
colonoscopy for CRC screening, even though patients have expressed discomfort at the
idea of this procedure and a preference for less invasive methods (Schwartz et al., 2013).
CRC Screening Guidelines For the average individual, current guidelines issued by the
National Cancer Institute endorse regular screening for CRC beginning at age 50 and
continuing until age 75, at intervals determined by family history, comorbidities and
previous findings. For individuals with a family history of CRC or irritable bowel
syndrome, it is recommended that screening begin earlier, as determined by a
gastroenterologist. Based on the more recent findings, however, the American Cancer
Society made a qualified recommendation to begin colorectal cancer screening at age 45,
for those individuals at average risk, and continue through age 75 (A. Wolf et al., 2018).
The recommendations of the United States Preventive Services Task Force (USPSTF)
and American Medical Society (AMS) remain to begin CRC screening at age 50. It is
unclear how individuals under the current age of screening understand screening
36
guidelines.
Current Study For this study, we sought to examine the perceived risk, severity, and
importance of screening for colorectal cancer among individuals below the current age of
screening, as well as their knowledge of CRC screening methods. Using Mechanical Turk
(MTurk) we recruited a national sample of older Millennials (25-35 years of age) and
younger Generation X-ers (36-45 years of age), as rates of colorectal cancer are
increasing in both groups and yet both groups are under the age at which current
guidelines recommend regular CRC screenings. The two age ranges allow for a
comparison between those who are approaching the age of currently recommended
regular CRC screening (36-45 years of age) and those who are farther removed from this
responsibility (25-35 years of age). The data for this study was collected in April and
early May of 2018, shortly before the America Cancer Society made the recommendation
to change the age of screening initiation to 45 years of age, but after the findings of
increased rates in younger individuals had received significant media coverage.
3.2 Methods
Study Sample
Amazon Mechanical Turk was used to recruit a national convenience sample of
individuals belonging to the two age groups described above. Based on methods
employed in previous research by team members (Nguyen et al., 2016), we recruited
workers with approval ratings equal to or greater than 98% (meaning previously
completed tasks were met with approval) and with more than 5000 Human Intelligence
Tasks (HITs) completed, as this ensures high quality work across a high number of
completed tasks. All elements of informed consent, including study team contact
37
information, description of the task, estimated time of completion, and compensation
were included in the recruitment statement. Inclusion criteria were acceptance of the task,
which implied consent, residing in the United States, and being between the ages of 25
and 45 at the time of the survey. Survey completion took approximately 10 – 15 minutes.
Workers received $3.00 for participating, which equated to between $12.00 and $18.00
per hour and was above the average compensation for MTurk Workers ($6.00 per hour)
and above the current national minimum wage of $7.25 per hour. The University of
Maryland, College Park Institutional Review Board reviewed and approved this study.
Measures
Generational Cohort was based on definitions from the Pew Research Center which
categorizes Millennials as individuals who, in 2017, were between the ages of 21 and 36,
and Generation X as those aged 37-52 in 2017 (Pew Research Center, 2018). The age
range for each cohort was slightly adjusted because of MTurk restrictions in specifying
the age of participants. We ran three surveys in MTurk, setting the inclusion criteria as
those born between 1987-1991, 1982-1986, and 1972-1981, respectively (categories set
by Amazon). In order to mirror the ten-year age ranges of previous research (Rebecca L.
Siegel et al., 2017) we then grouped individuals by generational cohort, with those aged
below 36 categorized as Millennial (0) and those aged 37 and above as Generation X (1).
Despite specifying the year of birth, there were seven participants below the age of
inclusion: 23 (N=2), 24 (N=2), and 25 (N=3). These individuals were included with the
millennial cohort.
Perceived risk and severity of CRC, and perceived importance of CRC screening, were
based on the following questions, with the answer option consisting of a sliding scale
38
from 1 (not at all likely) to 100 (very likely):
How likely do you think it is that you will develop colon or rectal cancer in your
lifetime?
If diagnosed with advanced stage colon or rectal cancer, how likely do you think it
is that you would die from it?
How important is it that you are screened for colon or rectal cancer in your
lifetime?
The same questions were asked regarding cancer in general (How likely do you think it is
that you will develop any type of cancer in your lifetime, etc.).
Knowledge of colorectal cancer screening methods was assessed by presenting seven
cancer screening tests (Mammography, Colonoscopy, Pap Smear, Stool Blood Test (also
known as FOBT), Prostate-Specific Antigen (PSA), Stool DNA test, and Sigmoidoscopy)
and asking participants: “Which of these screening tests detect colon or rectal cancer?
Check all that apply.” Knowledge of CRC screening guidelines was assessed through the
following question: “At what age should someone of average risk for colon or rectal
cancer begin regular screening?” The answer was open-ended but required a numeric
entry and was coded both categorically (with 10-year age intervals) and dichotomously
(correct/incorrect, with correct = 50). Perceived health literacy (HL) was collected using
the four-item measure by Chew (Chew, Bradley, & Boyko, 2004) and coded via
previously used categories, with a score between 4 and 15 indicating low health literacy,
16-19 indicating medium (or adequate) health literacy, and a score of 20 indicating high
health literacy (King-Marshall et al., 2015).
Previous CRC screening (“Have you ever received a screen for colon or rectal cancer?”
(yes/no) and provider relationship (“Do you have a provider you see at least once a year”
(yes/no) were also included in the analysis.
39
Previous experience with Cancer was a binary variable and required at least one positive
response to the following questions: “Have you even been diagnosed with cancer?”;
“Have you ever been diagnosed with a precancerous condition?”; “Has anyone in your
family ever been diagnosed with cancer?”; “Has cancer impacted your life in any other
way?”
Statistical analysis
All statistical analyses were conducted using SPSS 25. Initial bivariate analysis, using
Chi-Square Test or Mann-Whitney U Test (to account for uneven distributions) examined
the difference between generational groups by the main variables (gender, race, income,
health literacy, previous CRC screening, previous experience with cancer, perceived risk
and severity of cancer in general and CRC specifically, and perceived importance of
screening, as well as knowledge of CRC screening methods and recommended age of
screening initiation). Multivariable linear and logistic regressions examined the predictive
power of generational cohort and health literacy on perceived cancer and CRC risk and
severity, and knowledge of recommended age of screening initiation. A Poisson
regression model examined predictors for the number of correct CRC screens identified
by participants.
3.3. Findings
Demographics (Table 3.1)
624 individuals participated in the survey; of these, 44.7% individuals were classified as
belonging to Generation X (or Gen-Xers), and 55.3% as Millennials. The sample was
split between females (55.6%) and males (44.2%), with one person identifying as other
40
(genderqueer). 83.5% of participants identified as white or Caucasian, 6.7% as black or
African American, and 6.7% as Asian American; 5.6% identified as Hispanic. Most were
employed full-time (59.2%) and were not current students (94.4%). 48.4% were married
or legally partnered, and 87.1% had some college education or more.
Table 3.1: Participant Demographics (N=624) Participants (N (%))
Generational Cohort
Generation X 279 (44.7)
Millennial 345 (55.3)
Identified Gender
Female 347 (55.6)
Male 276 (44.2)
Other (Genderqueer) 1 (.2)
Race
Asian/Asian American 42 (6.7)
African American/Black 44 (7.1)
Caucasian/White 521 (83.5)
Other 17 (2.7)
Hispanic – Yes 35 (5.6)
Married or legally partnered- Yes 302 (48.4)
Education
Elementary School 3 (.5)
HS or GED 78 (12.5)
Some college or college grad 434 (69.6)
Grad school 109 (17.5)
Income
<20,000 53 (8.5)
20-49,000 229 (36.7)
50-79,000 192 (30.8)
>80,000 150 (24.0)
Employment
Full-time 466 (59.2)
Part-time 102 (13)
41
Looking for work 28 (3.6)
Homemaker 58 (7.4)
Current student – no 589 (94.4)
Health Literacy
Low 124 (19.9)
Medium 394 (63.1)
High 106 (17)
Have you previously been screened for CRC?
Yes 39 (6.3)
Have you ever been diagnosed with cancer?
Yes 15 (2.4)
At what age should the average person begin
screening for CRC?
>19 0
20-29 5 (.4)
30-39 65 (10.4)
40-49 292 (46.8)
50-59 247 (39.6)
60+ 15 (2.4)
At what age should the average person begin
screening for CRC?
Correct (50 Yo) 225 (36.1)
Which of these tests is a screening method for CRC?
identified NO correct screen 6 (1)
identified one correct screen 205 (32.9)
identified two correct screens 188 (30.1)
identified three correct screens 176 (28.2)
identified all four correct screens 49 (7.9)
CRC Knowledge
About one third of participants (36.1%) identified the correct currently recommended age
of screening initiation (50 years). Other age ranges that were frequently cited as the
recommended age of initiation of CRC screenings were 30-39 (10.4%), 40-49 (46.8%),
42
40 (28%), and 60+ (2.4%). Most individuals were aware (or could deduce) that a
colonoscopy is a common method for colorectal cancer screening (96.3%). However,
fewer were aware of FOBT testing (59.1%), Stool DNA tests (38%) and Sigmoidoscopy
(15.7%) as CRC screening methods. 22.3% of participants believed PSA to be a screen
for CRC. 99.8% and 98.7% of participants were aware that mammography and pap
smears, respectively, were not tests for CRC. Only 7.9% of participants correctly
identified all four CRC screens presented, and 21.5% identified at least one incorrect test
for CRC (Mammography, PSA or Pap).
Generational Differences (Tables 3.2 and 3.3)
No significant differences between generational cohorts existed in terms of education,
Health Literacy, previous CRC screening, and having a medical provider they see at least
once a year. Cohorts did not vary in their perceptions of severity of a CRC diagnosis.
Furthermore, no significant differences existed in the perceived importance of CRC
screening and the perceived importance of cancer screening in general. Cohorts rated the
importance of cancer screening in general at around 72.1/100, and the importance of
CRC screening in their lifetime at 71.6/100. Both groups identified on average one
correct CRC test out of the four presented, with no significant differences between
groups.
Cohorts did differ significantly in terms of gender, race, and previous experience with
cancer, with Generation X more likely to be female, identify as white or Caucasian, and
have previous experience with cancer. Furthermore, Gen X-ers reported higher rates of
perceived risk of CRC (25/100, compared with Millennials at 18/100) and cancer in
general (50/100; Millennials =4 3/100), and severity of cancer in general (55/100,
43
Millennials = 53/100). Significant group differences also existed in the knowledge of age
of CRC screening initiation, with 47% of Generation X correctly stating age 50 as the age
to begin screening, compared with 27.2% of Millennials.
Table 3.2: Group comparisons of continuous variables (Mann-Whitney U Test)
Median
Variable Range Millennial Gen X Mann- Z-score p
(N=345) (N=279) Whitney
U
How likely 1-100 43 50 55670 3.4 <.00
are you to
develop any
type of cancer
in your
lifetime?
If diagnosed 1-100 53 55 53089 2.2 0.03
with any type
of cancer,
how likely are
you to die
from it?
How 1-100 75 75 48476 0.2 0.88
important is it
to get
screened for
any type of
cancer in your
lifetime?
How likely 1-100 18 25 53458 2.3 0.02
are you to
develop colon
or rectal
cancer in your
lifetime?
If diagnosed 1-100 53 55 49,259 0.5 0.61
with colon or
rectal cancer,
how likely are
44
you to die
from it?
How 1-100 78 79 46824 -0.6 .88
important is it
to get
screened for
colon or rectal
cancer in your
lifetime?
Education 1-17 16 16 48,004 -0.1 0.95
Table 3.3: Group comparisons of categorical variables (Chi-Square Test)
Millennial Gen X
Variable N(%) N(%) χ2 P
Identified Gender (Female) 172 (49.9) 175 (62.7) 10.3 <.00
Previous CRC Screen (yes) 18 (5.2) 21 (7.5) 1.4 0.25
Previous experience with cancer 3 (.9) 12 (4.3) 7.7 <.00
(yes)
Relationship with a Provider 235 68.1) 202 (72.4) 1.4 0.25
(yes)
Race
African American or Black 29 (8.4) 15 (5.4) 11.8 <.00
Caucasian or White 273 (79.1) 248 (88.9)
Asian or Asian American 32 (9.3) 10 (3.6)
Other 11 (3.2) 6 (2.2)
Health Literacy
Low 73 (21.2) 51 (18.3) 1.9 0.4
Medium 219 (63.5) 175 (62.7)
High 53 (15.4) 53 (19.0)
Which tests identify CRC?
0 correct 3 (.9) 3 (1.1) 2.4 0.7
1 correct 110 (31.9) 95 (34.1)
2 correct 109 (31.6) 79 (28.3)
3 correct 100 (29) 76 (27.2)
45
4 correct 23 (6.7) 26 (9.3)
At what age should the average
person begin screenings for
CRC?
Correct (50) 94 (27.2) 131 (47.0) 26 <.00
At what age should the average
person begin screenings for
CRC?
>_19 0 (0) 0 (0) 46.5 <.00
20-29 5 (1.4) 0 (0)
30-39 54 (15.7) 11 (3.9)
40-49 176 (51.0) 116 (41.6)
50-59 105 (30.4) 142 (50.9)
60+ 5 (1.4) 10 (1.6)
Multivariable analysis (Table 3.4)
Stepwise Linear regression models produced significant predictors of perceived risk,
severity, and importance of screening, as well as knowledge of CRC screening guidelines
and methods. Independent variables for all models were Generational Cohort, Health
Literacy, Gender, Race, Income, Relationship with a Provider, and Previous CRC Screen.
Model 1 (“How likely are you to develop any type of cancer in your lifetime”) was
significantly associated with Generational Cohort, Race, and Gender, with Gen-Xers,
those identifying as white or Caucasian, and females more likely to give higher ratings of
perceived risk of cancer in general. Generational Cohort (Gen-X) and Gender (Female)
were also associated with the question “If diagnosed with any advance stage cancer, how
likely are you to die from it?” (Model 2). Health Literacy (High), Race (Asian) and
Relationship with Provider (Yes) significantly added to Model 3 (“How important is it to
be screened for any type of cancer in my lifetime?”) and Model 4 (“How likely are you
to develop colon or rectal cancer in your lifetime?”) was associated with Previous CRC
46
screen and Race, with those who have previously undergone CRC screening and those
identifying as non-black scoring their risk of colon or rectal cancer more highly. No
significant associations existed for Model 5: “If diagnosed with advanced stage colon
cancer, how likely are you to die from it?”. Model 6 (“How important is it to be screened
for CRC in my lifetime)” was significant, with Health Literacy (High), Previous CRC
Screen (Yes), Relationship with a Provider (Yes), Gender (Female) and Income adding to
the model. A logistic regression model (Model 7) assessed answers to the question of
when one should initiate regular CRC screening. Two of the six predictor variables were
statistically significant: Females (compared to males) and those belonging to Generation
X (compared with Millennials) were less likely to answer correctly (estimated ORs = .61
and .47, respectively; not displayed in table). Finally, a Poisson regression model was
used to examine factors associated with the number of correct responses to the question
“Which of these tests identify a screening test for CRC?”. This model (Model 8) did not
identify any significant association. The corresponding statistics for all significant models
(except model 7) can be found in Table 4.
Table 3.4: Stepwise Linear Regression
Models
95.0%
Confidence
Interval
Dependent Variable Significant Lower Upper
predictors B p Bound Bound
General Model 1: Gender 10.41 <0.00 6.39 14.43
Cancer How likely (Female)
Questions are you to Race (White) 11.53 <0.00 6.15 16.91
develop any
type of Generational 5.23 0.01 1.12 9.27
cancer in cohort (Gen
your X)
lifetime?
47
Model 2: If Gender 5.01 0.10 1.19 8.84
diagnosed (Female)
with any
Generational 4.25 0.03 0.43 8.07
advance
Cohort (Gen
stage cancer
X)
type of
cancer, how
likely are
you to die
from it?
Model 3: HL (High) 7.71 <.00 2.44 12.99
How
important is Race (Asian) 9.04 0.03 1.16 16.91
it to be
Relationship 4.67 0.03 0.35 8.99
screened for
with provider
any type of
(yes)
cancer in
my
lifetime?
CRC Model 4: Previous CRC 11.69 <.00 4.53 18.85
Specific How likely screen (yes)
Questions are you to
develop Race (Black) -7.77 0.03 -14.55 -1.00
colon or
rectal cancer
in your
lifetime?
Model 6: Previous CRC 13.35 <.00 5.03 21.67
How screen (yes)
important is
it to be
screened for HL (High) 8.24 <.00 2.86 13.61
CRC in my
lifetime? Relationship 6.13 <.00 1.60 10.67
with provider
(yes)
Gender -5.81 <.00 -9.86 -1.77
(Female)
Income .82 0.02 0.11 1.54
* All models included generational cohort, race (black, white and Asian), gender,
HL (high, medium, low), previous CRC screen, relationship with provider, and
income
3.4. Discussion
48
Generational research has previously been conducted to examine differences and
preferences among medical students (Borges, Manuel, Elam, & Jones, 2006, 2010), the
health care workforce (Piper, 2012), and patients (DeMaria & Hoffman, 2016; Johanson,
2017), and implications for academic instruction, mentoring, workplace culture, and the
general treatment of patients have been discussed. Generational cohort provided a
valuable lens through which to examine cancer perceptions and knowledge. Both groups
overestimated the risk of CRC, as the lifetime risk for CRC is between 4% and 5% (ACS,
2017b; Campos, 2017), while these cohorts rated the likelihood at 18/100 (Millennial)
and 25/100 (Gen X), with Gen-Xers reporting significantly higher CRC risk compared
with Millennials. Gen-Xers furthermore reported significantly higher risks of cancer in
general (50/100) and greater likelihood of dying of cancer (55/100), compared with
Millennials, who have a more realistic view when rating their risk of developing cancer
(43/100). However, both groups again severely overestimate the risk and severity, as the
actual lifetime risk of developing cancer is around 39.66% for men and 37.65% for
women, while the risk of dying from cancer is 22.03% for men and 18.76% for women
(ACS, 2017b). (Note that comparing these rates of severity is not completely accurate as
our severity question asked about the risk of dying if diagnosed with advanced stage
cancer, while the ACS statistics report the risk of dying from cancer in one’s lifetime).
The importance of cancer screenings in general and CRC screening specifically
was rated by both groups between 75/100 – 78/100, indicating that most individuals are
aware of the benefits of screening. However, if these rates in the future translate to
screening uptake at the age of recommended screening initiation, we will remain at the
same suboptimal screening rates for colorectal cancer, as currently only on estimated 60-
49
76% of age-eligible individuals undergo guideline-concordant CRC screening (CDC,
2019).
Generation X was more likely to be aware of current CRC screening guidelines
while Millennials believed CRC screening to begin at a younger age (between 40 and
49). If screening guidelines are changed based on rises in CRC incidence in younger
populations, these guidelines will have to be carefully disseminated to this population
using targeted interventions. Both groups furthermore showed a lack of awareness of all
available CRC screening methods, indicating that more discussion is needed around
available screening options, especially as awareness of less invasive methods might lead
to an uptake of screening. In addition to CRC screening methods, providers should be
prepared to discuss actual cancer and CRC risk and severity rates, especially with patients
approaching age 50.
Finally, regression models indicate that targeting subgroups for interventions may
be effective. Women, and Gen-Xers reported higher rates of risk and severity, while high
Health Literacy and having an established relationship with a provider led to a greater
appreciation of the importance of both cancer screening in general and CRC screening
specifically. In addition, the finding that individuals who identify as black or African
American see themselves at lower risk of developing colon or rectal cancer is concerning,
since the inverse is true (Augustus & Ellis, 2018).
Limitations
It is possible, as in any other online survey, that participants misrepresented themselves
in their profile to meet inclusion criteria, though this is considered unlikely. Furthermore,
participants recruited via MTurk are paid by task completed, and therefore may have
50
answered survey questions more quickly than general volunteers; however, previous
research that used MTurk workers found samples to be as reliable as other internet
samples (Buhrmester, Kwang, & Gosling, 2011; Chandler & Shapiro, 2016). Finally, all
data were collected via survey at one time point and therefore no conclusions about
causality can be drawn. However, we believe these data provide a good starting point for
further examination of generational differences in cancer knowledge.
3.5 Conclusion
Incidence and mortality rates of CRC are rising in those below the current recommended
age for regular screening, and this trend has been projected to continue for the
foreseeable future. While the ACA has called for a lowering of the age of CRC screening
initiation to 45 years of age, no formal modifications to the guidelines have been made as
of yet. If the recommended age of screening is lowered, careful dissemination is needed
to ensure the uptake of screening by those most in need. Physicians must be carefully in
tune with symptoms of CRC in this population, which will require attentive and careful
communication between patients and physicians so that symptoms are not broadly
dismissed. An awareness of the varying perceptions of risk, severity, and importance of
screening that may influence use of health care services, as well as the extent of
knowledge of screening methods and current guidelines, can guide these conversations.
51
Chapter 4: “Nothing too outrageous” v. “I am highly concerned”: Variation in
descriptions of medical information when presented differently.
Abstract
Background Current guidelines by the American Cancer Society recommend that regular
screening for colorectal cancer (CRC) begin at age 50; recent research however shows
rates of CRC rising among adults below the age of 50. Increasing rates in this population
are occurring despite declining overall rates of CRC. The mechanisms behind this remain
unclear, and guidelines for screening have not been updated to reflect new findings.
Current CRC screening reports can be difficult to interpret and consequently lead to non-
adherence to treatment recommendation. This study aimed to examine how individuals
below the age of 50 interpret a CRC screening report, and if slight modifications to this
report can ease the burden of interpretation.
Methods An online survey of individuals between the ages of 24 and 45 was conducted
in April and May of 2018. Participants were recruited using Amazon© Mechanical Turk,
a crowdsourcing internet marketplace which allows for the quick completion of tasks for
compensation. Participants were randomized to receive one of two colonoscopy result
reports, the first being a replicate of the report provided by a large, urban medical center
in Maryland, and the second a modified report containing a “gist” synthesis of findings.
Participants’ open-ended interpretations of these reports were analyzed qualitatively
using NVivo 12, and measures of perceived severity of the report, importance of follow-
up, and trust in a plan of care were assessed.
Findings Results 2 results showed that individuals who received the modified
colonoscopy report correctly rated the report as more serious and rated the importance of
follow-up higher than those who received the standard medical report; however, those
52
who received the standard report were more likely to believe that a plan of care was in
place for the fictional patient. Qualitative results showed participants who received the
modified report were more likely to understand that results from a biopsy were pending,
while those who received the traditional report were more confused about the content.
Significance Current screening reports are difficult to understand and add the burden of
translating medical information to the already full plate of the patient. Screening reports
that contain a synthesis of findings could help patients understand information more
easily and plan next steps accordingly. More research is needed to make
recommendations for what this synthesis should contain.
4.1 Introduction
It has long been known that poor literacy skills are predictive of poor health outcomes for
patients (Baker, Parker, Williams, Clark, & Nurss, 1997; Weiss, Hart, McGee, &
D'Estelle, 1992) and that even patients with adequate literacy can have inadequate Health
Literacy (HL) (Graham & Brookey, 2008). Though current estimates for national HL
rates are hard to come by, a 2003 national survey estimated that only 12% of United
States residents have high HL, with 49% having adequate and 39% low HL ("America's
Health Literacy: Why we Need Accessible Health Information. An Issue Brief from the
U.S. Department of Health and Human Services.," 2008). HL is increasingly viewed as a
characteristic reaching beyond the individual level to families and communities, as well
as the organizations that serve them and policies that guide health care implementation
and cost (Batterham, Hawkins, Collins, Buchbinder, & Osborne, 2016). HL interventions
are therefore focusing less on improving the HL of individuals but instead providing
populations with accessible services, health care staff that is trained in patient-centered
53
care, and health materials that are understandable and useable. Fuzzy Trace Theory (FTT)
has been used to guide the development of health materials to optimize understanding
(Samuel G. Smith et al., 2015).
Fuzzy Trace Theory (FTT)
FTT was developed as a grand theory in psychology in the late 1980s and early 1990s to
better understand sophisticated forms of human information processing, especially the
relationship between reasoning and memory (Reyna & Brainerd, 1995). One early tenet
was that mature reasoning uses synthesized, “gist” interpretations of materials, which are
formed by individual emotion, education, culture, experience, worldview, and level of
development, rather than memories of exact content (“verbatim” representation) to make
decisions or draw conclusions. That is, material is synthesized to provide an overall
estimate of meaning rather than deconstructed around particular arguments and
statements. This stood in contrast to the belief of the time that memory capacity was
positively associated with cognitive reasoning skills. This idea was based on then-recent
experiments that showed that adults and children recalled what they learned from an
experiment, rather than the actual activity or text that was presented, and the term “Fuzzy
Trace” was coined to describe the representation used for reasoning (Bjorklund &
Buchanan, 1989). FTT applies to written as well as numerical material; findings suggest
that most people, when presented with quantitative information, translate it to a
qualitative form, or gist representation. For example, percentages are often translated to
words like “most” or “less than half” for ease of interpretation and decision-making.
The implications of FTT for medical information processing and decision making
are vast. FTT may explain why communicating precise information about risk may not be
54
effective in preventing certain behaviors, and why some health materials are more
effective in communicating a message than others. Implications for informed consent also
exist; according to FTT, if a patient recalls an informed consent conversation verbatim,
i.e. repeats the facts or numbers presented, it does not necessarily imply a deeper
understanding (Reyna, 2008).
FTT and Numeracy
Numeracy has been an under-explored construct in health care, although numbers
feature prominently in medical information. Reyna et. al. (2008) questioned the unspoken
assumption that patients understand and benefit from numerical health information and
went further to question that adequate health literacy implies adequate numeracy. The
authors return to Fuzzy Trace Theory to suggest that like verbatim representations, which
are literal representations of information but contain little meaning and are therefore not
useful in medical decision making, the ability to “know” numbers, that is, be aware of
percentages of chance or risk, does not necessarily equate to the ability to use that
information to make informed decisions. As this research is still in its infancy, it is
suggested that further research on numeracy include examining the relationship between
numeracy and risk perceptions (Reyna et al., 2009).
Researchers have tied inadequate numeracy to an overestimation of risk and posited
that this heightened risk perception leads to avoidance of advantageous health behaviors
such as screening, thus impacting health outcomes (Gurmankin et al., 2004; Lipkus &
Peters, 2009). Lipkus et. al. summarize the implications of these findings as follows:
The main challenge for public health and health care is to educate the individual as
well as create methods of communicating quantitative information that are
understood, accepted and used by the public to encourage adaptive health behaviors
55
and decisions. (Lipkus, 2009)
Current Study
The aim for this mixed-method study was to examine how survey participants
described results from a fictional colonoscopy report that presented information in one of
two ways: 1. A “standard” report that contains descriptions of findings and
recommendations; or 2. A modified, or “gist”, report containing the same findings
included in the standard report, but with revised recommendations that summarize
findings and next steps. The purpose was to assess variations in how participants
summarize the findings. We were specifically interested in the role health literacy and
numeracy play in the interpretation of these reports.
4.2 Methods
Pilot Study
During a one-year pilot study, colonoscopy result reports were obtained from one large,
urban medical institution. Participants were recruited immediately prior to their
colonoscopy procedure (both screening and diagnostic cases were recruited), and
demographic and health literacy data collected. Within two months of their procedure,
participants were contacted and, if they agreed, interviewed in their home. These
interviews lasted between one and two hours, during which all were asked to describe the
information presented in the reports, highlight sections that were confusing or helpful,
and provide recommendations for improving the report. Participants narrated their
thought process while during the activity and interviews were audio-recorded. Among
other findings, researchers noted a pattern of participants noting that the medical
information presented to them in the reports was appreciated and may be handed to their
56
primary care physician for reference, but a summary statement of findings and
recommendations, written in lay terms, was desirable and would be helpful in
determining next steps for the patients themselves (Mueller NM, 2017). Based on these
findings, and guided by FTT, the research team constructed a “gist” report, which
contained all information from the original medical report but also a revised
recommendations section, which synthezied findings. The revised report was read by five
lay persons, to determine readability. The pilot study was approved by the University of
Maryland, College Park Institutional Review Board, and funded through the University
of Maryland MPower Initiative (PI: Curbow). Data were collecting August 2015 through
September 2016.
Larger Study
As part of a larger online survey (described in detail in Chapter 3), survey
participants (N=720) were randomized to one of two fictional colonoscopy reports; 1.
The standard version provided to patients after colonoscopy; or 2. The gist report created
by the research team. The reports described the same serious findings: likely stage IV
colorectal cancer, with two polyps that necessitated removal and one large mass, possibly
malignant, that was biopsied. Both the standard medical report and the gist report are
available in Appendix C: Questionnaire. Participants were then asked to provide an open-
ended description of findings and to indicate their beliefs about the seriousness of
findings, importance of follow-up, and existence of a plan-of-care for the fictional
person. The University of Maryland IRB approved this study. Data were collected in
April and May of 2018.
Measures
57
Report Descriptives
Participants were asked to read through the colonoscopy reports and answer the following
questions: 1. How serious is this report?; 2. How important is it that this patient
follow up with a doctor?; and 3. To what extent is there a plan of care in place for
this patient? All answers options ranged from 1 (not at all) to 10 (very much so).
Participants were then asked the following open-ended question: “If this colonoscopy
report belonged to a loved one, and he or she asked you what it meant, what would you
say?” Participants’ responses to this question were assessed both qualitatively and
quantitatively. For quantitative purposes, a binary verbatim variable was created to
capture those responses that contained verbatim representations of results: This was
coded as verbatim (1) or not verbatim (0). To qualify as verbatim, a response had to
contain two or more medical words used in the reports, such as the following responses:
“The results of the colonoscopy are abnormal and there is a reason to believe that you
may have a cancerous mass in your cecum.”; and “You have hemorrhoids and 2 colon
polyps, which can develop into colon cancer.” Finally, we asked participants to state
whether the recommendations of the report were written in medical language (0) or plain
language (1) to determine the success of our randomization.
Health Literacy and Numeracy
Perceived health literacy (HL) was collected using the four-item measure by Chew
(Chew et al., 2004) and coded via previously used categories, with a score between 4 and
15 indicating low health literacy, 16-19 indicating medium (or adequate) health literacy,
and a score of 20 indicating high health literacy (King-Marshall et al., 2015). Perceived
Numeracy was assessed using the Subjective Numeracy Scale (Fagerlin et al., 2007;
58
Zikmund-Fisher, Smith, Ubel, & Fagerlin, 2007), with a summary mean score ranging
from 1-6. Previous CRC screening was determined by asking “Have you ever received a
screen for colon or rectal cancer?”, with a binary scoring (yes/no).
4.3 Analysis
Quantitative analysis
All quantitative analyses were conducted using SPSS v25. Bivariate analyses were
conducted to assess differences between the two groups (Gist report and standard report)
in terms of age, health literacy, numeracy, income, education and sex. Bivariate analysis
was also conducted to compare responses on the three main variables (How serious is this
report; How important is it that the patient follow up with a doctor; and To what extent is
a plan of care in place for the patient) by report type (standard v. gist). Finally, logistic
regression was used to determine the predictive factors of a verbatim response, while
controlling for demographic variables.
Qualitative Analysis
Two members of the research team (NM and JY) read through responses to the question
“If this colonoscopy report belonged to a loved one, and he or she asked you what it
meant, what would you say?” An initial list of codes was created and subsequently
collapsed and refined through discussion. Report type was documented as an attribute but
arbitrary labels were used to blind coders (report x and report y). Reponses were
uploaded to NVivo 12 for coding. Questions continued to be resolved through
discussions and codes continued to be refined during the coding process. Upon
completion, kappa was calculated to assess reliability and was found to be high (>.93).
59
Coding was then split by report type. Researchers remained blind to report type because
of the labeling: However, due to the different content of the reports, report type could be
intuited through some responses. As a final step, the research team read through the
coding summary and created memos for each report type. The three most salient codes
for each report type were summarized and are presented in Table 4.
Regression Models
Logistic regression was used to determine the factors that contributed to a verbatim
response. Predictor variables were report type received, health literacy, numeracy, and
previous experience with cancer. We controlled for age, self-identified gender, and
income.
4.4 Results
Participant demographics are displayed in Table 4.1 and discussed in more detail in
Chapter 3.
Table 4.1: Participant Demographics
Participants (N =
720)
Identified Gender (%)
Female 369 (51.2)
Male 350 (48.7)
Other (genderqueer) 1 (.1)
Age (mean, SD) 35.9 (5.6)
Race
American Indian/Alaskan Native 9 (1.3)
Asian/Asian American 97 (13.5)
African American/Black 47 (6.5)
60
Caucasian/White 546 (75.8)
Other 21 (2.9)
Hispanic – Yes 47 (6.5)
Married or legally partnered- Yes 353 (49)
Education
Elementary School 5 (.7)
Some HS 3 (.4)
HS Graduate or GED 78 (10.8)
1-3 years of college or Associates Degree 218 (30.3)
4 years of College or Bachelor's Degree 277 (38.5)
Graduate School 139 (19.3)
Income
<20,000 100 (13.9)
20-49,000 251 (34.9)
50-79,000 209 (29)
>80,000 160 (22.2)
Employment
Full-time 462 (64.2)
Part-time 100 (13.9)
Looking for work 28 (3.9)
Homemaker 58 (8.1)
Other 72 (9.9)
Current student – no 664 (92.2)
Health Literacy
Low 185(25.7)
Medium 422 (58.6)
High 113 (15.7)
Bivariate Results
61
Group comparisons (t-test and Chi-square) were run ensure that randomization was
successful. Individuals who were randomized to receive the gist report and those who
received the medical report were similar in terms of health literacy, numeracy, education,
gender and age. A significant difference existed in income, with those receiving the
standard report reporting a slightly higher income than those who received the gist report
(t=2.06, CI=.1,1, p<.01).
Table 4.2 displays the results of bivariate analysis examining differences in perceptions
of the report based on which report version was read. The question “How would you
describe the wording of these impressions” was included as an additional check, to ensure
that randomization occurred correctly. Chi-square results indicate that it did, with just
under 80% of those who received the Gist report describing it as written in a way that
allowed most people to understand it, and just under 90% of those who received the
medical report indicating that it was written in medical language, which can be difficult
to understand for most people (χ 2 =334.3, p<.001). When asked about the report,
participants who read the gist report interpreted it as more serious (Mean=9.2, compared
with Standard Report Mean=8.0, p <.001), and the importance of follow-up higher
Mean=9.7, compared with Standard Report Mean=9.6, p <.001, than those who received
the standard report. Those who received the standard report were more likely to think that
a plan of care existed for the recipient of the report compared with those who read the
gist report (Mean = 7.5 and 7.1, respectively; p<.001).
Table 4.2: Bivariate Analysis of Report Impressions by Report Type
62
Gist Report Medical
(375) Report
(345)
N (%) χ 2 (p)
How would you describe the
wording of these impressions?
It was written in plain language 305 40 (11.6%) 334.9 (,.001)
(language that most people can (80.1%)
understand) (1)
It was written in medical 76 299 (88.2%)
language (language that a doctor (19.9%)
may use, but that can be difficult
to understand for most people)
(2)
Mean (sd) t (CI)
How serious is this report? 9.2 (1.2) 8.0 (1.8) -10.6 (-1.3,-
.9)*
How important is it that this 9.7 (.8) 9.6 (1.1) -4.1 (-.3,-.1)*
patient follow up with a doctor?
To what extent is there a plan of 7.1 (2.3) 7.5 (1.9) 2.8 (.1,.8)*
care in place for the patient?
*Significant at p<.001
Qualitative Results
Table 4.3 shows prominent responses by report type to the question: “If this colonoscopy
report belonged to a loved one, and he or she asked you what it meant, what would you
say?” For each report type, the three most salient response categories are displayed.
Responses from those who received the standard report were characterized as follows: 1.
Participants described the need for follow-up or more information 2. Findings were
described using medical terminology; 3. Participants expressed uncertainty or gave vague
responses. Responses to the gist report were characterized by the following: 1. The need
to wait for results of the biopsy and then follow up with a doctor; 2. The need to wait for
63
biopsy results (without mentioning further consultation with a provider); and 3.
References to the discussion of diet within the gist report. Both groups displayed
misunderstanding of certain words (“It looks like the abnormalities are probably
malignant, so I wouldn't worry.”; “I would tell them that the report was negative and
there is a serious risk of cancer”).
Verbatim Variable
Participants who received the standard medical report were significantly more likely to
explain results using a verbatim response; 81 (23.3%) of those who received the medical
report explained it using terminology taken directly from the report, compared with 49
(12.9%) of those who received the gist report (χ 2 =13.23, p<.001). Two logistic
regression models were used to further explore this variable; the first featured HL and
controlled for age, race, gender, income, and previous experience with cancer. Neither
numeracy nor HL predicted a verbatim response; however, allowing borderline
significance, individuals with previous experience with cancer were less likely to give a
verbatim response than those who reported having no previous experience with cancer
(HL Model: OR= .636, p=.055; Numeracy Model: OR=.630, p=.05). These results are not
displayed in a table.
64
Table 4.3: Qualitative Responses to "If this colonoscopy report belonged to a loved one, and he or she asked you what it
meant, what would you say?"
Standard (Medical) Report
Category Response
Follow-up with Doctor I can't understand the terminology. There are a number of questions that I would ask your
doctor in order to make the situation clear to us.
I'd say go get it looked at
I would try to help them as best as I could, but if there was anything I couldn't answer, I'd tell
them to call their doctor to request clarification.
I would say that they should talk to their doctor
Use of medical terminology The polyps that were biopsied may be malignant or cancerous. The tissue is being sent for
testing and we need to wait for the final results.
The doctors have found one large mass in the cecum, which they have already taken the
biopsies, we need to wait for the result.
there are some Internal and external hemorrhoids and colorectal surgery to be done and need
to check CEA today
Cecal mass, likely malignant. Biopsied.
Uncertain or vague It looks like the abnormalities are probably malignant, so I wouldn't worry. You might want
to ask your doctor though how risky things are.
I would say that there were some issues but I don't think it's anything too serious.
I'm glad it wasn't worse.
I would say it looks like it serious but I am not totally sure due to all the medical terms and I
would use the internet to find out more information on the report terms.
When performed by specially trained professionals, colonoscopies are extremely safe
Gist Report
65
Category Response
Wait for results and follow-up This report is very serious and they think you have colorectal cancer. You need to follow up
with the surgeon they set you up with and you will be getting blood tests in the mail.
You have a tumor and you need to see a surgeon also they took a biopsy and will let you
know the results. It's very important you follow up.
You have a mass that is possibly cancer. I know that is is scary to hear that but the doctors
have taken a biopsy so we will know for sure whether it is cancerous soon. In the mean time
we need to set up an appointment to see a surgeon and get his opinion. You also had a few
polyps in your colon that were removed today and there were hemoroids as well as infection
so it is important that you eat a healthy diet high in fiber to help with the infection. I will be
here for you as much as I can and if you feel like you want a second opinion I will help you
get that as well.
Wait for results I would go over it with them in detail, and advise that while it may not look promising, we
should wait for the biopsy to come back to see if there are more concerns regarding the mass.
There is a lot to be concerned about, but let’s wait for the results and then plan our attack
i would say that it's being tested to see if it's serious and be patient
The doctors think it might be cancer. They're going to send the stuff to a lab to get it checked
out, and then get back to you. For now just keep doing what you're doing, and get some
fluids and stuff into you. They'll let you know.
There's a possibility of cancer. Let the doctor perform follow-up tests. You'll know more
after.
Diet They are sending a biopsy to test for cancer, you have hemorrhoids and should eat a more
fibrous diet and drink plenty of fluids to reduce the effects of your hemorrhoids.
It's normal report but you have to follow proper diet
They said they found something that might be cancer, but they have to test and make sure.
They said to eat more fiber.
I will suggest him to consult a colorectal surgeon and advise him to taken fiber and protein
66
rich foods.
Start eating more fiber and drinking more water, and we'll deal with whatever else is ahead.
67
4.5 Discussion and Conclusion
Discussion
Findings indicate that even minor changes to medical texts can result in significant
differences in interpretation. Individuals receiving the gist report were better able to
register the seriousness of the findings and the importance of follow-up care. However,
those individuals were also less likely to feel that a plan of care was in place for the
hypothetical patient, perhaps due to the understanding that pathology results were still
pending and a plan would be decided on after the exact diagnosis was known.
Both qualitative and quantitative results indicate that those receiving the standard
report used more language directly taken from the report in describing the findings. FTT
would suggest that this is an indication that the standard report was less likely to be
understood by participants, though this was not directly shown in this study. Further
research should be conducted to examine the relationship of verbatim responses to patient
understanding, especially in terms of informed consent.
Qualitative results further indicated that those who read the standard report felt
the need for immediate follow-up with a provider, possibly to have the report explained
to them, while the participants who read the gist report stated they would wait for the
pathology results and then follow-up with a provider. While the standard report caused
more confusion and uncertain responses, both groups misused terms such as malignant,
which seemed to be confused with benign, and positive/negative findings, which were
used in the more colloquial sense (positive-good, negative=bad). Finally, the inclusion of
more specific information on beneficial changes to diet in the Gist report may have been
68
misguided, as participants tended to latch on to this piece of information, possibly
because discussions of diet are familiar and relatable, and provide something tangible to
focus on. This information may have overshadowed the more serious findings and the
exact content of the Gist synthesis should be further evaluated.
Limitations
Our sample consisted of individuals below the current recommended age of colorectal
cancer screening. Health Literacy and Numeracy may be more relevant factors in samples
with a wider age distribution. All data were collected via survey at one time point and
therefore no conclusions about causality can be drawn. However, we believe these data
add to our understanding of how patients interpret medical information.
Conclusion
Medical results should be constructed with the patient in mind. Results from this study
can assist in designing practical screening reports. While patients appreciate all medical
information provided, a synthesis is helpful to guide next steps. These statements should
focus on the most relevant aspects of the findings, saving recommendations for lifestyle
changes, if applicable, for discussions with providers. They should be stated in simple,
direct language in which patients can see a clear chain of necessary steps. Appropriate
services should be available to address patient concerns or anxiety,
69
Chapter 5: Awareness of Colorectal Cancer Trends Among Individuals Below the
Age of 50
Abstract
Background Current guidelines by the American Cancer Society recommend that regular
screening for colorectal cancer (CRC) begin at age 50; recent research however shows
rates of CRC rising among adults below the age of 50. Increasing rates in this population
are occurring despite declining overall rates of CRC. The mechanisms behind this remain
unclear, and guidelines for screening have not been updated to reflect new findings.
Younger individuals may not be aware of their CRC risk, and providers may be unaware
of the need to screen their younger patients.
Methods An online survey of individuals between the ages of 24 and 45 was conducted
in April and May of 2018. Participants were recruited using Amazon© Mechanical Turk,
a crowdsourcing internet marketplace which allows for the quick completion of tasks for
compensation. We randomized participants (N=720) to receive one of two statements
describing trends in the rates of colon cancer, one true and one false, and examined
participants’ ability to discern true from false information and certainty in the correctness
of their response.
Findings Results showed a lack of foundational knowledge among participants to
differentiate between correct and incorrect statements regarding colon cancer trends.
Interestingly, participants seemed to default to accepting the validity of the statement.
Those who believed the statement they read to be true were more confident in their
response than those who believed the statement to be false; women were overall less
confident in their responses than men.
Significance Individuals below the age of 50 seem unaware of current trends of CRC in
70
their cohort. Providers should intervene accordingly and provide accurate information on
trends and available screening options. Providers should furthermore take note of a bias
towards believing credible-sounding information, discuss with patients which medical
information they read online, and remind individuals, especially men, to take a more
critical stance.
5.1 Introduction
Overview
Increases in colorectal cancer (CRC) incidence and mortality among individuals below
the current recommended age of screening initiation have been noted both in the US and
developed countries in general (Campos, 2017; R. L. Siegel et al., 2017). A recent
projection predicted continued increases in CRC incidence rates among adults under the
age of 55 through at least 2040 (Araghi et al., 2018). When CRC is detected individuals
below the age of 50, it is usually found at an advanced stage and mortality rates are
higher (Campos, 2017). It is therefore critical that we begin discussing CRC in younger
populations and the actions we can take to optimally prevent this pathology. To our
knowledge, there has been no examination of what individuals below the age of 50 know
about current CRC trends.
In the United States, individuals have reported high levels of trust in information
received directly from providers; 69.5% of respondents to the Health Information
National Trends Survey (HINTS) trusted health information from a doctor a lot, while
1.4% reported trusting it not at all. Trust in medical information found online was fairly
low, with 13.7% of respondents reporting they would trust online information a lot, and
8.7% stating they would trust it not at all (Hesse & Moser, 2018). Despite their mistrust,
71
individuals in the US (and around the world) are using the computer more and more to
search for medical information, with women more likely to search online for health
information than men (Percheski & Hargittai, 2011). One in three adults is said to use the
internet to diagnose or learn about health issues (Fox & Duggan, 2012), and trust in
content increases when it is easy to use and the authority of the author is recognized
(Sbaffi & Rowley, 2017). Furthermore, despite their trust in provider information,
patients tend not to adhere to recommendations; for example, an estimated 50% of
medications are not taken correctly or at all (Brown & Bussell, 2011). This is likely
because trust does not equate to understanding, and despite healthcare professionals’
assumptions that medical instructions and information are easy to understand, patients
often struggle to do so (Cornett, 2009).
Current Study
This study randomized survey participants to receive one of two statement about rates
of colon cancer in the United States, one true and one false. Our goal was to ascertain if
younger adults were aware that CRC rates were rising in the under 50 population, while
declining for older Americans. We hypothesized that if aware, individuals would
recognize the false information as indeed false.
5.2 Methods
Sample
In April and May 2018, an online survey was conducted using Amazon Mechanical Turk
(MTurk). Participant demographics are presented in Table 8, and the details of this
survey have been described previously in Chapter 3. The University of Maryland,
72
College Park Institutional Review Board approved this study.
Measures
For this study arm, participants were randomized to one of the two below statements
about colorectal cancer rates in the United States:
1. From the mid-1980s through 2013, colon cancer rates:
declined in adults age 55 years and older
increased by 2.4% per year in adults 20 to 29 years
increased by 1% per year in adults age 30 to 39 years
or
2. From the mid-1980s through 2013, colon cancer rates:
increased in adults age 55 years and older
declined by 2.4% per year in adults 20 to 29 years
declined by 1% per year in adults age 30 to 39 years
Statement 1 is correct (Sigel 2017), while statement 2 is incorrect. Following presentation
of the statement, individuals were asked the following questions:
What do the data say is happening to colon cancer in the US? This questions
allowed for an open-ended response.
Do you think this statement about colon cancer is false (0) or true (1)?
On a scale of 0 (not at all sure) to 10 (completely sure) how sure are you that
your answer to the last question is correct?
(To note, at the conclusion of the survey, participants were automatically directed to an
informational page containing the correct statement about CRC trends.)
In addition to the demographic variables collected in Table 9, perceived health literacy
(HL) was collected using the four-item measure by Chew (Chew et al., 2004) and coded
via previously used categories, with a score between 4 and 15 indicating low health
literacy, 16-19 indicating medium (or adequate) health literacy, and a score of 20
indicating high health literacy (King-Marshall et al., 2015). HL was also used as a
continuous variable in the regression analysis, with a range of 4 (low health literacy) to
20 (high health literacy).
73
Analysis
Quantitative data were analyzed using SPSS 25. Sample descriptives are presented in
Table 1 and discussed in more detail elsewhere. The two groups (correct statement and
incorrect statement) were compared by demographic factors to ensure randomization was
successful. Chi-square and Mann-Whiney U Test were used to assess differences in
education, income, gender, race, age, marital status, previous experience with cancer,
previous colorectal cancer screen, health literacy and numeracy. Results are presented in
Table 2. Open-ended responses were analyzed by group to assess participants’
understanding of statements. Results and examples are summarized in the findings
section. Finally, a multivariable linear regression assessed the factors predictive of high
confidence in individuals’ assessments of statement correctness.
5.3 Findings
A description of the study sample can be found in Table 5.1. Bivariate analysis (Table
5.2) revealed no significant differences between participants who received the correct
statement and those who received the incorrect statement, indicating successful
randomization. Qualitative analysis of responses indicated that individuals were able to
understand the statement provided. Representative examples of responses to the false
statement are as follows: “Increasing in older adults, decreasing in younger adults”, and
“It is more likely in older people”. Participants who received the true statement described
it with the following terminology: “Declining in over 55s but increasing in 20-39 year
olds [sic]” and “Chances decrease as you get older.” Few individuals gave ambiguous
responses such as “After 1975 anyone can get cancer”, and “Increasing”, and even less
responded with “I don’t know.”
74
Table 5.1: Participant Demographics Participants (N =
720)
Identified Gender (%)
Female 369 (51.2)
Male 350 (48.7)
Other (genderqueer) 1 (.1)
Age (mean, SD) 35.9 (5.6)
Race
American Indian/Alaskan Native 9 (1.3)
Asian/Asian American 97 (13.5)
African American/Black 47 (6.5)
Caucasian/White 546 (75.8)
Other 21 (2.9)
Hispanic – Yes 47 (6.5)
Married or legally partnered- Yes 353 (49)
Education
Elementary School 5 (.7)
Some HS 3 (.4)
HS Graduate or GED 78 (10.8)
1-3 years of college or Associates Degree 218 (30.3)
4 years of College or Bachelor’s Degree 277 (38.5)
Graduate School 139 (19.3)
Income
<20,000 100 (13.9)
20-49,000 251 (34.9)
50-79,000 209 (29)
>80,000 160 (22.2)
Employment
Full-time 462 (64.2)
Part-time 100 (13.9)
Looking for work 28 (3.9)
75
Homemaker 58 (8.1)
Other 72 (9.9)
Current student – no 664 (92.2)
Health Literacy
Low 185(25.7)
Medium 422 (58.6)
High 113 (15.7)
Table 5.2: Bivariate Comparison of Groups (Chi-square for categorical
and Mann-Whitney U Test for continuous)
Statement Received X2 p
N(%)
TRUE FALSE
Health Low 87(23.8) 98 (27.7) 4.4 0.11
Literacy
Medium 212 (57.9) 210 (59.3)
High 67 (18.3) 46 (13)
Race Black 27 (7.4) 20(5.6) 2.5 0.5
White 280 (76.5) 266 (75.1)
Asian 47 (12.8) 50 (14.1)
Other 12 (3.3) 18 (5.1)
Gender male or 177 (48.4) 174 (49.2) 0.05 0.9
other
female 189 (51.6) 180 (50.8)
Statement Received
N(Median)
TRUE FALSE U z p
Age 366 (35) 354(35) 64,116 -0.1 0.8
Income 366(6) 354(6) 63,038 -0.6 0.5
ANOVA results (Table 5.3) showed that individuals who correctly identified the
statement as true (N=288) and those who incorrectly identified the statement as true
76
(N=307) were more confident in their response than those who identified their statement
as false, correctly or incorrectly. A two-way ANOVA (Table 5.4) shows a statistically
significant interaction between Statement Received and Correct Answer Identified for
Confidence scores (F=29.54, p<.001). Plots for this interaction are available in Appendix
A: Methods, Figure A4. Multiple linear regression examined the predictive power of
Health Literacy, Age, Gender and Income on confidence in the correctness of response.
Only gender was significant, with females reporting lower scores of confidence in their
response than males (B= -.094; CI=-.752,-.086; p=.014). Results are presented in Table
5.5.
Table 5.3: One-way ANOVA of Statement Received and Response Type.
Dependent Variable = Confidence in Response
Statement Response N(%) Mean 95% Confidence Interval for
received to (SD) Mean
statement Lower Upper Bound
Bound
Statement 1 Incorrectly 86 4.7 (2.4) 4.23 5.26
(True) identified (11.6)
as false
Correctly 280 6.1 (2.2) 5.88 6.39
identified (38.9)
as true
Statement 2 Incorrectly 299 6.2 (2.1) 5.98 6.44
(False) identified (41.5)
as true
Correctly 55 5.4 (2.6) 4.74 6.12
identified (7.6)
as false
Model significant at F=12.6, p<.001
Table 5.4: Two-Way ANOVA of Statement Received and Response Type.
Dependent Variable = Confidence in Response
77
Sum of Degree of
F p-value
Squares Freedom
Correct Statement
Identified (yes/no) 6.74 1 1.4 .24
Statement Received
(true/false) 14.96 1 3.12 .08
Correct Statement
Identification*Statement 142.22 1 29.54 <.001
Received
Table 5.5: Multiple Linear Regression Assessing Predictors of Confidence
in Response
95.0% Confidence Interval
B Lower Bound Upper Bound p
Health Literacy -0.03 -0.1 .04 .37
Age -0.02 0.05 .13 .27
Gender -0.44 -0.77 -0.10 .01
(Female)
Income -0.01 0.07 0.05 .71
5.5 Discussion
Qualitative review of participant responses indicate that the medical information
presented was understood. However, there was a lack of knowledge to differentiate
between correct and incorrect statements. Interestingly, participants seemed to default to
accepting the validity of the statement, with 579 out of 720 participants classifying the
statement as true, regardless of which statement they received. Two-way ANOVA
suggests that participants’ confidence in interpretation of the statement depended on their
response: those who received the true statement and correctly identified it as true were
more confident in their assessment than those who received the false statement and
78
correctly identified it as false. This indicates that individuals are hesitant to disbelieve
information when unsure, and has important implication for discussion of online medical
information. Providers should be aware of this bias among their patients and discuss
strategies for assessing the validity of medical information
Strengths and Limitations
It is possible, as in any other online survey, that participants misrepresented
themselves in their profile to meet inclusion criteria, though this is considered unlikely.
Furthermore, participants recruited via MTurk are paid by task completed, and therefore
may have answered survey questions more quickly than general volunteers; however,
previous research that used MTurk workers found samples to be as reliable as other
internet samples (Chandler & Shapiro, 2016). Finally, all data were collected via survey
at one time point and therefore no conclusions about causality can be drawn. However,
we believe these data provide a good starting point for further examination of
generational differences in cancer care.
Conclusion
Individuals below the current age of recommended regular CRC screening lack
foundational knowledge about CRC trends. This likely indicates that they are unaware of
their own risk status and may not recognize symptoms of CRC, should they arise.
Providers should take note of this lack of knowledge and intervene. Providers should
furthermore be aware of a bias towards believing credible-sounding online information,
especially among men. Taking the time to address this during medical visits could lead to
a more critical approach among patients when searching for online medical information.
79
Chapter 6: Summary
6.1 Overview of Findings and Implications
This dissertation was prompted by two factors. The first of these was previous research
by the University of Maryland colorectal cancer study team, led by Dr. Barbara Curbow,
which investigated how patients, when presented with colonoscopy reports, interpret and
use this information. For this research, which served as a pilot study for this survey, we
obtain a standard colonoscopy report from a nearby urban medical system and made
alterations, based on Health Literacy literature, so that we were left with three different
reports. The first was the standard report, which contained a text description of findings
and recommendations, as well as small pictures of the colon taken during the procedure.
For the second report, we removed the pictures and left only the text. Finally, for the third
report, we removed the pictures but added metaphorical descriptions of findings. For
example, the text would read as follows: “A 5mm polyps (about the size of a blueberry)
was found.” Metaphors have previously been found helpful in describing medical
concepts to patients with low HL (Gallagher, McAuley, & Moseley, 2013; Krieger,
Parrott, & Nussbaum, 2011). These reports were presented to patients who had recently
undergone a colonoscopy at the same medical system and had been recruited to
participate just prior to their procedure. Participants were asked to read through all three
reports and narrate their thought process as they underlined passages they thought were
helpful, confusing, or concerning. Interestingly, the results from this pilot study showed
us that rather than finding them helpful, participants across HL levels were confused by
the metaphors and mostly ignored the pictures. Participants repeatedly stated that they
appreciate the thorough findings provided, even if they were unable to understand some
80
or most of it, as they would take it to their provider for discussion. However, in addition
they would appreciate a synthesis statement at the end of the report that would present
important findings and recommended next steps in lay terms. Based on this feedback, and
guided by Fuzzy Trace Theory, I designed the gist report used in this survey.
At the same time, new findings reported that rates of colon and rectal cancer were
increasing in individuals below the current recommended age of screening (R. L. Siegel
et al., 2017). It has been known that when CRC is detected in younger people, it is
usually at a more advanced stage than when it is detected in those above the age of 50,
compounding the importance of this finding (Campos, 2017). I therefore designed a
survey for participants under the age of current recommended screening to assess the
following: 1. What does this population perceive and know about colon and rectal
cancer? and 2. When presented with a CRC screening report, what information does this
population take away from it, and which factors influence this? The goal was to create
provider and systems-level recommendations for addressing shortcomings in CRC
knowledge and producing medical information that is understandable and usable. The
below aims and hypotheses guided the inquiry
6.1.1 Aim 1
Aim: Compare differences in perceived risk to CRC and severity of CRC diagnosis
between generational cohorts.
H1: GenXers report higher levels of perceived risk to CRC, compared with
Millennials.
H2: Millennials are more likely to think a CRC diagnosis will result in death,
compared with GenXers.
Our hypotheses were based on the fact that even though rates of colorectal cancer are
increasing in adults below the age of 54 (Rebecca L. Siegel et al., 2017), national
81
guidelines still set the average age of CRC screening commencement at 50 (ACS, 2017a).
These guidelines may lead to the perception that CRC is exclusively a disease related to
aging, and younger individuals may not be aware of the risk of CRC (Rabin, 2017). At
the same time, lack of knowledge about a condition may lead to the overestimation of
severity (Reyna et al., 2009). We therefore posited that Millennials are less aware of
factors surrounding CRC and therefore are less aware of their risk to CRC, while at the
same time they overestimate the severity of a diagnosis. These hypotheses were tested
using Question 33 (perceived risk to CRC) and Question 34 (perceived severity of CRC
diagnosis) of the questionnaire.
These hypotheses were not supported by the research findings, as the data show
no group differences in perceptions of CRC risk and severity. Additionally, both groups
overestimated the risk of CRC, indicating an across-the-board lack of knowledge about
the condition and preventive measures.
Gen-Xers also reported significantly higher rates of risk and severity to cancer in
general. This may indicate that while CRC is not a high concern of those below the
current age of screening, a general sense of cancer risk and severity increases as one ages.
This information can be useful to providers in guiding discussion about risk and severity
of specific types of cancer, especially when carefully considered in terms of family
history and lifestyle.
6.1.2 Aim 2
Aim: Assess knowledge of CRC screening methods and guidelines between age
groups.
H3: GenXers score higher on measures of knowledge of CRC screening,
compared with Milllennials
82
This hypothesis was partially supported by the data. Significant group differences
existed in the knowledge of age of CRC screening initiation, with 46% of Generation X
correctly stating age 50 as the age to begin screening, compared with 27% of Millennials,
suggesting that as one approaches the age of recommended screening initiation,
knowledge of the guidelines increases. Besides belonging to Generating X, being female
was a predictor of knowledge of current screening guidelines.
Generation X and Millennials identified on average one out of four CRC
screening methods, indicating that a lack of knowledge of available methods spans
generational cohorts. This information supports revised guidelines with a younger age of
screening initiation, as this may increase knowledge of guidelines in younger cohorts.
Providers should also take note that men may need more targeted information about
guidelines.
6.1.3 Aim 3
Aim: Assess participants’ relationship with a provider and the impact the
relationship with a provider has on CRC screening knowledge and perceptions of
risk
H4: GenXers more often report having a provider whom they see regularly than
Millennials.
H5: Participants with an established relationship with a provider display more
CRC knowledge compared with participants without an established relationship.
The above hypotheses were not supported by these data. Surprisingly, high numbers of
both Millennials (67.8%) and Gen-Xers (73.1%) reported having a provider they see at
least once a year. In a multivariable linear regression model, the relationship with a
provider variable predicted higher ratings of the importance of CRC screening, along
with being male and higher income. Provider relationship did not predict perceived risk
or severity of CRC. The variable itself may not have been specific enough nor captured
83
the quality of the provider relationship. Future research may begin with a qualitative
exploration of a provider relationship and the CRC discussions that take place to better
understand the role this relationship takes in CRC prevention.
6.1.4 Aim 4
Aim: Assess whether participants who receive a colonoscopy report formatted to
contain Gist information report different perceptions about the report than
participants who receive a standard colonoscopy report.
H6: Participants who receive Report 1 (verbatim) provide lower scores of
perceived seriousness, compared with participants who receive Report 2.
H7: Participants who receive Report 1 (verbatim) rate the importance of follow-
up lower than participants who receive Report 2.
H8: Participants who received Report 2 (gist) are more likely to agree that there is
a plan of care in place for the patient, compared with participants who receive
report 1.
Exploratory: How would patients describe the reports to their loved ones, and
how does this differ by report type?
Fuzzy-Trace Theory (FTT) guided the design of Report 2 (Reyna, 2008, 2012; S.
G. Smith et al., 2015). Information presented in medical jargon has been shown to cause
confusion and anxiety among patients (Herber et al., 2014; Schnitzler et al., 2017). These
hypotheses stated that those who received synthesized findings (the Gist report) would be
better equipped to recognize the seriousness of both the report itself and the importance
of follow-up. Furthermore, we posited that those who received the Gist report would be
more likely to state that they feel a plan of care is in place for the hypothetical patient,
compared with those who read the standard report. These hypotheses were tested using
Question 23 of the report. Data supported H7 and H7; participants who received the Gist
report rated the seriousness of the report and the importance of follow-up more highly
than those who received the standard report. H8 was not supported by the data.
Participants who received the standard text report were more likely to feel that there is a
84
plan of care in place for the hypothetical patient. While the exact nature of this cannot be
intuited from these data, we suspect that this is related to the lower perceived seriousness
reported by the same group. That is, because these participants did not perceive the report
as quite so seriousness as the participants who read the Gist report, they were more likely
to state that they were in good hands. Further research should be conducted for the design
and optimization of the report and the resulting effects on patients.
Question 25 (If this report belonged to a loved one, and he or she asked you what
it meant, what would you say?) was used to test the exploratory aim. Results showed that
those who received the Gist report were more likely to take a two-pronged approach to
follow-up, that is, wait for pathology results and then follow-up with their doctor to
discuss next steps, while those who received the standard report stated that their first
move would be to follow-up with their doctor. Those who received the standard report
also displayed more confusion about findings and used more verbatim language to
describe the findings, with 81 (23.3%) of those who received the standard report
explaining results using terminology taken directly from the report, compared with 49
(12.9%) of those who received the gist report. Taken together, these findings suggest that
individuals who received the standard report had more difficulty understanding the
content and wanted to see a provider for clarification. Future research on the design of
medical reports should test whether a better understanding of the seriousness of findings,
and an appreciation for the importance of follow-up, translate to an actual increase in
follow-up and treatment adherence, when compared with standard of care.
6.1.5 Aim 5
Aim: Randomize participants to receive either a false or true statement describing
85
current trends of CRC and examine rates of belief and disbelief of the statement, as
well as confidence in the accuracy of the assessment.
H9: When asked if a statement about colon cancer trends is true or false,
participants with lower HL are more likely to believe in the correctness of their
assessment, compared to those with higher HL.
The final aim was explored through a two-way ANOVA and multiple linear
regression. ANOVA showed that regardless of which statement was received,
participants defaulted to believing in the validity of the statement, with 288 participants
correctly identifying the statement as true, and 307 participants incorrectly identifying the
statement as true, compared with 56 who correctly identified it as false and 86 who
incorrectly identified it as false. A multivariable linear regression explored H9 and results
did not support the hypothesis, as health literacy did not predict confidence in a response.
Being female was inversely related to confidence in a response. Further research should
examine the implications of confidence in one’s ability to correctly identify true and false
online medical information, paying special attention to gender differences.
6.3 Strengths and Limitations
The general limitations of cross-sectional survey data apply to this study, that is,
all data were collected from participants at one period of time. No inferences about
causality can therefore be made. In addition, the sample likely suffers from non-response
bias, as those individuals who did not participate in the study likely differed from those
who did, resulting in a non-representative sample. From previous research, we know that
individuals with access to the internet are, on average, more highly educated and have
higher incomes than those who are not, though this trend is equalizing with the
proliferation of smart phones (Rech Wachelke, Vizeu Camargo, Lincoln Bezerra Lins, &
Vieira de Lima Nunes, 2011). Furthermore, our sample consisted of individuals below
86
the current recommended age of colorectal cancer screening. Health Literacy and
Numeracy may be more relevant factors in samples with a wider age distribution. Finally,
because of the limited information shared by Amazon, it is unclear how some variables,
such as location or age, are reported and if they are updated or not. Participants recruited
via MTurk are paid by task completed, and therefore may have answered survey
questions more quickly than general volunteers; however, previous research that used
MTurk workers found samples to be as reliable as other internet samples. The limitations
of using MTurk are discussed in more detail in Appendix A: Methods.
Despite these limitations, cross-sectional data are useful for establishing
prevalence of beliefs and knowledge, in this case as related to CRC. Cross-sectional data
also ensure that there is no loss to follow-up, as participation is required only once.
Because participants were required to answer questions to proceed, and because open-
ended questions required responses of a certain type (i.e. the response to “How old were
you on your last birthday?” had to be a numeric entry between 1 and 100), missing data
and data errors were minimized or eliminated. Finally, this is the first study we know of
that examines beliefs and knowledge of CRC specifically in individuals below the current
age of recommended screening and tests the influence of health literacy and numeracy.
6.4 Future Research Directions
The research presented here provides many avenues for further research. Firstly, it
is clear that a lack of knowledge of CRC guidelines and methods is pervasive, and this
lack of knowledge may be translated to overestimates of risk of CRC. Provider
discussions should contain realistic explanations of risk and severity, carefully considered
in light of family history and current symptomatic status. We agree with previous
87
research that states the need for more caution when evaluating younger, symptomatic
patients and awareness of the consequences of a late diagnosis (Campos, 2017), and add
that a greater awareness of available methods of CRC detection may lead to an update in
screening rates. There are possibilities for educational interventions to address this, and
future research should examine whether tailored provider discussion of risk, severity, and
available methods of screening lead to an increase in screening utilization among younger
patients. Physicians should be reminded of the importance of taking a family history, as
the existence of younger family members with CRC or a first-degree relative with CRC
are considered risk factors for CRC, and the patient should consider early
screening(Forsberg et al., 2015). We should furthermore explore why, though individuals
seem to conceptually understand the importance of cancer and CRC screening, rates
among those currently at the age for CRC screening remain suboptimal: for example, is
awareness of the importance of screening on the rise, and does this translate to an uptake
of screening in practice? Future research should examine factors that promote critical
thinking in the wider population. Calls for critical thinking among healthcare
professionals proliferate and a universal definition that combines the popular constructs
of biomedical, humanist and social-justice oriented critical thinking is sought (Kahlike,
2018). At the same time, it may be worth considering how these critical thinking skills
can be passed on or encouraged in the patient population, in tandem with health literacy
efforts.
Policy Implications
As CRC rates are expected to continue to rise in this younger population, consideration
and planning is needed to ensure that disadvantaged groups benefit from any changes to
88
screening policies. Incidence rates will need to be monitored closely and examined for
racial disparities, as delayed decreases in CRC incidence in US blacks, when compared
with US whites, was noted in overall CRC incidence rates (Araghi et al., 2018). Health
care providers, health plans, insurers, employers, policy makers, and researchers should
focus resources on helping people to address lifestyle issues related to colorectal cancer,
beginning at an age much earlier than previously thought. Public health efforts should be
closely attuned to new medical findings concerning the etiology of CRC in younger
populations, and public health officials should be at the ready to translate these findings
into new or adapted strategies to reduce the burden of CRC. Thorough examinations of
the cost-effectiveness of CRC screening programs in younger populations is critical to
ensure the optimal use of available resources (Subramanian et al., 2016). Economic
measures should be devised to standardize the cost of CRC screenings.
Clinical Implications
Based on this research, we support the idea of lowering the recommended age of
screening initiation provided that screening options and risk of CRC are clearly
presented. Educational intervention for providers may be necessary to ensure that all
available screening methods are known and a shared decision between provider and
patient can be made about which method constitutes the best fit, taking into account
factors such as patient anxiety about invasive procedures, cost or extent of health
insurance coverage, and other variables that may bar completion of the screening
procedure.
In addition, we advocate for more research on the inclusion of a Gist statement in
CRC screening and diagnostic reports. This should explore whether or not a better
89
understanding of results translates into better follow-up and more decision-making in the
cancer care process. Furthermore, it should be examined if tailoring the content of the
Gist statement to specific populations, based on patient input, is feasible and effective.
Conclusion
Existing interventions that have increased rates of CRC screening could easily be
applied to a younger population. Successful interventions include creating a reminder
system for providers to discuss CRC screening with patients and handing out FOBT
screening kits as standard of care, for example when receiving an annual flu-shot (Rat et
al., 2018). A reminder system and list of questions to screen younger patients for CRC
family history and current symptoms could be implemented and FOBT kits could be
provided during well-visits. Additionally, public health officials should work closely with
the medical community to ensure that superior tests, such as the fecal immunochemical
test (FIT), are introduced efficiently into practice.
90
Appendix A: Methods
Sample Description
Participants consisted of a convenience sample recruited through Amazon’s Mechanical
Turk (MTurk), an online marketing place that allows researchers to quickly access large
numbers of individuals to complete tasks ranging from translation services to survey
completion. These tasks are called Human Intelligence Tasks (HITs), because they
require humans instead of machines to be completed, and individual “workers” or
“Turkers” are compensated per HIT completed. Individuals requesting the work to be
done are called “Requesters.” Once a task has been submitted by a Requester, it becomes
available to workers that meet the inclusion criteria, who can then choose whether or not
to complete the task. If the task is completed, the Requester can review the work and
decide whether it meets the standards set forth in the request: If it does not, payment can
be denied. This has negative consequences for the worker;when Requesters send out
tasks, they can specify that only workers with low payment denial rates can complete it
(Amazon, 2017).
The use of MTurk is becoming popular among social science researchers because
it allows for the recruitment of large sample sizes in a short period of timeand with
minimal expense. Previous studies have found the MTurk population to produce reliable
and valid results that are not significantly different from most other convenience samples,
though the MTurk population was found to be more diverse than standard internet
samples or American college samples (Buhrmester et al., 2011; Goodman, Cryder, &
Cheema, 2013; Nguyen et al., 2016). When 3,006 MTurk participants were compared
with those in another large internet sample, it was found that gender splits were
91
approximately the same in both (MTurk sample 55% female, internet sample 57%). Also,
36% of Turkers were non-white, compared with 23% in the internet sample; furthermore,
Turkers were older on average (32.8 years v. 24.3 years) and featured similar numbers of
non-Americans (31%). Previous research outlines the steps to take to ensure high-quality
responders and maintaining data security when collecting data using MTurk (Mason &
Suri, 2012; Nguyen et al., 2016).
Sampling Procedure
Based on previous research, we recruited workers with approval ratings equal to or
greater than 98%, and with more than 500 HITs completed, to ensure high quality results
and a familiarity with MTurk. Initial inclusion criteria were acceptance of the task, which
implies consent, having achieved Master Qualification on MTurk, and being between the
ages of 25 and 46 at the time of the survey. Achieving Master Qualification requires that
Turkers have an approval rating above 98, and have completed at least 500 HITs. These
scores are based on ratings given by previous Requesters, and indicate that Turkers
follow instructions and complete tasks, ensuring a high quality of responses. However,
after launching the survey and enrolling no more than 239 individuals, we removed the
Master Qualification as inclusion criteria. We retained the criteria that approval ratings
were equal to or greater than 98%, and Turkers had more than 500 HITs completed.
We aimed to recruit up to 800 individuals total that fit the above described inclusion
criteria, with 400 participants between the ages of 25-35 (older millennials) and 400
participants between the ages of 36-46 (younger GenXers). We initially ran two pilot
surveys to test the MTurk procedures and accessibility of the questionnaire, with 10
participants each. Subsequently, coding was entered into the larger survey that barred
92
those who completed the pilot from participating in the larger study. In April 2018, the
larger survey was launched in three arms, with each arm specifying a year-of-birth range
pre-set by Amazon (Amazon categories: Born 1972 to 1981 (Age 35-45); Born 1982 to
1986 (Age 30-35); Born 1987 to 1991 (Age 25-30)). 239 Turkers accepted the task, after
which enrollment stalled. The Master Qualification requirement was removed, and the
surveys were re-launched, again specifying that those who previously completed the
survey were not eligible to participate again. The subsequent Workers who accepted the
task were not necessarily non-Masters, but this variable was no longer available to us
once the requirement was removed. In total, 784 Workers accepted the task.
Missing Data
Missing data were minimal, as respondents were required to answer a question to proceed
in the survey and required to complete the survey to receive a payment code. 784 Turkers
accepted the HIT, 12 of these answered no questions, and a further 45 dropped off around
the time of the colonoscopy report reading. This could have been because the report
required time to read, the terminology was intimidating, individuals were interrupted, the
report was being viewed on a mobile screen and was difficult to read, or for multiple
other reasons. Because no demographic data had been collected on these individuals at
this point, they are excluded from the final analysis, resulting in an N=727.
Possible Bots
During the summer of 2018, MTurk Requesters noted a drop in data quality and an
increase in clusters of geolocation indicators. This raised flags that individuals were
running programs (or “Bots”) to complete HITs. While Amazon’s policy allows Workers
to run programs to more efficiently help select and preview HITs, it does not allow the
93
running of programs to complete them. While there is no way to guard against this, a way
to assess for Bot data is to carefully examine open-ended responses for nonsense answers
such as “nice” or “good” (Dreyfuss, 2018). Using this criterion, we examined open ended
responses and flagged seven participants as possible Bots. Bot answers included the
following: “hfffffffffffffffff”; “Nothing”; test test”, “none” and “nice work.” Six out of
the seven Bots were located outside of the United States. Three Bots identified as female,
all seven reported low health literacy, six identified as Millennials, and four identified as
Asian. These participants were removed from all analysis, further reducing the general
sample size (N=720).
Comparison of Master Qualification vs. no Master Qualification
As stated, above, no data were available as to whether or not the second batch of survey
participants had Master Qualifications or not. However, we wanted to compare
participants in the first survey batch to those in the second. Chi-square tests showed that
there were no significant differences in gender, race, perceived health literacy, numeracy,
income or location (US or non-US) between these groups. In addition, there was no
significant difference in belonging to a generational cohort (Millennial or Gen-X)
between masters and non-masters.
94
Figure A1: Boxplots of Health Literacy by Masters’ Qualification
Figure A2: Boxplots of Numeracy by Masters’ Qualification
95
Comparison of US Responders v. Non-US Responders
101 survey participants were identified as not being in the US at the time of their
survey,
Figure A3: Map of participant location created using the latitude and longitude variables.
Created by Travis Hyams using https://www.darrinward.com/lat-long/ and used with
permission
based on the geolocation variable automatically collected by Qualtrics (see Figure A3).
Of these participants, 3 identified as black (3%), 25 as white (24.8%), 59 as Asian (58.4)
and 14 as other (13.9%). These individuals differed significantly from US respondents,
with non-US respondents being more likely to identify as Asians, Millennials and male,
with low health literacy and low income. Because of these differences, non-US
respondents were omitted for the analysis in Chapter 3, as the questions pertained to US-
specific screening guidelines and methodologies, leaving a total sample size of 624
respondents for this analysis.
Variable Preparation
All variables included in this study were examined for: a) percentage of missing
96
data, b) frequency distribution, c) normal distribution, and d) outliers. Missing data are
described above. If data were found to be non-normally distributed, a non-parametric test
was instead chosen for bivariate analysis (for example, Mann-Whitney U Test rather than
t-test), and Generalized Linear Models (GLIM) in SPSS was used for linear regression
modeling, as GLIM does not have an assumption of normal distribution (Agresti, 2015).
Data were visually examined for outliers by creating boxplots. Outliers were removed
from analysis by calculating z-scores and removing data with z-scores ± 3, as 99.7% of
the data should be within three standard deviations of the mean. However, the need to
remove outliers was overall minimal.
Finally, collinearity was assessed with correlation matrices. Where statistically
significant correlation between two variables existed, decisions on which controls to keep
in each model were made based on the following two criteria: (1) statistical association of
the measure with the dependent variable and (2) importance of the measure to the model
based on theory and previous research.
97
Interaction Effects for Analysis III
Figure A4: Interaction Effects of Statement Received (true or false) and Correct Identification of
Results (0=incorrect, 1=correct). Dependent variable = Confidence in Response (Analysis III,
Chapter 5).
98
Human Subjects Procedure
a. Informed Consent
All study activities were approved by the University of Maryland, College Park
Institutional Review Board (IRB) (See Appendix B). A waiver of written informed
consent and HIPAA was granted. Participants were provided with details of the study and
possible risks and benefits, modeled on the University of Maryland Informed Consent
Form template, in the description of the task. When Workers accepted the task, they were
giving consent for the data to be used. To ensure anonymity of Workers to the best of our
ability, the questionnaire was contained in Qualtrics, instead of within Amazon, and
workers’ information, besides geolocation, was not saved. After accepting the task,
workers received an external link to the Qualtrics survey. Once the task was completed,
Workers received a completion code to enter into MTurk to receive payment.
b. Compensation
Compensation for MTurk tasks, or HITs, is determined by the Requester. A
frequently cited rule by workers is that work should be compensated at a rate of at least
10 cents per minute (MoneyPantry, 2017). For completing this approximately 15-minute
questionnaire, Turkers were compensated at a rate of $.20 per minute, for a total of $3 per
individual. In addition to this, Amazon charges a 25% fee, which includes a 5% fee for
the use of individuals with Master Qualification, as well as $.50 per participant for the
Premium Qualification of specifying birth year range. Including the pilot study, the
overall cost of this study was $3724.80.
c. Funding
99
Funding for this study has very generously been provided by Dr. Barbara Curbow
and the Department of Behavioral and Community Health.
100
Table A1: Questionnaire constructs and corresponding question numbers
Construct Corresponding Source and Psychometrics
questions in
survey
Health Literacy Questions 2-5 The BRIEF Health Literacy Screening Tool
(Haun, Noland-Dodd, Graham-Pole, Rienzo, &
Donaldson, 2009)
Concurrent validity for the BRIEF was assessed and
found to be acceptable:
Pearson correlation results were:
r (378) = .40 (P < .01) for the BRIEF and the
REALM
r (378) = .42 (P < .01) for the BRIEF and
the STOFHLA
r (378) = .61 (P < .01) for the REALM and
the STOFHLA
Numeracy Questions 6-13 The Subjective Numeracy Scale
(Fagerlin et al., 2007; Zikmund-Fisher et al., 2007)
The SNS is both reliable and highly correlated with
the Lipkus, Samsa & Rimer (2001) numeracy
measure
It has been validated in both risk communication and
utility elicitation domains
Perceived Question 33 Adapted from the Health Information National
Susceptibility Trends Survey (HINTS)
Perceived Question 34 Adapted from the Health Information National
severity Trends Survey (HINTS)
Perceived Question 35 Created by the study team
Importance of
Screening
Knowledge of Question 14 Adapted from the Health Information National
CRC screening Trends Survey (HINTS)
Previous CRC Question 16 Adapted from the Health Information National
screen Trends Survey (HINTS)
101
Provider Question 74 -77 Adapted from the Health Information National
Relationship Trends Survey (HINTS)
Ability to Question 23 Created by study team based on previous research
Interpret Report
Demographics Questions 74-89 Adapted from CDC measures and previously used
by the research team
102
Appendix B: IRB Documentation
103
Appendix C: Questionnaire
Note: Headers were not visible to the person completing the questionnaire. Please use the
following link to see the actual formatting of the questionnaire: Mueller_Survey_Draft_9.25.17.
Intro
Q1 Hello!
This survey examines people's thoughts about cancer risk and related subjects. Please read each
question carefully and answer truthfully.
We are looking for your honest and immediate responses to the information. We ask that you do
not use any outside resources to help you answer the questions.
Also, you may learn new information as you go through the survey. Please do not go back to
change your answers based on information you learn from the survey.
Thanks!
End of Block
Health Literacy
Q2 How confident are you filling out medical forms by yourself?
o Extremely (1)
o Quite a bit (2)
o Somewhat (3)
o A little bit (4)
o Not at all (5)
Q3 How often do you have someone help you read medical materials?
o Always (1)
o Often (2)
104
o Occasionally (3)
o Sometimes (4)
o Never (5)
Q4 How often do you have problems learning about your medical condition because of
difficulty understanding written information?
o Always (1)
o Often (2)
o Occasionally (3)
o Sometimes (4)
o Never (5)
Q5 How often do you feel you don’t understand what the doctor tells you?
o Always (1)
o Often (2)
o Occasionally (3)
o Sometimes (4)
o Never (5)
End of Block
Subjective Numeracy
Q6 How good are you at working with fractions?
o 1 Not at all good (1)
o 2 (2)
o 3 (3)
105
o 4 (4)
o 5 (5)
o 6 Extremely good (6)
Q7 How good are you at working with percentages?
o 1 Not at all good (1)
o 2 (2)
o 3 (3)
o 4 (4)
o 5 (5)
o 6 Extremely good (6)
Q8 How good are you at calculating a 15% tip?
o 1 Not at all good (1)
o 2 (2)
o 3 (3)
o 4 (4)
o 5 (5)
o 6 Extremely good (6)
Q9 How good are you at figuring out how much a shirt will cost if it is 25% off?
o 1 Not at all good (1)
106
o 2 (2)
o 3 (3)
o 4 (4)
o 5 (5)
o 6 Extremely good (6)
Q10 When reading the newspaper, how helpful do you find tables and graphs that are parts of a
story?
o 1 Not at all helpful (1)
o 2 (2)
o 3 (3)
o 4 (4)
o 5 (5)
o 6 Extremely helpful (6)
Q11 When people tell you the chance of something happening, do you prefer that they use words
("it rarely happens") or numbers ("there's a 1% chance")?
o 1 Always prefer words (1)
o 2 (2)
o 3 (3)
o 4 (4)
o 5 (5)
o 6 Always prefer numbers (6)
107
Q12 When you hear a weather forecast, do you prefer predictions using percentages (e.g., “there
will be a 20% chance of rain today”) or predictions using only words (e.g., “there is a small
chance of rain today”)?
o 1 Always prefer percentages (1)
o 2 (2)
o 3 (3)
o 4 (4)
o 5 (5)
o 6 Always prefer words (6)
Q13 How often do you find numerical information to be useful?
o 1 Never (1)
o 2 (2)
o 3 (3)
o 4 (4)
o 5 (5)
o 6 Very often (6)
End of Block
Colon and rectal cancer
Q14 Which of these screening tests detect colon or rectal cancer? Check all that apply.
▢ Mammography (1)
▢ Colonoscopy (2)
▢ Pap Smear (3)
108
▢ Stool blood test (aka FOBT) (4)
▢ Prostate Specific Antigen (PSA) (5)
▢ Stool DNA test (6)
▢ Sigmoidoscopy (7)
Q15 Has a doctor ever recommended you to get a screening test to check for colon or rectal
cancer?
o No (1)
o Yes (2)
o I don't know (3)
Q16 Have you ever had a screening test for colon or rectal cancer?
o No (1)
o Yes. Which one? (2) ________________________________________________
o I don't know (3)
Q17 Which statement is closest to where you are now in your plans to get a screening test for
colon or rectal cancer?
o I have already been tested (1)
o I am not thinking of getting tested (2)
o I think I need to consider getting tested (3)
o I think I should get tested but I’m not quite ready (4)
o I think I will probably get tested (5)
o I am committed to getting tested (6)
109
Q18 At what age should someone of average risk for colon or rectal cancer begin regular
screening?
________________________________________________________________
Q19 Rank the causes below from very important (1) to not important (7) in increasing the
likelihood that you will develop colon or rectal cancer. Click on an answer and drag it into the
position you think appropriate.
______ Behavior (1)
______ Character (2)
______ Social Environment (3)
______ Physical Environment (4)
______ Genetics (5)
______ Chance (6)
______ Society (7)
End of Block
Colonoscopy Reports
Q20 On the next page, you will be asked to read the results of a fictional colonoscopy. A
colonoscopy is a medical procedure that examines your rectum and colon with a small scope. We
then ask you to answer some questions about the information in the report. We are looking for
your honest and immediate responses to the information. Again, please do not use any outside
resources to help you answer the questions.
GIST Q21
These are the findings from your colonoscopy. They are separated into Findings, Impressions
and Recommendations. The Findings will be useful for your primary care doctor or specialist.
The Impressions summarize what we found today, and the Recommendations outline your next
steps.
Findings
The perianal exam was abnormal. Findings included non-thrombosed external hemorrhoids. A
multilobed non-obstruction large mass was found in the cecum, arising from the area of the
appendiceal orifice. The mass was non-circumferential. The mass measured 5-6cm in its largest
dimension. No bleeding was present. Biopsies taken with cold forceps for histology. A semi
pedunculated polyp was found in transverse colon. The polyp was 10mm in size. The polyp was
removed with hot snare. Resection and retrieval were complete. Estimated blood loss: none. A
pedunculated non-bleeding polyp was found in the descending colon. The polyp was 12mm in
110
size. The polyp was removed with a hot snare. Resection and retrieval with a roth net were
complete. Multiple small and large-mouthed diverticula were found in the entire colon. Non-
bleeding internal hemorrhoids were found in retroflexion and were small.
Impressions
We found a mass in your colon and think it is cancer. We are running a test to see if it is cancer
and will let you know as soon as we find out.
Your colon has small pouches which are inflamed or infected. This is a condition called
Diverticulitis.
We also found hemorrhoids on both the inside and outside of your anus.
Recommendations
We have referred you to a colorectal surgeon. It is important that you see this surgeon as soon as
possible.
Results about your mass will be sent to you in a few days. Be on the lookout for this in the mail.
You can go back to eating as you did before. However, a diet with lots of fiber and extra liquids
(such as water) may help with your diverticulitis and hemorrhoids. Please talk to your doctor
about this if you have concerns.
VERBATIM REPORT
Q22 These are the findings from your colonoscopy.
Findings
The perianal exam was abnormal. Findings included non-thrombosed external hemorrhoids. A
multilobed non-obstruction large mass was found in the cecum, arising from the area of the
appendiceal orifice. The mass was non-. The mass measured 5-6cm in its largest dimension. No
bleeding was present. Biopsies taken with cold forceps for histology. A semi pedunculated polyp
was found in transverse colon. The polyp was 10mm in size. The polyp was removed with hot
snare. Resection and retrieval were complete. Estimated blood loss: none. A pedunculated non-
bleeding polyp was found in the descending colon. The polyp was 12mm in size. The polyp was
removed with a hot snare. Resection and retrieval with a roth net were complete. Multiple small
and large-mouthed diverticula were found in the entire colon. Non-bleeding internal hemorrhoids
were found in retroflexion and were small.
111
Impression
Cecal mass, likely malignant. Biopsied.
One 10mm polyp in transverse colon. Resected and retrieved.
One 12mm, non-bleeding polyp in descending colon. Resected and retrieved.
Diverticulitis in entire examined colon.
Internal and external hemorrhoids.
Recommendation
Referral to colorectal surgery.
Check CEA today.
Resume previous diet as tolerated.
Observe until discharge criteria are met.
Discharge home.
Await pathology results.
Q23 Please answer the questions below by moving the slider to the appropriate number.
How serious is this report? (4)
How important is it that this patient follow
up with a doctor? (5)
To what extent is there a plan of care in place
for this patient? (3)
Q24 Please rank the following by how important it would be for you to have, with (1) indicating
most important and (6) indicating least important.
______ A table or graph showing how my findings compare to normal findings (1)
______ The percentage of masses like this one that are cancerous (2)
______ The percentage of times that biopsy results like this are accurate (3)
______ Whether the doctor thinks this is cancer (4)
______ Whether the doctor thinks the situation is serious (5)
______ How much trust the doctor will have in the results of the biopsy (6)
Q25 If this colonoscopy report belonged to your parent, and he or she asked you what it meant,
what would you say?
112
________________________________________________________________
Q26 Please tell us why you would explain it that way.
________________________________________________________________
Q27 Think about the recommendations the doctor gave you in the report. How would you
describe the wording of the recommendations?
o It was written in medical language (1)
o It was written in plain language (2)
Q28 A1 and A2 Cancer statement
Q28 Please read the summary below about colon cancer rates and answer the questions on the
next page. You can use the red back button to return to the statement at any time.
From the mid-1980s through 2013, colon cancer rates:
declined in adults age 55 years and older
increased by 2.4% per year in adults 20 to 29 years
increased by 1% per year in adults age 30 to 39 years
Q29 Please read the summary below about colon cancer rates and answer the questions on the
next page. You can use the red back button to return to the statement at any time.
From the mid-1980s through 2013, colon cancer rates:
increased in adults age 55 years and older
declined by 2.4% per year in adults 20 to 29 years
declined by 1% per year in adults age 30 to 39 years.
End of Block
Cancer Statement A1 and A2
113
Q30 What do these data say is happening to colon cancer in the US?
________________________________________________________________
Q31 Do you think the statement about colon cancer is true or false?
o True (1)
o False (2)
Q32 On a scale of 0 (not at all sure) to 10 (completely sure) how sure are you that your answer to
the last question is correct? Please move the slider to indicate your choice.
0 (0)
1 (1)
2 (2)
3 (3)
4 (4)
5 (5)
6 (6)
7 (7)
8 (8)
9 (9)
10 (10)
End of Block
114
Risk of cancer
Q33 How likely do you think it is that you will develop any of the cancers listed below in
your lifetime? Please be sure to indicate the likelihood for each type of cancer by moving the
slider to the right.
Skin (1)
Lung (2)
Human-Papilloma (HPV) Related (anal,
penile, cervical or head and neck) (3)
Colon or Rectal (4)
Breast (5)
Any Type (6)
Q34 If diagnosed with advanced stage cancer, how likely do you think it is that you would die
from any of the cancers listed below? Please be sure to indicate the likelihood for each type of
cancer by moving the slider to the right.
Skin (1)
Lung (2)
Human-Papilloma (HPV) Related (anal,
penile, cervical or head and neck) (3)
Colon or Rectal (4)
Breast (5)
115
Any Type (6)
Q35 How important is it that you are screened for any of the types of cancer listed below in
your lifetime? Please be sure to indicate the importance for each type of cancer by moving the
slider to the right.
Skin (1)
Lung (2)
Human-Papilloma (HPV) Related (anal,
penile, cervical or head and neck) (3)
Colon or Rectal (4)
Breast (5)
Any Type (6)
Q36 Have you ever been diagnosed with cancer?
o No (1)
o Yes. What type? (2) ________________________________________________
Q37 Have you ever been diagnosed with a pre-cancerous condition?
o No (1)
o Yes. On what location on body, or what type of cancer? (2)
________________________________________________
o I don't know (3)
116
Q38 Has anyone in your family ever been diagnosed with cancer?
o No (1)
o Yes. Who, and what type of cancer? (2)
________________________________________________
o I don't know (3)
Q39 Has anyone close to you died of cancer?
o No (1)
o Yes. Who, and what type of cancer? (2)
________________________________________________
o I don't know (3)
Q40 Has cancer impacted your life in any other way?
o No (1)
o Yes. How so? (2) ________________________________________________
Genetic Questions
Q41 Researchers have found that some people have distinct genetic changes, also referred to as
mutations. These mutations may raise a person’s risk for some types of cancer. When someone is
at risk, they are often referred to as having “positive mutation status”, meaning they carry that
mutation.
Q42 Have you ever had a genetic test to check your mutation status that was conducted by a
medical professional?
o No (1)
o Yes (2)
o I don't know (3)
117
Q43 What was your test result?
o Positive mutation status. Type of cancer(s): (1)
________________________________________________
o Negative mutation status. Type of cancer(s): (2)
________________________________________________
o Uninformative/Indeterminate/Inconclusive (3)
o I prefer not to answer (4)
o I don't know (5)
Q44 Have you ever had a direct to consumer genetic test (for example, 23andMe). In other
words, you sent a sample of DNA to a private company and you received information about
your DNA?
o No (1)
o Yes (2)
o I don't know (3)
Q45 Did your direct to consumer genetic test provide information about your cancer mutation
status?
o No (1)
o Yes. About what type of cancer(s)? (2)
________________________________________________
o I prefer not to answer (3)
o I don't know (4)
Q46 What was your test result?
o Positive mutation status. What type of cancers? (1)
________________________________________________
118
o Negative mutation status. What type of cancers? (2)
________________________________________________
o Uninformative/Indeterminate/Inconclusive (3)
o I prefer not to answer (4)
o I don't know (5)
Q47 To what extent do you believe it is ethical to provide cancer risk data to people?
(1) not at all likely (10) very likely (1)
Evaluation Of Treatment Strategies To Reduce Drug Craving And Harm In People With Opioid
Use Disorder Protocol Development And Implementation Task Order
End of Block
Social Desirability
Q48 I am always courteous even to people who are disagreeable.
o Definitely true (1)
o Mostly true (2)
o Don't know (3)
o Mostly false (4)
o Definitely false (5)
Q49 There have been occasions when I took advantage of someone.
o Definitely true (1)
o Mostly true (2)
o Don't know (3)
119
o Mostly false (4)
o Definitely false (5)
Q50 I sometimes try to get even rather than forgive and forget.
o Definitely true (1)
o Mostly true (2)
o Don't know (3)
o Mostly false (4)
o Definitely false (5)
Q51 I sometimes feel resentful when I don't get my way.
o Definitely true (1)
o Mostly true (2)
o Don't know (3)
o Mostly false (4)
o Definitely false (5)
Q52 No matter who I'm talking to, I'm always a good listener.
o Definitely true (1)
o Mostly true (2)
o Don't know (3)
o Mostly false (4)
o Definitely false (5)
120
End of Block
Social Isolation
Q53 I experience a general sense of emptiness.
o Yes! (1)
o Yes (2)
o More or less (3)
o No (4)
o No! (5)
Q54 I miss having people around.
o Yes! (1)
o Yes (2)
o More or less (3)
o No (4)
o No! (5)
Q55 I often feel rejected.
o Yes! (1)
o Yes (2)
o More or less (3)
o No (4)
o No! (5)
Q56 There are plenty of people I can rely on when I have problems.
121
o Yes! (1)
o Yes (2)
o More or less (3)
o No (4)
o No! (5)
Q57 There are many people I can trust completely.
o Yes! (1)
o Yes (2)
o More or less (3)
o No (4)
o No! (5)
Q58 There are enough people I feel close to.
o Yes! (1)
o Yes (2)
o More or less (3)
o No (4)
o No! (5)
End of Block
PC Questions
Q59 On an average weekday, about how many hours do you spend in the physical presence of
other people?
122
________________________________________________________________
Q60 On an average weekday, about how many hours do you spend in the virtual
(online) presence of other people?
________________________________________________________________
Q61 On an average weekend day, about how many hours do you spend in the physical presence
of other people?
________________________________________________________________
Q62 On an average weekend day, about how many hours do you spend in the virtual presence
of other people?
________________________________________________________________
General Health
Q63 In general, how would you describe your health?
o Excellent (1)
o Very good (2)
o Good (3)
o Fair (4)
o Poor (5)
o Very poor (6)
Q64 Now thinking about your physical health, which includes physical illness and injury, for
how many days during the past 30 days was your physical health not good?
Q65 Now thinking about your mental health, which includes stress, depression, and problems
with emotions, for how many days during the past 30 days was your mental health not good?
________________________________________________________________
123
Q66 During the past 30 days, for about how many days did poor physical or mental health
keep you from doing your usual activities, such as self-care, work, or recreation?
________________________________________________________________
Q67 During the past 30 days, for about how many days did pain make it hard for you to do
your usual activities, such as self-care, work, or recreation?
________________________________________________________________
Q68 Do you have a primary medical provider you see at least once a year?
o No (1)
o Yes (2)
Q69 How would you describe your relationship with this provider?
o Excellent (6)
o Very good (5)
o Good (4)
o Fair (3)
o Poor (2)
o Very poor (1)
Q70 Please tell us why you would describe your relationship that way.
________________________________________________________________
Q71 Do you have health insurance?
o No (1)
o Yes (2)
Q72 How would you rate the quality of your health insurance?
124
o Excellent (6)
o Very good (5)
o Good (4)
o Fair (3)
o Poor (2)
o Very poor (1)
Q73 Please tell us why you would rate your insurance this way.
________________________________________________________________
End of Block
Demographics
Q74 With which gender do you identify?
o Female (1)
o Male (2)
o Other (3) ________________________________________________
Q75 How old were you on your last birthday?
________________________________________________________________
Q76 Are you of Hispanic or Latino descent?
o No (1)
o Yes (2)
Q77 With what race do you identify?
125
o White/Caucasian (1)
o Black/African American (2)
o American Indian or Alaskan Native (3)
o Asian/Asian American (4)
o Native Hawaiian/Pacific Islander (5)
o Other (6) _______________________________________________
Q78 About how many hours a week do you work for MTurk or in similar jobs?
________________________________________________________________
Q79 Other than MTurk and similar jobs, are you employed for pay?
▢ Yes, full-time (1)
▢ Yes, part-time (2)
▢ No, I'm disabled (3)
▢ No, I'm retired (4)
▢ No, I'm looking for work (5)
▢ No, MTurk or similar jobs are my only jobs (6)
▢ I'm a homemaker (7)
▢ Other (8) ________________________________________________
Q80 Are you currently attending school?
o No (1)
126
o Yes, college (2)
o Yes, graduate or professional school (3)
o Yes, other: (4) ________________________________________________
Q81 Do you routinely care for another person who has a disability, serious illness or advanced
age?
o No (1)
o Yes. For hour many hours a week? (2)
________________________________________________
Q82 What is your relationship to this person? He/she is...
o My grandparent (1)
o My parent (2)
o My child. How old is this child? (3)
________________________________________________
o My spouse or partner (4)
o Other (5) ________________________________________________
Q83 Do you consider yourself the primary caregiver for this person?
o No (1)
o Yes (2)
Q84 How much stress does routinely caring for this person cause you? Please indicate your
choice by moving the slider to the appropriate number. 1 indicates no stress, 10 indicates high
stress.
127
Emotional stress (1)
Physical stress (2)
Financial stress (3)
Q85 What is your marital status?
o Married/legal domestic partnership (1)
o Living with partner (2)
o Divorced (3)
o Widowed (4)
o Single (5)
o Other (6) ________________________________________________
Q86 Do you have dependent children under the age of 18 living with you?
o No (1)
o Yes (2)
Q87 What is the highest grade of school you have completed? Please move the slider to the
appropriate number.
Elementary (0-8)
High School (9-12)
GED (12)
College (13-16)
Graduate School or above (17)
Please indicate the highest year (1)
128
Q88 Please estimate your yearly household income.
o less than $10,000 (1)
o $10,000-$19,999 (2)
o $20,000-$29,999 (3)
o $30,000-$39,999 (4)
o $40,000-$49,999 (5)
o $50,000-$59,999 (6)
o $60,000-$69,999 (7)
o $70,000-$79,999 (8)
o 80,000-$89,999 (9)
o $90,000-$99,999 (10)
o $100,000 or above (11)
Q89 How many people including yourself are supported on this income?
________________________________________________________________
129
Appendix D: MTurk Interface
Coding
The University of Maryland requests the below task to be completed. The
task consists of a survey which will take about 15 minutes to complete. The compensation for
this task is $3.00.
The survey will ask questions about your experience with health and health care, and
will ask you to interpret some medical information. There are no right or wrong answers to any
of the questions.
This task is being requested by researchers in the School of Public Health at the
University of Maryland, College Park. There are no known risks to participating in this task. Any
answers you provide as part of this task will be kept confidential to the best of our ability and not
shared with Amazon. Any potential loss of confidentiality will be minimized by using Qualtrics,
not Amazon, to collect and store data; by storing the final data on a secure UMD drive, to which
only the research team has access.
If we write a report or article about this task, your identity will be protected to the
maximum extent possible. Your information may be shared with representatives of the
University of Maryland, College Park or governmental authorities if you or someone else is in
danger or if we are required to do so by law.
Your participation in this task is completely voluntary. You may choose not to take
part at all. If you decide to participate in this task, you may stop participating at any
time. If you decide not to participate in this study or if you stop participating at any time,
you will not receive payment for the task.
130
If you decide to stop taking part in the study, if you have questions, concerns, or
complaints, or if you need to report an injury related to the task, please contact the
investigator:
Nora Mueller
Department of Behavioral and Community Health
School of Public Health
University of Maryland, College Park 20742
If you have questions about your rights as a participant or wish to report a task-related
injury, please contact:
University of Maryland College Park
Institutional Review Board Office
1204 Marie Mount Hall
College Park, Maryland, 20742
131
Telephone: 301-405-0678
This task has been reviewed according to the University of Maryland, College Park
IRB procedures for research involving human subjects.
By accepting this task, you indicate that you are at least 18 years of age; you have
read this form or have had it read to you; your questions have been answered to your satisfaction
and you voluntarily agree to participate.
133
MTurk Interface
The University of Maryland requests the below task to be completed. The task consists of a
survey which will take about 15 minutes to complete. The compensation for this task is
$3.00.
The survey will ask questions about your experience with health and health care and will ask you
to interpret some medical information. There are no right or wrong answers to any of the
questions.
This task is being requested by researchers in the School of Public Health at the University of
Maryland, College Park. There are no known risks to participating in this task. Any answers you
provide as part of this task will be kept confidential to the best of our ability and not shared with
Amazon. Any potential loss of confidentiality will be minimized by using Qualtrics, not
Amazon, to collect and store data and by storing the final data on a secure UMD drive, to which
only the research team has access.
If we write a report or article about this task, your identity will be protected to the maximum
extent possible. Your information may be shared with representatives of the University of
Maryland, College Park or governmental authorities if you or someone else is in danger or if we
are required to do so by law.
Your participation in this task is completely voluntary. You may choose not to take part at all. If
you decide to participate in this task, you may stop participating at any time. If you decide not to
participate in this study or if you stop participating at any time, you will not receive payment for
the task.
If you decide to stop taking part in the study, if you have questions, concerns, or complaints, or if
you need to report an injury related to the task, please contact the investigator:
Nora Mueller
Department of Behavioral and Community Health
School of Public Health
University of Maryland, College Park 20742
nmueller@umd.edu
(703) 581-5285
If you have questions about your rights as a participant or wish to report a task-related injury,
please contact:
134
University of Maryland College Park
Institutional Review Board Office
1204 Marie Mount Hall
College Park, Maryland, 20742
E-mail: irb@umd.edu
Telephone: 301-405-0678
This task has been reviewed according to the University of Maryland, College Park IRB
procedures for research involving human subjects.
By accepting this task, you indicate that you are at least 18 years of age; you have read this form
or have had it read to you; your questions have been answered to your satisfaction and you
voluntarily agree to participate.
Survey link: https://umdsurvey.umd.edu/jfe/form/SV_50dnv2RAhzxFbjT
Provide the survey code here:
135
References
. (2007).
ACS. (2017a). American Cancer Society Recommendations for Colorectal Cancer Early Detection.
Retrieved from https://www.cancer.org/cancer/colon-rectal-cancer/detection-diagnosis-
staging/acs-recommendations.html
ACS. (2017b). Colorectal Cancer Fact & Figures 2017-2019. Atlanta: American Cancer Society.
ACS. (2017c). The Cost of Cancer. American Cancer Society Cancer Action Network Public Policy
Resource. doi:https://www.acscan.org/policy-resources/costs-cancer
Adams, R. J., Smith, B. J., & Ruffin, R. E. (2001). Patient preferences for autonomy in decision making in
asthma management. Thorax, 56(2), 126-132. doi:10.1136/thorax.56.2.126
Agresti, A. (2015). Foundations of Linear and Generalized Linear Models. Hoboken, NJ: John Wiley &
Sons.
AHRQ. (2016). Total Expenses and Percent Distribution for Selected Conditions by Source of Payment:
United States, 2014. Medical Expenditure Panel Survey Household Component Data, Generated
Interactively.
Al Sayah, F., Majumdar, S. R., Williams, B., Robertson, S., & Johnson, J. A. (2013). Health literacy and
health outcomes in diabetes: a systematic review. J Gen Intern Med, 28(3), 444-452.
doi:10.1007/s11606-012-2241-z
Alberti, L. R., Garcia, D. P. C., Coelho, D. L., De Lima, D. C. A., & Petroianu, A. (2015). How to improve
colon cancer screening rates. World Journal of Gastrointestinal Oncology, 7(12), 484-491.
doi:10.4251/wjgo.v7.i12.484
Allen, S. V., Solberg Nes, L., Marnach, M. L., Polga, K., Jenkins, S. M., Files, J. A., . . . Pruthi, S. (2012).
Patient understanding of the revised USPSTF screening mammogram guidelines: need for
development of patient decision aids. BMC Women's Health, 12(1), 36. doi:10.1186/1472-6874-
12-36
America's Health Literacy: Why we Need Accessible Health Information. An Issue Brief from the U.S.
Department of Health and Human Services. (2008).
Araghi, M., Fidler, M. M., Arnold, M., Jemal, A., Bray, F., & Soerjomataram, I. (2018). The Future Burden
of Colorectal Cancer Among US Blacks and Whites. JNCI: Journal of the National Cancer Institute,
110(7), 791-793. doi:10.1093/jnci/djx287
ASGE. (2014). Colorectal Cancer Screening [Press release]
Atkin, W. S., Edwards, R., Kralj-Hans, I., Wooldrage, K., Hart, A. R., Northover, J. M. A., . . . Cuzick, J.
(2010). Once-only flexible sigmoidoscopy screening in prevention of colorectal cancer: a
multicentre randomised controlled trial. The Lancet, 375(9726), 1624-1633.
doi:https://doi.org/10.1016/S0140-6736(10)60551-X
Augustus, G. J., & Ellis, N. A. (2018). Colorectal Cancer Disparity in African Americans: Risk Factors and
Carcinogenic Mechanisms. Am J Pathol, 188(2), 291-303. doi:10.1016/j.ajpath.2017.07.023
Aune, D., Chan, D. S. M., Lau, R., Vieira, R., Greenwood, D. C., Kampman, E., & Norat, T. (2011). Dietary
fibre, whole grains, and risk of colorectal cancer: systematic review and dose-response meta-
analysis of prospective studies. BMJ, 343. doi:10.1136/bmj.d6617
Baker, D. W., Parker, R. M., Williams, M. V., Clark, W. S., & Nurss, J. (1997). The relationship of patient
reading ability to self-reported health and use of health services. Am J Public Health, 87(6),
1027-1030.
Baron, R. C., Rimer, B. K., Breslow, R. A., Coates, R. J., Kerner, J., Melillo, S., . . . Briss, P. A. (2008). Client-
Directed Interventions to Increase Community Demand for Breast, Cervical, and Colorectal
Cancer Screening: A Systematic Review. American Journal of Preventive Medicine, 35(1,
Supplement), S34-S55. doi:http://dx.doi.org/10.1016/j.amepre.2008.04.002
136
Basch, C. H., Basch, C. E., Zybert, P., & Wolf, R. L. (2016). Fear as a Barrier to Asymptomatic Colonoscopy
Screening in an Urban Minority Population with Health Insurance. Journal of Community Health,
41(4), 818-824. doi:10.1007/s10900-016-0159-9
Batterham, R. W., Hawkins, M., Collins, P. A., Buchbinder, R., & Osborne, R. H. (2016). Health literacy:
applying current concepts to improve health services and reduce health inequalities. Public
Health, 132, 3-12. doi:https://doi.org/10.1016/j.puhe.2016.01.001
Bentley, F. R., Daskalova, N., & White, B. (2017). Comparing the Reliability of Amazon Mechanical Turk
and Survey Monkey to Traditional Market Research Surveys. Paper presented at the Proceedings
of the 2017 CHI Conference Extended Abstracts on Human Factors in Computing Systems,
Denver, Colorado, USA.
Berinsky, A. J., Huber, G. A., & Lenz, G. S. (2012). Evaluating Online Labor Markets for Experimental
Research: Amazon.com's Mechanical Turk. Political Analysis, 20(3), 351-368.
Berkman, N. D., Sheridan, S. L., Donahue, K. E., Halpern, D. J., Viera, A., Crotty, K., . . . Viswanathan, M.
(2011). Health literacy interventions and outcomes: an updated systematic review. Evid Rep
Technol Assess (Full Rep)(199), 1-941.
Bjorklund, D. F., & Buchanan, J. J. (1989). Developmental and knowledge base differences in the
acquisition and extension of a memory strategy. Journal of Experimental Child Psychology, 48(3),
451-471. doi:http://dx.doi.org/10.1016/0022-0965(89)90052-0
Blalock, S. J., & Reyna, V. F. (2016). Using fuzzy-trace theory to understand and improve health
judgments, decisions, and behaviors: A literature review. Health psychology : official journal of
the Division of Health Psychology, American Psychological Association, 35(8), 781-792.
doi:10.1037/hea0000384
Borges, N. J., Manuel, R. S., Elam, C. L., & Jones, B. J. (2006). Comparing millennial and generation X
medical students at one medical school. Acad Med, 81(6), 571-576.
doi:10.1097/01.acm.0000225222.38078.47
Borges, N. J., Manuel, R. S., Elam, C. L., & Jones, B. J. (2010). Differences in motives between Millennial
and Generation X medical students. Med Educ, 44(6), 570-576. doi:10.1111/j.1365-
2923.2010.03633.x
Brown, M. T., & Bussell, J. K. (2011). Medication adherence: WHO cares? Mayo Clin Proc, 86(4), 304-314.
doi:10.4065/mcp.2010.0575
Buhrmester, M., Kwang, T., & Gosling, S. D. (2011). Amazon's Mechanical Turk. Perspectives on
Psychological Science, 6(1), 3-5. doi:10.1177/1745691610393980
Campos, F. G. (2017). Colorectal cancer in young adults: A difficult challenge. World journal of
gastroenterology, 23(28), 5041-5044. doi:10.3748/wjg.v23.i28.5041
Cancer Facts & Figures 2013: Atlanta: American Cancer Society; 2013
CDC. (2013). Retrieved from http://www.cdc.gov/cancer/colorectal/statistics/race.htm
Carcaise-Edinboro, P., & Bradley, C. J. (2008). Influence of Patient-Provider Communication on Colorectal
Cancer Screening. Med Care, 46(7), 738-745. doi:10.1097/MLR.0b013e318178935a
Cassese, E. C., Huddy, L., Hartman, T. K., Mason, L., & Weber, C. R. (2013). Socially Mediated Internet
Surveys: Recruiting Participants for Online Experiments. PS: Political Science and Politics, 46(4),
775-784.
CDC. (2019). Colorectal Cancer Screening Rates.
Chandler, J., & Shapiro, D. (2016). Conducting Clinical Research Using Crowdsourced Convenience
Samples. Annu Rev Clin Psychol, 12, 53-81. doi:10.1146/annurev-clinpsy-021815-093623
Chew, L. D., Bradley, K. A., & Boyko, E. J. (2004). Brief questions to identify patients with inadequate
health literacy. Fam Med, 36(8), 588-594.
137
Cho, R. N., Plunkett, B. A., Wolf, M. S., Simon, C. E., & Grobman, W. A. (2007). Health literacy and patient
understanding of screening tests for aneuploidy and neural tube defects. Prenatal Diagnosis,
27(5), 463-467. doi:10.1002/pd.1712
Clouston, S. A. P., Rubin, M. S., Chae, D. H., Freese, J., Nemesure, B., & Link, B. G. (2017). Fundamental
causes of accelerated declines in colorectal cancer mortality: Modeling multiple ways that
disadvantage influences mortality risk. Social Science & Medicine, 187(Supplement C), 1-10.
doi:https://doi.org/10.1016/j.socscimed.2017.06.013
Coleman, C., Kurtz-Rossi, S., & McKinney, J. (2009). Calgary Charter on Health Literacy.
Cornett, S. (2009). Assessing and Addressing Health Literacy. OJIN: The Online Journal of Issues in
Nursing, 14(3).
Curbow, B., King-Marshall, E., Mueller, N., Sultan, S., & George, T. (under review). Social vs. practical
problems in attaining a colonoscopy: Different profiles?
Curbow, B. A., Dailey, A. B., King-Marshall, E. C., Barnett, T. E., Schumacher, J. R., Sultan, S., & George, T.
J., Jr. (2015). Pathways to colonoscopy in the South: seeds of health disparities. Am J Public
Health, 105(4), e103-111. doi:10.2105/ajph.2014.302347
Cutilli, C. C., & Bennett, I. M. (2009). Understanding the Health Literacy of America Results of the
National Assessment of Adult Literacy. Orthopaedic nursing / National Association of
Orthopaedic Nurses, 28(1), 27-34. doi:10.1097/01.NOR.0000345852.22122.d6
Davis, D. M., Marcet, J. E., Frattini, J. C., Prather, A. D., Mateka, J. J., & Nfonsam, V. N. (2011). Is it time to
lower the recommended screening age for colorectal cancer? J Am Coll Surg, 213(3), 352-361.
doi:10.1016/j.jamcollsurg.2011.04.033
Davis, T. C., Dolan, N. C., Ferreira, M. R., Tomori, C., Green, K. W., Sipler, A. M., & Bennett, C. L. (2001).
The role of inadequate health literacy skills in colorectal cancer screening. Cancer Invest, 19(2),
193-200.
Davis, T. C., Williams, M. V., Marin, E., Parker, R. M., & Glass, J. (2002). Health literacy and cancer
communication. CA Cancer J Clin, 52(3), 134-149.
Deen, K. I., Silva, H., Deen, R., & Chandrasinghe, P. C. (2016). Colorectal cancer in the young, many
questions, few answers. World J Gastrointest Oncol, 8(6), 481-488. doi:10.4251/wjgo.v8.i6.481
DeMaria, J., & Hoffman, C. (2016). Generational Infuences on Patient Perspectives Impacting Patient
Engagement. Paper presented at the Healthcare Information and Management Systems Society,
Las Vegas.
Difallah, D. E., Catasta, M., Demartini, G., Ipeirotis, P. G., Cudr, P., #233, & -Mauroux. (2015). The
Dynamics of Micro-Task Crowdsourcing: The Case of Amazon MTurk. Paper presented at the
Proceedings of the 24th International Conference on World Wide Web, Florence, Italy.
Dreyfuss, E. (2018, 08.18.2018). A Bot Panic Hits Amazon's Mechnical Turk. Wired.
Edwards, B. K., Ward, E., Kohler, B. A., Eheman, C., Zauber, A. G., Anderson, R. N., . . . Ries, L. A. G.
(2010a). Annual report to the nation on the status of cancer, 1975-2006, featuring colorectal
cancer trends and impact of interventions (risk factors, screening, and treatment) to reduce
future rates. Cancer, 116(3), 544-573. doi:10.1002/cncr.24760
Edwards, B. K., Ward, E., Kohler, B. A., Eheman, C., Zauber, A. G., Anderson, R. N., . . . Ries, L. A. G.
(2010b). Annual Report to the Nation on the Status of Cancer, 1975–2006, Featuring Colorectal
Trends and Impact of Interventions (Risk Factors, Screening, and Treatment) to Reduce Future
Rates. Cancer, 116(3), 544-573. doi:10.1002/cncr.24760
Fagerlin, A., Zikmund-Fisher, B. J., Ubel, P. A., Jankovic, A., Derry, H. A., & Smith, D. M. (2007). Measuring
numeracy without a math test: development of the Subjective Numeracy Scale. Med Decis
Making, 27(5), 672-680. doi:10.1177/0272989x07304449
Forsberg, A. M., Hagel, E., Jaramillo, E., Rubio, C. A., Bjorck, E., & Lindblom, A. (2015). Predicting
Outcome in Colonoscopic High-risk Surveillance. Anticancer Res, 35(9), 4813-4819.
138
Fox, S., & Duggan, M. (2012). Mobile Health 2012. Retrieved from
http://www.pewinternet.org/files/oldmedia//Files/Reports/2012/PIP_MobileHealth2012_FINAL
.pdf.
Gallagher, L., McAuley, J., & Moseley, G. L. (2013). A randomized-controlled trial of using a book of
metaphors to reconceptualize pain and decrease catastrophizing in people with chronic pain.
Clin J Pain, 29(1), 20-25. doi:10.1097/AJP.0b013e3182465cf7
Giaj Levra, N., Cuniberti, F. A., Rava, A., Vietti, G., & Sciascia, S. (2012). Health Literacy and Discharge
Instruction Adherence. Journal of General Internal Medicine, 27(3), 273-273.
doi:10.1007/s11606-011-1969-1
Goodman, J. K., Cryder, C. E., & Cheema, A. (2013). Data Collection in a Flat World: The Strengths and
Weaknesses of Mechanical Turk Samples. Journal of Behavioral Decision Making, 26(3), 213-224.
doi:10.1002/bdm.1753
Graham, S., & Brookey, J. (2008). Do patients understand? The Permanente journal, 12(3), 67-69.
Gurmankin, A. D., Baron, J., & Armstrong, K. (2004). Intended Message Versus Message Received in
Hypothetical Physician Risk Communications: Exploring the Gap. Risk Analysis, 24(5), 1337-1347.
doi:10.1111/j.0272-4332.2004.00530.x
Ha, J. F., & Longnecker, N. (2010). Doctor-Patient Communication: A Review. The Ochsner Journal, 10(1),
38-43.
Haggar, F. A., & Boushey, R. P. (2009). Colorectal Cancer Epidemiology: Incidence, Mortality, Survival,
and Risk Factors. Clinics in Colon and Rectal Surgery, 22(4), 191-197. doi:10.1055/s-0029-
1242458
Hanoch, Y. (2004). Improving Doctor-Patient Understanding of Probability in Communicating Cancer-
Screening Test Findings. Journal of Health Communication, 9(4), 327-335.
doi:10.1080/10810730490468496
Hassan, C., Rossi, P. G., Camilloni, L., Rex, D. K., Jimenez-Cendales, B., Ferroni, E., . . . Group, H. (2012).
Meta-analysis: adherence to colorectal cancer screening and the detection rate for advanced
neoplasia, according to the type of screening test. Alimentary Pharmacology & Therapeutics,
36(10), 929-940. doi:doi:10.1111/apt.12071
Haun, J., Noland-Dodd, V., J, Graham-Pole, J., Rienzo, B., & Donaldson, P. (2009). Testing the BRIEF
Health Literacy Screening Tool. The Federal Practitioner.
Herber, O. R., Gies, V., Schwappach, D., Thürmann, P., & Wilm, S. (2014). Patient information leaflets:
informing or frightening? A focus group study exploring patients’ emotional reactions and
subsequent behavior towards package leaflets of commonly prescribed medications in family
practices. BMC Family Practice, 15(1), 163. doi:10.1186/1471-2296-15-163
Hesse, B., & Moser, R. (2018). Health Information National Trends Survey (HINTS), 2018.
Holt, C. L., Litaker, M. S., Scarinci, I. C., Debnam, K. J., McDavid, C., McNeal, S. F., . . . Martin, M. Y.
(2013). Spiritually based intervention to increase colorectal cancer screening among African
Americans: screening and theory-based outcomes from a randomized trial. Health Educ Behav,
40(4), 458-468. doi:10.1177/1090198112459651
IOM. (2003). Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care. In S. BD, S.
AY, & N. AR (Eds.), RACIAL AND ETHNIC DISPARITIES IN DIAGNOSIS AND TREATMENT: A REVIEW
OF THE EVIDENCE AND A CONSIDERATION OF CAUSES. Washington, DC: National Academies
Press (US).
Ipeirotis, P. G. (2010a). Analyzing the Amazon Mechanical Turk marketplace. XRDS, 17(2), 16-21.
doi:10.1145/1869086.1869094
Ipeirotis, P. G. (2010b). Demographics of Mechanical Turk. Available at SSRN:
https://ssrn.com/abstract=1585030. NYU Working Paper No. ;CEDER-10-01.
139
Issaka, R. B., Avila, P., Whitaker, E., Bent, S., & Somsouk, M. (2019). Population health interventions to
improve colorectal cancer screening by fecal immunochemical tests: A systematic review. Prev
Med, 118, 113-121. doi:10.1016/j.ypmed.2018.10.021
Johanson, L. S. (2017). Caring for Patients of the Millennial Generation: Considerations for Nurses. Nurs
Forum, 52(3), 207-210. doi:10.1111/nuf.12190
Johnson, C. M., Wei, C., Ensor, J. E., Smolenski, D. J., Amos, C. I., Levin, B., & Berry, D. A. (2013). Meta-
analyses of colorectal cancer risk factors. Cancer Causes & Control, 24(6), 1207-1222.
doi:10.1007/s10552-013-0201-5
Jones, R. M., Devers, K. J., Kuzel, A. J., & Woolf, S. H. (2010). Patient-Reported Barriers to Colorectal
Cancer Screening: A Mixed-Methods Analysis. American journal of preventive medicine, 38(5),
508-516. doi:10.1016/j.amepre.2010.01.021
Jones, R. M., Vernon, S. W., & Woolf, S. H. (2010). Is discussion of colorectal cancer screening options
associated with heightened patient confusion? Cancer epidemiology, biomarkers & prevention :
a publication of the American Association for Cancer Research, cosponsored by the American
Society of Preventive Oncology, 19(11), 2821-2825. doi:10.1158/1055-9965.EPI-10-0695
Kiechle, E. S., Bailey, S. C., Hedlund, L. A., Viera, A. J., & Sheridan, S. L. (2015). Different Measures,
Different Outcomes? A Systematic Review of Performance-Based versus Self-Reported Measures
of Health Literacy and Numeracy. Journal of General Internal Medicine, 30(10), 1538-1546.
doi:10.1007/s11606-015-3288-4
King-Marshall, E. C., Mueller, N., Dailey, A., Barnett, T. E., George, T. J., Jr., Sultan, S., & Curbow, B.
(2015). "It is just another test they want to do": Patient and caregiver understanding of the
colonoscopy procedure. Patient Educ Couns. doi:10.1016/j.pec.2015.10.021
King-Marshall, E. C., Mueller, N., Dailey, A., Barnett, T. E., George, T. J., Jr., Sultan, S., & Curbow, B.
(2016). "It is just another test they want to do": Patient and caregiver understanding of the
colonoscopy procedure. Patient Educ Couns, 99(4), 651-658. doi:10.1016/j.pec.2015.10.021
Klabunde, C., Vernon, S., Nadel, M., Breen, N., Seeff, L., & Brown, M. (2005). Barriers to Colorectal
Cancer Screening: A Comparison of Reports From Primary Care Physicians and Average-Risk
Adults. Med Care, 43(9), 939-944.
Krieger, J. L., Parrott, R. L., & Nussbaum, J. F. (2011). Metaphor use and health literacy: a pilot study of
strategies to explain randomization in cancer clinical trials. J Health Commun, 16(1), 3-16.
doi:10.1080/10810730.2010.529494
Lambert, M., Luke, J., Downey, B., Crengle, S., Kelaher, M., Reid, S., & Smylie, J. (2014). Health literacy:
health professionals’ understandings and their perceptions of barriers that Indigenous patients
encounter. BMC Health Services Research, 14, 614. doi:10.1186/s12913-014-0614-1
Lansdorp-Vogelaar, I., Knudsen, A., & Brenner, H. (2010). Cost-effectiveness of colorectal cancer
screening – an overview. Best practice & research. Clinical gastroenterology, 24(4), 439-449.
doi:10.1016/j.bpg.2010.04.004
Lee, J. K., Liles, E. G., Bent, S., Levin, T. R., & Corley, D. A. (2014). Accuracy of fecal immunochemical tests
for colorectal cancer: systematic review and meta-analysis. Annals of internal medicine, 160(3),
171-171. doi:10.7326/M13-1484
Leggett, B. A., & Hewett, D. G. (2015). Colorectal cancer screening. Internal Medicine Journal, 45(1), 6-
15. doi:10.1111/imj.12636
Levin, B., Lieberman, D. A., McFarland, B., Smith, R. A., Brooks, D., Andrews, K. S., . . . Winawer, S. J.
(2008). Screening and Surveillance for the Early Detection of Colorectal Cancer and
Adenomatous Polyps, 2008: A Joint Guideline from the American Cancer Society, the US Multi-
Society Task Force on Colorectal Cancer, and the American College of Radiology*†. CA: A Cancer
Journal for Clinicians, 58(3), 130-160. doi:10.3322/CA.2007.0018
140
Lipkus, I. M., & Peters, E. (2009). Understanding the Role of Numeracy in Health: Proposed Theoretical
Framework and Practical Insights. Health Education & Behavior, 36(6), 1065-1081.
doi:10.1177/1090198109341533
Makoul, G., Cameron, K. A., Baker, D. W., Francis, L., Scholtens, D., & Wolf, M. S. (2009). A multimedia
patient education program on colorectal cancer screening increases knowledge and willingness
to consider screening among Hispanic/Latino patients. Patient Educ Couns, 76(2), 220-226.
doi:10.1016/j.pec.2009.01.006
Marchand, L. L. (1999). Combined influence of genetic and dietary factors on colorectal cancer incidence
in Japanese Americans. J Natl Cancer Inst Monogr(26), 101-105.
Mason, W., & Suri, S. (2012). Conducting behavioral research on Amazon’s Mechanical Turk. Behavior
Research Methods, 44(1), 1-23. doi:10.3758/s13428-011-0124-6
May, F. P., Whitman, C. B., Varlyguina, K., Bromley, E. G., & Spiegel, B. M. (2016). Addressing Low
Colorectal Cancer Screening in African Americans: Using Focus Groups to Inform the
Development of Effective Interventions. J Cancer Educ, 31(3), 567-574. doi:10.1007/s13187-015-
0842-z
Miller, T. A. (2016). Health literacy and adherence to medical treatment in chronic and acute illness: A
meta-analysis. Patient Education and Counseling, 99(7), 1079-1086.
doi:http://dx.doi.org/10.1016/j.pec.2016.01.020
MoneyPantry. (2017). 10 Amazon Turk Tips. Retrieved from http://moneypantry.com/amazon-
mechanical-turk-tips/
Mousavinezhad, M., Majdzadeh, R., Akbari Sari, A., Delavari, A., & Mohtasham, F. (2016). The
effectiveness of FOBT vs. FIT: A meta-analysis on colorectal cancer screening test. Medical
journal of the Islamic Republic of Iran, 30, 366-366.
Mueller NM, K.-M. E., Hyams T, Christie J, Chun J, Curbow BC. (2017). "'You wanna avoid it, but they
make it so pleasant': Patient Experience of the Colonoscopy Procedure". Paper presented at the
American Public Health Association, Atlanta, GA.
Muir, K. W., Santiago-Turla, C., Stinnett, S. S., Herndon, L. W., Allingham, R. R., Challa, P., & Lee, P. P.
(2006). Health Literacy and Adherence to Glaucoma Therapy. American Journal of
Ophthalmology, 142(2), 223-226.e222. doi:http://dx.doi.org/10.1016/j.ajo.2006.03.018
Murphy, D. A., Lam, P., Naar-King, S., Robert Harris, D., Parsons, J. T., & Muenz, L. R. (2010). Health
literacy and antiretroviral adherence among HIV-infected adolescents. Patient Education and
Counseling, 79(1), 25-29. doi:http://dx.doi.org/10.1016/j.pec.2009.07.014
Nguyen, J., Moorhouse, M., Curbow, B., Christie, J., Walsh-Childers, K., & Islam, S. (2016). Construct
Validity of the eHealth Literacy Scale (eHEALS) Among Two Adult Populations: A Rasch Analysis.
JMIR Public Health Surveill, 2(1), e24. doi:10.2196/publichealth.4967
Ntiri, D. W., & Stewart, M. (2009). Transformative Learning Intervention: Effect on Functional Health
Literacy and Diabetes Knowledge in Older African Americans. Gerontology & Geriatrics
Education, 30(2), 100-113. doi:10.1080/02701960902911265
Nutbeam, D. (2000). Health literacy as a public health goal: a challenge for contemporary health
education and communication strategies into the 21st century. Health Promotion International,
15(3), 259-267. doi:10.1093/heapro/15.3.259
O'Keefe, S. J., Li, J. V., Lahti, L., Ou, J., Carbonero, F., Mohammed, K., . . . Zoetendal, E. G. (2015). Fat,
fibre and cancer risk in African Americans and rural Africans. Nat Commun, 6, 6342.
doi:10.1038/ncomms7342
Parker, R. M., Baker, D. W., Williams, M. V., & Nurss, J. R. (1995). The test of functional health literacy in
adults. Journal of General Internal Medicine, 10(10), 537-541. doi:10.1007/BF02640361
Percheski, C., & Hargittai, E. (2011). Health information-seeking in the digital age. J Am Coll Health,
59(5), 379-386. doi:10.1080/07448481.2010.513406
141
Peterson, D. (2015). All That Is Solid: Bench-Building at the Frontiers of Two Experimental Sciences.
American Sociological Review, 80(6), 1201-1225.
Peterson, N. B., Dwyer, K. A., Mulvaney, S. A., Dietrich, M. S., & Rothman, R. L. (2007). The influence of
health literacy on colorectal cancer screening knowledge, beliefs and behavior. J Natl Med
Assoc, 99(10), 1105-1112.
Pew Research Center. (2018). Generations and Age.
Piper, L. E. (2012). Generation Y in healthcare: leading millennials in an era of reform. Front Health Serv
Manage, 29(1), 16-28.
Pittman, M., & Sheehan, K. (2016). Amazon’s Mechanical Turk a Digital Sweatshop? Transparency and
Accountability in Crowdsourced Online Research. Journal of Media Ethics, 31(4), 260-262.
doi:10.1080/23736992.2016.1228811
Pleasant, A. (2014). Advancing Health Literacy Measurement: A Pathway to Better Health and Health
System Performance. Journal of Health Communication, 19(12), 1481-1496.
doi:10.1080/10810730.2014.954083
Rabin, R. C. (2017). What yound people need to know about colon cancer. New York Times.
Rat, C., Latour, C., Rousseau, R., Gaultier, A., Pogu, C., Edwards, A., & Nguyen, J. M. (2018). Interventions
to increase uptake of faecal tests for colorectal cancer screening: a systematic review. Eur J
Cancer Prev, 27(3), 227-236. doi:10.1097/cej.0000000000000344
Rech Wachelke, J. F., Vizeu Camargo, B., Lincoln Bezerra Lins, S., & Vieira de Lima Nunes, A. (2011). Data
collection in Internet environment in social representations studies. Avances en Psicología
Latinoamericana, 29(1), 11-18.
Redaelli, A., Cranor, C. W., Okano, G. J., & Reese, P. R. (2003). Screening, prevention and socioeconomic
costs associated with the treatment of colorectal cancer. Pharmacoeconomics, 21(17), 1213-
1238.
Reyna, V. F. (2008). A Theory of Medical Decision Making and Health: Fuzzy Trace Theory. Medical
decision making : an international journal of the Society for Medical Decision Making, 28(6), 850-
865. doi:10.1177/0272989X08327066
Reyna, V. F. (2012). A new intuitionism: Meaning, memory, and development in Fuzzy-Trace Theory.
Judgement and Decision Making, 7(3), 332-359.
Reyna, V. F., & Brainerd, C. J. (1995). Fuzzy-trace theory: An interim synthesis. Learning and Individual
Differences, 7(1), 1-75. doi:http://dx.doi.org/10.1016/1041-6080(95)90031-4
Reyna, V. F., Nelson, W. L., Han, P. K., & Dieckmann, N. F. (2009). How numeracy influences risk
comprehension and medical decision making. Psychol Bull, 135(6), 943-973.
doi:10.1037/a0017327
Rothman, R. L., Montori, V. M., Cherrington, A., & Pignone, M. P. (2008). Perspective: The Role of
Numeracy in Health Care. Journal of Health Communication, 13(6), 583-595.
doi:10.1080/10810730802281791
Rudd, R., AT, M., & D, N. (2012). Health Literacy and Definitions of Terms. In Health Literacy in Context:
International Perspectives. Canada: Nova Science.
Rudd, R. E. (2013). Needed action in health literacy. J Health Psychol, 18(8), 1004-1010.
doi:10.1177/1359105312470128
Rust, C., & Davis, C. (2011). Health Literacy and Medication Adherence in Underserved African-American
Breast Cancer Survivors: A Qualitative Study. Social Work in Health Care, 50(9), 739-761.
doi:10.1080/00981389.2011.585703
Sag, A. A., Selcukbiricik, F., & Mandel, N. M. (2016). Evidence-based medical oncology and interventional
radiology paradigms for liver-dominant colorectal cancer metastases. World J Gastroenterol,
22(11), 3127-3149. doi:10.3748/wjg.v22.i11.3127
142
Sanguinetti, J. M., Lotero Polesel, J. C., Iriarte, S. M., Ledesma, C., Canseco Fuentes, S. E., & Caro, L. E.
(2015). Informed consent in colonoscopy: A comparative analysis of 2 modes. Revista de
Gastroenterología de México (English Edition), 80(2), 144-149.
doi:https://doi.org/10.1016/j.rgmxen.2015.03.004
Sarfaty, M., & Wender, R. (2007). How to increase colorectal cancer screening rates in practice. CA
Cancer J Clin, 57(6), 354-366. doi:10.3322/ca.57.6.354
Sbaffi, L., & Rowley, J. (2017). Trust and Credibility in Web-Based Health Information: A Review and
Agenda for Future Research. J Med Internet Res, 19(6), e218. doi:10.2196/jmir.7579
Schnitzler, L., Smith, S. K., Shepherd, H. L., Shaw, J., Dong, S., Carpenter, D. M., . . . Dhillon, H. M. (2017).
Communication during radiation therapy education sessions: The role of medical jargon and
emotional support in clarifying patient confusion. Patient Education and Counseling, 100(1),
112-120. doi:http://dx.doi.org/10.1016/j.pec.2016.08.006
Schwartz, P. H., Edenberg, E., Barrett, P. R., Perkins, S. M., Meslin, E. M., & Imperiale, T. F. (2013).
Patient understanding of benefits, risks, and alternatives to screening colonoscopy. Fam Med,
45(2), 83-89.
Semuels, A. (2018). The Internet Is Enabling a New Kind of Poorly Paid Hell. The Atlantic.
Seurer, A. C., & Vogt, H. B. (2013). Low health literacy: a barrier to effective patient care. S D Med, 66(2),
51, 53-57.
Siegel, R., DeSantis, C., & Jemal, A. (2014). Colorectal cancer statistics, 2014. CA: A Cancer Journal for
Clinicians, 64(2), 104-117. doi:10.3322/caac.21220
Siegel, R. L., Miller, K. D., Fedewa, S. A., Ahnen, D. J., Meester, R. G. S., Barzi, A., & Jemal, A. (2017).
Colorectal cancer statistics, 2017. CA Cancer J Clin, 67(3), 177-193. doi:10.3322/caac.21395
Siegel, R. L., Miller, K. D., Fedewa, S. A., Ahnen, D. J., Meester, R. G. S., Barzi, A., & Jemal, A. (2017).
Colorectal cancer statistics, 2017. CA: A Cancer Journal for Clinicians, 67(3), 177-193.
doi:10.3322/caac.21395
Smith, S. G., Raine, R., Obichere, A., Wolf, M. S., Wardle, J., & von Wagner, C. (2015). The effect of a
supplementary (‘gist-based’) information leaflet on colorectal cancer knowledge and screening
intention: a randomized controlled trial. Journal of Behavioral Medicine, 38(2), 261-272.
doi:10.1007/s10865-014-9596-z
Smith, S. G., Vart, G., Wolf, M. S., Obichere, A., Baker, H. J., Raine, R., . . . von Wagner, C. (2015). How do
people interpret information about colorectal cancer screening: observations from a think-aloud
study. Health Expect, 18(5), 703-714. doi:10.1111/hex.12117
Smith, S. G., Wolf, M. S., Obichere, A., Raine, R., Wardle, J., & von Wagner, C. (2013). The development
and testing of a brief (‘gist-based’) supplementary colorectal cancer screening information
leaflet(). Patient Education and Counseling, 93(3), 619-625. doi:10.1016/j.pec.2013.08.013
Subramanian, S., Tangka, F. K., Hoover, S., Nadel, M., Smith, R., Atkin, W., & Patnick, J. (2016).
Recommendations From the International Colorectal Cancer Screening Network on the
Evaluation of the Cost of Screening Programs. J Public Health Manag Pract, 22(5), 461-465.
doi:10.1097/phh.0000000000000386
Taggarshe, D., Rehil, N., Sharma, S., Flynn, J. C., & Damadi, A. (2013). Colorectal cancer: are the "young"
being overlooked? Am J Surg, 205(3), 312-316; discussion 316.
doi:10.1016/j.amjsurg.2012.10.016
Weiss, B. D., Hart, G., McGee, D. L., & D'Estelle, S. (1992). Health status of illiterate adults: relation
between literacy and health status among persons with low literacy skills. J Am Board Fam Pract,
5(3), 257-264.
Whitley, J., Smith, J. D., & Vaillancourt, T. (2013). Promoting Mental Health Literacy Among Educators.
Canadian Journal of School Psychology, 28(1), 56-70. doi:10.1177/0829573512468852
143
Williams, M. V., Davis, T., Parker, R. M., & Weiss, B. D. (2002). The role of health literacy in patient-
physician communication. Fam Med, 34(5), 383-389.
Williams, R., White, P., Nieto, J., Vieira, D., Francois, F., & Hamilton, F. (2016). Colorectal Cancer in
African Americans: An Update: Prepared by the Committee on Minority Affairs and Cultural
Diversity, American College of Gastroenterology. Clinical and Translational Gastroenterology,
7(7), e185. doi:10.1038/ctg.2016.36
Wolf, A., Fontham, E., Church, T. R., Flowers, C. R., Guerra, C. E., LaMonte, S. J., . . . Smith, R. A. (2018).
Colorectal cancer screening for average-risk adults: 2018 guideline update from the American
Cancer Society. CA: A Cancer Journal for Clinicians, 68(4), 250-281. doi:doi:10.3322/caac.21457
Wolf, M. S., Baker, D. W., & Makoul, G. (2007). Physician–Patient Communication about Colorectal
Cancer Screening. Journal of General Internal Medicine, 22(11), 1493-1499. doi:10.1007/s11606-
007-0289-y
Wolf, M. S., Curtis, L. M., Wilson, E. A. H., Revelle, W., Waite, K. R., Smith, S. G., . . . Baker, D. W. (2012).
Literacy, Cognitive Function, and Health: Results of the LitCog Study. Journal of General Internal
Medicine, 27(10), 1300-1307. doi:10.1007/s11606-012-2079-4
Yardley, C., Glover, C., & Allen-Mersh, T. G. (2000). Demographic factors associated with knowledge of
colorectal cancer symptoms in a UK population-based survey. Annals of the Royal College of
Surgeons of England, 82(3), 205-209.
Yee, L. M., & Simon, M. A. (2014). The Role of Health Literacy and Numeracy in Contraceptive Decision-
Making for Urban Chicago Women. Journal of Community Health, 39(2), 394-399.
doi:10.1007/s10900-013-9777-7
Zikmund-Fisher, B. J., Smith, D. M., Ubel, P. A., & Fagerlin, A. (2007). Validation of the Subjective
Numeracy Scale: effects of low numeracy on comprehension of risk communications and utility
elicitations. Med Decis Making, 27(5), 663-671. doi:10.1177/0272989x07303824
Zolnierek, K. B., & Dimatteo, M. R. (2009). Physician communication and patient adherence to
treatment: a meta-analysis. Med Care, 47(8), 826-834. doi:10.1097/MLR.0b013e31819a5acc
144